Ubiquiti Networks – cash flow charts up to Q2 2016

Disclosure – I’m long Ubiquiti Networks, Inc. (UBNT).


• With low investment needs, Ubiquiti generates good free cash flow.
• Swings in non-cash working capital add a lot of volatility to the cash flow from operations (CFO). When the effect of the swings in non-cash working capital are removed, the growth of CFO is more steady, but still with dips in some years.
• In the latest half year, 34.7% of the rise in CFO has come from a drop in non-cash working capital (relative to the average rate of increase).
• The drop is likely to be reversed at some point. Investors need to look at CFO adjusted for the swings in non-cash working capital rather than raw CFO.
• If non-cash working capital goes up, bringing CFO down, the market might not make the logical adjustment to CFO, but that shouldn’t matter to patient long-term investors.

Closing stock price on April 14, 2016: $33.72. Market cap: $2.8 billion.

The company and the business model

From the company’s 10-K – “Ubiquiti Networks develops high performance networking technology for service providers and enterprises.”. The service providers are typically Wireless Internet Service Providers (WISPs). “Our technology platforms deliver highly-advanced and easily deployable solutions that appeal to a global customer base, particularly in under-networked markets.” Manufacturing is outsourced, resulting in low capex. Sales costs are low as the company relies on an internet forum/community to evangelise the products rather than a costly sales force, which allows disruptive pricing, while the pricing encourages the evangelism. Headcount is low and Ubiquiti are able to concentrate on R&D. Ubiquiti has a policy of focusing on the quality of R&D staff rather than quantity, and the CEO has blamed some recent glitches on straying a little from the policy. Results have been fairly volatile, probably because of the lack of recurring revenue, exposure to emerging markets and the variable impact of product launches. The CEO holds a majority of the stock, and only draws a nominal salary.

About this piece

This is about Ubiquiti’s cash flows, with some comparison to two competitors. There are different vizualisations of cash produced by the business versus cash put in, some of which are highly unconventional. I don’t cover the company’s products, markets or prospects, about which there is relevant material on Seeking Alpha in articles and comments.

The cash is held overseas

96% of Ubiquiti’s cash is held outside of the United States ($476.9 million out of $496.7 million), and there’d be a tax bill if the cash was repatriated. The long-term debt of $115.5 million is the result of borrowing in the US to cover the gap between cash generated by operations in the US and the sum of stock buybacks, head office costs and investment in the US.

Cash flow and revenue

The chart below lets you compare Ubiquiti’s cash flows to revenue. The red bars represent CFO (Cash From Operations), which is one of the three “bottom line” figures in the cash flow statement. The others are cash into investment, shown as blue bars, and the financing cash flow which is not shown. The small size of the blue bars shows how small the investment flow has been compared to the inflow from operations. Although CFO is a prominent item in the cash flow statement, it’s affected by swings in non-cash working capital. The brown line represents CFO adjusted to remove the effect of the change in working capital, with the average change in working capital added back in. It represents the underlying CFO better than ‘raw’ CFO. A positive change in working capital actually means less cash tied up in things like receivables and inventories, which means more CFO. I explain more about it later. I’ve also indicated some ‘bad news’ items along the top. The BEC fraud was an email-based impersonation, and two employees were blamed for incompetence.

Ubiquiti cash flow revenue and news 2Q16

A chart further down shows CFO as a percent of revenue, for Ubiquiti and two other companies (find “Ubiquiti turns more of its revenue into cash flow”).

Volatile quarterly cash from operations

The next two charts show how volatile quarterly CFO has been, with no obvious seasonality.

Ubiquiti quarterly cash flows 2Q16

The per-share view –

Ubiquiti quarterly cash flows per share 2Q16

The difference between the two charts above is the share count used in the ‘per share’ version. This has come down recently due to buybacks. It’s possible that the market sees the buybacks and the dividend as confirming the company is no longer in a growth phase. I believe that both are a reasonable use of the cash the company generates, and the fact that free cash flow has been generated in years of high growth and low growth is positive for the business.

Ubiquiti share count

Walk charts

For cash flows I prefer ‘walk charts’ to the more conventional time series charts. The next chart is just to illustrate the principle.

Walk chart illustration

The next chart is the real thing, with an absolute and a per-share version. The small investment flow compared to CFO results in near vertical graphs. In both versions, the latest two quarters have each been multiplied by 4 to annualize them. Without the multiplication, the upturn in CFO would have looked like a big drop, but obviously the annualized CFO does not accurately represent the amount added to the cash stash.

Ubiquiti cash flow 2Q16

An upturn in free cash flow can be bad in some cases, if it’s the result of under-investing, perhaps to impress Wall Street with the free cash flow figure. That does not apply to Ubiquiti, because substantial growth in CFO has been achieved over several years with little investment, and with the CEO holding a majority of the stock, he would not squeeze out cash in the short term at the expense of long term growth.

Adjusting for the change in working capital

One problem with looking at CFO is the effect of the change in working capital, and the terminology around the “change in WC” can be confusing. For example, if Accounts receivable goes up, it pushes up non-cash working capital (at least if nothing else pushes it down, like accounts payable going up by the same amount). When non-cash working capital goes up, you might expect that “the change in working capital” will be positive, but actually it’s negative (at least according to common usage). That’s because the phrase “the change in working capital” is used in the context of the cash flow statement, and it’s a kind of shorthand for “the effect of the change in non-cash working capital on cash flow”. If non-cash working capital goes up, the effect is to decrease CFO (cash from operations). So, a positive “change in WC” means that non-cash working capital has gone down, with a positive effect on CFO.

The change in WC has two characteristics. First, there’s the average working capital needs, which depends on revenue and the nature of the business. Most businesses need more working capital as their revenue grows, although a few (such as retailers that pay suppliers AFTER the goods have been sold) have the advantage of negative working capital needs, which means that someone else (like the suppliers) helps to finance the company’s growth. The second characteristic is the variability of the change in WC – a company that sells consumer staples will have much less variability than a shipbuilder with cyclical revenue.

Ubiquiti does not have the advantage of negative working capital needs, but that doesn’t matter much because the cash flow easily covers the needs (or, it would take a massive growth rate to tie up all the cash generated in non-cash working capital). Ubiquiti’s working capital is quite variable, because inventory has to be built up ahead of product launches to minimize the chance of stock-outs, and sales of new products are not very predictable, so inventory might shift faster or slower than expected. The company has enough cash and cash flow to cover the variability in the change in WC, but investors need to be aware of how the variability has affected the CFO number.

In the latest two quarters, a positive change in WC was good for the CFO figure, and it’s likely to swing back at some point in the future making CFO less than otherwise. Until the latest half year, looking at the annual figures, the change in WC has happened to smooth out CFO before the change in WC, making CFO less volatile than either net income or CFO before the change in WC. The next chart shows that the (possibly random) smoothing effect went into reverse in the latest half year. Investors should probably not get too excited about the big jump in CFO (for H1 2016, which jumps up when annualized), and should look instead at the panels on the right in the graphic after this one, where I replaced the change in WC with its average since 2010. In the chart, “Net cash provided by operating activities” is the same as CFO, it’s just closer to the term used in the cash flow statement.

Ubiquiti cash flows and income 2010 to 2Q16

For the change in working capital I included everything under “Changes in operating assets and liabilities” in the cash flow statement. That might be contentious and technically wrong, but to me what matters is the variable changes in value or quantity that affect the CFO number. “Variable changes” excludes amortization and depreciation where the reduction in value is scheduled in advance and included in the ‘add back’ part of the cash flow statement.

In the next graphic, I show versions of CFO plotted against cash into investment (with absolute amounts on the top, and per-share amounts below). On the left I show normal CFO in blue, and CFO before the change in working capital in red. Some of the detail isn’t clear, but you should be able to see that for the first half of fiscal 2016, CFO has been significantly ahead of CFO before the change in WC. You could say that CFO happened to be flattered by the change in WC, so the figure for H1 2016 exaggerates the underlying CFO. In the middle, the blue line for CFO has been replaced with a version where CFO has been adjusted for the average change in working capital, and those figures are less than for unadjusted CFO because the average change in working capital is negative (because the company has required more non-cash working capital over the period). The average is just based on the average increase per year, in other words I did not try to match working capital to revenue.

Because it wasn’t easy to clearly show both the red line and the blue line in the middle panel, I show just the blue line in the panel on the right. It shows the CFO adjusted for the average change in WC, in other words without any random swings from the change in WC, but using the average rather than just ignoring it (which would mean ignoring the non-cash working capital needed to support growth). I believe the panels on the right are the best representation so far of the history of the company’s underlying CFO in relation to the investment that generated it.

Ubiquiti walk chart with average change in WC applied 2Q16

The following table shows the data the chart above was based on. CFO hardly grew from 2013 to 2015 (in thousands, from $131,891 to $134,547, or 2.0%, or $1.46 per share to $1.50 per share, or 2.7%). After adjusting for the average change in working capital, the growth was 82.1%, or 84.0% per share (34.9% and 35.6% CAGR). On that basis, the only two year period with a drop was 2011 to 2013, from $86,700 to $73,438, or -15.3%, or $0.84 per share to $0.81 per share, or -3.6%. The 2013 results were affected by a serious counterfeiting problem from which the company recovered quickly, and in the same year the company had its biggest ever positive change in working capital. While there was some comfort in seeing cash freed from being tied up in working capital at a time of crisis, the pleasing CFO figure was not as informative as the dip in net income, and if you don’t adjust CFO to remove the effect of swings in working capital, you are probably better off just looking at net income, at least if you are looking at a year or two rather than the overall trend. (To get the change in working capital, add the items under “Changes in operating assets and liabilities:” in the cash flow statement, and see how it compares to the average of -$12,541 thousand per year.)

Ubiquiti table for CFO and change in WC 2Q16

I used the figures in the table to calculate that 34.7% of the latest half year’s rise in CFO has come from a drop in non-cash working capital (relative to the average rate of increase). Here’s the calculation –

The CFO for H1 2016 is 125,074 (this is all in $ thousands). That’s 57,800.5 more than the full year figure for 2015 divided by 2. The change in WC for the half year was 13,781, which is 20,051.5 more than the average (of -12,541 for a full year, which is -6,270.5 for a half year). The 20,051.5 ‘excess’ change in WC is 34.7% of the rise in the half year’s CFO of 57,800.5.

One way of viewing the working capital in relation to revenue is to look at the items expressed as days of sales, for example DSO (days sales outstanding, in receivables), or DSI (days of inventory). For a full year’s revenue, the calculations are like DSO = 360 * receivables / revenue. The result is only a poor approximation when sales are lumpy or volatile, for example a big order just before the year end would give no time to collect the cash, resulting in an artificially high number for DSO from the calculation, while the actual DSO would be unusually low. Even though there’s volatility in Ubiquiti’s sales, the following chart shows a clear general decline in the total of the items charted (Receivables + Inventories + Vendor deposits – Payables – Customer deposits) expressed as days of sales. That’s surprising, as management have said that more inventory is held to avoid stockouts, and the use of larger distributors means higher receivables because they take longer to pay (whereas a small distributor with poor or less verifiable finances might be required to pay up front). The half year’s revenue for H1 2016 has not been annualized in the chart.

Ubiquiti working capital as days sales

An increase in the days of inventories can signal a problem in selling stock, although in Ubiquiti’s case it could be a planned increase ahead of a product launch or anticipated sales. An increase in the days of receivables can signal problems including the manipulation of earnings, but that’s unlikely for Ubiquiti as the CEO is the majority shareholder and the cash and free cash flow means there’s no pressing need to raise capital. The cash flow and cash held overseas are likely to be good enough to enable borrowing in the US on reasonable terms, and if for some reason the terms become onerous then the company can stop the buybacks. Therefore there’s little pressure to flatter the accounts to make raising capital cheaper or easier. If days of receivables or inventories go up a lot, it’s still worth checking earnings call transcripts and the prepared remarks, and the “Liquidity and Capital Resources” section of the 10-Q or 10-K.

Comparison to Ruckus and Cisco

Ruckus and Cisco both have growing free cash flow, but neither has the vertical graph that Ubiquiti has, which shows they needed to invest more, relatively, to get their CFO. Cisco is a big old mature company, with less growth in the period charted. There’s noticeably more growth in Cisco’s ‘per share’ chart, which suggests that buying back stock has been a big factor in the ‘per share’ growth of their CFO. Ruckus and Cisco were just the first two competitors I thought of. That could imply some kind of selection bias, but I can promise I did not systematically choose competitors that made Ubiquiti look better or worse in comparison. I did not adjust for the average change in WC (only because of the time it would take).

Ubiquiti cash flow comparison 2Q16

Ubiquiti’s investment flow consists of capital expenditure and the purchase of intangible assets, with the purchase of intangibles actually consisting of the cost of trademark applications. For Ruckus and Cisco, the investment flow contains items which are not relevant to the operating business, such as parking cash in short term investments. That’s why the X axis is labeled “PPE plus acquisitions” for them, while Ubiquiti simply has “Cash into investment”. The purpose of acquisitions is ultimately to generate CFO, which is the same purpose as for PPE, so the sum of both is a relevant quantity to chart CFO against. Ruckus’s PPE includes “internal use software”. Purchase of intangibles is not a named item in Ruckus’s cash flow statement, while for Cisco it might be included in the small “Other” category. Ubiquiti could conceivably make an acquisition in future to acquire technology or an R&D team, but many targets would be ruled out because they would increase the headcount by a massive proportion compared to the acquired earnings or cash flow.

Plenty more competitors are listed in Ubiquiti’s 10-K, and possibly none have a very similar product mix, similar customers and a similar breakdown of sales by geography. The comparison to Cisco and Ruckus won’t be perfect, but I still think it’s relevant to show Ubiquiti’s near vertical graphs next to the less vertical graphs of the other two companies, because the steep slopes are the result of the business model rather than the products, customers or sales geography.

By comparing Ubiquiti and Ruckus to Cisco for various items, the next chart gives some idea of the relative sizes. Cisco is by far the biggest, while Ubiquiti has rather more revenue than Ruckus. Ubiquiti beats Ruckus on net income and cash from operations. Ruckus spends more on R&D and PPE + acquisitions, and Ubiquiti needs to continue to get more from less to stay ahead.

Ubiquiti compare size rel to Cisco

The next chart shows that Ubiquiti has invested less as a percentage of revenue than the other two companies. It has turned more of its revenue into cash flow than Ruckus. Ubiquiti’s conversion of revenue to cash flow has usually been behind Cisco, but it was well ahead in 2011, 2013 and H1 2016. (Adjusting for swings in working capital might show a small consistent outperformance, but I didn’t have the time to do that for Cisco.) Because the chart shows ratios, there was no need to annualize the quarters.

Ubiquiti compare cash flow percent 2Q16

Since making the charts, it’s been announced that Ruckus is being acquired by Brocade, see “Cisco Rivals Brocade And Ruckus Are Merging. Who’s Next?” by Eric Jhonsa, SA Eye on Tech, Apr. 4, 2016 (seekingalpha.com).

A bad cash flow walk chart

Just to show what a bad chart looks like, here’s one from the oil patch. Negative free cash flow is not necessarily bad, in fact it’s good so long as the return on investment is sufficiently ahead of the cost of capital, and the balance sheet is not overstretched. It didn’t work out like that for Chesapeake and many other resource stocks. From the chart, you didn’t need to predict the oil price crash to see that capex in excess of CFO was not increasing CFO.  A history of positive free cash flow reduces the risk, so long as the history stays relevant (e.g. vital patents aren’t going to run out), although it doesn’t eliminate risks, and there’s always plenty of them listed in a 10-K.

Chesapeake Energy walk chart

Considering R&D as investment

The R&D spend has the same purpose as investment in property, plant and equipment – to maintain and increase the cash flow from operations. Spending on PPE tends to be lumpy whereas it’s not a good idea to vary R&D a lot (which would mean alternately hiring and losing staff). However it’s not a very clear distinction because a massive project like a gas to liquids plant can incur costs over many years, with possibly less certainty about future costs, operational success and markets than for R&D in some cases.

Because of the similarity of purpose, and because Ubiquiti’s R&D spend is much bigger than its conventional investment flow, in the next chart I’ve included R&D with investment, and added R&D to CFO. That’s not conventional, and is irregular because I’ve added an expense to cash flows, which technically mixes accrual accounting with cash accounting, but the R&D expense is likely to be fairly close to the cash cost because it’s a regular item rather than a volatile one, although there are non-cash costs associated with R&D in the form of stock-based compensation. That was $2.9 million out of total stock-based compensation of $5.0 milllion in 2015, or 5.3% of the $54.6 million R&D expense, with the proportion falling to 4.2% in H1 2016. IMO the chart is probably a reasonable approximation to CFO before the cash cost of R&D, plotted against cash invested including the cash cost of R&D, providing a different and arguably more complete view of cash generated plotted against cash put in.

Ubiquiti adding RnD to cash flows 2Q16

This shows just the ‘R&D included’ green line without the other stuff –

Ubiquiti cash flow plus RnD 2Q16

Even with R&D included in investment, and without removing the non-cash stock-based compensation, the green line is more vertical than the lines for Ruckus and Cisco shown earlier (where R&D was not included).

Considering R&D and stock buybacks as investment

The main effect of Ubiquiti’s stock buybacks is to increase the cash flow from operations per share (along with EPS), although it also reduces the total cash cost of paying the dividend. The effect is sufficiently ‘investment like’ that I’ve included stock buybacks in the next chart. Buybacks are not exactly like other investments, because they depend on the success of the other investments, whereas an investment in PPE or an area of R&D has more scope for success or failure independently of the success or failure of other investments. However, it’s still a use of capital which ought to be compared to alternatives when allocating capital.

So long as Ubiquiti generates positive free cash flow, then all previous buybacks increase the free cash flow per share. In that sense, buybacks might be seen as a relatively safe investment, which justifies flattening the graph. Because Ubiquiti’s CEO is a majority shareholder, he can be relied on to make buybacks only when he believes it’s a good use of capital, which is likely to be whenever the stock is cheap enough due to the good free cash flow and the difficulty in finding acquisitions that would make sense, although having to borrow in the US to finance buybacks could be a limiting factor. While a steeper graph in a walk chart is generally better, the steepness isn’t the only consideration.

Up til now, the per-share charts have included the benefit of stock buybacks on CFO per share while ignoring the cost of the buybacks, so it’s inevitable that the red ‘buybacks included’ line in the next chart is not as nicely steep as the green line where they are not included. I’ve also switched to a cumulative view, which has the advantage of smoothing out noise and showing the long term progress. Some caution is needed, because a horrible looking drop to zero in a normal chart would look like a less scary flat line in a cumulative chart.

Ubiquiti cash flows - buybacks on top of RnD to 2Q16 - cumul per share

The next chart is also cumulative and per-share. Each panel shows three lines. The steepest (dark blue) is the relatively standard CFO vs cash invested graph. The green line has the R&D expense included in investment and added back to the cash from operations, and the light blue line is the same except that stock buybacks have been included in investment, with the effect that points on the light blue line are like points on the green line shifted to the right. There are two panels because there are two different ways of accumulating per-share quantities. Rather than decide which is best, I prefer to conclude that however you look at Ubiquiti’s figures, they’ve got a lot of cash out of their operations compared to what they’ve put in, although it takes longer for the effect of stock buybacks to pay back (in the per-share quantities).

Ubiquiti adjusted cash flows cumul per share

The highlighted figures in the bottom two rows of the following table show that the per-share cash generated (as measured by variations on CFO adjusted for the average change in working capital) in the latest half year accounted for 16.9% to 18.8% of the total from 2010 to H1 2016, while the half year is only 7.7% of the period. I’d call that an excellent half year.

Ubiquiti flows per share table

The cash stash

In thousands

Q2 2016 (December 31, 2015)
Cash and cash equivalents $496,672
Total liabilities $209,188
Cash minus Total liabilities = $287,484

You can think of “Cash minus Total liabilities” as a highly conservative break-up value, which ignores the value of all non-cash assets.

2011 (June 30, 2011)
Cash and cash equivalents $76,361
Total liabilities $39,703
Cash minus Total liabilities = $36,658

Growth in Cash minus Total liabilities, 2011 to Q2 2016
= $287,484 – $36,658
= $250,826

Dividing the last figure by the Q2 2016 Weighted average shares, Diluted, of 86,091 (thousand), gets:
$250,826 / 86,091
= $2.91 extra cash minus total liabilities per share between 2011 and Q2 2016.

Cumulative quantities

In this part I give a very rough idea of where the accumulated CFO has gone.

In thousands

Total stock buybacks + Total cash invested + Total dividends
= $199,226 + $29,558 + $30,672
= $259,456

Total CFO

Total CFO – (Total stock buybacks + Total cash invested + Total dividends)
= $631,484 – $259,456
= $372,028

I’ll call that $372,028 “Total free cash flow after buybacks and dividends”. The figure includes dividends but not other financing flows, because the company borrows in the US while holding (much more) cash outside of the US. It’s bigger than the $250,826 increase in cash minus total liabilities (over a shorter period, because I haven’t seen a balance sheet for 2010). It’s smaller than the increase in Total stockholders’ equity, which increased by $481,439 between 2011 (June 30) and Q2 2016 (December 31, 2015). The increase was from minus $53,872 to a positive $427,567.

While the breakdown is extremely rough, it’s apparant that the biggest use of cash has been simply to keep it, with:
Cash (and cash equivalents) increasing by $496,672 – $76,361 = $420,311.
Cash minus total liabilities increasing by $250,826.
Total stock buybacks = $199,226.
Total dividends = $30,672.
Total cash invested = $29,558.

(More recently, buybacks have been the biggest use of cash.)

I expect many analysts would regard keeping the cash as having a lazy balance sheet. My view is that for a company with no recurring revenue and dependence on emerging markets and product launches, with various risks (including any trick that Cisco might pull), money is the best insurance that money can buy. It might also be handy if an acquisition opportunity pops up or buybacks become very attractive, but I suppose I shouldn’t praise the insurance and opportunity aspects for the same dollars.


The analysis here is not deep because Ubiquiti’s cash flows are relatively simple. A high proportion of revenue drops down to net income. Most of the net income drops down to cash from operations, with a small reduction on average due to the growing working capital needs, and quite a lot of variation due to the swings in working capital. I recently wrote about cash conversion at a big pharma company (which I’m long), and there was both the material (in various reconciliations) and the need, with a poor record of converting profit to free cash flow which cast doubt over the ability to maintain the dividend in coming years. After a complicated analysis I concluded that their cash conversion was very likely to improve. I don’t see the scope or need for such an analysis in Ubiquiti’s case.

Other material

I’ve made a kind of waterfall chart which illustrates Ubiquiti’s cash flow statements from 2013 to H1 2016, including a breakdown of the change in working capital. I don’t think the detail is necessary or that most readers will want it, which is why the graphic is only linked to. There’s  also a ‘bad news’ table and a few spreadsheet tables on the link which I decided weren’t worth including in this piece.

For someone else’s explanation of the change in working capital, see oldschoolvalue.com (when the author refers to a previous error, he hadn’t grasped that a positve “change” meant working capital had gone down, but it’s a rare slip from him).

The main source

All the figures for Ubiquiti were taken from the SEC filings on their site.

DISCLAIMER: Your investment is your responsibility. It is your responsibility to check all material facts before making an investment decision. All investments involve different degrees of risk. You should be aware of your risk tolerance level and financial situations at all times. Furthermore, you should read all transaction confirmations, monthly, and year-end statements. Read any and all prospectuses carefully before making any investment decisions. You are free at all times to accept or reject all investment recommendations made by the author of this blog. All Advice on this blog is subject to market risk and may result in the entire loss of the reader’s investment. Please understand that any losses are attributed to market forces beyond the control or prediction of the author. As you know, a recommendation, which you are free to accept or reject, is not a guarantee for the successful performance of an investment.


GlaxoSmithKline’s record of cash conversion

Disclosure: I’m long GlaxoSmithKline plc (GSK)

About this piece

The rate of conversion from income to cash flow matters because the more cash a company generates, the more cash it can invest or return to shareholders without increasing debt. The cash conversion ratio has various definitions, but here I look into the ratio of Free cash flow to Profit after tax, before making adjustments that deviate from those metrics.

In this piece I try to stick to historic facts rather than predict future cash conversion, and I don’t cover the effect of the deal with Novartis on future cash conversion.

Where core earnings have leaked before reaching free cash flow

I’ve made two tables for different periods, because the nine months reported for 2015 include the deal with Novartis, where Novartis acquired the oncology business (but not the pipeline), GSK acquired vaccines, and the consumer businesses were pooled, with GSK holding the controlling interest. Also the nine-months-2015 figures are not reported in as much detail as the annual figures.

Cash conversion summary table

Major restructuring

The category “Major restructuring” has been –

  1. Big
  2. Relatively consistent
  3. Deducted from core results
  4. Not deducted from free cash flow
  5. Thefore it’s dragged cash conversion down
  6. It’s been the biggest factor in dragging cash conversion down.

I don’t speculate in this article on how big Major restructuring will be in the future, but I think that to have an opinion about it you probably ought to know about those six points.

There’s more to GSK’s cash conversion than Major restructuring, but much of it is complicated. Because the item is significant and relatively simple, I’ve taken the opportunity to spell out the facts.

Looking back further

GSK started reporting core results in 2012, so I can’t look back to earlier years and use the same reconciliations as a data source. It’s still possible to look at some key items.

GSK legal and major restruc

IMO the chart above shows that major restructuring and legal expenses have not been unusually high since 2012. (If they continue at their historic levels, their negative effect on cash conversion will continue.)

Intangibles are a fairly complicated area regarding cash conversion, with several related items in the reconciliations from core profit after tax to free cash flow. They are listed on the left in this image from a spreadsheet –

GSK intangibles conversion - spread

The three minus signs in the bottom row indicate a negative effect on cash conversion from intangible assets (by “negative effect” I mean tending to pull the conversion rate down below 100%).

The lack of reconciliations before 2012 meant I couldn’t backtrack for intangible assets in a comparable way. Instead I made the chart below. It shows net cash charges lower than the net non-cash charges in most years, contrary to the results since 2012 based on all the reconciliations. The difference is mostly because I included the cash investment flow “Proceeds from sale of intangible assets” (in the red line for purchases minus proceeds). GSK’s free cash flow includes the purchases but not the proceeds, and in that sense it’s a conservative measure.

GSK intangibles expenses cash and difference

The chart above shows that in 2012 there was an exceptionaly big positive contribution from intangibles to cash conversion, which was the result of big proceeds from sales. The cash proceeds and purchase are shown in the next chart, which also has the non-cash items (I made those less prominent). The quantities in the chart are highly variable, with Purchase of intangibles looking the most regular.

GSK intangibles expense and cash

To sum up about the proceeds from sales of intangible assets – they aren’t included in the reconciliations, and the figures in the reconciliations show a negative effect on cash conversion from intangibles. When the proceeds are included, there’s a positive effect on cash conversion from intangibles. However, the proceeds have been irregular and they spiked in 2012, so when I later adjust for the proceeds, there could be some doubt over the justification for doing so.

The effect of non-controlling interests

The cash conversion in the table at the top is between core profit after tax and free cash flow. I’ve already mentioned that GSK’s definition of free cash flow excludes the sale of intangibles. There’s also an issue due to starting with core profit after tax – it excludes the core profit attributable to non-controlling interests (NCI). It’s the core profit attributable to shareholders (after deducting core profit attributable to NCI) that is divided by the number of shares to get core earnings per share. So, if you estimate a rate of cash conversion and you want to apply it to core EPS to get FCF per share, the cash conversion needs to be between core profit attributable to shareholders, and FCF. Although not the most relevant quantities, I started with core profit after tax and FCF (by GSK’s definitions) and the conversion between them, because all the data for the conversion is in reconciliations, which makes errors less likely and checking easier.

Adjusting conversion for NCI and the sale of intangible assets

Effect of NCI and purchase of intangibles

The table above shows that if you already have an estimate for the rate of conversion between core profit after tax and free cash flow (by GSK’s definitions), and you want the rate of conversion between core profit attributable to shareholders and FCF adjusted to include proceeds from the sale of intangibles, then you could add about 13.5% to the estimate (or reduce the leakage by 13.5%). However, there’s no guarantee that the combined effect of non-controlling interests (NCIs) and the sale of intangibles will be the same in the future as over the periods in the table. If you are thinking about the future, you might prefer to use the 9-months-2015 figure of 10.63% for the profit attributable to NCIs, but there’s a complication – while bigger NCIs make profit after tax bigger for any given profit attributable to shareholders, it also leads to more dividends paid out to the NCIs, and the cash payments reduce free cash flow. That’s a complicated point – the takeaway is that you should be wary of assuming the recently higher profit attributable to NCIs will mean better cash conversion from core EPS to free cash flow per share.

Refering to the previous table, the approximate leakage-reducing adjustments are in the rightmost column, in cells Q24 and Q26, with values 13.72% and 13.22% (of core PAT). The total leakage (including “other”) adjusted for non-controlling interests and the sale of intangibles is –

2012 to Q3 2015 ~ 44.67%
2012 to 2014 ~ 30.20%

[I had that wrong originally. Subtracting the adjustments in the previous table only gave approximate results, which were 42.12% and 28.65%. Amended January 31, 2016.]

Those leakages are between:

Core profit attributable to shareholders (which core EPS is based on)
Free cash flow as defined by GSK plus Proceeds from sale of intangible assets.

I believe the two metrics are more relevant than the core PAT and FCF that I started with, and I consider the figures for the two periods to be the main result here.

Relating core EPS, leakage and FCF per share

GSK FCF per share -no count no formulas - spread

The table above is just simple math and is not meant as any kind of prediciton. For example, the top left shows the obvious fact that if core EPS is 100p and the leakage before hitting free cash flow is 30%, then free cash flow per share will be 70p. The table is only useful if you have some idea of what core EPS and the leakage will be. If you think cash conversion will improve so that FCF per share will be 80% of core EPS, then you look at the row for 20% leakage. If you think core EPS will be less than 100p or more than 130p, then the table isn’t much use to you. I walled off the cells in the top left where FCF per share was less than 80p because GSK have stated their intention to pay an annual ordinary dividend of 80p for each of the years 2015 to 2017 (which excludes a special dividend).

GSK’s M-Score for earnings manipulation

According to gurufocus, GSK’s Beneish M-Score signals that the company is likely to be an accounting manipulator. If so, this analysis would be seriously undermined. I don’t think management would take the risk after the company’s trouble on the ethics front (see “Three Take-Home Messages from China’s Glaxo Verdict” by Yanzhong Huang, Sep 25, 2014 (forbes.com), and from 2012, “GlaxoSmithKline to pay $3bn in US drug fraud scandal” (bbc.co.uk)). There have also been measures to be more ethical (see “GSK confident halting payments to doctors will pay off” by Andrew Ward, January 7, 2016 (ft.com)). It could also be argued that GSK have less incentive to manipulate earnings because they can emphasise core results where various costs are excluded. Even so, my view that GSK have probably not manipulated earnings is only an opinion, and while I believe the M-Score is flawed, it is at least objective.

The rest is just details

I’ve given the adjusted leakages which I regard as the main result regarding the record. Readers who don’t like much detail should probably not read the rest of this piece, although I suggest following the comments.

Coming up, I give GSK’s definitions of core results and free cash flow, and the reasons why I used those metrics. Then I say why the areas of leakage were not easy to spot, and describe my methods. After that most of the piece consists of various breakdowns of the info I summarized in the chart at the top.

GSK’s definitions

GSK use the term “core results” for everything from turnover to earnings per share, which they arrive at by excluding many items. It’s the conversion of core “Profit after taxation” to free cash flow that I wrote about before applying adjustments.

“Core results exclude the following items from total results:
amortisation and impairment of intangible assets (excluding
computer software) and goodwill; major restructuring costs,
including those costs following material acquisitions; legal
charges (net of insurance recoveries) and expenses on the
settlement of litigation and government investigations; other
operating income other than royalty income; disposals of
associates, products and businesses, and acquisition
accounting adjustments for material acquisitions, together
with the tax effects of all of these items.” (Annual report for 2014)

“Free cash flow is the net cash inflow from operating activities less
capital expenditure, interest and dividends paid to non-controlling
interests plus proceeds from the sale of property, plant and
equipment and dividends received from joint ventures and
associated undertakings.” (Annual report for 2014)

Reasons for starting with core profit after tax and free cash flow

The reason for using a measure of core profit rather than just profit is that GSK have made statements about the growth of core EPS they expect from 2014 to 2020. (I’ve already explained that by using core profit after tax I only needed to use the reconciliations, until adjusting for the profit attributable to NCIs).

The reason for using free cash flow rather than cash from operations is that free cash flow has deductions for investment, and the same dollar can’t be used for capex and paid out in dividends. There’s room for argument about that, for example if you believe capex could be reduced without too much effect on future earnings, or if it could be reasonably financed through debt. It could also be reasonable to use definitions of free cash flow other than GSK’s own definition. I settled on using free cash flow as defined by GSK (before adjusting for the sale of intangibles) without too much agonizing, because no metric is guaranteed to be acceptable to everyone.

Leak spotting

GSK reconciliation rollercoaster

The reconciliations in GSK’s reports don’t make it immediately clear where cash conversion has been failing. I’ve lined up the various reconciliations between core earnings and GSK’s FCF, and in the chart above the biggest and most consistent drops have been in the last stage, which is GSK’s “Reconciliation of free cash flow”. However, looking into that reconciliation does not usefuly identify any leakage. For example, in 2014 there was £707 milliion interest paid, and the payment is included in the reconciliation of FCF. But, there’s a £727 million “Finance expense” in the income statement (and a smaller amount of “Finance income”), which is also in the core earnings that the series of reconciliations starts with. What has happened is that the non-cash item “Finance income net of finance expense” (of £659 million) is added back in the cash flow statement, before the cash payment of £707 milliion interest is deducted in the free cash flow reconciliation. Given the procedure, it’s not surprising that there’s a drop-off in the FCF reconciliation.

It’s the same for other deductions in the FCF reconciliation. For example there’s the cash item Purchase of property, plant and equipment, which corresponds to the non-cash item “Depreciation” which is added back in the cash flow statement. The upshot is, you don’t learn where the leakage is between core profit after tax and FCF just by looking at the FCF reconciliation, even though it looks like that’s where the problem is from the “reconciliation rollercoaster” chart.

To find where cash conversion was failing, I grouped the adjustments differently. Instead of grouping them according to which reconciliation they were in, I grouped together everything which related to “Interest”, and everything which related to “Depreciation”, etc. For each grouping I calculated a “conversion” quantity, simply by adding up the grouped quantities. The addition was possible throughout because the signs were already correct for the reconciliations, and the reconciliations formed an unbroken chain from core profit after tax to GSK’s free cash flow. For “everything related to depreciation”, the items are: Depreciation, Purchase of property, plant and equipment, and Disposal of property, plant and equipment. The first of the three is positive because it was added back in the cash flow statement, the second is negative because it represents a cash outflow, and the third is positive because it represents a cash inflow. After grouping related items, some “stand-alone” items remained, such as “Major restructuring”.

The procedure is not from a book and my terminology is likely to be non-standard. I called the sums “conversions”, but “net adjustements” (between core profit after tax and GSK’s FCF) might have been better.

I could only take that particular analysis back to 2012 when core results were first reported. The figures for 9 months 2015 are lacking in detail, for the reconciliation between “Profit after tax” and “Cash generated from operations”. I’ve ‘kludged’ them into the same categories as for the whole years, by setting non-disclosed quantities to zero. The non-disclosed quantities are not likely to have been ignored by the accountants, and they could be in “Depreciation and other adjusting items”. The period “9 months 2015” includes seven months after the Novartis transactions.

Leakage broken down by year and type

I’ve charted the spreadsheet results in various formats, either absolute amounts or as percentages, with different splits of the 3 year and 9 months period. I’ll start with a cumulative chart which shows how the adjustments or “conversions” eat away at the core earnings at the top to end up with the smaller amounts of free cash flow at the bottom.  A negative adjustment pulls a line to the left, and a positive adjustment pushes it to the right.

GSK conversion

The next chart is the same except the quantities are expressed as a percentage of the year’s core earnings. It means the series all start at 100% at the top, making it easier to compare years.

GSK conversion percent

The two charts above were cumulative, which meant they were good for showing how the various accounting items eroded core profit after tax on it’s way to being reconciled to free cash flow, but the impact of each item could only be represented by quite a small horizontal distance. The next two charts are not cumulative and it’s easier to see how the effect of the items compare to each other.

In the next chart, quantities are scaled as a percentage of “total conversions” (for the period), for example the red bar for “Major restructuring” reads over 25%, which means that over 2012 to 2014 the category accounted for more than a quarter of the erosion between core profit after tax and GSK’s FCF. Note that “Share-based incentive plans” (at the top of the chart) is negative, at about -13%. The plans are good for cash conversion because they are a non-cash expense, making earnings lower (the higher the expense) without affecting FCF. Because the expense is good for cash conversion, it’s a negative percentage of the erosion or leakage between core earnings and FCF.

GSK conversion period totals PC

The next chart shows absolute amounts rather than percentages. Visually, it’s a mirror image of the previous chart, because the negative adjustments just stay negative (instead of being a positive percentage of the total negative adjustment).

GSK conversion period totals

The next chart breaks the absolute amounts down into four distinct periods, which means you can tell how consistent or variable the effect of each item is. None of the “conversions” are constant, but some are more consistent than others. For example “Major restructuring” has been consistently high, although in the two low periods the item was only about half the size as in the two high periods. In contrast, the “Acquisition and disposal” conversion has not been consistent, being highly positive in 2012, while being negative in the other periods, and very highly negative in the 9 months 2015 period.

GSK conversion items


I’ve written “GSK’s cash conversion – systematic guesswork“, where I try to assess the future rate of cash conversion. As the title implies, I’m not confident about being able to predict the future accurately.

Also, “Checking the overall leakage results in my cash conversion spreadsheets for GSK“.

For the tables where I calculated the “cash conversion factors”, see “GSK cash conversion – excess charts and selected spreadsheet images“.

DISCLAIMER: Your investment is your responsibility. It is your responsibility to check all material facts before making an investment decision. All investments involve different degrees of risk. You should be aware of your risk tolerance level and financial situations at all times. Furthermore, you should read all transaction confirmations, monthly, and year-end statements. Read any and all prospectuses carefully before making any investment decisions. You are free at all times to accept or reject all investment recommendations made by the author of this blog. All Advice on this blog is subject to market risk and may result in the entire loss of the reader’s investment. Please understand that any losses are attributed to market forces beyond the control or prediction of the author. As you know, a recommendation, which you are free to accept or reject, is not a guarantee for the successful performance of an investment.

GlaxoSmithKline – core EPS projected to 2020

Disclosure: I’m long GlaxoSmithKline plc (GSK).


• I’ve improved my projections of core EPS based on GSK’s statements.

• The projections for 2020 range from 102.38p to 128.20p with a mean of 114.85p.

• The projections are spread fairly evenly, not bunched around the mean.

About this piece

GlaxoSmithKline have projected the growth of core EPS through to 2020 in statements like “CAGR of mid-to-high single digits” (for 2016 to 2020). I made numeric projections based on those statements, in “GlaxoSmithKline – upside in the pipeline but some risk to the dividend” December 4, 2015 (on my blogsite, on Seeking Alpha). Here I’ve refined those projections. I’ve based a long piece on just a few statements from GSK, because I want to cover the area properly before I write about cash conversion from core EPS to free cash flow.

Two changes of method

The first difference is that I’ve made a single set of four different assumptions about growth from 2015 to 2016. Previously I’d made low, mid and high assumptions, then I thought they might be too high and recalculated using lower values. With low, mid and high assumptions for the other two periods, there are now 36 different possible combinations. The second difference from my previous work is that I’ve calculated the outcome for each combination, which let me see how the outcomes are spread.

I’d originally estimated low, mid and high values of 103.32p, 116.25p, and 130.49p for core EPS in 2020, and then after making a more pessimistic assumption I estimated 102.38p, 114.17p and 127.06p. Now I’m projecting a range from 102.38p to 128.20p with a mean of 114.85p.

2020 core EPS

The first chart shows 2020 projected core EPS for each combination of assumed growth (low, mid, high, and for 2016, higher).

GSK core EPS 2020 variation

The chart shows that most of the variation depends on the growth assumed for 2016 to 2020, which is not surprising because the period covers four years while the other two periods are only one year long. Visually, the sawtooth pattern has more variation between its lows and highs than there is for the general rise from left to right. The sawtooth pattern is the result of cycling through “low, mid and high” for 2016 to 2020 every time the combination is advanced to the next along the X axis.

The absolute effect of changing the assumed growth for one of the three periods depends on the growth assumed for the other periods, but I can give a rough idea of the effect:

Period of growth ~ Change in 2020 core EPS

2014 to 2015 ~ 5.0p to 6.0p (L to H)
2015 to 2016 ~ 2.8p to 3.4p (L to H2)
2016 to 2020 ~ 16.5p to 17.9p  (L to H)

For example assuming HML instead of LML increases the 2020 projection by 5.10p (from 103.32p to 108.42p), which is within the range 5.0p to 6.0p for 2014 to 2015.

Comparing the midpoints, the periods account for about 21.3%, 12.0% and 66.7% of the total variation (between the lowest growth assumption LLL and the highest, HH2H).

The quotes in the graphic are from the Q3 2015 results announcement, and the core EPS of 95.4p to which the growth rates are applied is from the annual report for 2014.

The spread of results

Sorting the projected 2020 values shows how evenly spread the results are.

GSK core EPS 2020 projections - size order

Cumulative probability

GSK cumul unweighted 2020 core EPS

The next chart is like the previous except I weighted the outcomes, giving “mid” cases twice as much weight as “low” and “high” cases (and for 2016 growth, giving “mid” and “high” cases twice as much weight as “low” and “higher” cases). The weighting is arbitrary, but it does at least reflect the fact that outcomes should usually be less likely the further they are from a projection, for example the 2016 to 2020 “CAGR of mid to high single digits” is probably more likely to be 7% than 5% or 9% if GSK’s statement is accurate.

GSK cumul weighted 2020 core EPS

If you prefer looking up values in a table to reading from the chart, find “Cumulative probability tables” below.

Here I’ve fitted straight lines, and for the unweighted case the line explains 99.85% of the variation, while for the weighted case the line explains 99.72% of the variation.

GSK cumul prob 2020 core EPS

For many situations in which a range of outcomes is possible, outcomes at the low end or at the high end are much less likely than outcomes in the middle. That’s not the case here, because both of the charts above are close to linear. The fact that low-end outcomes are not particularly unlikely matters, because at the low end the dividend cover is not great, and cash conversion from core EPS to free cash flow has not been great either.

I’m not saying that the probability distribution for 2020 core EPS does not have a peak in the middle, I’m only investigating the implications of GSK’s statements and my numeric interpretations of them. So far as I can tell, if you rely on GSK’s statements and you believe my low, mid and high (and higher) assumptions are reasonable, then to be consistent you can’t rely on the bottom-of-the-range outcomes being unlikely. If you believe GSK’s statements are conservative rather than accurate without positive or negative bias, then it’s reasonable to believe that bottom-of-the-range outcomes are unlikely.

Frequency charts

There are problems with frequency tables and charts based on them when the data set is small, and I don’t believe they are as useful as the cumulative probability charts above. There aren’t enough data points to get a good frequency chart (which is why there are a few different ones), but it’s clear that there isn’t much bunching around the average.

GSK projection frequency charts

GSK wide frequency chart

I’ve also tried weighting the results that went into the frequency analysis, as for the cumulative probability, so that a “mid” assumption carries twice the weight of a “low” or “high” assumption (for 2016, “mid” and “high” were given double weight, relative to “low” and “higher”).

GSK weighted projection frequency charts


GSK wide weighted frequency chart with trend

I fitted a line in the chart, using the best fit out of linear, logarithmic, exponential and power functions. The value of R squared is 0.037 which means it only explains 3.7% of the variation.

The weighted frequency charts are still more flat than bell-shaped.

Rescaling for dividend cover

From the Q3 2015 results announcement, “The Group also stated its intention to pay an annual ordinary dividend of 80p for each of the next three years (2015-2017).”). Dividing core EPS by 80p gets a measure of how well that dividend would be covered.

GSK div cover from core EPS 2020 projections - size order

How projected core EPS fans out

Next I project core EPS for each combination of assumptions. The result is a series of lines fanning out with no obvious concentration inside the range. (The units are pence per share.)

GSK core EPS projections - fan

Projected CAGR – the difference a year makes

GSK core EPS CAGR from 2014

GSK core EPS CAGR from 2015

There are fewer lines in the chart above because the three possibilities for growth from 2014 to 2015 are not relevant. The apparent convergence towards just three lines is the result of two facts – the assumptions for 2016 to 2020 core EPS have more influence in the later years, and three different assumptions were made for the period: low, mid and high. There’s no reason to expect a ‘tri-modal’ distribution, and a 2020 value in the gaps is probably about as likely as a value inside one of the three groups of lines.

The math behind the even spread

For anyone interested in the math, there’s a simple explanation of why the 2020 projections are spread evenly. If you flip an unbiased coin four times and count the number of heads, two heads are more likely than no heads or four heads, because there are six ways of getting two heads (TTHH, THTH, HTTH, THHT, HTHT and HHTT), but only one way of getting no heads (TTTT) or four heads (HHHH). That’s also true for your winnings if you won one dollar for each head. But, if you won $8 if the first flip was heads, $4 for the second flip, $2 for the third and $1 for the fourth, there’s no bunching around the average. Possible winnings range from $0 to $15, and for each whole number of dollars in that range, there’s exactly one way of getting it. For example, the only way to get $6 is from THHT where the winnings are $0 + $4 + $2 + $0. The outcomes can be converted to binary numbers which give the value of the winnings, e.g. THHT becomes 0110 which is 6 in base 10. No other binary number equals 6, and the range of binary numbers 0000 to 1111 can’t bunch around the average any more than for the range of decimal numbers 0 to 15.

The relevance to the projections here is that the result is sensitive to the assumptions about the 2016 to 2020 period, because it lasts four years. That’s like the ‘eight dollar bet’ in the example, except that growth multiples are multiplied while the winnings in the example were added. I was still able to give rough percentages for the variation accounted for by the three periods (about 21.3%, 12.0% and 66.7%). I don’t prove that the effect of the assumptions is weighted just right to give the linearity in the results, but maybe it’s evident due to the results following mathematically from the assumptions. I expect there would be a bell curve if I modeled growth over 2016 to 2020 separately for each year, but a wider range of assumed growth for each year might be required.

In the rest of the piece I show and explain the work which supports the results given above. Although I avoid complex math there could still be too much information for many readers.

Frequency charts

Wikipedia have a quite readable page about Frequency distributions, and you can find many elementary texts by searching for “frequency tables”.

I’ve used the term “sub-range” for the parts produced by chopping the total range into equal parts, but “bin” is a common term with the same meaning. A sub-range or bin is also a class, but “class” is also used for discrete cases where you wouldn’t use those terms, for example if the classes are “Democrat” and “Republican”.

It would have been conventional to divide the 36 cases into six classes (from the square root of 36, or the more accurate formula C = 1 + 3.3 * log(n)), but I like to get various views of the same data and I used 4, 5 and 18 classes. For a small data set, a value can be very close to the edge of a class and a tiny difference in the value would make a noticeable difference to the table. If a single value is not the minimum or maximum value, it’s hard for it to be close to the edge of a sub-range in three different ways of dividing the total range into equal sub-ranges, especially when one of the divisions is into a prime number of classes (as in the bar chart with 5 bars).

Some of the frequency charts shown earlier (the ones with 4 or 5 bars representing sub-ranges) were derived from versions of the chart showing projected core EPS in size order (after “shows how evenly spread the results are”, above). To make room on the right, I rescaled the X axis to start at 90p (core EPS), which makes the variation look greater (if you ignore the scale on the X axis). I then put the 2020 core EPS values in ranges of equal size, and made various notes on the right.

GSK core EPS 2020 in 4 subranges

In the chart above, there are ten purple bars in the bottom range (from 102.38p to 108.835p). That’s noted on the right by “10 projections”, and the number 10 was used in the unweighted frequency chart for four sub-ranges. Getting the figures for the weighted frequency charts was more complicated, and the scheme is shown in the next chart. One of the purple bars is labelled “MH2L”. The “M” counts as “inner” because it’s the “mid” possibility out of low, mid or high. The “H2” counts as “outer” because it’s the “higher” on the end of the possibilities low, mid, high and higher, and the “L” for “lower” counts as “outer” because it’s at the begining of low, mid and high (if it’s the lowest or highest growth case, it’s “outer”). So MH2L contributes two to the “outer” count and one to the inner count.

GSK symbol weighting

In the previous bar chart, the counts are shown to the right of the purple bars, as –

21 outer
9 inner

The weighted score of 39 was obtained by multiplying the “inner” 9 by 2, and adding the “outer” count 21, to get 9*2 + 21 = 18+21 = 39. The figure 39 is used for the sub-range in the four bar weighted frequency diagram. I also checked that the results agreed with the formula: projections = 3 * (outer + inner). The multiplication by 3 is because there are three symbols for each combination (one symbol for each period), and each of them is categorized as either outer or inner.

In the bar chart above, the total range was split into four, and I’ve described the procedure for the bottom sub-range represented by the purple bars. It’s similar for the other sub-ranges (represented by other colors), and the same kind of procedure was applied in the next chart to the five equal sub-ranges that the total range was divided into.

GSK core EPS 2020 in 5 subranges

The other frequency charts have 18 sub-ranges, and the procedure for getting the data for those charts was more complicated. It might be possible to figure it out from the spreadsheet images shown later (find “the tables I used for the 18-bar charts” below), but it might be easier to make your own frequency charts if you want to check the results.

There might be a subtle selection bias in the frequency tables – by considering the total range between the lowest and highest values, when the range is divided into sub-ranges, the lowest sub-range is guaranteed to include at least one item (the lowest value), and the highest sub-range will also include at least one item (the highest value). Arguably, there’s no such selection bias because the data is not a sample. In any case, even if 1 is subtracted from the top and bottom sub-ranges (or 3 for the weighted versions), the results do not look like bell curves. I’ve checked that, but rather than go into the complications here, I’ll point out that the cumulative probability charts are enough to show that outcomes for 2020 core EPS at the lower end of the range or at the upper end are not particularly unlikely. I also use that fact to only write a little about skewdness (how asymmetric a distribution is) and kurtosis (how peaked it is). Both are low. The distribution is skewed very slightly to the right of the peak, but I found no free software that gives skewdness and also says where that peak is (you can’t assume it’s the mean or the median). Kurtosis is particularly problematic as different software uses different measures of it, nearly always without saying which.

Spreadsheet tables

In the rest of the piece I show the spreadsheet tables the charts are based on, and some of the formulas. In the first image you can see I labelled the fourth assumed value of growth in 2016 as “higher”. I could just as well have shifted the values to the left in the table and labelled them “lower”, “low”, “mid” and “high”.

GSK core EPS growth multiples spread

GSK core EPS growth from 2014 spread

GSK core EPS growth from 2015 spread

GSK core EPS projections spread

There are no formulas for the next chart. I needed to copy, transpose and sort, and one of those turned the formulas into figures.

GSK core EPS sorted on 2020 value - spread

GSK dividend cover using sorted core EPS  - spread

I didn’t use a spreadsheet for the calculations for the four bar and five bar frequency charts. I show the tables I used for the 18-bar charts, but some of the procedure was manual and not easy to explain (or follow, probably). The cells under “Nearest middle, first” were used to check for a peak but were not used above.

GSK frequency spread

Cumulative probability tables –

GSK cumul 2020 core EPS spread

The next image shows a selection of formulas. Showing all of them would take a lot of space, so I’ve only shown the first few columns. The formulas follow a pattern, and most of the formulas were put in the cells by dragging.

GSK core EPS projections formulas

My calculation of the arithmetic mean of the 2020 core EPS projections uses:
with the result (in pence):

OpenOffice’s documentation does not state that the function gives the arithmetic mean. I checked that OpenOffice’s AVERAGE gives the same result as the arithmetic mean using:


in cell BA53, with the result 4134.59, then I divided:


with the result 114.85 which agrees with the AVERAGE calculation. For more precision than is warranted, it’s 114.849840. The SUM and division calculations are not in the spreadsheet image.

GSK core EPS averages formulas

This image shows the formula view of the tables used for the 18-bar frequency charts.

GSK frequency formulas

The formulas for cumulative probability (of 2020 core EPS) –

GSK cumul 2020 core EPS formulas

That’s all, thank you for reading this.

DISCLAIMER: Your investment is your responsibility. It is your responsibility to check all material facts before making an investment decision. All investments involve different degrees of risk. You should be aware of your risk tolerance level and financial situations at all times. Furthermore, you should read all transaction confirmations, monthly, and year-end statements. Read any and all prospectuses carefully before making any investment decisions. You are free at all times to accept or reject all investment recommendations made by the author of this blog. All Advice on this blog is subject to market risk and may result in the entire loss of the reader’s investment. Please understand that any losses are attributed to market forces beyond the control or prediction of the author. As you know, a recommendation, which you are free to accept or reject, is not a guarantee for the successful performance of an investment.

GlaxoSmithKline – upside in the pipeline but some risk to the dividend

Disclosure: I’m long GlaxoSmithKline plc (GSK), Gilead Sciences, Inc. (GILD)


• Slow growth is projected (I’ve made numeric projections of core EPS based on GSK’s statements).
• The record of approvals from the FDA suggests upside in the pipeline, but that needs to be weighed against the risk to the dividend.
• The Novartis “put” option for the consumer business is booked as a liability without recognizing the benefit of buying out Novartis.
• Growth of the consumer business could increase the liability for the “put” option.
• A large number of miscellaneous points complicate the investment case.

About this piece

GSK seemed like an undervalued stock but as I researched I found a few more negative points than positive, and a lot of complication. I see too much risk for any margin of safety, and this long piece is not a good use of time for investors who just want to find good investments. I’ve listed the 58 headings here, which might help some readers find their way around.

The company and stock

GSK is a large British healthcare company with international operations and revenue, split between Pharmaceuticals (59%), Vaccines (16%), and Consumer Healthcare (25%) (2014 sales restated to reflect a shake-up). Morningstar put “ADR” after the company name, while Seeking Alpha have “NYSE” after the name and stock ticker. This UK broker’s site says each ADR represents two ordinary shares.

Big changes

Andrew Witty became CEO on May 21, 2008, and from about 2008 to 2011 R&D staff were reorganized into small teams, designed to mimic the nimble and entrepreneurial nature of small bio-tech companies. Annual reviews could result in projects being stopped and teams broken up. There have been reasons to be sceptical about the “small teams” structure, which I outline under “Negative articles” further down. The quantity and quality of approvals from the U.S. Food and Drug Administration (FDA) in recent years are evidence of the change working, but there have also been failures. Pharma sales have been hit by patent expiry, and GSK’s big seller Advair has patent expirations and has had pricing pressure.

A deal completed in March 2015 meant Novartis got GSK’s oncology business while GSK acquired vaccines from Novartis. GSK and Novartis also pooled their consumer healthcare products into a joint venture, where GSK has control with 63.5% of the equity. The joint venture combines GSK’s sales capability in emerging markets with Novartis’s in Central and Eastern Europe and CIS (a trading bloc with old Soviet Union countries). The deal was immediately dilutive to earnings.

Some of the expected cost savings have already been realized, through eliminating duplication in sales, admin, and infrastructure, through less need of third party contractors, and other economies of scale. Sales of consumer products and vaccines don’t fall off a cliff when patents expire, so making those segments bigger should provide some long term stability (vaccine sales are lumpy, so they won’t reduce short term variability).

GSK still has oncology products in the pipeline, and they were not offered to Novartis. At least six of them are planned to receive approval in 2018 to 2020, and oncology is one of six areas GSK say they are focused on – HIV/Infectious Disease, Respiratory, Vaccines, Oncology, Immuno-inflammation and Rare Disease. Novartis get “first refusal” on new oncology drugs but they can’t stop GSK from commercialising oncology drugs themselves or accepting a better offer.

Sales and innovation breakdown, and Rx/Cx switching

The figures are from the Financial outlook and guidance presentation, 6 May 2015. They are “pro forma” 2014, as if the Novartis transaction had been completed before 2014.

Sales by geography
US 30%
EU 29%
International 41%

Sales by segment
Pharma 59%
Vaccines 16%
Consumer healthcare 25%

Segment innovation
GSK use “Rx” for prescription medication, “Vx” for vaccines, and “Cx” for consumer healthcare. A chart headed “Balanced innovation” gives:

Vx 3, Rx 8, Cx 10%+ Innovations sales.

The Vx and Cx numbers are explained with –

“Includes key recent and near-term launches plus late-stage assets. Rx: Breo, Anoro, Incruse, Arnuity, Tanzeum, Nucala. Tivicay, Triumeq, Vx: Menveo, Bexsero, Shingrix”

From the associated transcript –

“innovation sales being products that were launched on a rolling basis in the last three years. But sometimes that number will spike, as it has done in the first quarter this year, especially if you are able to switch a product from prescription to over the counter.”

Advair and generics

The patent has expired in Europe, where the product is known as “Seretide” in most countries (see Wikipedia’s “Fluticasone/salmeterol“). In the USA the diskus device patent expires in 2016, while the HFA-device expires 2015-2026 (according to the table “Pharmaceutical products, competition and intellectual property” in the annual report for 2014). The diskus device is a dry powder inhaler and “HFA” is an inhaler propellant.

In 2014, sales of the Seretide/Advair respiratory products were down 15% to £4,229 million, which was 18.4% of “Turnover” (revenue).

GSK’s guidance assumes some impact from generics. The impact is hard to predict, because replication of Advair is not easy, but not impossible either. From the Q3 results transcript, about Advair –

“we have seen a lot of price pressure and we have seen increasing generic competition ex-US in emerging markets and, to a lower degree, in Europe.”

Generic manufacturers seem to have a problem either in making a close copy of Advair, in controlling quality, or at least in getting FDA approval. From the Q3 results transcript, about Advair in the US –

“If you project forward and look at the average review times of the FDA and all of those, particularly given that this is a reasonably complicated product to generate a generic for, to state the obvious, it seems unlikely to me that this would go very quickly. Could it conceivably start sometime in 2017? Conceivably, yes. Could it be later than that? Yes. Could it fail? Yes. So, unfortunately, I am not really the man to help you too much. What we have tried to do to help you is what we said back in May, when we gave you an indication of how we saw the growth rate of the company running all the way through to 2020, essentially to signal to you that in that period when we dialled in, just for the purposes of that assumption, a genericisation in America, it reassured us that, even if we do have a genericisation of America during that period, we can deliver good, solid, sustainable sales growth and earnings growth for the company between May 2015 and 2020.”

IMO the statement suggests more upside from slower generic competition, than downside from faster generic competition, compared to GSK’s assumptions about Advair, but I have not seen the exact assumptions so I can’t be sure. Sales have declined in the US due to price reductions agreed with the giant Pharmacy Benefit Manager (PBM) Express Scripts, which got Advair back on the “preferred” list, see “Note to Big Pharma: Discounts work. GSK price cuts score Advair a payer boost” by Carly Helfand, August 5, 2014 (fiercepharma.com). From the Q3 results transcript –

“US Pharma sales were down 10% proforma, primarily driven by Advair which was done 18%. It is down 19% year-to-date as we absorb the price reductions we agreed last year, but also as we transition our portfolio to the new products.”

The price cut was forced, but it reduces the incentive to generic manufacturers to get an FDA-approved copy of Advair.

GSK mention “the sales decline in Advair/Seretide of £182 million in the quarter.”. The decline isn’t over – “despite the continued decline of Seretide/Advair sales”, but it’s slowing, “you will see a continued reduction in Advair price in the US in terms of the net price that we are charging, but the rate of decline is decelerating.”.

“As far as Seretide in Europe is concerned, essentially we’re continuing to see generics launch in different countries. The bottom line is that, with a few exceptions, the generics are taking relatively low volume shares, so perhaps 3-5% market shares. Most of the hit – not all as there is clearly a volume hit as well – is price that we are taking and we are taking price to retain share in a number of countries. However, I would fully expect that pressure to continue.”

Other competition from generics

From the Q3 results transcript (my bold) –

Avodart going generic in the US at the start of Q4″

“The overall decline in the US was also affected by a continuing tough comparator for Lovaza which was down 66% post the introduction of generics last year. The business had £44 million of sales in Q4 last year, so this will continue to be a headwind into the fourth quarter.”

There’s patent expiration info for the top four sellers under “Legal proceedings”, below.

After Advair

“As you then rotate through the potential generic Advair scenario, one way or another, as I have described, and you look at our business the other side of that – wherever you choose to put that window. If you look at the business on the other side of that, you have a business where there is no material intellectual property rights threat to the company’s portfolio until 2026/27.” (Q3 results transcript)

GSK’s statements about core EPS and its growth

GSK expect core EPS to dip and then grow. That’s partly the result of products coming off patent while new products gain traction, and also because the deal with Novartis was earnings dilutive, but with cost reductions accumulating until 2017. Core results exclude –

“amortisation and impairment of intangible assets (excluding computer software) and goodwill; major restructuring costs, including those costs following material acquisitions; legal charges (net of insurance recoveries) and expenses on the settlement of litigation and government investigations, and acquisition accounting adjustments for material acquisitions, disposals of associates, products and businesses, other operating income other than royalty income, and other items, together with the tax effects of all of these items.”

My projections are based on statements like “CAGR of mid-to-high single digits”. The figures for EPS depend on core EPS for 2014, which is 95.4p in the annual report. The basic EPS and the diluted EPS were 57.3p and 56.7p, a lot less than the core EPS. The figure is also in this quote from a transcript –

“The flexibility we have built into our cost base in recent years enabled us to respond quickly and effectively to these challenges during the year, so that we were able to deliver core EPS of 95.4 pence, down 1% on an ex-divestment constant currency basis despite a top line decline of 3% on the same basis.” (q4-2014-analyst-presentation-transcript.pdf)

I’ve stuck to British currency, and 95.4 pence is also 95.4p or £0.954, or $1.44 (USD).

These quotes from the Q3 2015 results announcement give the percentage change expected. CER is Constant Exchange Rate. I’ve made the most relevant words bold –

“Guidance for 2015

Core EPS for 2015 is expected to decline at a percentage rate in the high teens (CER), primarily due to continued pricing pressure on Seretide/Advair in the US/Europe, the dilutive effect of the Novartis transaction and the inherited cost base of the Novartis businesses. The 2015 guidance excludes potential income from the proposed divestment of ofatumumab, which was announced on 21 August 2015. The Group now intends to treat income generated from the proposed divestment as a non-core item, as set out on page 45.”

“2016 outlook

In 2016, GSK expects to see a significant recovery in core EPS with percentage growth expected to reach double digits on a CER basis as the adverse impacts seen in 2015 diminish and the sales and synergy benefits of the Novartis transaction contribute more meaningfully.”

“At its Investor Day on 6 May 2015, GSK outlined a series of expectations for its performance over the five year period 2016-2020. This included an expectation that Group core EPS would grow at a CAGR of mid-to-high single digits on a CER basis. The introduction of a generic alternative to Advair in the US was factored into the Group’s assessment of its future performance. The Group also stated its intention to pay an annual ordinary dividend of 80p for each of the next three years (2015-2017). For more information see:” gsk.com/en-gb/investors/investor-event

The CFO has said “we will see variation year to year around those averages” (Financial outlook and guidance, May 6, 2015, transcript).

GSK’s words translated into figures

I’ve made low, mid and high assumptions, which I put into a spreadsheet.

low ~ mid ~ high

2015, “decline at a percentage rate in the high teens”
-19% ~ -17% ~ -15%

2016, “percentage growth expected to reach double digits”
10% ~ 12% ~ 14%

2016-2020, “CAGR of mid-to-high single digits”
5% ~ 7% ~ 9%

Low, mid, high and probability

In the spreadsheet, the “low” outcome for core EPS assumes all three low cases. If low, mid and high were all equally likely (for the actual core EPS to be closest to), then the probability of all three cases being low is only 0.037, or 3.7% (assuming the cases are independent). That’s also true for “mid” and “high”. I’d guess there’s at least 90% probability of the 2020 EPS being between the low and high values shown, with figures near “mid” more likely, but this is fundamentaly not a rigorous statistical excercise due to reliance on GSK’s statements and translating the words into low, mid and high cases.

The spreadsheet

GSK core EPS CER growth to 2020

The resulting CAGR of core EPS for 2014 to 2020 is:

low ~ mid ~ high
1.34% ~ 3.35% ~ 5.36%

I also tried reducing the 2016 assumed growth (“percentage growth expected to reach double digits”) to 9%, 10% and 11%, with the results:

low ~ mid ~ high

Core EPS 2020
102.38 ~ 114.17 ~ 127.06

CAGR from 2014 to 2020
1.18% ~ 3.04% ~ 4.89%

CAGR from 2015 to 2020
5.79% ~ 7.59% ~ 9.40%

Dividend cover from core EPS
2015 0.97 ~ 0.99 ~ 1.01
2016 1.05 ~ 1.09 ~ 1.13
2017 1.11 ~ 1.16 ~ 1.23

I’ve charted core EPS based on the lower assumptions for 2015 growth.

GSK core EPS CER chart

Core EPS and the dividend

The figures are hardly exciting, but there is at least the benefit of some growth in core EPS on top of the dividend.

The yield was 6.07% as at Dec 4, 2015. Based on “80p for each of the next three years” ($1.21) at the current share price of 1,319.50p (the “buy” price) ($19.94), there’s a 6.06% yield, which is in the future. There are complications, with the “future yield” slightly lower than the yield now, and a special dividend due for payment on April 14, 2016. (You can see the special dividend, share price, yield etc. on Hargreaves Lansdown. The 2014 EPS shown is 95.4p, which is GSK’s core EPS, not the Basic or Diluted EPS.)

Both the current and future yield I described are slightly above “US High Yield BB Effective Yield:” (ycharts.com), the second-highest grade (below BB+) for junk bonds (Wikipedia), where the risk is higher than for investment grade, and for bonds there’s no prospect of growth. Because transatlantic trade isn’t hard, the relative pricing suggests a lack of confidence in GSK’s ability to maintain the dividend. GSK’s debt is rated investment grade, see “Credit ratings” below.

Core EPS and other per share quantities

Core EPS needs to be examined because excluding costs doesn’t make them disappear. I’ve charted EPS and cash flow per share quantities in various ways, from 2012 (when GSK started giving core EPS) to 9 months 2015.

The period “9 months 2015” includes seven months since the transactions with Novartis, so that’s seven months without the oncology business, seven months with the Vaccines business acquired from Novartis, and seven months with the new consumer business which includes Novartis’s former consumer business, and where Novartis has a 36.5% minority shareholding.

My “Cash from operations” (or “Cash from ops” or “CFO”) is always after deducting “Taxation paid”, which is different to GSK’s terminology (their “Cash generated from operations” is before the deduction). I also deduct net interest paid, and payments to minority shareholders net of dividends from associates.

I show two series of charts. The first three have CFO with the adjustments described above, and the investment flow is subtracted to get a kind of free cash flow. Subtracting the investment flow is not conventional, it includes acquisitions and disposals, and purchase and sale of equity investments. At least some of the equity investments are in companies where GSK have research collaborations, and that’s likely to account for all of the equity investments.

The investment flow includes “Decrease in liquid investments” which is not relevant, but it’s been miniscule except for £224 million in 2012 (4.15p per share). It also includes “Dividends from associates and joint ventures” and “Interest received” which results in some double counting in free cash flow, because those amounts are also in my adjusted OCF. Those dividends and interest received only sum to £286 million (5.9p per share), over 2012 to nine months 2015.

In the second set of three charts, CFO does not include changes in working capital. Free cash flow is calculated by starting with GSK’s own version and subtracting changes in working capital. From the annual report for 2014 –

“Free cash flow is the net cash inflow from operating activities less capital expenditure, interest and dividends paid to non-controlling interests plus proceeds from the sale of property, plant and equipment and dividends received from joint ventures and associated undertakings. It is used by management for planning and reporting purposes and in discussions with and presentations to investment analysts and rating agencies.”

GSK EPS and cash flow per share

The chart above shows a great deal of variability in everything except core EPS. The annualization of the nine month period is only reasonable for core EPS, due to the effect of a big one-off investment inflow, with “Disposal of businesses” worth £10,253 million mostly from selling oncology – about the “Acquisition accounting” item, “This included the profit on disposal of the Oncology business to Novartis of £9,233 million” (or 172p per share, my calculation). The annualization consisted of multiplying by 4/3, and the movements would still be sharp without it.

GSK per share quantities cumulative

The “cumulative” chart above shows that core EPS has been less than reported EPS over the period, so it has not just been a way to ignore costs, although it would have been the other way round without the profit on selling Oncology, which is excluded from core EPS. Cash from operations was only 72% of core EPS, showing low cash conversion. There’s actually more free cash flow, due to the investment inflow in the 9 month 2015 period.

GSK cumulative core EPS and FCF

The ratio of free cash flow to core EPS over 2012 to 2014 is 64%. The period excludes the big investment inflow from selling oncology.

The variability in the figures and the change with the Novartis deal mean these ratios cannot be projected with confidence. There is scope for them to be higher through better cash conversion.

My original estimates of core EPS in 2020 were:

low ~ mid ~ high
103.32p ~ 116.25p ~ 130.49p

If cash from operations per share is 72% of core EPS (as averaged since 2012), then CFO per share in 2020 =

low ~ mid ~ high
74.39p ~ 83.70p ~ 93.95p

If free cash flow per share is 64% of core EPS in 2020 (as averaged 2012 to 2014), then FCF per share in 2020 =

low ~ mid ~ high
66.12p ~ 74.40p ~ 83.51p

Dividends have to be paid from free cash flow to be sustainable, and the low CFO estimate and the low and mid FCF estimates are less than the 80p dividend set for 2015, 2016 and 2017. It would take an improvement in the ratio of FCF to core EPS to get FCF per share of 80p in 2020, for the low and mid projected core EPS. In fact the ratio would need to be 68.82% (instead of the historic 64%) for the 2020 “mid” core EPS of 116.25p to convert to 80p FCF per share.

That seems achievable, as the problem is not that GSK have to make massive investments just to stand still, instead the problem has been low conversion of profit to cash from operations. However, “achievable” is not the same as certain. The ratio FCF to core EPS will be better if GSK can avoid non-core costs like big fines, restructuring charges and other “one-offs”, preferably for a long time, and the ratio won’t need to grow as much if core EPS exceeds expectations.

In the next three charts, the effect of changes in non-cash working capital is excluded. Following convention, in the context of cash flow I omit “non-cash”, although stricly speaking “working capital” is just current assets minus current liabilities, while non-cash working capital excludes cash and other liquid assets (like short term securities) used to park cash in rather than in the business operations, and it excludes loans. In the context of cash flow and owner’s earnings, “change in working capital” is like the decrease in it, not the increase, and a more logical term would be “the effect on cash of the change in (non-cash) working capital”.

The usual rationale for excluding changes in working capital is that big changes in it are not usually sustainable and the company has some control. If inventory fell from $100 million to $90 million, it couldn’t keep falling by $10 million for another ten years. For other cases such as an increase in receivables it’s not quite so clear that it’s unsustainable, for example sales growth could allow further increases even if Days Sales Outstanding stayed constant.

There’s more detail in “Changes in Working Capital and Owner Earnings – The Complete Guide” November 19, 2015 (oldschoolvalue.com). The guide isn’t quite complete. It’s right to concentrate on the working capital cycle, but it doesn’t say how the cycle can be affected by growth, seasonality or window-dressing synchronised to the reporting period, or how factors like a changing mix of products sold affects working capital, or how selling more through distributors or into Asian countries where payment takes longer can affect the cycle.

In the next three charts I use GSK’s own figures for free cash flow, and subtract the effect on cash of changes in working capital, from GSK’s FCF and from my adjusted cash from operations.

I’ll also mention here that Basic earnings per share were 82.6p, 95.5p and 94.4p in 2005, 2006 and 2007, so you can see there hasn’t been much growth in EPS since 2006.

GSK EPS and cash flow ex WC per share

GSK per share quantities cumulative CFO ex WC

GSK cumulative core EPS and FCF ex WC

Excluding the effect of changes in working capital, the ratio of cash from operations (with my adjustments) per share to core EPS over the period 2012 to Q3 2015 is 86% (using the figures in the bar chart). Applying that ratio to my original estimates of core EPS in 2020:

low ~ mid ~ high
103.32p ~ 116.25p ~ 130.49p

I get CFO per share:

low ~ mid ~ high
88.86p ~ 99.98p ~ 112.22p

Subtracting the effect of changes in working capital from GSK’s own figures for free cash flow, the ratio of FCF per share to core EPS over the period 2012 to 2014 is 56%. Applying that ratio to my original estimates of core EPS in 2020, I get FCF per share:

low ~ mid ~ high
57.86p ~ 65.1p ~ 73.07p

Those figures are less than the 80p dividend planned for 2015 to 2017. Because FCF by GSK’s definition excludes acquisitions, it can’t be increased by not making them. The ratio would need to be 68.82% (instead of the historic 56%) for the 2020 “mid” core EPS of 116.25p to convert to 80p FCF per share. There are still reasons to believe the dividend can at least be maintained, which I’ve outsourced in a link under “The sustainability of the dividend”, below.

GSK have large current assets and current liabilities, and changes in them which are relatively small as a percentage of current assets or current liabilities can be big in absolute terms or compared to income. The decline of short term provisions had a big effect in 2012.

GSK current assets and liabilities

The items which figure in GSK’s cash flow statement under “Changes in working capital” are –

Current assets:
Trade receivables
Other receivables
Current liabilities:
Trade payables
Other payables
Pension and other provisions

I reordered the list to match the chart above. The items correspond to the blue, brown, red and yellow areas. The other colors are for items which are too like cash or loans to be considered “working capital” (it would be more correct to call it “non-cash working capital”).

The next images show the spreadsheets that the charts in this section are based on. Everything is in £ millions exect per share amounts, dates, % etc. I found the reports by googling “GlaxoSmithKline annual report 2012” (etc., without the quotes). GSK don’t make it easy to find them, and the UK’s company information is way behind the US SEC’s EDGAR database of company filings.

The next spreadsheet image (and the formula view that follows it) is a version I used to check the original which the six cash flow and EPS charts are based on. The version for checking came out a lot neater than my first attempt, and the differences between both have been resolved.

GSK EPS core EPS and cash flow spread

GSK EPS core EPS and cash flow formulas

For convenience (if anyone wants to check the work), the next image has three cash-related screenshots from the annual report for 2014, one of which shows the movements in working capital. The screenshots don’t have all the data for 2012 to 9 months 2015, but the titles should help with finding the data in GSK’s documents.

GSK cash flow from screenshots

GSK current assets and liabilities spread

Cash conversion and investment

The figures here are all in £ millions.

I said that higher cash conversion seems achievable, as the problem is not that GSK have to make massive investments just to stand still. I show that over 2012 to 2014, capex has only taken a small (5.20%) bite out of the core EPS, above the amount covered by including depreciation. I also show that capex and purchase of intangibles at historic levels relative to core EPS will not by themselves swallow up enough cash to stop free cash flow from covering the dividend. I have to admit that the reasoning towards those conclusions is hard to follow because of the number of steps taken.

I’ve calculated cash invested divided by my adjusted net cash inflow from operations (adjusted CFO), using figures shown in the spreadsheet (without subtracting the change in working capital) –

2012 ~ 75.15%
2013 ~ 8.30%
2014 ~ 24.88%
9mth 2015 ~ -1,371.97%

Using totals for 2012 to 2014, for cash invested / adjusted CFO –
3,185 / 14,145 = 22..52%
and including 9 months 2015:
-3,277 / 14,616 = -22.42%.

Apart from acquisitions, most of the investing flow is property, plant and equipment or PPE (followed by purchase of intangibles), at a fairly steady £1.1 to £1.2 billion a year. This quote from the annual report for 2014 suggests to me there’s some scope for PPE to come down.

“In 2014, we invested £1,261 million in new and renewal property, plant and equipment. This is mainly related to a large number of projects for the renewal, improvement and expansion of facilities at various worldwide sites. Property is mainly held freehold. New investment is financed from our liquid resources. At 31 December 2014, we had contractual commitments for future capital expenditure of £459 million and operating lease commitments of £701 million. We believe that our facilities are adequate for our current needs.”

However, the PPE needs to be kept to a high standard for productive research and for the quality of production, and it can’t be neglected. The freehold property could be used to support leaseback finance if necessary (if that would not break terms in existing debt agreements).

Adjusted CFO per share / core EPS was 82.39% over 2012 to 2014 (from 263.91p / 320.30p as in the spreadsheet), so in the period, cash invested / core EPS was 22..52% * 0.8239 = 18.56%. (I’d calculated cash invested / adjusted CFO = 22..52% a while back.)

Here some of the figures are from the annual report for 2014 and are not in my spreadsheet.
Net purchase of PPE totalled 3,274 in the period, making Net purchase of PPE / cash invested = 3,274 / 3,185 = 102.8% (the result is possible because of disposal of businesses in 2013).
Net purchase of PPE / adjusted CFO = 3,274 / 14,145 = 23.15%, and
Net purchase of PPE per share / core EPS = 23.15% times 82.39% = 19.07%. (I calculated adjusted CFO per share / core EPS to be 82.39% a few lines above.)

Depreciation and Amortisation of intangible assets amounted to 2,383 and 1,960 over 2012 to 2014, a total of 4,343, with depreciation equalling 72.79% of net purchase of PPE. Depreciation plus amortisation was 4,343 / 3185 = 136.36% of cash invested. Core EPS excludes most amortisation but not depreciation. Only 27.21% of net purchase of PPE was not covered by the depreciation included in core results. Multiplying that 27.21% by the 19.07% for “Net purchase of PPE per share / core EPS” gives 5.20% for the ratio “Net purchase of PPE per share not accounted for by depreciation in core EPS” divided by core EPS. In other words, capex has only taken a small (5.20%) bite out of the core EPS (above the amount covered by including depreciation in core EPS).

That still leaves the purchase of intangibles, which sums to 1,545 in the period, which are not amortised in core EPS, and which averaged 10.65p per share per year (at the Q3 share count of 4,835 (million)). The purchase of intangibles per share also averaged 9.98% of core EPS in the period (2012 to 2014). Per share, the excess of capex over depreciation, plus the purchase of intangibles, has averaged 5.20% + 9.98% = 15.18% of core EPS. Cutting the purchase of intangibles would risk missing opportunities for growth, or could leave awkward gaps in the company’s IP, although partnering to develop IP can be an alternative to buying it.

The lower core EPS I projected for 2020 was 102.38p, and 15.18% off that is 86.84p. That would be enough to pay an 80p dividend out of (but with not much cover), given the assumed core EPS and if free cash flow was not less than core EPS minus the two items: 1) capex in excess of depreciation, which is assumed to  be 5.20% of core EPS, and 2) the purchase of intangibles, which is assumed to be 9.98% of core EPS. (It’s not usual to subtract capex from EPS, but I’m looking into the conversion of earnings to free cash flow, and the effect of capex and purchase of intangibles on that.) I don’t mean that those assumptions are likely to hold, just that capex and purchase of intangibles at historic levels relative to core EPS would not by themselves swallow up enough cash to stop free cash flow from covering the dividend. I’ll repeat that this is based on the low case assumptions for core EPS growth, and it’s the second and more pessimistic version where I assumed only 9% growth for 2016 based on GSK’s expectation – “percentage growth expected to reach double digits”.

Still some uncertainty about Q4

Here I’m quoting from GSK’s Q3 2015 transcript. About volatility in 2015 quarterly results –

“Q4 will be no exception, with the biggest quarter last year for Oncology sales dropping out, along with Avodart going generic in the US at the start of Q4 as well as the usual lumpy comparisons for vaccine sales, depending on the timing of tenders. We also expect continued growth in the minority interest, given the increasing contributions from HIV sales in the Consumer joint venture, and a much higher tax rate than in Q4 last year.”

Offsetting that there are new launch products and the benefit of cost restructuring programmes. Management are sticking to their 2015 guidance, which does not include the proposed divestment of remaining ofatumumab rights to Novartis, which could happen in Q4 or in 2016.

A question about Q4 EPS (and therefore full year 2015 EPS) was not answered with figures –

“The guidance for high-teens EPS decline this year, on a full-year basis, seems to imply that for Q4 you will have a really sharp decline, like at least 30% EPS decline. I would have thought that there would be further progress made on cost savings. Could you talk about the magnitude of the upward pressures in Q4 this year on opex, and why the guidance could not be a little better now? “


“Q3 was pretty light for Consumer, but Q4 is going to be pretty heavy for Consumer because of the cough and cold season and the shifts around, and as you know, post the transaction we have a much bigger cough and cold portfolio than we had before the transaction, so you are going to see a few movements like that.

To a general point, we are very pleased with the performance in the quarter. Frankly, we still have quite a few moving parts for Q4 as the new businesses all settle down. Simon listed some of those in his commentary.

Let’s see how the fourth quarter plays out. We just felt actually it was a little early to declare victory. This is one step at a time. We are very focussed on delivering the best number we can and let’s see what Q4 looks like.”

Seasonal sales for flu and consumer caused an increase in receivables in Q3, which is expected to reverse in Q4. The expected reversal seems to contradict the light Q3 and heavy Q4 in the quote. Even if sales are early in Q4, I wouldn’t expect a rush to pay before year end.

The sustainability of the dividend

I’ve already looked into the ability to pay at least 80p in dividends per year in 2020, based on projected core EPS and the relation between core EPS and cash flow in the past (find “Core EPS and other per share quantities”, above). The results did not look good, but there’s scope for better cash conversion and for core EPS to exceed expectations. Presumably, management are confident of the ability to pay the 80p dividend they said will be paid for 2015, 2016 and 2017.

The current ratio (current assets/current liabilities) is 1.33 (as at Q3 2015) which is reasonable, but I mention below that the Altman Z score for bankruptcy risk is in the grey zone and not very far from the distress zone.

See “No, GlaxoSmithKline Is Not Going To Cut Its Dividend” by Ben Strubel, Nov. 3, 2015 (seekingalpha.com). While being mostly reassuring, the “however” at the end acknowledges the possibility of more risk at GSK than at pharma companies doing much better. My opinion is that a dividend cut would have a big effect on the share price and investors who can’t take such a loss of capital and income should avoid the stock.

There’s a piece on fool.co.uk where the author believes the dividend can be paid as planned – “How Safe Is GlaxoSmithKline plc’s 6% Dividend Yield?” by Rupert Hargreaves, July 6, 2015. The author mentions the plan to cut costs by £3 billion a year. The savings were in the Q1 results release, just before outlining the expected EPS growth, so the EPS growth should include the savings –

“Total annual benefits of £3 billion from combination of existing restructuring and synergy programmes, now expected to be largely delivered by end of 2017 within existing cost estimates but with an accelerated rate of expenditure”

GSK’s Q3 transcript has –

“Then, most obviously in our cost base where we are on track or slightly ahead of our plans and have delivered total incremental savings in the first nine months of this year of over £700 million compared to the same period last year, £300 million of that in the third quarter alone.”

The articles I linked to don’t consider the possible problem with the Novartis “put” option, where GSK might have to reserve more if growth in the value of the business is faster than expected. For more on that find – The Novartis “put” option – below. It’s a long section and I don’t write about growth until near the end.

GSK’s Beneish M-Score

According to gurufocus the M-Score signals that GSK are a manipulator (on gurufocus you have to click away a sign-in box before you see anything). The score is the highest (worst) in the last 20 quarters. I don’t have a lot of confidence in the M-Score, except that the breakdown gives a list of things to check. The Novartis transactions are the kind of thing that can affect the score quite innocently, but GSK’s score turned up in the latest quarter, as well as in the previous two. Part of the score is the Days Sales in Receivables Index (DSRI) and for GSK it implies a 16% rise over the previous year. In the Q3 results announcement GSK say –

“Working capital increased by £707 million in the quarter, primarily reflecting an increase in receivables in the quarter related to seasonal sales of vaccines and other products. This increased the working capital conversion cycle by one day. In the nine months, working capital was also significantly impacted by the inclusion of inventory acquired with the former Novartis Vaccines business. The increase was partly offset by favourable exchange effects.”

The Working capital conversion cycle is the same 216 days as at Sept 30, 2015 as it was on Sept 30, 2014, although it dipped in December and increased since then. However, “Working capital percentage of turnover” increased like this (the latest quarter on the right) –

24%, 22%, 24%, 25%, 27%

I suppose the latest increase was covered by GSK’s statement about Novartis Vaccines. Working capital affects the M-Score through an index called TATA.

It’s a quirk of the M-Score that companies are viewed with more suspicion if they reduce leverage (the opposite of what Beneish expected, and few people have noticed). GSK’s reduction of leverage gets them a low LVGI index, and multiplying by the coefficient of -0.327 gives 0.2392. Taking that off the M-Score gives -1.9392, which is still above the commonly used threshold of -2.22 (below which management are supposed to be honest).

My guess is that while management will want to give a positive impression, they’ll be reluctant to stray into earnings-manipulation territory. On the ethics front, GSK have good data transparency and make virtually no profit on vaccines sold to poor countries, but have been fined for bribery. A scandal over earnings-manipulation would confirm impressions that the company is corrupt, while the positive points would probably be ignored by the media. I expect management to be aware of that. I’ve never met the management and they might all be extremely honest people, but I feel I don’t serve readers by simply assuming a company’s management are more honest than the average for management.

Also on gurufocus, GSK’s Altman Z score (for bankruptcy risk) puts it in the grey zone, and closer to the distress zone than the safe zone. The current score is the lowest (worst) by far in the last 20 quarters.

For comparison, Pfizer’s M-Score suggests they are not a manipulator. Their Altman Z score is in the grey zone, but about in the middle, which beats GSK’s score.

Currency risk

The combination of head office costs in the UK and high international sales contributes to GSK’s exposure to the British currency. I haven’t found a breakdown of costs by region but I can quote figures under “Employees by region” in the 2014 annual report –

USA 17%
Europe 39%
Emerging Markets 38%
Japan 4%
Other 2%

The figures will have changed with the Novartis transactions, and –

“We also announced plans to change our geographical R&D footprint by bringing our significant R&D operations together into two global centres – one in Philadelphia in the USA and the other in the Stevenage area of the UK.”

See also a piece about planned job cuts under “Negative articles” below.

GSK segments 9 mth 2015

About the Novartis vaccines now acquired by GSK, the Annual report for 2014 has –

“The combined business will also benefit from increased exposure in key markets such as the USA where Novartis has a strong presence and track record of regulatory approvals.”

and the proportion of vaccine sales in the US increased to 44.5% in Q3 2015, although vaccine sales are lumpy so the figure is not a reliable guide to the future.

Derivatives are used to manage risk, but for currencies, “Foreign currency transaction exposures arising on internal and external trade flows are not generally hedged”, instead they match local currency income with local currency costs where possible. However, this looks like hedging –

“During 2014, the Group continued entering into forward foreign exchange contracts which it designated as cash flow hedges of its foreign exchange exposure arising on Euro and US dollar denominated coupon payments relating to the Group’s European and US medium term notes. This is a continuation of the initial hedging put in place in 2013.”

About borrowings:

“we seek to denominate borrowings in the currencies of our principal assets and cash flows. These are primarily denominated in US dollars, Euros and Sterling. Certain borrowings can be swapped into other
currencies as required.”

When GSK aim to preserve earnings that’s likely to be in the British currency, not the US dollar, but the “matching” principle mitigates the company’s currency exposure as much for a US investor as for a UK investor.

I’ve shortened a table that described GSK’s foreign exchange sensitivity. It’s now only a rough guide, after the Novartis deal –

Change in income, £m

2014 ~ 2013

10 cent appreciation of the US dollar
(263) ~ 40

10 cent appreciation of the Euro
11 ~ 8

10 yen appreciation of the Yen
– ~ 1

About not speculating –

“We do not hold or issue derivatives for speculative purposes. Our Treasury policies specifically prohibit such activity.” (2014 Annual report)

HIV success

“Dolutegravir-based regimens have done so well, because a) it is an excellent molecule and b) the team did the most phenomenal job of coming to market not just with one or two pivotal studies showing base competitiveness. It came to the marketplace with a full dataset comparing itself to most of the other regimens and demonstrated extraordinary performance. It is very rare that I have seen a product which essentially hit its mark in pretty much every trial it did in every class it was put up against. That is what has really made this product cut through into what is obviously a competitive market dominated by one company, and it is where we have been able to take very significant market shares very quickly and deliver, as I have said already, the most successful product launch in the category in the US.”

The “one company” is Gilead. Dolutegravir is in the ViiV joint venture, which some critics say should be spun out. For some confirmation, though without as much enthusiasm, see “GSK takes on Gilead with new two-drug HIV treatments

While justifying the formation of the two joint ventures, the CEO acknowledged the effect of competition from Gilead. About forming ViiV –

“It was that move which gave the company and, frankly, the external stakeholders, the confidence that GSK was going to stay in the HIV market in the face of the onslaught from Gilead. That has allowed us to get through the desert, get to the other side, and now build something very, very exciting.” (Financial outlook and guidance, May 6, 2015, transcript)

With increasing longevity in patients with HIV, the focus is on reducing the toxicity of medications. Gilead’s “TAF” has completed Phase 3 studies (find “For more than a decade” to get past the technical stuff). The key benefit of TAF is a reduced adverse effect on bone density and kidneys. Gilead describe their progress in HIV in “Gilead Sciences’ (GILD) Presents at Piper Jaffray Healthcare Conference Transcript” Dec 1, 2015 (seekingalpha.com).

Glaxo’s chief strategy officer can see the day when there’s no more profit in HIV – “Success at Glaxo’s HIV Unit May Mean Having to Call It Quits” by Oliver Staley, July 9, 2015 (bloomberg.com). That’s great for HIV patients but not so good for Gilead and GSK, although the price per pill would be higher for a cure, and the research will be directed elsewhere when the potential for improvement becomes marginal.

The SUMMIT study and other negative results

GSK’s SUMMIT study was big, with 16,590 patients. It was meant to prove that Breo gave a survival advantage for severe COPD (chronic obstructive pulmonary disease), but results did not show that with statistical significance. Success had not been assumed or incorporated into forecasts, which indicates that the forecasts are conservative, with upside –

“I would just like to reiterate that SUMMIT was never in any of our forecasts that we shared with the street. So far, for example, when we said that we would deliver at least £6 billion of new product sales by 2020, that did not include SUMMIT – that was always an upside to those numbers.” (Q3 results transcript)

GSK said the failure had no adverse effect on Breo sales, and claimed useful results –

“Some of the outcome studies, as you say, have failed. Others, for example SUMMIT which missed its primary endpoint, contained within it an awful lot of incredibly useful information which the scientific community are picking up on: reduction in exacerbations, the ability to show that the safety signals are rather reassuring and in fact the decrease in the decline of lung function in patients on the treatment there.” (R&D event transcript)

There’s a less rosy view in “Glaxo Shoots Itself In The Foot… Again” by EP Vantage, Sep. 9, 2015 (seekingalpha.com), where it’s implied that with no survival advantage over either of Breo’s two components, prospects were poor for Breo, and also for turning around GSK’s respiratory division.

About failed outcome studies –

“I don’t think the failures we have had are molecule quality failures. There is no doubt where we have had failures, there are issues around targets and that is why there is such a big investment we have put in target sciences.” (R&D event transcript)

Either the failed drugs didn’t latch on to the targets or poor targets were selected. It doesn’t sound great but this is science – if you knew the result for sure you wouldn’t need to run the experiment. GSK say –

“We absolutely have introduced new designs which include interim analyses, utility analyses and ability to get early read-outs, and I gave you an example for a smaller study here in respiratory with a real in-stream data analysis. We think we’ve got actually a rather sophisticated way of looking at outcome studies.” (R&D event transcript)

GSK has two failures in “Top 10 Clinical Trial Failures of 2014” Feb 9, 2015 (genengnews.com). So do Merck. Only drugs that did not meet their primary endpoints were considered. More recently there’s  “GlaxoSmithKline’s cardio R&D group records another big PhIII flop” by John Carroll, October 27, 2015 (fiercebiotech.com). The author is critical of the company’s cardio group, implying they push projects along the pipeline that ought to be killed.

Previously – (the link has irrelevant video that starts automaticaly) “Glaxo’s Utter Stinkbomb Of A Drug” by Matthew Herper, Forbes Staff, Mar 31, 2014 (forbes.com). Cardio drug darapladib failed, and made some patients smell. The author criticizes GSK for developing a drug that no-one would take even if it worked. Adding the percentages, at worst 9% of patients on the drug had a diarrhea or odor problem compared to 1.1% on the placebo. 9% is high, but if it didn’t take too long for the effect to wear off, IMO a patient could reasonably take the drug, stop if there was a problem, and stay indoors for a while. Drugs have had worse side effects. However, pursuing development was probably a mistake, though it’s easy to say that with hindsight.

Cardiology is not one of the six areas GSK say they are focused on, and there isn’t much about it in the R&D event presentation or transcript, with only one cardio drug in the selection from the pipeline (GSK2998728 TTR production inhibitor TTR Cardiomyopathy, subject to exercise of option). Cardiovascular, metabolic and urology had revenue of £965 million in 2014, 5.2% of the drug and vaccine sales (of £18,670), not counting any cardio drugs in Established products (£3,011 million). Most of the £965 million had nothing to do with cardio, as it included Avodart revenue of £805 million and Avodart is used to treat benign prostatic hyperplasia (BPH).

Recent failures and return on investment

From the R&D event transcript –

Q – “Then, finally, a question on the return on investment and, obviously, you showed us what you believe your return on investment to be, a couple of years ago. In that time darapladib and losmapimod have fallen over and Breo and Anoro are doing, perhaps, less well than we all expected, so would you expect that return on investment to fall dramatically when you restate it next year?”

A – “we will be publishing the IRR next year, as we said we would, and, no, I don’t, and I think if you look at the value that is created from the Oncology pipeline you will see that there have been some very startling successes as well, in terms of the income.”

Darapladib and losmapimod are both heart drugs, see the previous section “The SUMMIT study and other negative results” for criticism of GSK’s cardio division.

IRR is Internal rate of return. It was announced in February 2014 that the IRR on R&D was estimated to be 13%. A long term target of 14% was announced in 2010 when IRR was 10%. Those IRR figures are about ten times higher than the return on R&D given for GSK in an industry ranking in May 2014, find “22 Companies Ranked”, below. IRR is non-GAAP so regulators don’t take much interest in the figures.

Upside in the pipeline

1) The pipeline looks good, if this from the R&D event presentation is reliable –

“Of the ~40 assets profiled today, 80% of new molecules, biologicals and vaccines are potentially 1st in class”.

The figure will be lower for assets not selected for the event. Obviously, there’s a difference between “1st in class” and “potentially 1st in class”. You can’t expect “definitely 1st in class”, but they could have added something like “we believe at least 60% will be first in class” if they had enough confidence. (BTW a cynic might believe the R&D day was prompted by criticism and calls for the company to be broken up.)

2) There’s a good record of approvals from the FDA. After talking about reducing fixed R&D costs, the CEO said –

“That is why, over the last five years, nobody else has developed or filed or gained registration approval by FDA for more products than GSK, so we have been able to deliver a very substantial amount of product, while we have been able to control our R&D cost, because we have been taking out fixed cost and dedicating the resources to flexibility project cost. We will continue to do that.” (GSK’s R&D day transcript)

Later in the PDF –

“Quality: Andrew has already alluded to the fact we have had more approvals than any other company. The number isn’t what matters, the fact is that actually we have more first-cycle approvals and we have had nothing but first-cycle approvals since 2012. That is a marker of quality; you don’t get that just by doing things fast, although you will see we actually do, do things fast in terms of development organisation.”

The performance could be less good over different periods. I won’t critize the remarks for selection bias, because R&D was reorganized around 2008 to 2011, so the periods seem fairly reasonable. Ideally the “quantity” and “quality” periods would have been the same, but “nothing but first-cycle approvals” is a tough metric because it would be ruined by a single exception. About competition to the pipeline –

“we think we have an extremely competitive profile not just against the current standard of care out there on the marketplace, but also, much more importantly, we think it is pretty future-proofed about anything from anything that is coming down the pipe.” (the Q3 results transcript)

The last statement is an opinion and its usefulness depends on how far you trust the management.

3) Some of the pipeline is large molecule or biological, which might be harder for generic manufacturers to copy, but I can’t be more specific than “some” and “might”.

4) Nucala has been approved by the FDA, after the R&D day. It’s for severe asthma, in combination with other drugs. The approval was more broad than expected, according to “GlaxoSmithKline plc (ADR): Why Nucala’s FDA Approval Is Significant“. Also, “GSK receives European marketing authorisation for Nucala® (mepolizumab) in 31 countries” December 2, 2015 (gsk.com).

5) Upside from the pharma pipeline is underlined by the projected growth in pharma trailing vaccines and consumer. From the Q1 2015 results release –

“Group revenue expected to grow at a CAGR of low-to-mid single digits over the five year period 2016-2020 on CER basis:

– Vaccines sales expected to grow at a CAGR of mid-to-high single digits

– Pharmaceuticals sales expected to grow at a CAGR of low single digits with the possible introduction of generic Advair in the US factored into this assessment

– Consumer Healthcare sales expected to grow at a CAGR of mid single digits”

The low growth of pharma could combine optimistic assumptions about the pipeline, with the decline of off-patent drugs, but that’s limited by some impact already for Advair and Lovaza, and the statement about Advair – “I do not believe for a second that that erosion will go to zero” (Q3 results transcript).

Hepatitis C

The R&D event transcript has this about Hepatitis C and HCV (the Hepatitis C virus) –

“HCV, of course – there are cures, and here I will just focus on one thing when I get to it, which is the way of thinking about the single administration cure for HCV.”

The problem here is that Gilead (GILD) already have good cures. Even if the single dose is attainable, that’s not likely to be enough to compensate for getting late to the party, when Gilead have cured many cases, leaving fewer for GSK and probably establishing a record of safety and efficacy. I can’t say that GSK should definitely have dropped their HCV NS5B inhibitor. I don’t want to be a “back seat driver” when I don’t have all the information or expertise. Maybe most of the cost is sunk and it’s worth spending a little more even if the payoff is not all that big (they can’t tell shareholders that the HCV product won’t be good enough for high sales without jeopardizing sales or negotiations). Other companies including Gilead and Merck are still developing HCV products, and while that represents competition it also suggests it isn’t entirely “game over”.

Searching the transcript for “cure”, I found no other cases of pre-existing cures. Because the cures for HCV were acknowledged, for the assets profiled at the R&D day it’s fairly likely that other cures would have been acknowledged if GSK was competing with them, at least if they were as good as Gilead’s HCV cures.

Asthma in Africa

This is relevant to Nucala which was approved recently. Nucala is for severe asthma, in combination with other drugs. I don’t like saying that disease in Africa is positive for GSK, but if any investors believe that asthma is only a “rich world” disease they need to read this – “South Africa has world’s fourth highest asthma death rate” Sept 11, 2012 (health24.com). Pharma products are typically sold at much lower prices in poor countries.

Negative articles

I dispute some of the negativity, under “Reasons to like the small team structure”, below.

This is old – “3 Things to Loathe About GlaxoSmithKline” by G A Chester, June 21, 2013 (fool.com). The ten year total return was under 6% and lagged most European pharma companies. With the share price now lower than when the article was published, the picture is not likely to have changed.

The next link covers corruption, an issue I haven’t seen mentioned recently – “GlaxoSmithKline: The Bad Isn’t All That Bad” by stockpucker, Aug. 20, 2014 (seekingalpha.com).

UPDATED: GlaxoSmithKline rips into RTP R&D group, chopping 900-plus jobs in global reorganization” by John Carroll, December 3, 2014 (fiercebiotech.com). The piece gives space to GSK’s claims about sharpening focus etc., before blaming poor performance on the reorganization of R&D into small teams. While 2013 was a good year for the number of FDA approvals, they were for drugs with low commercial potential. It’s implied that researchers were expected to behave like entrepreneurs, an unreasonable expectation for most scientists. IMO someone has to be able to assess both the science and the commercial prospects, for the efficient allocation of funds, without which income won’t grow, and ultimately that would also be bad for the R&D spend and drug discovery.

Being too keen to punish failure can reduce collaboration (as in Microsoft’s organizational chart). Some scientists are entrepreneurial and are likely to prefer the small team structure. The article “GSK’s Biotechy World” by Derek Lowe, July 1, 2010 (blogs.sciencemag.org) describes the problem – the huge pharma companies that resulted from consolidation had R&D strangled by bureaucracy, with low return on the cost.

There’s nothing new about GSK wanting to be nimble like biotech – that was the motivation claimed for splitting R&D into six units in 2001. See “GlaxoSmithKline pushes its labs towards ‘biotech’ future” – you are likely to see something about “access” and only see one paragraph, but it’s enough to confirm the story. The logic seems to be that if splitting into six didn’t achieve the objective, splitting into smaller units (and initially, adding job insecurity) would do the job. While the previous failure is not reassuring, bureaucracy is a big problem for big companies, and at least GSK have been trying to do something about it.

The author of “Acronym-Fest: GSK and Its DPUs” by Derek Lowe, December 2, 2011 (blogs.sciencemag.org) has deja-vu about claims for the new model such as being nimble. It’s a short article with many comments, some from ex-Glaxo/Wellcome scientists who say research went downhill after merging with SmithKline Beecham in 2000. Other comments are about the failure of similar models in other companies. Researchers from various companies bemoan the many visions they had to try to adapt to, such as labs should be like factories.

Back in May 2014, GSK were in the bottom quarter in “Who’s The Best In Drug Research? 22 Companies Ranked” by Matthew Herper, Forbes Staff, May 22, 2014 (forbes.com). The score used is a rather odd combination of economic return and other metrics like patents per million dollars of R&D, with the top company having a low economic return (but higher than GSK’s). Approvals from the FDA are not included, the reason given is they are too random. The charts showing a long term decline in return on R&D might put you off investing in the industry.

The R&D teams

The R&D teams are called “Discovery Performance Units” (DPUs). They are interdisciplinary, it’s claimed they are given sufficient time to produce results, and they seem to be left to get on with their work without too much interference, although there are annual reviews of R&D which in theory should weed out any projects that have become non-viable. From the R&D event transcript, after being asked why peers didn’t use the DPU model –

“DPU model: well actually the DPU model is really pretty straightforward. It is about saying “Have you got leaders in the organisation that you trust to try and get on and make medicines in interdisciplinary groups?” I think if you look at what many companies have done they have pretty much gone down that sort of road one way or another; not fully, not always calling it a “DPU model”. I don’t think it is as unusual now as it was when we first introduced it and it is based on a pretty simple idea, which is you need the disciplines inter-related. They need to be able to talk to each other on a daily basis, you need a leader who can lead that. It is rather like the biotech model in that respect and you need to give people time to pursue their areas. Perhaps the best example that you are going to see today is the unit that John Bertin leads, which has been absolutely relentlessly focussed on innate immunity and pattern recognition receptors for the last six or seven years and has become world-leading.”

For more on the annual review of R&D, find “prioritisation and stopping things” in the next section.

This is a balanced piece about GSK’s DPUs, even though the title is a bit dramatic – “Glaxo Scientists ‘Live or Die’ With Project in Research Overhaul” by Albertina Torsoli, November 30, 2011 (bloomberg.com).

It seems difficult to devise a structure or system that gets the right balance between science and profit, bearing in mind that a relentless focus on profit could kill the science without which there’d be no profit. Teams should probably be rewarded for finding problems like toxicity early, but that could be hard to implement in practice. There’s a risk of teams becoming uncooperative silos, and a risk that teams will be rewarded for their ability to “sell” their work to reviewers. A lot depends on the project reviews being fair, with competence in both business and science needed. One problem with the small biotechs GSK are trying to imitate is that by over promising they can keep good salaries for staff and management going for years.

If the R&D structure proves to be fundamentally flawed as implied in some of the negative articles, one possibility is a further reorganization, which could either work or make things worse. If the pipeline underperformed by enough to threaten the dividend, the stock would fall, but with some chance of GSK being acquired. I see those possibilities as relevant to considering the long term downside, but I can’t put probabilities to them.

The R&D budget and job cuts

Asked about the possibility of cutting R&D after losing the oncology business and gaining OTC (consumer) and vaccines, the president of pharmaceuticals R&D said no, because jobs had already been cut while restructuring R&D. From the R&D event transcript –

“In terms of the R&D budget, clearly, as Andrew has said, we have actually undergone some quite substantial reduction in headcount, we have done restructuring to end up with the right sites, with the right footprint, and that has freed up money to put into projects and I think that we have got a budget which allows us to deliver what we have described here, we have got a budget which allows us to drive these things forward fast. I will remind you that we took BRAF from powder on the bench to approval in under five years, we took dolutegravir from powder on the bench to approval in six years, we have done the same with actually Anoro for inhaled. So I think we have got a budget which is designed to allow us to deliver what we have got there, I don’t think we are in the stage where we are going to need an uplift unless all of this turns out to work that I am going to talk about today and I don’t expect that all of it is going to work and I don’t see opportunities for reducing that. I think that would be a way of actually diminishing the chance of getting the returns that we’ve got from these medicines.

And the final point to say is we every year, and this year particularly knowing the excitement that we’ve got in the pipeline, really go through a process of prioritisation and stopping things and one of the key things in R&D is knowing what to stop and making sure that you don’t spend money where you don’t need to.”

Keyur Parekh from Goldman Sachs made the point –

“Glaxo has more R&D sites than Roche for a budget that is 50% lower than Roche.”

In reply Patrick Vallance, the President of Pharmaceuticals R&D, said that the rationalization to two major global sites was addressing that. As well as recent closures,

“We also closed a major site in the UK about seven years ago”

In answer to –

“your spend on facilities and central support functions is now about 15% of total R&D, up from 11% in 2012.”

the R&D spend varied with opportunities and especially with the number of Phase 3 studies. I’m not sure about that reply, at least not if the 15% “now” included the non-central cost of the failed SUMMIT study, with 16,590 patients. While a higher proportion of spending on facilities and central support is contrary to GSK’s aim of reducing fixed costs, scientists need good labs to be productive.

It seems to be assumed that fewer, bigger sites are better for R&D. In academia a UK study concluded the most productive lab size was 10 to 15 members – “Lab Size Study Stirs Debate” by Bob Grant, February 9, 2015 (the-scientist.com). About Pfizer’s huge lab in Connecticut – “Mergers don’t make good medicines” by Scott Gottlieb, May 9, 2014 (independent.co.uk). The article claims the failed megalab is now mostly mothballed.

I have a theory (which I don’t wholly believe). Change such as a reorganization or a merger disrupts research. Early stage work becomes subject to review, and scientists who just want to get on with discovery are frustrated and some of them leave. The disruption leads to less discovery for a long time. Before there’s enough time for scientists to settle back into productive research, the lack of commercial success prompts more disruption, in the form of mergers, consolidation of research sites, or reorganization, and the cycle repeats. The first half of that is more or less from the ‘Mergers don’t make good medicines’ article, and the second half depends on management being reactive rather than doing nothing, which seems fairly likely. I don’t mean that hopeless projects should continue, but productive teams should be left to get on with their work with the minimum of disruption.

Reasons to like the small team structure

Against the claim of poor performance from GSK’s drug R&D, there have since been good results for HIV, in the ViiV joint venture. The previous quote “the team did the most phenomenal job of coming to market” suggests GSK’s usual team structure applied to the joint venture, and about ViiV and consumer healthcare –

“In the case of both of these JVs, we made it super-clear that GSK is operationally in control of everything to do with these businesses on a day-to-day basis.” (Financial outlook and guidance, May 6, 2015, transcript)

and I conclude that ViiV’s success is evidence that the team structure is at least compatible with success.

The reviews on glassdoor.co.uk (later) are fairly good, suggesting that GSK’s scientists are not demoralized by the reorganization.

See also “Recent failures and return on investment” (above), where oncology is cited as a successful area, and there’s enough success to compensate for the failures to ensure that the return on R&D will not have to be revised down.

Allegations about the drugs industry

There have been a few books on this theme. On Amazon,  with reviews:

The Truth About the Drug Companies: How They Deceive Us and What to Do About It” by Marcia Angell, August 9, 2005. An anonymous reader who claims to be a Vice President in one of the biggest drug companies, gives the book a five star rating.

Bad Pharma: How Drug Companies Mislead Doctors and Harm Patients” by Ben Goldacre, February 5, 2013
Possibly too UK-centric for US readers.

Big Pharma: Exposing the Global Healthcare Agenda Paperback” by Jacky Law, February 6, 2006

A common theme is the way companies spent a lot on lobbyists and funnelled money to doctors and journals to promote drugs. It’s alleged that more was spent on marketing than on R&D, and the marketing cost was included in the high cost of developing a successful drug.

An honest model

GSK stopped paying doctors to speak for them, and stopped paying sales reps for hitting sales targets. From GSK’s transcript for the Investor Event, May 6, 2015 –

“The changes we have made in our healthcare practitioner payment model, where we are the first company to walk away from paying doctors to speak on our behalf, where we have already stopped paying any representative anywhere in the world bonuses associated with short-term sales, are all designed to establish a commercialisation model which is in step with the external stakeholders of this industry: the payors, the governments, the regulators. Make no mistake, the business model which has historically been prosecuted by this industry, the commercialisation model, is not in step with regulator, government or payor expectations. We made the call two years ago to start that journey, and we reiterate today that we see this not just as something we have to; it is something we want to do. It is something that we know is difficult, we know is challenging, but we believe that it is potentially an extraordinary defining competitive advantage if we are able to succeed. The early progress we are making is very reassuring and convinces us that we are on the right track.”

I can’t help liking that, but there could be some risk as well as reward in being the first mover, rather than waiting to see how it works out for a peer. About not paying sales people for hitting sales targets –

“We altered the way we reward our customer-facing pharmaceutical and vaccine sales professionals in 2011.

This new system rewards sales professionals for the quality of their interactions with healthcare professionals, including an element of customer evaluation, rather than for achieving individual sales targets.

This approach aligns with our core value of putting the interests of patients first and the goal we share with healthcare providers of improving patient health.” (Annual report for 2011)

One abuse I’ve heard of is the funding of patient advocacy groups who demand that the patron company’s drugs are supplied by the healthcare system. It’s natural for patients to want the best drugs, regardless of the expense to the state or insurers, so patients are natural allies of pharma companies with the best drugs. I expect some patient groups do a lot of good work, whether or not they are demanding wider prescribing for some drugs. I’ve found nothing about GSK funding patient groups. This alleges a different abuse of patient groups – “Big pharma mobilising patients in battle over drugs trials data” by Ian Sample, July 2013 (theguardian.com).

Woodford, break-up, Berkshire and insiders

Neil Woodford is a highly regarded UK fund manager who has invested fairly heavily in GSK, see “Neil Woodford: GlaxoSmithKline’s troubles are only temporary” by Denise Roland, August 5, 2014 (telegraph.co.uk). Since then he’s called for the company to be broken up – “City Star Woodford Urges £67bn Glaxo Break-Up” October 24, 2015 (news.sky.com). If a trader called for GSK to be broken up, it would not be surprising, but Woodford’s record suggests he would be patient so long as he remained confident about long term success. He might genuinely believe that GSK’s parts would perform better as independent businesses, or he might now have doubts about GSK’s long term success and see a break up as providing a full or partial exit. Woodford still liked GSK after meeting the management, which is a point in favor of being able to trust management, as Woodford would not have got where he is if he was easily fooled.

In – “Is Neil Woodford Right About GlaxoSmithKline plc & HSBC Holdings plc?” by Dave Sullivan – Tuesday, 1 December, 2015 (fool.co.uk) – skip to “Breaking up, or spinning off?” – one point made is that Woodford believes the CEO is managing too many business, so “consumer” should be spun off. I don’t believe having multiple businesses is necessarily a problem, if competent managers with integrity can be found. Warren Buffett famously headed many disparate businesses where control was decentralized to trusted managers, and Berkshire Hathaway head office only had 24 employees according to “Here Are All Of The People Warren Buffett Works With At Berkshire Hathaway’s HQ” by Sam Ro, March 1, 2014 (businessinsider.com). GSK only need a trustworthy manager at Consumer to avoid the JV being much of a distraction to the CEO.

If it’s true that Woodford said GSK has much unlocked value, it would be hard for him to unload without harming his reputation, unless news makes a change of mind credible.

The fact that with the Novartis deal, GSK reduced its exposure to prescription drugs, implies low confidence in the prospects for them, or that prospects might be good but are uncertain due to pressure on pricing, particularly from buyers in the US. If the prospects are uncertain, GSK did the right thing, and if GSK got it wrong, they should at least benefit from the drugs they retained.

Berkshire Hathaway sold all their GSK stock in the fourth quarter of 2013. It’s reported on insidermonkey.com (the page might jump a lot while loading).

This broker’s page is the only convenient source I found for “Director deals” (the UK version of insider transactions). A large sale and purchase almost cancel out. Those are by Abbas Hussain, President of Global Pharmaceuticals. The purchase shows he owns over £1.9 million worth (about $2.9 million USD). Patrick Vallance, the President of Pharmaceuticals R&D, also bought in September, though in a much smaller quantity, see the link for more detail. There could easily be a lot more insider ownership that I did not find.

The consumer business

The main brands are –

Pain relief – Voltaren, Panadol, Excedrin, Fenbid
Respiratory – Theraflu, Otrivin, Flonase, Contac, Smokers’ Health
Oral Health – Sensodyne, Polident, Parodontx, Biotene, Aquafresh
Nutrition/Gastro intestinal – Horlicks, Eno, Tums
Skin Health – Physiogel, Abreva, Zovirax, Lamisil, Fenistil

From “GSK announces major 3-part transaction with Novartis …” April 24, 2014 (us.gsk.com) –

“Following completion of the transaction, GSK will be a global leader in Consumer Healthcare with revenues of £6.5 billion, on a 2013 pro forma basis. The new business will hold category leading positions and brands in Wellness, Oral health, Nutrition and Skin health, combining OTC and FMCG capabilities and expertise.”

(OTC – Over The Counter medicines. FMCG – Fast Moving Consumer Goods)

“In Wellness, the new combination’s £3.4 billion complementary portfolio will create the world’s largest OTC business with the leading position in more than 35 countries around the world.”

I’ll repeat that the consumer healthcare business combines GSK’s strength in emerging markets with Novartis’s strength in Central and Eastern Europe and CIS. The Q3 2015 pre-announcement aide memoire gives Consumer revenue of £6.1 billion, 12 month pro forma 2014 (£bn at actual rates).

Having both prescription and consumer products means GSK can sometimes switch a product from prescription to consumer (“Rx/Cx switch opportunities”, in the jargon) –

“This is a proven capability for GSK, it is one that we are investing in and we expect and plan to target a switch every five years” (Financial outlook and guidance, May 6, 2015, transcript)

I’m not certain that when a prescription product is switched into OTC, ownership is transferred to the joint venture. If that’s the case, there would need to be compensation from Novartis or they’d be getting a free ride via their minority interest. Because GSK manage the “day to day” business of the joint venture, there’s no need for the set of products managed by the same team to be the same as the set of products owned by the joint venture. It’s likely that Novartis could also switch prescription products to OTC.

“A wonderful example of this and the incrementality it can bring for us and our retailers is Flonase. Flonase was a bestselling prescription drug in North America, a clinically proven superior product that treats not just one allergic response but six. We could not have switched this product without our Pharma colleagues, because we were tapping into 40 clinical studies and submitted 600,000 pages of data to the FDA. We launched it with a military FMCG style approach, 23 miles of shelf space and nearly a million assets at point of sale, and I am really pleased to say that we were able to achieve, just in a couple of months, over 11% market share. We are the top one and two SKUs in Health and Beauty across the US at the moment. We are 0.3% off being the number one ‘recommended by doctor’ allergy product and, most excitingly for us and the retailers, 70% or our source of volume is from people who are new to the OTC category.” (Financial outlook and guidance, May 6, 2015, transcript)

Starting with an 11% operating margin, GSK aim for at least 20% by 2020, which would put them in the top quartile of FMCG and OTC players. From the same transcript, about savings in two areas –

“In non-working Media simply by adopting best practices around origination and global asset management I think we will get between 15% and 30% savings on production. In Market Research, simply taking out the duplication in similar categories, modernising and digitising and focusing everything behind those Power Priority Brands will also release savings.”

To explain the jargon, a  TV ad is “working” because it reaches the target audience, while measuring the effect of it is “non-working” (see “Non-working marketing spend – A story a CFO should read” by Andrew Challier on September 4, 2012 (blog.ebiquity.com)).

“Over time, whilst our priority in the immediate term has been to stabilise the supply chain, there are obvious network consolidation opportunities, whether that be within our own sites or within the very large number of third party manufacturers we are still using.”

About market research –

“We will also be making sure we capitalise on our Shopper Science Lab. This is a really state-of-the-art facility that allows us to research these innovation products in both simulated digital and real retail environments with our retail partners, to test packaging, claims and our shopper materials, just as we have done with the launch of Sensodyne mouthwash through the first quarter.” (Financial outlook and guidance, May 6, 2015, transcript)

About making the business agile, the CEO added –

“I am going to use the opportunity not just to save costs but to really improve agility, by removing management layers and reducing interfaces, traditionally in complex, matrixed organisations, between regions, areas and local markets.”

I like the stated intention. The cases I know of where bureacracy and corporate silos were a clear problem are from the tech industry, but change isn’t slowing down and I expect the need for agility to be general.

An old (2013) downloadable report from Accenture claims the consumer healthcare industry faces change that will challenge traditional suppliers of OTC,  vitamins & dietary supplements (VDS) and herbal products. New growth areas including fortified foods will drive growth in the industry, as consumers take control of their health (and body image, I’d say) and aim to prevent disease. GSK seem likely to miss out on that. They are busy with efficiency and extracting synergies. They are firmly entrenched in OTC. The debt, dividend and big minority interest make extra investment in consumer healthcare difficult. There’s a case for saying GSK should stick to what they know in “consumer”, which is science-based OTC products.

The Novartis “put” option

The deal where GSK gained Novartis Consumer products is complicated by Novartis retaining a minority interest, with a “put option” that could be exercised from 2018 to 2035, forcing GSK to buy them out. The detail is in the link above, “GSK announces”, but the next quote is from the Q1 2015 results announcement, under “Statement of changes in equity” –

“In certain circumstances, Novartis has the right to require GSK to acquire its 36.5% shareholding in the Consumer Healthcare joint venture at a market-based valuation. This right is exercisable in certain windows from 2018 to 2035 and may be exercised either in respect of Novartis’ entire shareholding or in up to four instalments. If exercised, GSK would not be able to avoid this obligation, and so has recognised a financial liability of £6,204 million in Other non-current liabilities. This represents the present value of the estimated amount payable by GSK in the event of full exercise of the right by Novartis.”

Everyone calls the option a “put” option, but it’s at estimated market value so it’s not like an option to “put” at a fixed price. The time value is different to the usual traded options, and if Novartis think the value of the business will go down, they have an incentive to exercise the “put” before it happens.

The term “present value” implies that a discount rate has been applied to the estimated future liability. If the discount rate is 10% (for example), then if the value of the business grows as expected, the amount reserved for the “put” option has to increase by 10% per year. GSK call that “the unwinding of the discounting effects” (it’s in a quote further down).

GSK also has obligations relating to the ViiV Healthcare joint venture (for HIV drugs). Following the previous quote –

“In certain circumstances, the other shareholders in ViiV Healthcare, Pfizer (11.7%) and Shionogi (10%) may require GSK to acquire their shareholdings at a market based valuation. Pfizer may request an IPO at any time and if either GSK does not consent to such IPO or an offering is not completed within nine months, Pfizer could require GSK to acquire its shareholding. Shionogi may also request GSK to acquire its shareholding in ViiV in certain circumstances and limited windows in 2017, 2020 and 2022.”

Consenting to an IPO is an alternative to buying out Pfizer. It looks like GSK might not have that alternative for Shionogi, because of this from the Q3 2015 results (my bold) –

“Total EPS was 181.7p, compared with 35.8p in 2014, the increase primarily reflecting the profits on disposal of the Oncology business and the Aspen Pharmacare shares, partly offset by the increase in the liability for the contingent consideration for the acquisition of the former Shionogi-ViiV Healthcare joint venture and increased major restructuring expenditure.”

The liability is likely to be included in the “Other non-current liabilities” of £9,653 million, along with the £6,204 million for the Novartis “put” on the consumer business.

A shareholder who looked at the balance sheet and did not know about the liability recorded for the Novartis “put” option, ought to be pleased to learn about it (until realizing it might grow too fast). That’s because, if Novartis exercised the option for the entire minority holding, for the same amount as booked for the liability, GSK would have £6.2 billion less cash, while the £6.2 billion liability would disappear from “Other non-current liabilities”, and Novartis’s minority interest would be cancelled. The first two of those effectively cancel out on the balance sheet, leaving the third item as a gain, which I describe:

On the balance sheet, Shareholders’ equity would go up as a result of Non-controlling interests going down. For Q3 2015, the figure for non-controlling interests was £3,900 million. The non-controlling interests were valued at £3,900 million on the balance sheet, while the £6,204 million in Other non-current liabilities for the “put” option is presumably bigger because it allows for some growth in the “market-based valuation” which would be applied to the “put” option. There’s also the effect in the income statement, where less “Profit/(loss) attributable to non-controlling interests” implies more “Profit attributable to shareholders”, but this is hard to pin down, because the share given to non-controlling interests was less for the latest nine months than for the latest quarter, while profit for the latest quarter was only 7.4% of the profit for the latest nine months (you’d normally expect about 33% plus some seasonal variation). There’s a simple conclusion from all these figures –

(£ millions)
Q3 2015 ~ 9 months 2015

647 ~ 8,800

Profit/(loss) attributable to non-controlling interests
109 ~ 24

Profit attributable to shareholders
538 ~ 8,776

This is how Total comprehensive income splits –

Total comprehensive income
(87) ~ 7,617

Attributable to shareholders
(201) ~ 7,593

Non-controlling interests
114 ~ 23

For “Core results”:

Profit attributable to non-controlling interests
141 ~ 331

From “Reconciliation of free cash flow and adjusted free cash flow”:

Q3 2015 ~ 9 months 2015

Distributions to non-controlling interests
(24) ~ (234)

The takeaway from all that is, only the “Core results” figures look reliable for estimating the benefit to income from having to buy out Novartis. The nine month results only include seven months after the Novartis transactions, and it looks right for the Q3 figure to be more than a third of the nine month figure (it’s 42.6% of it). £141 million could serve as a rough estimate of the core profit per quarter that would accrue to shareholders if Novartis were bought out, but there are ‘moving parts’ that create variability – the benefit of further cost cuts, some growth (hopefully), there’s seasonality with cold and flu products selling more in winter, and non-controlling interests include the ViiV joint venture, not just Novartis.

Whatever the benefit, the point I made earlier is that it’s not in the accounts, while the obligation for the “put” and the cash held for it just cancel each other out on the balance sheet (so far as Shareholders’ equity is concerned). Either the benefit will be realised eventually, or the Novartis “put” option will expire (after 2035), leaving GSK with reduced liabilities while the assets held against them are freed up. Until then, the liability can be regarded as partly similar to a short-tail insurance float, in the sense that it can be deployed, but only in low-risk assets where the value can be realized quickly, which limits the return, but the cost of capital (in this case) is essentially zero. Longer term investments can be made with the “float” so long as they fit with the windows during which the “put”can be exercised.

About the “put” option, GSK say –

“bringing in that minority will be significantly credit enhancing of itself, given the profitability and cash generation capability of that company.” (Q3 results transcript)

There’s a complication I’ve not described yet. Presumably, the terms of the “put” option protect GSK from having to pay out more than they can handle. Otherwise, suppose sales, profits and cashflow all doubled in a few years, and the market value also doubled. Depending on how much growth is built in to the liability, GSK would be liable for up to twice £6.2 billion = £12.4 billion, and the joy of success would be tempered by the problem of how to raise enough capital in case the “put” option was excercised. While management must surely have insisted on terms to mitigate that kind of problem, it’s still likely that the liability will increase as the acquired business grows and the discount rate unwinds, and (without knowing the terms), Total comprehensive income could conceivably fall as a result of success, with some strain on the balance sheet.

Another way to describe the possible problem is to assume a fixed PE of 10 for the business, and calculate the effect of £10 million more earnings. 63.5% of the extra earnings accrues to shareholders, and 63.5% of £10 million is £6.35 million. The earnings accruing to Novartis are 36.5% of £10 million, or £3.65 million. The value of Novartis’s share of the business goes up by 10 * £3.65 million = £36.5 million, and the liability for the “put” option increases by the same amount. The change in Total comprehensive income would be £6.35 million – £36.5 million = -£32.85 million, and the effect on Shareholders’ equity would be the same. A £32.85 million hit to “Total comprehensive” is not a big deal, but it’s £32.85 million for every £10 million increase in the consumer business earnings, and a bigger hit the higher the PE is.

The business will not be valued simply by assuming a PE and applying it, but earnings won’t be ignored and the problem remains for any reasonable method of valuation, unless there are sufficiently strong terms in the agreement designed to mitigate the problem. The £6.2 billion reserved may be a conservative estimate, but it won’t be conservative enough to cover all the possible growth between now and 2035.

If necessary, it’s likely that GSK could raise cash by selling some of the ViiV interest, or floating the consumer business (subject to agreed terms, for both) and maybe buying it back in stages as finances allow. Another option is a deal where Novartis get another piece of GSK’s pharma business. However, that might have to be done ahead of Novartis exercising the put, depending on how quickly GSK have to pay. The implications of a successful consumer business for the Novartis “put” might be a factor in Neil Woodford wanting the business to be broken up (although that’s speculation).

IMO the next quote suggests GSK are looking at options to fund the “put” –

“I think, as you pointed out, the agencies have already factored into their view of our balance sheet the liability of taking that ‘put’ when it comes and remember the window doesn’t open up until 2018, exactly how we fund it and how we deal with it we will have to decide when we get there, but it is already factored into the credit metrics for the company” (Q3 results transcript)

There’s a lot of conjecture in the next two paragraphs, which I prefer to simply ignoring the possibilities. The simplest option for financing the “put” is to swap more drugs for Novartis’s minority interest, preferably before 2018. The problem with any deal with Novartis is, they are likely to realize how the “put” could be a long term problem for GSK, which gives them a good bargaining position. What GSK need is a long term guarantee of a loan that would finance the “put” and give enough time for GSK to raise capital, by selling or floating some of the consumer business if necessary.

Because Novartis have a 36.5% stake, if GSK are obliged to pay £10 billion (for example), then the value of GSK’s 63.5% holding is £10 billion * 63.5/36.5 = £17.4 billion and the whole business is worth £27.4 billion. In theory the business would provide good collateral for a loan, however much it takes to buy out Novartis. GSK have a lot of debt and I have not checked the terms, and the terms could make such a deal difficult or impossible. If not ruled out by the terms attached to existing debt, GSK would need finance from a reliable source with deep pockets. Berkshire Hathaway might be the best source from GSK’s point of view. A big investment bank would not be as reliable but they could be promised business if parts of GSK were floated to repay the loan. Novartis would not be in such a good negotiating position if finance for the “put” was in place, or even if it looked likely to happen. While the finance I’ve described would reduce risk, it’s complicated enough that perceptions and credit ratings might not reflect that. I’ve never heard of a long term “put” financing facility, and innovative or specialist finance is not likely to be cheap.

This from the Q3 2015 results relates to the problem –

“Acquisition accounting and other adjustments resulted in a net charge of £229 million (Q3 2014: £579 million). This included remeasurement of the liability and the unwinding of the discounting effects on both the contingent consideration for the acquisition of the former Shionogi-ViiV Healthcare Joint Venture and on the Consumer Healthcare Joint Venture put option.”

The Novartis deal hadn’t happened in 2014, but ViiV existed then and the 2014 annual report mentions a similar charge, with the timeframe for paying Shionogi –

“Net other operating expense of £700 million (2013 – £1,124 million income) included, following the improved sales performance of Tivicay and Triumeq, an increase in the liability for the contingent consideration for the acquisition of the former Shionogi-ViiV Healthcare joint venture which has increased to £1.7 billion, resulting in a charge for the year of £768 million (2013 – £253 million). The liability represents the present value of expected future payments to Shionogi. These will be paid over a number of years and will vary in line with sales of products that contain dolutegravir.”

“These will be paid over a number of years” should mean there’ll be no sudden shock. Hopefuly, any big payments to Novartis would also be over a number of years.

There’s nothing unusual about earnout clauses and milestone payments, which have a similar effect of having to pay more to previous owners when a business is successful. It’s the lifespan of the Novartis “put”, the size of the consumer business, the size of Novartis’s stake, and the unwinding of the discount rate that pose a potential risk to investors, at least until it can be proved otherwise. Rapid growth of consumer products may be unlikely, but the twenty year lifespan of the “put” option gives time for a big gap to develop between the £6.2 billion reserved now and the market value of Novartis’s stake later, given revenues of £6.1 billion, on a 2014 pro forma basis, and the size of the minority shareholding (36.5%). If it’s a problem, it’s a problem of success. No-one wants a business to fail just to minimize earnouts.

The “market-based valuation” for the consumer healthcare business will presumably change with the valuation of companies with exposure to consumer healthcare, but the amount reserved for the “put” is not likely to be reduced if valuations fall, because such a fall could be temporary. The amount reserved for the “put” is a lot more than the minority interest on the balance sheet, so it isn’t just reflecting the market value currently. If the “put” is valued conservatively, it won’t be valued symmetrically, and a higher valuation could lead to a higher estimate of the liability.

The analyst Kerry Holford might have alluded to the problem with the Novartis “put” when a complicated question started with the dividend and finished with –

“What are your options from 2018 onwards if they do put the option and generic Advair does come?”

The CEO didn’t answer it –

“As far as your dividend question is concerned,” … “we have now given you a forecast forward for three years and now you want four or five or six years. On that particularly cheeky question I am going to decline to answer for obvious reasons.”

I can’t predict how the consumer healthcare joint venture will grow over ten or twenty years. Instead I’ve made a table which covers a wide range of assumptions, and (not surprisingly) gives a very wide range of outcomes. I call it a “worry table”, because you can exclude cells (the blue figures) where you think the assumptions are too unlikely to be worth worrying about, and you can exclude cells where the projected value of the “put” is too small to be worth worrying about (which should include anything less than the reserved amount of £6,204 million), and see if there’s anything left. I realize it’s unconventional and many readers might prefer that I’d chosen a narrow range. In the charts I’ve only charted for growth rates up to 10%, because double digit growth for a big enterprise over long periods seems unlikely, but the growth rates go up to 16% in the table.

I have not discounted the future values. If you assume growth of 8% and a discount rate of 5% for 20 years, the multiple for Net present value / Initial value is given by (1.08/1.05)^20 = 1.757. The multiple for 3% growth with no discounting is 1.03^20 = 1.806, which isn’t too far off 1.757. In other words, you can approximate growth of 8% and a discount rate of 5% by subtracting 8% – 5% = 3% and get a reasonable estimate by looking up 3% in a chart, since none of this is certain anyway, although I’ve only shown the error in the approximation for a single case. There’s more reason to apply a discount rate to the future liability if you believe GSK will be able to pay a few billion in the future more easily than it can now. GSK already has charges in the accounts from the unwinding of discount rates. If the consumer healthcare business generates free cash flow and pays a dividend to its owners, that could be offset against the liability, but modeling that would mean more assumptions and more spreadsheeting.

The highest initial value of the Novartis interest is £2,925 million, derived from assuming that it makes up 75% of the total non-controlling interest. That values the consumer joint venture at £2,925 million * 63.5/36.5 = £5,089 million, which is 83% of the £6.1 billion sales (2014 pro forma), while the low initial value of £1,950 million implies the JV is valued at £3,392 million, or 57% of sales. With a 20% operating margin targeted for 2020, with no sales growth that implies a £1.22 billion operating profit. If the JV was valued at 10 times operating profit, the value of £12.2 billion is much higher than any of the values I calculated from the controlling interest. However there are quite a few factors in that, with the operating profit being for 2020, assuming no growth in revenue and a 20% operating margin, and the operating profit is before interest and tax. If you increase the asumed initial value of Novartis’s minority interest, the value after the same amount of growth increases in the same proportion, which increases the liability for the “put”, although the value of GSK’s controlling interest also increases by the same proportion.

GSK OTC 10yr growth

GSK OTC 20yr growth

GSK worry table spread

GSK worry table spread formulas

The debt schedule in the 2014 annual report will not be up to date due to the deal with Novartis. It’s under “Contractual obligations and commitments” on page 69. It’s still worth showing, because it shows that most of the obligations did not have to be paid in the next five years. The totals (as at year end 2014, in £ millions, not broken down by type) are –

Total ~ Under 1 year ~ 1 to 3 years ~ 3 to 5 years ~ > 5 years
37,952 ~ 4,744 ~ 6,137 ~ 5,300 ~ 21,771

It’s something to check when the 2015 report is published. See also the bar chart for “Maturity profile of gross debt”, which gives more detail but without figures. In the 2014 report the chart goes up to 2045, when very roughly £800 million is due.


GSK closed its defined benefit pension plan to new entrants in 2001, going by –

“Sir Andrew Witty is a member of the Glaxo Wellcome defined benefit pension plan with an accrual rate of 1/30th of final pensionable salary. This plan has been closed to new entrants since 2001. The section of the plan that Sir Andrew is a member of provides for a normal retirement age of 60 and a maximum pension value of 2/3rds of pensionable salary. Since 1 April 2013, pensionable earnings increases are limited to 2% per annum for all members, including Sir Andrew.” (Annual report for 2014)

Many big companies closed such plans, because of rising life expectancy, and falling stock markets (in the UK, many companies withdrew supposedly surplus funds from schemes as stock investments rose in the 1990s).

While actuaries ought to be giving unbiased estimates of the eventual cost, that does not rule out the estimated liability increasing. From a table in the annual report with the heading “Movements in defined benefit obligations”:

(£ millions)
Pensions ~ Post-retirement benefits

Obligations at 1 January 2012
(14,334) ~ (1,616)

Obligations at 31 December 2014
(17,801) ~ (1,397)

The pensions obligation grew by £3,467 million in three years, an average of £1,155.67 million per year. Taking that average rise as a percentage of profit after tax –

2012 ~ £1,156 million is 24.7% of £4,678 million PAT
2013 ~ £1,156 million is 20.5% of £5,628 million PAT
2014 ~ £1,156 million is 40.8% of £2,831 million PAT

I don’t mean to imply that the pension obligation will keep growing by £1.2 billion a year, just that past experience shows the potential for it to grow a lot, unless the possibility can be disproved. In the annual report for 2007, the “Present value of scheme obligations” for pensions was 10,338. That implies an increase of 17,801-10,338 = 7,463 over seven years, an average of 1,066 per year, slightly less than the 1,156 average increase since January 2012 (in £ millions).

Taking the seven year average of 1,066 per year, at 4,835 million shares, that’s an average annual increase of 22p per share in the pension obligation. While the actual increases subtract from Other comprehensive income, they must at some point subtract from profit to avoid a pensions deficit from building up. The pensions costs are not excluded from GSK’s definition of free cash flow, and they don’t seem to be excluded from core EPS, except possibly if they’re in the “other adjustments” in “Acquisition accounting and other adjustments”.

(£ millions)
2014 ~ 2013 ~ 2012

Remeasurement (losses)/gains on defined benefit plans
(1,181) ~ 847 ~ (685)
That cost is in Other Comprehensive Income.

Pension and other post-employment costs
403 ~ 170 ~ 95

Most of the above cost is:

Defined benefit schemes (pension and healthcare)
371 ~ 138 ~ 51

As a percentage of profit after tax, the cost of Defined benefit schemes was 1.1% (2012) and 13.1% (2014).

From “Pension and other post-employment costs” to here, the costs don’t include the remeasurement losses. They are spread between Cost of sales, Selling, general and administration, and Research and development. There would generally be a remeasurement loss in OCI either before a sustained rise in the costs hitting the income statement, or at the same time (similar to the way an insurance company increases its reserves to reflect more insurance losses than previously expected).

The estimated pension liability is based on long term assumptions which are revised each year. For the UK, 2014 only:

Rate of increase of future earnings 2.00%
Discount rate 3.60%
Expected pension increases 3.00%
Inflation rate 3.00%

The biggest number in the sensitivity analysis is the £645 million extra liability for a 0.25% decrease in the discount rate. That should work in GSK’s favor if interest rates rise.

Assumed life expectancy beyond age 60, projected for 2034, ranges from 28.7 years (male, USA) to 32.2 years (female, UK). I expect GSK’s actuaries know their job, but it isn’t completely clear to me that the projections are for people with a good pension and post-retirement healthcare (where relevant), rather than just the average life expectancy.

From the sensitivity analysis, each extra year of life expectancy costs GSK an estimated £57 million.

An employee review on glassdoor complained about management being old (and having conservative attitudes). Old management who have been employed long enough to be in the defined benefit scheme are not likely to leave.

Maybe I should just trust the actuaries, but I can’t rule out the assumptions proving to be too optimistic. That would not immediately increase the cash paid out, but at some point higher future costs have to be recognized as non-cash charges, which the market would probably not ignore. At some point the obligation has to stop rising. When a scheme member retires, at least they stop accumulating pension rights.

Rising life expectancy is good for business, assuming elderly people continue to need medication. However, big old companies with legacy defined benefit schemes will generally be at a disadvantage relative to younger companies with no such liability, or from the expense angle, without the hit to Cost of sales, SG&A and R&D.

Credit ratings

The “put” obligation to Novartis is one reason given for a downgrade from Moody’s of the long-term credit rating to A2. More recently Moody’s changed the outlook, see “Rating Action: Moody’s changes outlook on GSK’s A2 issuer rating to negative” 29 May 2015 (moodys.com). There’s quite a lot of detail in their “ratings rationale”, most of it not cheerful reading for shareholders, but it’s the job of a ratings agency to consider what could go wrong rather than what might go right. The most positive opinion is about liquidity, due to GSK’s cash and a revolving loan. The cash balance in the Q3 results is £5,908 million, about 43% less than the figure in the Moody’s piece.

On Market Realist, find “credit ratings”, and you’ll see two A grades for the long-term credit rating. See also “GlaxoSmithKline plc Earns A+ Credit Rating from Morningstar (GSK)” by Mark Watkins, November 4, 2015 (dakotafinancialnews.com).

Legal proceedings

There are five pages about this in the annual report for 2014 which you can find with “45 Legal proceedings” (“45” as in “Note 45”). I don’t go into a lot of detail here beyond reproducing headings to give some idea of the range of legal proceedings.

Intellectual property

Product liability
Paxil/Seroxat and Paxil CR (relating to pregnancy, and attempts to commit suicide or acts of violence)

Sales and marketing and regulation
China investigation
SEC/DOJ and SFO Anti-Corruption enquiries
US State Sales and Marketing Investigations
Average wholesale price
Cidra third-party payer litigation

EU sector enquiry
UK Competition and Markets Authority investigation

Securities/ERISA class actions – Stiefel
Commercial and corporate
Environmental matters

About “Intellectual property”, like most big pharma companies, GSK don’t have high product concentration, except for Advair which is off-patent anyway. The next table uses the figures under “Pharmaceuticals and Vaccines turnover by therapeutic area 2014” in the annual report for 2014. It presumably excludes OTC because the total is less than group “Turnover” of £23,006 million.

Drug (D) or Vaccine (V)
2014 Sales (millions) ~ 2014 Sales % (my calculation)

Total drugs and vaccine sales
Seretide/Advair (D)
£4,229 ~ 22.7%
Infanrix/Pediarix (V)
£828 ~ 4.4%
Avodart (D)
£805 ~ 4.3%
Epzicom/Kivexa (D)
£768 ~ 4.1%

Legal and patent status –

Seretide/Advair ~ no disputes, off patent (see note)
Infanrix/Pediarix ~ no disputes, off patent USA, expires 2016 Europe (see about vaccines)
Avodart ~ IP dispute, off-patent (so the case refers to past rather than future sales)
Epzicom/Trizivir/Kivexa ~ IP disputes ~ Epzicom/Kivexa expires USA 2016 (combination) and EU 2019 (combination). Trizivir is the same except the EU date is 2016.

I’ve described Advair as off-patent because sales are already declining and the message from management is that generic sales in the US depend on overcoming the difficulty of making the product and getting FDA approval rather than on patent expiration. See “Advair and generics” above for details.

For GSK’s vaccines and OTC businesses, patent disputes are less likely and less serious for the same reasons that the products don’t fall off a patent cliff – difficulty of production, and branding.

The successfuly launched HIV drugs Tivicay and Triumeq are not in the “Legal proceedings” section.

From the Q3 2015 results announcement –

“Legal charges of £72 million (Q3 2014: £318 million) included settlements of existing matters and litigation costs. Legal charges in Q3 2014 included the £301 million fine payable to the Chinese government. Cash payments in the quarter were £43 million (Q3 2014: £341 million).”

R&D day

The R&D event was held on November 3, 2015 in New York, featuring a selection of “assets” in the pipeline (39 “new molecular entities” or NMEs, and 15 new formulations or combinations, known as PLEs). I’ve already quoted extensively from the R&D event documents, but there are a few other points worth making. Patrick Vallance, President of Pharmaceuticals R&D, said before getting to the drugs –

“Publications are a surrogate for how well-connected we are, how well-respected we are in the academic community. Not only do we publish a lot but actually we publish in the very highest quality journals, highest citations, 35 publications annually in the very top journals.”

“I will just highlight one programme and that is the programme we have for chemists, where we have 50 chemists in training doing PhDs whilst being part of GSK. I think the 19th one was awarded last week, but the fellowship schemes go throughout GSK with individuals being allowed to focus on science.”

“One final thing to say about our external advice. We have recently formulated an immunology network to really try and bring in immunology challenge throughout the organisation because it is such a key theme through many of the medicines I am going to talk about and the latest example is actually bringing in six external academic investigators to have labs based inside GSK, not only to challenge, but of course to stimulate and give new ways of thinking, a so-called immunology catalyst.”

The common thread there is communication between company scientists and academia, which I believe increases the chance of getting the best idea from whoever has it. I can’t promise that they all collaborate effectively rather than try to defend silos, or that the initiatives are unique to GSK, but at least management are doing the right kind of thing. Partnership between companies is the norm (as well as suing each other), because microbes are often hit with two or more compounds with different action to make it harder to evolve resistance, and a single company very often doesn’t have all the compounds good enough for a combination.

A slide in the presentation shows a piece of paper, about which Vallance said –

“This bit of paper was a note on which one of our chemists, Brian Johns, in 2005, sitting down with his counterpart from Shionogi, drew some structures of a molecule that he thought they might want to make, and some approaches to chemistry – in this case, trying to look at how they can make a third ring in the molecule close. That was a pivotal step in what then became dolutegravir”

The case shows how ideas can be small enough to fit on a page and powerful enough to underpin a blockbuster drug, while underlining the need for communication and collaboration.

A table in the presentation shows the planned filing dates for the NMEs and PLEs. The filing dates are for New Drug Applications or NDAs, although that’s FDA terminology and GSK did not say that all the filings were with the FDA, so some planned filing dates might be for Europe. In the transcript, “FDA” occurs twelve times and European regulators are mentioned five times. Summing up the numbers for NMEs and PLEs under the planned filing dates –


2014 to 2017
7 ~ 5

2018 to 2020
11 ~ 5

2021 to 2025
21 ~5

39 ~ 15

A third of those were discovered at GSK –

“Of the work you are going to see presented today about a third was discovered by Discovery Performance Units inside GSK and about two thirds are worked on by those Discovery Performance Units, obviously in collaboration with outsiders.”

That’s only for the featured items. A fuller list of drugs and vaccines in the pipeline can be downloaded from the Product pipeline page. New projects in pre-clinical development are not disclosed for competitive reasons, and some project types might not be identified.

See also the quotes under “Upside in the pipeline”, above.

The R&D event was about the science rather than about the commercial potential of the drugs and vaccines profiled. An analyst who asked about the pricing of the Bexsero vaccine in relation to the competition was reminded that the event was about R&D and told to talk to “our Commercial colleagues” in the lunch break. Investors did not have that opportunity and are left with uncertainty about the commercial potential of the pipeline.

For an example of press comment (you’ll need to answer a few survey questions) see “GSK unveils new drugs to boost growth” Andrew Ward, Pharmaceuticals Correspondent,  November 3, 2015 (ft.com).

R&D in 2007

The annual report for 2007 had some cheerful reading –

“The best year for pharmaceutical R&D since the merger

2007 saw GSK’s best year for R&D since the Group was formed in 2000. We have undoubtedly made great strides in the last seven years – but there remains more to achieve and more benefits which we can look forward to as our investment in the pipeline delivers.

During the year, three new chemical entities and one new vaccine were approved; Veramyst for allergic rhinitis, Tykerb for breast cancer, Altabax for skin infections and Cervarix to prevent cervical cancer.

We have progressed a range of products through the pipeline, positioning us well for the future. A total of nine new phase III programmes started. These are the large scale trials where we seek to ascertain safety and also to prove unequivocally the efficacy of the medicines before submitting them for approval.

Our initiative to in-license potential treatments continued. We brought three new late-stage programmes into GSK and moved a further four into late-stage development, improving our ability to reload and sustain the pipeline we need. By its nature, R&D carries inherent risk. We were pleased that 2007 was a year of few disappointments, with the most notable termination being that of odiparcil, to prevent blood clots. A number of product line extensions were delayed which we had hoped would gain final regulatory approval in the USA, including Lamictal XR and Requip XL.”

I read about an old efficiency drive with a growing sense of deja-vu –

“In October 2007, GSK announced a significant new £1.5 billion Operational Excellence programme to improve the effectiveness and productivity of its operations. This new programme is expected to deliver annual pre-tax savings of £700 million by 2010. GSK expects to realise the majority of annual savings within the first two years of the programme, with approximately £350 million expected by 2008 and £550 million by 2009. These savings will partly mitigate the expected impact to 2008 earnings from generic competition and lower Avandia sales and the associated adverse impact on GSK’s gross margin.”

Find “Basic earnings per share were 82.6p, 95.5p and 94.4p in 2005, 2006 and 2007” above, so you can see the figures above a chart for 2012 to nine months 2015. EPS has been volatile recently, so I’ll give core EPS for 2014 which was 95.4p, and for nine month 2015 it was 57.7p. I have not checked for corporate actions since 2007 that might have skewed the comparison, but the steadily increasing dividend and steadily decreasing share count suggest there hasn’t been anything drastic. My point is that the “best year for R&D” and “more benefits which we can look forward to as our investment in the pipeline delivers” in the 2007 report did not translate into EPS growth. The CEO started in May 2008.

Cervarix failed commercially in the US, find “with this vaccine is because we were late”, below.

The vaccine alliance

GAVI is a broad alliance which promotes the accessability of vaccines. Poor countries get cheap vaccines under a GAVI agreement. Now GSK has frozen prices for ten years for any previously qualifying country that gets too rich for GAVI, see “GSK extends its price-freeze commitment to ten years for countries graduating from Gavi support” January 26, 2015 (gsk.com). I’ll quote –

“Four of every five of GSK’s vaccines are provided to developing countries at a substantial discount to western prices. We offer our lowest prices to Gavi which can be as little as a tenth of developed world prices. At these levels, we are able to just cover our costs, which is key to making our GAVI offer sustainable. We continue to look at ways to reduce production costs and any savings we make we will pass on to Gavi.”

The good news in the statement for the vaccines business is that when prices just cover costs and are a tenth of developed world prices, it implies a 90% margin over production cost in the developed world, though like drugs there needs to be a sufficient mark up on production cost to give a reasonable return on the investment. The 2014 annual report has “Investment decisions are rigorously benchmarked using a Cash Flow Return on Investment (CFROI) framework.”, but I have not seen any CFROI figures disclosed, for vaccines or anything else.

Vaccines sales and supply

Orders and payment are lumpy for vaccines, causing short term variability. About supply constraints –

“if you look at Vaccines in the short-term we’re bumping up against some capacity constraints.” (Financial outlook and guidance, May 6, 2015, transcript).

That seems to be mostly resolved. From the Q3 2015 results transcript –

“Progress is evident in a number of areas” … “in our supply chains, particularly Vaccines and Consumer, where the investments we have made to improve capacity and reliability allowed us to move early on the important seasonal businesses of flu vaccines and cough and cold in Consumer, creating the opportunity to take share and improve pricing, delivering a significant step up in profitability and growth as a result.”

“Vaccines were at £1.2 billion, up 13%, reflecting continued progress of our new meningitis franchise and a strong performance in the US, which also benefitted from sales of flu vaccine.”

However, from the same transcript –

“Hepatitis sales were down 15% mainly because of the supply constraints that we have previously talked about.”

“We continue to make investments in the supply chain to improve overall reliability and expand capacity for the future, but it will take into 2017 before the programme is fully complete.”

Barriers to entry in vaccines

The significant barriers to entry and the lack of a patent cliff are worth quoting extensively. From GSK’s transcript for the Investor Event, May 6, 2015 –

“we have been investing substantially in our capital base, especially in the Vaccine business, and you will hear a little more about that. This remains a highly capital-intensive industry, there is no doubt that is a barrier to entry, which is why there are only four global Vaccine companies. That is an area on which we shall continue to focus.”

Some capital-intensive industries such as steel making are cyclical and destroy capital. Fortunately there’s more to vaccines than high capital needs.

“If I take pneumococcal vaccine that we have developed, Synflorix, we invested almost £400 million before we had the first Phase III data available to us.

Complexity of manufacturing is another very significant barrier. If I take again Synflorix, it takes 24-28 months to release a batch. It takes 550 quality control checks before you are able to release a batch. Trust me, you have 550 opportunities to fail your batch; very complex.

Finally, combination vaccines are very important. If you take infant vaccine, combinations are a critical means to enhance coverage through immunisation and in the pertussis area. If you don’t have five or six vaccines combined in one shot, you can’t enter the game. Significant financial portfolio and technical barriers to entry, which explains why there are very few players and very few players are able to enter this game.”

“in contrast to the pharmaceutical industry there is no patent cliff, there is no cliff upon patent expiry. The vaccine business is a very long term, very sustainable business where life-cycle management plays a very important role. As a consequence for that, for instance, if I take our R&D budget in GSK Vaccines, about 50% is allocated to active life-cycle management and the other 50% to new product discovery and development.”

Life-cycle management is about expansion beyond the initial geographic area and the initial indication, and improving the manufacturing. As an example of a new indication for an existing vaccine –

“so this is news – Phase III data with our flu vaccine, quadrivalent, in the paediatric, six-month old population, where we successfully completed the Phase III trial and will be filing for this indication in 2016 here in the US.”

Manufacturing improvements include new formulations, better manufacturing reliability, and lower cost of production.

Vaccines breakthrough

Vaccines work by injecting substances which look like a pathogen to the immune system, while being harmless or much less harmful (attenuated). They are frequently combined with an “adjuvant” which is not a vaccine but enhances the effect, and adjuvants can vary in form from aluminum to an oil in water immersion. A problem with vaccines is they usually have less effect in the elderly. GSK have an adjuvant which they claim will make many vaccines effective in elderly people. About the Shingrix vaccine for shingles –

“A second point about this vaccine is about the adjuvant formulation, AS01. We believe that this is a breakthrough in vaccinology: all elderly vaccines have moderate efficacy, whether it is flu, or pneumovax or others, and it’s having an adjuvant formulation like AS1 that will allow us, or will enable the feasibility of a number of elderly vaccines. I will be describing to you one of those, but there are more to come.” (R&D event transcript)

About shingles, most people have zoster virus (aka “cow pox”) without it doing any harm, but the risk of it causing shingles increases from age 50 to about 80 to 90. A severe complication of shingles called post herpetic neuralgia is a leading cause of suicide in the elderly.

Shingrix passed Phase 3, see “GSK’s candidate shingles vaccine demonstrates 90% efficacy against shingles in people 70 years of age and over” 27 October 2015. GSK also claim 91% to 97% efficacy compared to 51% for Merck’s Zostavax, in the R&D event presentation. The age groups are probably not identical in that comparison, but GSK claim Shingrix is better in elderly patients due to the adjuvant. Also its storage requirement is “fridge” whereas Zostavax is “freezer”.

An analyst pointed out that GSK were the only company not to get a pandemic vaccine approved in the US a few years ago, because of the adjuvant. The reply described the variety of adjuvants and their safety profiles. The adjuvant in question is suspected of causing narcolepsy. Other adjuvants have well established safety profiles, and –

“The very same adjuvant formulation AS1 has been in thousands of newborn babies as a malaria vaccine in a population where health is fragile in sub-Saharan Africa, again with a very good safety profile. We feel confident that this vaccine adjuvant is very safe”

About the failure of Cervarix in the US, it was not due to the adjuvant. GSK’s failure …

“… with this vaccine is because we were late. I believe that if we had been first, it would have been very difficult to displace a cancer vaccine with a genital wart vaccine. That is how things unfolded but I just wanted to correct the perception that the adjuvant formulation has anything to do with the performance of that vaccine.”

I don’t doubt the replies, but there have been failures in vaccines which management would not have mentioned if it wasn’t for those pesky analysts.

Meningitis B vaccine

Vaccines for meningitis B have been approved without the usual proof of efficacy, because the unpredictable incidence made Phase III trials impractical. GSK’s Bexsero was approved in Canada and used in the Saguenay Lac province of Quebec, with 55,000 people aged two months to 20 years immunised between March and May of 2014. The cases of MenB stopped. Within a year there were two cases, both from people who had moved in from other provinces. GSK claim that data from the province shows Bexsero is the only meningitis B vaccine with effectiveness data in real-life use.

The UK has started vaccinating 700,000 children a year with Bexsero. If the success in Saguenay Lac is repeated, data from the UK should give GSK a strong case for approval in the US –

“That study will read out in 2019 and will inform our decision to fight for this indication in the US and also, importantly, as happened with meningitis C in the 1990s, should inform other countries’ decisions for implementation of universal mass vaccination in their paediatric population.”

Respiratory syncytial virus (RSV) vaccine

Respiratory syncytial virus (RSV) is highly infectious and affects infants and the elderly. GSK’s RSV vaccines are at an early stage with a Phase 3 study not expected to start until 2019. In infants RSV can cause pneumonia and bronchiolitis and can be a factor in the development of severe asthma. A problem in vaccinating against it is that babies can get the infection before their immune system is developed enough to benefit from vaccination. GSK’s solution is two vaccines, the first is to vaccinate pregnant women with, and the baby retains enough antibodies to provide protection for three to four months, by which time it can be vaccinated with the second vaccine.

An RSV vaccine in the 1960s did more harm than good, probably by inducing the wrong kind of inflammation, wiping out lung tissue as well as the virus. GSK believe they have overcome the problem, but the details are fairly technical. About the infant (not maternal) vaccine –

“We believe that we have a new platform immunisation technology in the form of a replication-defective chimpanzee adenovirus vector that we acquired from a biotech company called Okairos in 2013 that is able to induce that type of T-cell response, in fact this vector has been, in clinical trials in neonates with a malaria vaccine candidate and shown to be safe and able to induce the right T-cell responses. It is also the vaccine vector that we use in our Ebola vaccine, that, as you know, is in Phase III trials in a few thousand individuals.”

“Everybody is expressing – including ourselves in the past – what is called the Post-Fusion form of glycoprotein F, we have been successful and we have proprietary protection around it in expressing what is called the Pre-Fusion form of glycoprotein F.”

The “Pre-Fusion” form absorbed about 85% of the neutralising antibodies in a serum from an infected person, which means they look like RSV to the immune system. The “Post-Fusion” form only managed about 15%.

The term “new platform immunisation technology” suggests wider application than RSV, and the Ebola vaccine is mentioned, but if there were strong prospects for wider application it’s likely that more would have been said.

I’m not sure how Dr Moncef Slaoui can say the programme is “very high risk” while the vaccine will “undoubtedly” be recommended across the globe –

“We believe that this is the most credible vaccine in the industry against RSV, this is a very high risk programme, but this is also a very high reward programme. I believe that the vaccine against RSV will undoubtedly be a vaccine universally recommended across the globe for immunisation in infants, because of the incidence and importance and severity of the disease.”

Vaccinating COPD patients

GSK describe “A new vaccine concept” for chronic obstructive pulmonary disease (COPD). COPD is the result of irreversible lung injury caused by exacerbation episodes, which are often associated with lung infections. Two kinds of bacteria cause 30% to 50% of the infections that lead to exacerbation episodes (nontypeable haemophilus influenza and moraxella catarrhalis).

A vaccine against the first infection has completed Phase 1 and Phase 2 studies in a healthy elderly population. If further studies are successful, an antigen will be added for the second infection. COPD particularly affects elderly patients, and the previously mentioned adjuvant AS01 is used with the vaccine. Vaccination would be for COPD sufferers to prevent further infection from causing more lung damage. Filing for the COPD vaccine is not expected until after 2025.


Benlysta is for Systemic Lupus Erythematosis (SLE), a chronic immuno-inflammation disease affecting musculoskeletal, haematological, cutaneous & renal systems, resulting in poor quality of life. It’s a kinase inhibitor which is good at hitting it’s target (RIP1 kinase) with very little ‘collateral damage’ to anything else. Benlysta  was developed by GSK and Human Genome Sciences, and received FDA approval in 2011.

“when Benlysta was approved it was the first medicine for 50 years; if you look at the number of other people who have tried in SLE, there have been three other failures in Phase III from other compounds in SLE, so the fact this has hit again on a third pivotal study is not chance, it is actually a rather important observation.”(R&D event transcript)

Benlysta doesn’t outperform the placebo massively. For “Proportion of patients with SLE Responder Index (SRI) response at week 52”, it’s placebo 48.4%, Benlysta 60.8%, however the placebo is actually a high standard of care, not a sugar-pill. The statistical significance of the higher responder index is given by “p=0.0011”, which means if there was no real advantage over the placebo, there would only be 1.1 chance in 1,000 of getting results that good.

“first medicine for 50 years” is an achievement. Benlysta could move beyond Lupus (“subcut” is subcutaneous, or under-the-skin) –

“so definitely hit the target, got a nice molecule and got the potential to go into multiple areas where you have not only inflammation, but you have cell death driving the inflammation.”

“I think Benlysta with the subcut has the potential not only now to move into a different population, in terms of more people wanting to use this, but also in terms of different diseases that we are looking at and we will be filing for this very shortly, either at the end of this year or the beginning of 2016.”

“So we are going to go next year into parallel early experimental medicine studies, rheumatoid arthritis, ulcerative colitis and psoriasis. We will pick early from these”

A patient who suffered the effects of autoimmune B-cells had not responded to rituximab or Benlysta alone, but responded dramatically when both were given. Rituximab knocks out B-cells, and the bone marrow makes more of them. The theory is that in the presence of Benlysta, the new B-cells were not autoimmune, and if that’s right, maybe Benlysta could be used to reset a faulty immune system. GSK will be exploring the effect in a clinical study starting in 2016.


Amyloidosis is a disease where proteins don’t fold properly and they accumulate in tissues, including vital organs and nerves. One approach is to switch off the protein production. There’s a more novel approach which depends on a hypothesis from an academic GSK are working with. The theory is that the amyloids get a coating which stops them from being “phagocytosed”, which means the body’s garbage removal system doesn’t touch them. The approach based on that hypothesis is to make the coating (serum amyloid P) visible to macrophages so they can “throw out the trash”, although I’ve skipped some detail.

I’ve described a few vaccines and drugs I found interesting, but the selection misses out a lot, including all of oncology.


From GSK’s Data transparency page –

“Back in 2004 we introduced our publicly accessible Clinical Study Register online – a place where we could post information about the clinical research we carry out on our existing medicines, and also the ones that we are developing. We were the first pharmaceutical company to do this.”

Most media stories are positive about this, but not “Transparency measures forced on pharma” by Nuala Moran, 30 September 2014 (rsc.org/chemistryworld). The piece acknowledges cases where avoiding a conflict of interest is bad for communication and cooperation.

In “Drug firms far from transparent with trial data watchdog group finds” by Douglas W. House, SA News Editor, Nov 12 2015 (seekingalpha.com), GSK came out well.

Trial on antidepressant neglected suicide attempts” is about an antidepressant which was reported to be safe and beneficial for adolescents, after a clinical trial in 2001. The claims became controversial, and it was possible to re-evaluate the drug because GlaxoSmithKline made the data available. It turned out there was no benefit to adolescents, and there were eleven cases of attempted suicide or self harm in the trial, compared to just one in the control group. It’s shocking that the data was originally interpreted in such a biased way, but good that the data was opened up.

Public discussions about not keeping Established Products and ViiV

For Established Products and ViiV, management decided to publicly discuss whether or not to hold on to the businesses, so they could involve shareholders as well as banks and potential buyers. That was democratic, but while the discussion was taking place, “expectations of the HIV new products literally almost grew exponentially as the product began to launch” (Q3 2015 results transcript). Discussing the ownership of ViiV could be seen as either a lack of leadership or a lack of confidence in ViiV. Established Products are described as “very distributed” and “the complexity of extraction is very, very material”, making discussion about divestment seem pointless.

Partnering with Pfizer’s POD

There isn’t a lot left to say after giving the title for this press release from Pfizer – “Pfizer Announces Collaboration with GSK on Next-Generation Design of Portable, Continuous, Miniature and Modular (PCMM) Oral Solid Dose Development and Manufacturing Units” October 29, 2015. Pfizer already have a prototype “POD”. The PODs take up less space than conventional production equipment for solid pills. The same PODs can be used for development, clinical trials and commercial manufacturing, and are set up in about one year rather than two or three. GSK’s input is their “experience and expertise in continuous processing”. I found no news about it on GSK’s website, and I expect they’ll be the minor partner, but it’s still good for them to be involved.

Partnering with Zymeworks

GlaxoSmithKline partners with fast-growing Zymeworks in $440M-plus deal” by John Carroll, December 3, 2015 (fiercebiotech.com). Zymeworks have technology relating to antibodies, and proteins that last a long time in the bloodstream. The $440 million is potential milestone payments, and there was probably an up-front payment as well. Other big pharma companies have already cut deals with Zymeworks so there’s no reason I know of to expect the deal to give GSK a competitive advantage.

The Trans-Pacific Partnership trade deal

The TPP (Wikipedia, find “Cost of medicine”) has implications for pharma companies. For biologics, see “TPP trade deal: Even more bad news for biotech industry” by Laura Lorenzetti , October 5, 2015 (fortune.com). The agreement includes a common patent life for biologics which is much shorter than in the US currently. The rules are complicated, and companies can choose between two schemes, but complaints from bio companies point to the overall effect (they probably underspent on lobbyists). Possible effects include less investment in biologics, and higher prices over a shorter span in the US. However, biologics are generally hard to copy, so the effect might not be as big as the complaints suggest.

MSF (aka Médecins Sans Frontières or Doctors Without Borders) are concerned “about the inclusion of dangerous provisions that would dismantle public health safeguards enshrined in international law and restrict access to price-lowering generic medicines for millions of people”, see “Statement by MSF on the Official Release of the Full Text of the Trans-Pacific Partnership Trade Agreement” November 05, 2015 (doctorswithoutborders.org).

There’s also the Transatlantic Trade and Investment Partnership (TTIP) with implications for the UK’s National Health Service (NHS) and clinical trial data, according to The Independent, “What is TTIP? And six reasons why the answer should scare you“. TTIP is still being negotiated and agreement seems difficult because EU member countries can veto it.

Industry problems

When I blogged about Gilead, I said that many of their problems could affect other successful pharma companies, partly depending on circumstances. One problem is pressure to cut or contain costs, from governments or other large buyers. Competitors might develop superior products, or products good enough to keep the margin down. Competitors, buyers and activists can challenge patents. There’s competition from generic copies (at least for small molecules) when patents expire, unauthorised sales from producers in emerging markets, and sales on Alibaba. There’s also uncertainty over the future of “Obamacare” and the effects if it’s stopped.

As the low-hanging fruit is increasingly picked and effective high volume therapeutics go off patent, at some point the remaining opportunity will increasingly be in rare diseases and personalized medicine. We are still some way off from having problem free cures for every major disease, but in the long run it’s a problem for pharma companies, except those specializing in rare diseases or aiming to commercialize personalized medicine.

Glassdoor reviews

The employee reviews are here (glassdoor.co.uk). They are mostly quite good. IMO, anyone who has not checked a company by the employee reviews should not be too shocked by the negative comments so long as they are in the minority, as a wide range is usual. There were complaints about bureaucracy and time consuming admin processes, but that’s fairly common for big companies. It’s disappointing that the only review of the US operations is very bad, although it’s common for employees not based in the home country to range from disatisfied to very unhappy. I suggest looking at the US based glassdoor reviews (which I can’t get), although you might be redirected there anyway from the link I’ve given.

For comparison, for AstraZenica there’s enthusiasm for the company’s commitment to science. There was only one complaint about bureaucracy, but quite a lot of complaint about too many reorganizations, and change that either did not happen or took too long, leaving employees feeling insecure about their jobs or the projects they were working on.

For Pfizer, I was redirected to the UK glassdoor. Employees seemed happy and liked the friendly working environment and personal development. The word “bureacracy” was not used but it was implied, and management were criticized for being slow to adapt and too risk-averse to innovate. One employee implied that sites outside the US were isolated. Reviews on the US glassdoor might be a lot more positive.

GSK had the lowest overall score, with Pfizer having the highest, but there was only 0.3 difference (the maximum score is 5).

BBC radio programmes

This is just from memory and therefore less reliable. Using the links to make a methodical transcript would not be respecting the BBC’s intellectual property. You probably have to be in the UK to get the podcasts on the links.

In “Bitter pills” (bbc.co.uk), the CEO was interviewed in 2010. Soon after becoming CEO in May 2008, Andrew Witty realized that no-one was taking reponsibility for killing projects, which meant expensive support for developing compounds that were likely to fail. That’s what prompted the reorganization in into teams for discovery and development, and a policy of axing projects when the evidence was against success.

There’s a recent BBC radio interview with Patrick Vallance, the President of Pharmaceuticals R&D – “Problems of Developing Drugs“. When he held a top university post, Vallance had been a critic of “big pharma” R&D, saying it was too focused on marketing. Having made the transition from science to management, Vallance believes mobility of scientists between academia and industry would benefit both sides, though he emphasised academics moving to industry and back. He doesn’t miss science because he’s immersed in it in a broad way, looking at data or discussing clinical trial setup, for example.

He said about personalized medicine, that the disease names are based on 19th century science, and no one thinks of breast cancer as a single disease any more. While the sub-categories inevitably have fewer patients, understanding a disease, particularly its genetics and biochemistry, increases the chance of success. If a 3% success rate improves to a 6% success rate, it decreases the cost of development per success from $1 billion to $500 million. The effect can be seen in approvals for the industry as a whole. I’ll add that targetting a modern sub-category of a broad “19th century” category improves the chance of a cure, simply because a medicine for a disease is less likely to be effective if the disease is actually a broad category with much variation. If the sub-categories aren’t recognized, then understanding of the disease will be poor.

For some diseases, better understanding has led to a “golden era” of discovery. He mentioned depression and schizophrenia as two ‘non-golden’ cases where the understanding was not so good. I’ll add that much of the progress will be due to Moore’s law (the exponential increase in the power of digital chips), just as the invention and develpment of the microscope gave biology and then medicine a tailwind. When scientific advance opens up new opportunities, there’ll be more companies targeting the same area, the bigger the potential profit (as for hepatitis C). Still my own thoughts, Gilead’s patents relating to HCV have been challenged in some countries on the grounds that they did not add much to the academic research. Similar objections seem likely in other areas unlocked by academia, though maybe in countries that signed up to the TPP agreement, TPP can be used to extract compensation if patents are not upheld.

In malaria research, while thousands of drug candidates had been identified, only two were being followed up, due to the cost. The action taken was to make the data public, under an arrangement where GSK would get a small royalty in the event of development through to sales. I’m not sure of the details, and enforcement without patents seems difficult. Pharma companies tend to avoid malaria because it’s mostly poor uninsured people who get it. None of that is mentioned on “Our fight against malaria” (gsk.com).

I’d guess that Vallance moderated the small team “be like a biotech” policy developed under the previous president, where initially careers were closely linked to the success of the team.

Summing up

Before Andrew Witty become CEO, GSK had a corruption problem, and a reorganisation had not sufficiently addressed the problem of low productivity from R&D. Since then there’s been the continuing drag of reduced returns on R&D throughout the industry, the pension obligation rising by £7.5 billion since 2007, and the burden of paying the dividend and meeting expectations for it to rise, although the special dividend due in April 2016 might look like self harm if cash gets tight.

Bold moves by management include the reorganisation around the “DPU” team structure, ending payments to doctors, ending incentives to sales staff to hit sales targets, leading data transparency, the Novartis transaction, and the consolidation of sites along with big job cuts in R&D. The intention to cut layers of management from the consumer business and increase agility is good. There may be a problem in driving change through the organization, for example if the allegation about the cardio division over-claiming for their projects and pushing poor medications into Phase 3 is true. Also there were complaints about bureaucracy in the glassdoor reviews, but as I said that’s true for almost any big old company.

A key factor for a pharma/bio company to outperform in the long run is efficiency all along the pipeline, by which I mean from generating ideas to commercializing approved products. Efficiency along the pipeline is supported by the quotes under “Upside in the pipeline”, above, but there have also been recent failures. Previous years where FDA approvals looked good in the annual report were followed by a disappointing financial performance.

Low growth of core EPS is projected up to 2020, and unless growth is faster or cash conversion improves, the dividend will be at risk.

The Novartis “put” option represents a complex risk, with GSK having a higher liability the more successful the consumer healthcare business is, plus the liability will increase automatically with the unwinding of the assumed discount rate. On the other hand, if the “put” is exercised and enough has been reserved for it, there will be an increase in shareholders’ equity. One possibility is that GSK will have to reserve more for the Novartis “put” at the same time as the pension liability increases. That would not affect core EPS but would affect the balance sheet and the ability to pay the dividend.


If the dividend can be maintained, I see the combination of yield, some EPS growth, and possible upside as being enough to be reasonably attractive. I see quite a big upside through pipeline efficiency, but that’s just my opinion. Other investors could easily give more weight to the negative points and the history of underperformance, and less to management’s recent quotes about FDA approvals. The case won’t be proved until there’s sufficient successful commercialization, but waiting for proof is likely to miss any opportunity.

The high yield suggests the share price is supported by the dividend, not the prospects. If the pipeline really has got more efficient, that could have a substantial effect on the stock price. There has to be some risk that the long-term underperformer won’t turn round, but the approvals record with the FDA suggests to me that GSK will have more to offer than just the current and planned dividend.

My opinions are based on the research you see here, and I don’t claim they’re better than other investors’ opinions.

Some points for investors to consider

• Risk to the dividend
• How good is the pipeline
• The Novartis “put”
• Also, check the table under “Contractual obligations and commitments” when the annual report for 2015 comes out.

I kept the list short. GSK have ten pages of risk factors under the headings Patient safety, Intellectual property, Product quality, Supply chain continuity, Financial reporting and disclosure, Tax and treasury, Anti-Bribery and Corruption, Commercial practices and scientific engagement, Research practices, Environment, health and safety and sustainability, Information protection, Crisis and continuity management, and Third-Party Oversight. A lot could go wrong that I’ve hardly touched on.

A scenario

One possible scenario is that GSK need to reserve so much for the Novartis “put” option that they have to cut the dividend. That’s unlikely to happen in the next few years because the consumer business is not likely to grow in value by a lot in just a few years. Remember, £6.2 billion has been reserved in Other non-current liabilities, while the figure for all the non-controlling interests is much less, at £3,900 million, so some growth has already been allowed for. The risk exists because a lot of growth could happen in the next 20 years, though there could be terms in the agreement which mitigate the risk.

If that scenario happens, GSK would have a controlling interest in a stronger consumer business, and cutting the dividend would be good for the financial position and credit ratings. Income investors tend to ignore the positive side when the dividend is cut, and a sharply lower share price could represent a buying opportunity. Obviously I can’t say in advance what the overall condition of the company would be in such circumstances, except that cash would have to be tight for GSK to cut the dividend. It’s just a scenario, which might never happen.

That’s all. Thank you for reading this.

DISCLAIMER: Your investment is your responsibility. It is your responsibility to check all material facts before making an investment decision. All investments involve different degrees of risk. You should be aware of your risk tolerance level and financial situations at all times. Furthermore, you should read all transaction confirmations, monthly, and year-end statements. Read any and all prospectuses carefully before making any investment decisions. You are free at all times to accept or reject all investment recommendations made by the author of this blog. All Advice on this blog is subject to market risk and may result in the entire loss of the reader’s investment. Please understand that any losses are attributed to market forces beyond the control or prediction of the author. As you know, a recommendation, which you are free to accept or reject, is not a guarantee for the successful performance of an investment.

IPG Photonics charts up to Q3 2015

Disclosure – I’m long IPG Photonics Corporation (IPGP).


My main source of data is IPG’s SEC filings page, with some info from the Prepared remarks on the Presentations page and the results releases on the Press Releases page. Some “net sales” figures were reported as “revenue” outside of SEC filings, and in previous quarters I’ve confirmed that when both terms were used for the same quantity, they had the same figures.

I’ve also quoted from “IPG Photonics (IPGP) CEO Valentin Gapontsev on Q3 2015 Results – Earnings Call Transcript” Oct. 27, 2015 (www.seekingalpha.com).

Annual data

There’s more annual data charted in my piece “IPG Photonics – charts up to Q1 2015”, on Seeking Alpha and on my blog site, including the growth of sales in China and assets in Russia.


I don’t mean to imply that the quarterly twitches matter more than the general trend. The best entry points might be when quarterly results disappoint, although the market is capable of worrying ahead of results. For example recent share price weakness has been attributed to concern about China, see “IPG Photonics up 9.2% following mixed Q3 results, soft Q4 guidance” Oct 27 2015 (seekingalpha.com). The results were reassuring about China, which explains the rise, but that followed a 5% drop (see IPG Photonics misses by $0.03, beats on revenue).


Charts showing guidance are based on this from the prepared remarks:

    “We currently expect revenues for the fourth quarter to be in the range of $215 million to $230 million. We anticipate Q4 earnings per diluted share in the range of $1.00 to $1.15. The mid-point of this guidance represents quarterly revenue growth of approximately 7% and EPS remaining flat, respectively, year-over-year.

    The EPS guidance is based upon 53,392,000 diluted common shares, which includes 52,675,000 basic common shares outstanding and 717,000 potentially dilutive options at September 30, 2015.”

Income, revenue and costs

IPG results to Q3 2015 bars and line

About seasonality, the 10-K for 2014 has –

    “Historically, our net sales have been higher in the second half of the year than in the first half of the year.”

and from the Prepared remarks –

    “While the book-to-bill was slightly below 1 in Q3 2015 this is not unusual because we typically expect the fourth quarter to be seasonally weaker.”

Taken together, Q3 must be seasonally strong. That’s consistent with the chart above, although there was also a strong Q4 in 2014 due to exceptional orders as pointed out in the middle of this –

    “The Q4 guidance represents about 7% revenue growth at the midpoint over Q4 2014. I want to point out that our guidance range includes double digit growth in Germany, Japan and China. Year over year we expect Q4 revenue in China to grow at about 15%. While this represents a lower rate of growth as compared to the year-to-date it is still pretty good. What is pulling down the overall Q4 expected growth rate is the U.S. which had a very strong Q4 2014 with shipments of several super high-power lasers and a large automotive order. Revenue in the U.S. is more evenly spread between Q3 and Q4 this year rather than being weighted to Q4. In addition, we are expecting a slightly weaker quarter in Turkey and Korea where macro-economic conditions have softened.” (Prepared remarks)

In the next chart, “Operating expense plus tax” is actually Operating income – Net income, and includes a small amount of “Other income, net”.

IPG results to Q3 2015 stacked

IPG EPS to Q3 2015

IPG EPS growth to Q3 2015

Revenue and costs are affected by currency exchange rates –

    “IPG delivered yet another strong quarter, growing revenues by 22% year-over-year to $243.5 million in the third quarter of 2015. We translated that growth into a gross margin of 54.7% and we reported EPS of $1.18 which includes a $0.06 per share impact of foreign exchange transaction losses. Year-over-year, foreign currency losses reduced our sales by 12 percentage points. In other words, sales would have increased 34% if Q3 2015 exchange rates remained the same as a year ago.” (Prepared remarks)

The only hedge I found was an interest rate swap associated with the U.S. long-term note, worth a notional $11.3 million in 2014, which matured in June 2015. I’ve seen no evidence of currency hedging with derivatives, although the company said they analyze exposure and might use financial hedges. Encouraging faster payment in China was presumably more cost-effective, and also benefits customers rather than giving business to financial counterparties.

Other comprehensive income, exchange rates and dollar debt

As well as affecting how revenue and costs are translated into dollars, exchange rates also affect the dollar value of overseas assets, and the change in the value is recognized in Other Comprehensive Income –

IPG Net income and OCI to Q3 2015

The dollar is probably in long term decline, though with big swings (for the dollar index since 1971, click here then click “Historical” and “Max”). That suggests that in the long term, IPG will accumulate an OCI gain instead of the current accumulated OCI loss of $155.8 million. There’s some risk of further dollar gains before reversion to the trend.

I’ve warned for nearly a year that overseas dollar denominated debt could trigger a flight to safety that could push up the US dollar. The debt problem is explained in “Bond Issuers and Investors Are About to Find Out It’s Not Easy Being Green(back-Denominated)” by Tracy Alloway, August 24, 2015 (bloomberg.com). I don’t have the author’s certainty about it, due partly to a lack of evidence that borrowers don’t earn enough dollars to repay the debt. If it all goes bad, many companies would be affected. I’d rather be holding IPGP with the industry-leading margins, good free cash flow and big stash of cash than many other stocks.

The concern includes China, as in “China’s dollar debts come under pressure” by Michael Mackenzie and James Kynge, August 11, 2015 (ft.com) (you probably have to answer a market research question to see the article). There may be less impact on US firms if the debt is mostly held by local government and real-estate firms (see “China’s Big-Dollar Borrowers Hold Off on Hedging Foreign-Currency Debt” by Fiona Law and Esther Fung, Aug. 28, 2015 (wsj.com)).

Share count

The jump in the number of shares in 2007 looks like it’s the result of converting preferred stock to common stock in connection with the IPO in late 2006.

A public offering in 2012 raised $167.9 million after expenses (see optics.org for various details). It might look unnecessary given the growth in cash since then, but IPG’s cash is probably held overseas, mostly, with only about 16% of sales in North America, while head office costs are in the US. From the 2014 10-K –

“At December 31, 2014 and 2013 , the cumulative unremitted earnings that are reinvested in non-U.S. subsidiaries are approximately $522,000 and $346,000 , respectively.”

That’s in thousands, and compares to 2014 figures of $522,150 for cash and $1,210,887 for total assets (in thousands).

IPG share count Q3 2015

IPG change in shares Q3 2015

DSO and days of inventory

I calculated trailing twelve month values for Q3 amounts in the next two charts, which means that Q4 2014 is included twice. The low value of R-squared for the Days Sales Outstanding trendline indicates that it fluctuates around an average, rather than having any trend (technically, it’s wrong to fit a trend line to a series which puts a third quarter result on the end of annual results). The DSO figure for the latest quarter includes the effect of incentives for early payment in China. While IPG have reduced the average days of inventory during the period, the trendline is not reliable for prediction. IPG hold sufficient stocks of parts to be reasonably sure that assembly can begin as soon as orders are received, and thorough testing before dispatch adds to the inventory.

IPG DSO and days inventory Q3 2015

The next chart adds lines for sales, net income and cash from operations. It isn’t surprising if cash from operations is affected by DSO and days of inventory, but I haven’t tested for it.

I made various tests which aren’t shown, with negative results, for relations between the change in sales, and DSO and days of inventory. The tests included the change in next year’s sales, and both the absolute DSO and days of inventory, and the percentage change. The maximum variation explained was 16.4% for the change in inventory and change in next year’s sales. I’d thought that maybe inventory was built up ahead of an expected sales increase (e.g. for product launches), but the figures for the period don’t support that. As well as explaining little of the variation, the trend line sloped the wrong way, as if more inventory predicted lower sales growth. I did not control for management statements, and if it’s ever stated that inventory has been increased in preparation for anticipated sales, my results won’t apply.

IPG DSO days inventory - sales income and CFO Q3 2015

Cash flow

My chart of cash from operations against cash used in investment shows good growth of free cash flow. Annual cash flow can be highly variable, and quarterly cash flow is even more variable. The latest jump in cash from operations includes the effect of higher collection in China, as a result of incentives to pay early, motivated by IPG wanting to reduce exposure to the Yuan. A fall back in cash from operations would not be surprising. I have not separated out the effect of changes in working capital (receivables, inventory etc.) which is usually the most variable component.

The investing cash flow includes cash used for acquisitions. There are two cases where IPG briefly held short term investments, with a cash inflow following a cash outflow. “Purchases of short-term investments” for $25.5 million in 2011 were followed by “Proceeds from short-term investments” in 2012 for the same amount. $6.95 million of short term investments (in marketable securities) made in 2007 were sold in 2008 for $5.45 million. Those outflows and inflows are not investments in the business, but the effect in the charts is mainly on the timing of investment. The overall picture regarding free cash flow is not affected, except that they happen to make investment look a bit more lumpy.

“Property, plant and equipment” generally dominates, with figures of $68.2 million and $53.0 million in 2012 and 2011 (i.e. bigger than the short term investment), $88.6 million and $70.9 million in 2014 and 2013, and $18.2, $18.6 and $14.0 million in Q3, Q2 and Q1 2015 (with Q3 and Q2 calculated). Capex is expected to hold steady, excluding acquisitions which tend to be opportunistic –

    “We’re running, what is it about, 7% to 7.5% of revenue this year. So even that represents a substantial amount of spending activity. I don’t expect to increase as a percentage of revenue in next year at all.” (transcript on Seeking Alpha)

IPG quarterly cash flows to Q3 2015

Balance sheet charts

Where I include Q3 2015, there has obviously only been nine months between the quarter’s figures and the end-of-year 2014 figures. Cash was boosted by about $160 million raised from new equity in 2012.

IPG assets Q3 2015

IPG assets per share Q3 2015

IPG assets per cent of total Q3 2015

IPG liabilities and equity Q3 2015

IPG liabilities and equity per share Q3 2015

IPG liabilities cash and current assets Q3 2015

(The “calculated liabilities” are the same as “Total liabilities” from 2011 onwards, when they are reported.)

IPG liabilities as percent of cash Q3 2015

Materials processing and Other applications

In the Prepared remarks the CEO said –

    “In addition, we continue to pursue projects in non-materials processing applications, such as advanced, medical and telecom, because they present significant opportunities for us as evidenced by their strong growth in Q3 when sales for non-materials processing increased by 168% year-over-year.”

The CFO was less upbeat, with –

    “As a reminder, advanced applications sales are typically large and uneven from quarter to quarter.”

I expect he meant all the “Other applications”, having said –

    “Sales to other markets, including advanced applications, telecom and medical applications,”

When “Other applications” decline for a few consecutive quarters, you only get the “typically large and uneven” comment. While it’s possible the segment will see stronger or more even growth, without further information my attitude is that I’ll believe it when I see it.

IPG materials processing and other Q3 2015


IPG results to Q1 2015 spread

The annual growth in the number of shares is not shown.

IPG balance sheet spread Q3 2015

IPG DSO days inventory - sales income and CFO - spread Q3 2015

IPG quarterly cash flow Q3 2015 spread

IPG sales breakdown 15Q3 spread

Thanks SA

With thanks to Seeking Alpha for their policy about quoting from transcripts, which can be found at the end of the transcript.

DISCLAIMER: Your investment is your responsibility. It is your responsibility to check all material facts before making an investment decision. All investments involve different degrees of risk. You should be aware of your risk tolerance level and financial situations at all times. Furthermore, you should read all transaction confirmations, monthly, and year-end statements. Read any and all prospectuses carefully before making any investment decisions. You are free at all times to accept or reject all investment recommendations made by the author of this blog. All Advice on this blog is subject to market risk and may result in the entire loss of the reader’s investment. Please understand that any losses are attributed to market forces beyond the control or prediction of the author. As you know, a recommendation, which you are free to accept or reject, is not a guarantee for the successful performance of an investment.

A low cost world

Disclosure: I’m long Cash America International Inc (CSH), General Electric Company (GE), Gilead Sciences, Inc. (GILD), IPG Photonics Corporation (IPGP), Main Street Capital Corporation (MAIN), SunEdison (SUNE), Winmark Corporation (WINA).


Human ingenuity is leading to low-cost energy, low-cost production and more efficient business, which could lead to increased discretionary income and higher discretionary spending, although the technology which brings the benefits could also reduce lower and middle incomes.

About this piece

I make connections across many fields, I don’t claim specialist knowledge in any of them and I doubt if anyone has specialist knowledge of many of them.

The title is not meant to imply that everything is low cost or soon will be, and anyone paying for dental treatment in the US is likely to see why.

For the same level of consumer confidence, more discretionary income means more discretionary spending. I don’t try to predict future consumer confidence.

Much of the piece is about innovation in a variety of fields, with little specifically about the cloud and biotech because they are well known. As well as lowering costs, innovation will produce profound changes, but that’s not the point of this article.

Factors against rising discretionary income

Further down I write about consumer debt, and about technology discplacing jobs. There are other risks and factors that affect discretionary income (and more), some of which I list here although any detail is beyond the scope of this article: war, natural disasters, European debt and Europe’s immigration situation, and anything else that would massively disrupt trade. Other factors are mentioned where they are relevant, for example the claim that fracking businesses operate as a Ponzi scheme is in the following section about fracking. I don’t list all the negative factors every time I say something is positive for discretionary income.


Cheap energy is good for discretionary income any place where energy consumption outweighs production. While the consumption roughly equals the production, cheaper energy is good for the world economy (the “oil shock” in the 1970s when OPEC was formed and exercised its cartel power did not make the world richer). There are doubts about the commercial viability of fracking (technically, “hydraulic-fracturing”), as in Shale Fracking Is a “Ponzi Scheme” (September 19, 2014, but with older quotes), and I haven’t found a fracker that has not racked up a load of debt. “Einhorn Slams Mother Frackers” Sep 10, 2015 (econmatters.com) has less detail about the fundamentals but it shows that the concern still exists. If the debt goes bad, I can’t say how widely the consequences would spread. See also “Wall Street’s Biggest Banks May Have To Make Good On $26 Billion In Oil Hedges” by Tyler Durden, 04/09/2015 (zerohedge.com) – the piece names a few banks, who were also lenders.

About recent bankruptcies and closures, see “Fracking Firms That Drove Oil Boom Struggle to Survive” by Alison Sider, Sept 23, 2015 (wsj.com). If a well generates cash, it’s worth something and will have a buyer in the event of bankruptcy. It’s the bankrupts that hit the news, but it’s the survivors that will shape the future of the industry. There are about 39 bankrupt solar power companies in the linked list, from 2009 to 2014. It didn’t stop the price of solar PV modules from falling or stop installation from rising.

It’s claimed that the technology is advancing rapidly, for example with smart drill bits. See “How Big Data Can Save the Shale Boom” Jun 03, 2015 (the-american-interest.com), and there’s more in “Saudi Arabia may go broke before the US oil industry buckles” by Ambrose Evans-Pritchard Aug 5, 2015 (telegraph.co.uk).

The piece “How Long Can the U.S. Oil Boom Last?” by Dennis Dimick, Dec 2014 (nationalgeographic.com) gives 5 to 8 years before shale oil runs out. President Obama said about gas, “We have a supply of natural gas that can last America nearly 100 years” in 2012. The U.S. Energy Information Administration used to underestimate the amount of gas that could be extracted, but seems to have over-compensated and been too optimistic in 2012. From “Natural gas: The fracking fallacy” by Mason Inman, December 3, 2014 (nature.com), a three year study at the University of Texas concluded gas production would peak in 2020 and then decline 50% by 2030.

The problem with the predictions is that the easiest gas is naturally extracted first. (The term “sweet spots” has been picked up by detractors, it implies a sharp division between resources easy to extract and resources hard to extract, which may be the case but I haven’t seen data for how sharp the division really is.) Next year, the gas won’t be quite as easy to extract, but technology and expertise will have improved. That describes the situation for any year, and it’s difficult to project which factor will win the race in the years to come. It’s also hard to know if or when the finance will dry up, with the availability and terms depending on perception and attitude as well as the facts, and on the decisions of ratings agencies. Finance has obviously dried up for some operators.

The arguments I’ve seen over profitability were polarized, and you can spin the same facts to say there’s rapidly advancing technology, or that fracking depends on uncertain future technological advances.

Some sources claim there’s a vast amount of shale oil and gas elsewhere in the world. If US shale runs out soon, either it will discourage extraction overseas, or America’s frackers will tour the world and there’ll be more total oil and gas extraction, during which technology will advance and expertise will develop, eventually allowing production in the US to be resumed.

Solar PV and energy storage

While there’s some doubt about the future of cheap oil & gas in the US, solar photovoltaic energy has been steadily getting cheaper. The industry has followed Swanson’s Law, which says that the price of solar modules drops 20% for every doubling of the total capacity ever produced. It’s a bit like Moore’s Law (which drives the falling price of computing power), except that Moore’s Law depends on time rather than cumulative production. Installation costs will be stickier.

I checked the US holdings in the solar ETF TAN’s top ten holdings and I give a very sketchy financial summary for the three most profitable ones. I only looked at the latest 10-K and 10-Q, and used Morningstar when needed for a three-years-ago figure.

First Solar Inc (FSLR) (SEC filings) has had positive Net income for two full years and for six months in 2015, and three full years of positive free cash flow but a negative six months operating cash flow in 2015. They have a strong balance sheet with more current assets than total liabilities. The cash flow can vary due to the timing of revenue recognition. They sell solar modules and the solid results are not representative of solar project companies. The stock price looks volatile on the ten year chart.

Advanced Energy Industries Inc (AEIS) (SEC filings) do power conversion products and are also not representative of solar project companies. They have consistently positive free cash flow numbers for three full years (if you don’t count acquisitions) and for six months in 2015, also current assets are more than total liabilities, but they lost so much market share in the solar inverter business in Q1 2015 that they decided to exit it, with layoffs expected.

SunPower Corp (SPWR) (SEC filings) has had two whole years of positive Net income but a negative six months in 2015. Cash from operations has been positive for three whole years, but free cash flow has generally been negative. The balance sheet seems reasonable, but is complicated by many items refering to footnote 1, about related-party balances for transactions made with Total. From the Q2 10-Q, in thousands, Total current assets of $2,172,113 are 1.95 times the Total current liabilities of $1,113,960, and are 80.3% of the Total liabilities of $2,706,213. Long term debt is small at $225,338, and the biggest long term liability is Convertible debt, net of current portion $693,938. Some of the debt is non-recourse, with no claim on the company beyond the assets used as collateral. SunPower are engaged in solar projects. From Morningstar’s company profile, you’d think SPWR maintained the solar system (a big ask).

Solar projects need finance for development, the official lifetime is about 20 years and they are likely to last longer, so high debt is expected when they are retained. In contrast, fracked wells deplete quickly so high debt seems more likely to be a problem. The equipment is not likely to last anything like twenty years (and at least some of it is now a lot cheaper, probably the older stuff). Debt has been incurred in both industries to finance fast growth (or to stay afloat, as alleged for fracking).

The ten year chart of the ETF TAN probably overstates the volatility of major US solar stocks because it includes the OTC stock Hanergy Energy, see “TAN Gave It’s Investors A Sunburn – Consider ICLN Instead” by Bruce Vanderveen, May 29, 2015 (seekingalpha.com) and the comments.

Casual Analyst on Seeking Alpha has questioned the value of retained projects (April 2014, seekingalpha.com), given the falling costs, doubting that the long term Power Purchase Agreements will hold. I disclosed at the top that I’m long SUNE, but it’s speculative and I’m not recommending it. What matters here is, I don’t see valuation or other issues stopping solar PV from growing and contributing to low energy costs. The implications of negative free cash flow, financing etc. can be argued over, as can the retained projects, but solar PV does not have shale’s depletion issue.

There is also high interest in energy storage, mainly battery technology, with Elon Musk (CEO of Tesla Motors and SpaceX, and chairman of SolarCity) betting on conventional Lithium-ion technology for his Gigafactory (and he’s betting against “over-hyped” breakthroughs). The Gigafactory should reduce costs through economies of scale.

At $3,000 to $7,140, Tesla’s Powerwall for domestic energy storage is currently not cost-effective compared to a net metering plan, but otherwise would usually break even on a “levelized cost of electricity” basis (which includes things like installation costs). That’s according to “Tesla Battery Economics: On the Path to Disruption” by Ramez Naam, April 30, 2015 (rameznaam.com) (probably based on a $3500 price). That’s without considering the benefit of keeping the lights on in a power outage. Tesla also makes the Powerpack for industrial customers. Mercedes-Benz are starting to sell batteries for the same purpose, for domestic and business use, see “Move over Tesla: Mercedes-Benz takes on Powerwall with a home battery of its own” by Ellie Zolfagharifard, June 10, 2015 (dailymail.co.uk). You get more power per up-front dollar from a home generator, but you have to pay for gas, and other factors complicate a comparison.

Ramez Naam also wrote “Why Energy Storage is About to Get Big – and Cheap” April 14, 2015 (rameznaam.com), where the author makes a case for expecting energy storage to soon be more cost effective than activating gas peaker power plants to satisfy peak demand (that doesn’t sound good for General Electric, who sell gas turbines).

About a year ago I heard a technologist on the radio complaining that nearly all the R&D in battery technology was incremental, i.e. likely to achieve it’s goal but with little chance of achieving a breakthrough. I’ve just googled “new battery technology” (without quotes) and got 17 relevant results dated 2015, on a page of 20 results. The variation includes quantum dots and solid state lithium-ion. Of course a quick search doesn’t measure the research budget.

Electric vehicles will be a lot more popular when you can charge them up with low cost solar power and go further with cheaper, lighter, more energy-dense batteries. Households can use the grid less or maybe not at all when they can have solar PV panels on the roof and store the energy. There are complications, such as climate and seasons, smog in some of the world’s cities, the relative merits of thermal (non-PV) solar, politics, and the effect on the grid, power utilities, and people who don’t own rooftops. The advantage of solar PV and local energy storage is bigger in poorer countries where the grid is less developed, and it can be leapfrogged for a system where energy generation and storage is close to energy consumption. A distributed system based on renewable energy avoids much of the capital investment needed for a grid, most of the power transmission and distribution losses (6% in the US), and the purchase and transportation of fuel. Obviously there are still capital, maintenance and replacement costs for the generation and storage of energy.

Even if fracking fails through a lack of profit, finance or resources, renewables (mainly solar PV) and battery technology will mean low cost energy, which is positive for discretionary income. Countries with the right kind of desert (sunshine, location, and not too many sand storms) could become energy producers. Saudi Arabia plans on exporting its sunshine, see “Saudi Arabia solar power exports ‘absolutely realistic’” by Ed King, May 31, 2015 (rtcc.org). Countries like Libya and Iraq could see their economies transformed if they can improve their politics, but that’s quite a big “if”, and there may be more chance of Spain or Greece making good use of their solar PV potential. There are many more countries with high insolation (sunshine), I’ll just mention Mexico where solar PV could power manufacturing for export to the US. While wide area transmission grids already span countries and continents, electricity transmission over long distances is not cost free, and security is likely to add to the cost in the Middle East. Wherever it works, there will be customers benefiting from cheaper power.

A bit off-topic, where ground water is scarce but humidity is high (Californian fog?), it may be possible for solar powered refrigeration to condense water out of the atmosphere. See this piece about a Dyson Award winning invention for filling a bicycle water bottle. Obviously that would need to be scaled up massively to have an economic impact, it might have to wait for solar power to get cheap enough, and water would have to be transported inland from the coast.

Electric vehicles and vertical integration

Electric vehicles have fewer moving parts, with savings in servicing and maintenance, see “Electric vehicles offer big SMR cost savings” by Andrew Ryan, August 12, 2015 (fleetnews.co.uk). Fewer parts could result in lower production costs than for gas burners if the battery cost can be brought down. Fewer parts also makes vertical integration easier. The quality-control downside of outsourcing is described in “Why Tesla’s Vertical Manufacturing Move Could Prove Essential To Its Success” by Jim Gorzelany, Feb 27, 2014 (forbes.com).


85% of the wolrd’s generating capacity is still steam powered, but advances are possible even in this old area. The title “Breakthrough in plant efficiency” (September 10, 2015, theengineer.co.uk) may be an exaggeration at this stage, but watching the behavior of vapor bubbles with high speed cameras has increased understanding of the boiling process, which may allow safe operation at higher temperatures and greater efficiency. The problem with putting more heat into the same boiler is that bubbles prevent the efficient transfer of heat. Too many bubbles leads to some very hot hotspots that damage metal in a “boiling crisis”. One conclusion is that surfaces in the boiler should be rough, but not as rough as suggested by previous research into heat disappation in electronic devices.

Commodities, currencies and China

The low price of commodities generally should also be positive for higher discretionary income. Commodities are priced in dollars and exchange rates affect prices (and vice versa). I’ve been saying since November 2014 that there’s a risk that $6 to $9 trillion of offshore dollar denominated debt could trigger a flight to the safety of the dollar (and to other currencies percieved as safe). One possibility is that the debtors are mostly businesses that earn dollars, so the currency of the debt matches the currency they earn, with little currency risk. Countries with currencies that fall faster than world prices won’t experience the “low cost world” of the title, if they import enough to create inflation.

The US has long been a net importer, allowing exporting countries to build up high dollar reserves, especially China with $3.8 tillion worth. Russia had high currency reserves, some of which they used to try to prop up the ruble, but the market had it’s way. However, Russia is highly dependent on oil and faced sanctions, making it different to other emerging economies. China could trash the dollar by liquidating part of its dollar reserves, but that would disrupt trade and reduce the value of its reserves. In contrast, a market driven flight of capital is not a plan based on the consequences. That’s why a flight to the dollar seems more likely than a flight from the dollar. It isn’t impossible for traders to get nervous about the US trade deficit and low reserves, but there’s no obvious reason for them to suddenly get excited over a situation that’s lasted for decades. That may change if China successfuly challenges or undermines the US dollar’s status as a reserve currency, in which case a lower dollar is likely to help the US trade balance. Wikipedia’s List of countries by foreign-exchange reserves puts the US at number 18 with $121 billion, after the UK, France and Italy.

Rising wages in China don’t favor low cost production and low cost imports from China, but a big effect on import costs seems unlikely. See “A tightening grip” March 14, 2015 (economist.com), it’s a little old but it’s about long term issues. The piece gives the double digit growth per year in manufacturing wage rates since 2001, and the rise in productivity which isn’t far behind. The concentration of various manufacturers in one area keeps the supply chains short (literally), with what I’ll call “economies of concentration”. It’s an excellent piece for anyone interested in China.

China is a big and growing market for IPG Photonics’ lasers, and there’s some risk that China will push the development of their own lasers. I mentioned it in “IPG Photonics cutting edge”, May 2015 (on Seeking Alpha, on blog site, find “Vertical integration in China”). I quote a claim that the Chinese government aims to achieve the level of vertical integration that IPG Photonics has. The laser sales are in line with China’s manufacturing becoming more sophisticated. That shouldn’t be surprising – the first Korean car’s body was made from oil drums (see “Meet the first ever Korean car” topgear.com), and their industry is rather more sophisticated now. I doubt if rising wages in Korea caused inflation in the countries they exported to. Although China is bigger, industries like clothing where low wages matter most will move to the low wage countries.

For the long term trend in commodities, see “COMMODITIES & THE 130+ YEAR BEAR MARKET” by Cullen Roche, 12/16/2010 (pragcap.com). The author used “euphoria” in describing commodity markets at the end of 2010, and warned that financial products wrapped around commodities were speculative. You need Flash for the images, if you don’t have it there’s a chart for non-oil commodities on “Commodity prices: Over a hundred years of booms and busts” by Andrew Powell, 28 April 2015 (voxeu.org) showing an overall decline and much boom and bust.

A more mixed price history is described here – “Commodity prices in the (very) long run Mar 12th 2013 (economist.com). Since 1900, the risers have been energy products, precious metals, and steel and its ‘ingredients’. Aluminum and zinc fell. The piece draws on a study of thirty commodities by David Jacks. There are charts and tables in the 39 page PDF “From Boom to Bust: A Typology of Real Commodity Prices in the Long Run” by David S. Jacks (Simon Fraser University and NBER), May 2014 (nber.org).

For innovation in mining, see “Rio Tinto talks up autonomous trucks, innovation cred” by Allie Coyne, Jan 13, 2015 (itnews.com.au), which describes how the miner operates mines from a remote center. A tire blowout on a huge mining truck is very expensive. Tires can be monitored with surface acoustic waves, allowing replacement before a blowout, without the inconvenience of wires or batteries on a rotating wheel or shaft. This link is mostly about checking for under-inflated tires in light vehicles – “Acoustic Wave Sensor Market worth $1,183.52 Million by 2020 (marketsandmarkets.com).

Recycling reduces the need for mining, and when a metal or other resource gets too expensive, substitutes can be found. Beyond occasional switching to palladium instead of platinum in catalytic convertors, I can’t think of good examples of price substitution, probably because high prices soon stimulate production. We have learned to live without asbestos, toxic metals like cobalt have been dropped from many products, and many soft plastics are now phthalate-free. Nanotechnology is giving us new materials like carbon nanotubes, graphene and metamaterials. There’s some effort to develop better construction materials, after years of little change, but the only good story I could find is from 2009 – “Better Materials Could Build a Green Construction Industry” by Mark Fischetti (scientificamerican.com). As suggested in a comment, building codes that don’t move with the times could be a bottleneck. I expect it will be a long time before we have to start mining asteroids, as in “Single asteroid worth £60 trillion if it was mined – as much as world earns in a year” (dailymail.co.uk). That much metal would crash prices and not sell for £60 trillion ($92 trillion).

One innovation in recycling is labels that come off PET plastic bottles easier, so they don’t contaminate the plastic during recycling. For a range of recycling machinery see edgeinnovate.com. A “trommal screen” is a rotating drum with holes that only small bits drop through, and the smallest bits are called “trommel fines”.

Soft commodities

The Jacks study concluded that soft (agricultural) commodities have been in long term decline since 1850. Beef was an exception, having gone up over the period. Given the limited area of fertile land, innovation must have enabled production to rise faster than the population. Major factors include the use of farm machinery instead of horses, oxen and manual labor, and the “green revolution” (pesticides, herbicides and crop breeds). New productivity drivers are arriving.

Satellite images have been available to farmers, sometimes for free. This PDF from fsa.usda.gov (April 2014) proves it for the US, but it’s a bit dull for non-farmers. The piece from digitalglobeblog.com about Africa and South Asia is more readable but has nothing surprising.

Now drones are available, although their use is restricted. With the right drone and an app, a survey can incorporate images using infra red radiation as well as the visible spectrum. The relatively ‘pinpoint’ information can be used to apply whatever’s needed to the patch that needs it, saving on insecticides, herbicides, fertilizer and water. Using drones for crop spraying avoids the cost of either employing a pilot or taking flying lessons, and avoiding the pilot’s weight saves on fuel and the material used in the drone, though opposition to permitting it is likely. For more detail see “Meet the New Drone That Could Be a Farmer’s Best Friend” by Rachel Rohr, January 21, 2014 (modernfarmer.com).

The 7 Best Agricultural Drones on the Market Today” by Andrew Amato, 2014 (dronelife.com) has the “Lancaster” drone from PrecisionHawk at number 3, with LIDAR (a sort of laser radar) among other features. There’ll be more drones on the farm as restrictions ease – “Why 2015 is the year agriculture drones take off” by Clay Dillow, May 18, 2015 (fortune.com).

My amateur guess is that Precision Agriculture (as it’s sometimes called) is at an early stage, and someone needs to integrate everything, bringing drones, artificial intelligence, agricultural equipment etc. together to make the whole operation smooth for farmers, who want to sit in the tractor and not spend time grappling with the IT (until someday the tractor doesn’t need them).

Instead of automating work in the fields, agriculture can be brought into the factory, as in “Japanese plant experts produce 10,000 lettuce heads a day in LED-lit indoor farm“. It’s in a former semiconductor factory, and looks like a mutant hybrid of that and a greenhouse. Apparently it’s not advanced enough for the Japanese, see “Fully Automated Lettuce Factory to Open in Japan” August 21, 2015 (wsj.com). In one sense it’s an inefficient use of energy, because if the electricity used to power the LED lights was generated by solar PV, there’s a huge energy loss between the sunlight falling on the panels and the light coming out of the LEDs, and the conversion to electricity and back to light can never be 100% efficient. On the other hand, the panels can capture frequencies (i.e. colors) that plants don’t use, and the LEDs can be designed for the frequencies they do use. Solar panels can be on non-fertile land or unused areas like rooftops. Solar PV converts energy more efficiently than photosynthesis (Solar PV, 10% to 20%, Photosynthetic efficiency 0.1% to 8%), but photosynthesis is present in both cases.

Fish farms are a fast-growing source of food, and aquaculture is probably more sustainable than using fishing fleets. Unfortunately the farms are plagued by sea lice, and pesticides from fish farms cause toxic pollution. One idea is to put a farm further out in the sea where currents are stronger and it’s harder for the infestation to spread. Most of this PDF – “Landscape/seascape capacity for aquaculture: Outer Hebrides pilot study” is dull reading if you aren’t a Hebridean fish farmer, but the paragraph with “up to three times the size of the average fish farm being located further out to sea in deeper water locations that have stronger currents, which helps reduce potential adverse environmental effects” proves I’m not making it up. See Wikipedia’s “Aquaculture of salmonids” for more info and an old chart showing the rapid growth of production.

Good long term returns in primary producer countries

There’s a table in “World stock markets: How historic returns have varied by country” February 19, 2014 (monevator.com) which shows that the annualized real return with dividends reinvested since 1900 for Australian and South African stocks is over 7%, beating US stocks where the return was 0.9% less. Both countries are primary producers, which suggests that either commodity stocks outperformed, or the sector was good for other businesses, contrary to the effect of resource exports making industry uncompetitive through a higher exchange rate (part of the ‘resource curse’). If commodities prices have been in long term decline, then the returns from Australia and South Africa suggest that commodities have experienced good deflation, with lower costs and higher ouput more than compensating for lower prices. You could say they were “emerging markets” in 1900, but not in the sense where low wages encourage manufacturing for export in labor intensive industries. In the absence of finding more data, there’s a complication – valuation affects total returns, so the high total return does not necessarily reflect the performance of the businesses. I cover the issue with hypothetical cases in “The effect of valuation on long term total return” September 20, 2015 (wordpress.com).

Inflationary emergencies

The green revolution increased crop yields at the expense of genetic diversity, creating vulnerability. It’s also been claimed to have reduced variety in the diet, risking malnutrition, for example with less millet being eaten in India.

The most popular banana in the 1950s, Gros Michel, is no longer available after being hit by Panama Disease. The replacement, Cavendish, would not be easy to replace because few of the varieties of banana survive shipping in good condition. I’d guess a banana crisis could be solved by genetic modification (GM).

Nearly all grape vines have European tops grafted onto American roots, due to the insect phylloxera which attacks the roots of the original European vines. See “DIY: Grafting Grapes by Maria Finn” October 11, 2012 (gardenista.com).

If anything happens to wheat, a GM fix won’t be easy due to its complex “hexaploid” genome, though I’m not saying a fix is impossible (if Monsanto did the fix, don’t expect the IP to be universally respected). See “Bread wheat’s large and complex genome is revealed“. The grafting solution used for grapes seems impractical for wheat. The vast “monoculture” wheat fields that allow efficient production can make the crop vulnerable to weeds, pests and disease, and monocropping (the same crop in the same field every year) increases the risk. Innovation might mitigate the problem, for example in tomato greenhouses they’ll release parasitic wasps if they have a whitefly problem.

Climate change or just long droughts also threaten the food supply.

The immediate financial result of food shortages is food price inflation. So far that’s only been a major problem in emerging economies, fixed by the passing of bad weather or by higher prices encouraging more production.

In general, I think high inflation is likely to be mostly the result of temporary shortages rather than endemic, unless central bankers go nuts. It wasn’t always independent central bankers who dominated economic policy, and it’s possible that governments could go crazy and cause high inflation.

Although I focused on agriculture, natural disasters could affect supply chains in other industries. That has happened for hardware depending on Japanese semiconductors after a quake (WSJ 2011), but if it caused any inflation I didn’t notice it.

Orbital Insight and AI

Orbital Insight use AI (artificial intelligence) to derive market sensitive information from satellite images. They cover retail, real estate, agriculture and commodities. They can measure the height of a building by the length of the shadow and the time of day, which allows them to see how construction is progessing. Apparantely the same technique allows them to see how full a big oil tank is. There are thousands of the tanks across the world and no-one collates the data on how full they are. Now Orbital Insight will sell you the information. It’s a young industry and not proven, but I expect the demand is there. Orbital Insight don’t say much on their site, they want customers to make contact.

See “A stunning new look from space at nature, North Korea and Chipotle” by Ana Swanson, July 23 (washingtonpost.com). It’s long but with plenty of images which are worth seeing. Other earth-monitoring firms are mentioned. It implies that the cost of a satellite is about 1/500 of what it used to be, but without a start date. Sometimes human ingenuity takes a rest – I googled “falling cost of satellites” (without quotes) and the results were like this – “Duck! Falling Satellite Arrives on Sunday”.

If Orbital Insight helps stock traders, then traders who can’t afford the service will be at a disadvantage and they need to know about it. I don’t expect long term investors to be affected. The service should also allow large enough businesses to operate more efficiently, where the data is relevant, contributing to the “low cost world”.

The automatic identification of objects in images is possible due to a recent approach in artificial intelligence called “deep learning”. The advance followed decades of disappointing progress, leaving Sci-Fi fans frustrated by the lack of killer robots. This – “Everything you need to know about deep learning and neural networks” by Margi Murphy, Aug 19, 2015 (techworld.com), is about as simple an explanation as you’re likely to see, with key terms explained at the bottom. Obviously the implications of deep learning go well beyond Orbital Insight’s service.

Efficient air travel

In my piece “Air Travel – an overview” (on Seeking Alpha, on blog site) I described “free routing”, the ability to put planes on more direct routes than currently. That includes steadier ascent and descent than today’s stepping up and down between pre-set heights, with level flight in between, which is not fuel-efficient. However it requires government investment and the US program has problems and political opposition. I haven’t seen such bad news about the European version. I also wrote about how satellite technology will identify the position of planes more accurately, allowing them to fly closer without compromising safety.

The satellite systems depend on cooperation, for example when Russian jets fly close to airliners in the Baltic, they don’t transmit ID or position information. Radar has the advantage that cooperation is not necessary, although military aircraft can be designed for stealth (the Russian jets in the Baltic show up on radar).

A new kind of radar can tell planes from wind turbines, and monitor drones. The UK firm Aveillant realized that instead of electronicaly filtering out excess information, digital technology meant it could all be processed (with chips from Nvidia). They call it “holographic radar” (and sometimes slip “3D” in front). A 40nm version was trialled in July.

In my Air Travel piece I covered tensions in Asia that could affect the future of air travel, and interrupt the progress driven by human ingenuity. Find “territorial disputes in Asia”.

The Internet of Things and GE’s Predix

Moore’s law continues to drive efficiencies, and now the Internet of Things (IoT) could start driving them. IoT is about everything from cars to thermostats communicating through the internet without human intervention, allowing data to be analyzed in the cloud and acted on. If that allows the failure of a machine part to be anticipated (for example), it can be replaced at a convenient time, rather than risk failure at an inconvenient time.

For RFID Pallet Tracking, a pallet fitted with an RFID (radio frequency ID device), GPS (global positioning system) and Bluetooth (low power, short range wireless digital communication), reports where it is and what’s on it (presumably with a list matching inventory to pallet ID). When a pallet load leaves a warehouse, the event is reported to the cloud within half a minute, where the information is available to the client or the client’s software. DHL and Cisco claim that IoT will be worth nearly $2 tillion to supply chains (through avoiding under stocking or over stocking, I suppose).

For an intro to IoT see “The Internet of Things Is Far Bigger Than Anyone Realizes” Part 1 and Part 2. For a big problem, see “The Internet of Things Is Wildly Insecure — And Often Unpatchable” (all on wired.com, 2014).

General Electric seem well positioned for their IoT platform becoming widely used, see “5 things you need to know about General Electric’s IoT cloud service: Is GE’s Predix the platform for industrial IOT?” by Matthew Finnegan, August 5, 2015 (computerworlduk.com), and “5 Reasons Why General Electric Has Much Bigger IoT Potential Than Intel Does” by Anchorite, Sep. 25, 2015 (seekingalpha.com) (I don’t mean to endorse the title). As a large industrial conglomerate, GE are in a position to “eat their own cooking”, and sell products ready to connect to the cloud. The platform’s claimed security will help adoption, and the actual security will be crucial for adoption in the long term. As I’m not a lawyer or security expert I have to assume there’s a risk of liability for big security failures, and my GE holding is small.

They aren’t doing it all themselves – see “Intel & Cisco Developing “Predix-Ready” Devices for the Industrial Internet” October 9, 2014 (genewsroom.com).

In “GE Predix Article : The Industrial Cloud is Here” September 10, 2015 (gereportsasean.com), under “What Can Efficiency Save Us?”, the first two paragraphs start with “Imagine”. There are no concrete examples of cost savings. I expect savings are likely (if the security holds up), but there’ll be a learning curve, and I can’t say when savings will become apparent. They’ll come earlier if Predix really improves safety. For anyone new to IoT, I don’t mean to imply that it all currently revolves around GE and Predix.

While Moore’s Law reduces the cost of computing and creates efficiency generally, increasing computing power enables medical advances, for example through modeling or easier gene-sequencing, and while I’m not saying that’s bad for the economy, if you don’t class medical spending as discretionary, expensive new treatements are not good for discretionary income.

US Medical inflation

CNBC highlight record medical inflation in April, which could be just a monthly blip. They also suggest that medical inflation could return to the double digit percentages that prevailed before the financial crisis and recession.

According to ycharts.com, the inflation rate has been less than half the long term average (which is under 6% pa) for about two years (on average, with some months higher than others).

An infographic from accounting firm PWC projects an over 6% rise in health spending in 2016. That’s about level with 2014 and 2015, after a decline from double digit growth in 2007 which is attributed to inflation (rather than more spending at constant prices).

The high cost of US healthcare may be indicated by this – “Americans Are Going to Juarez for Cheap Dental Care“, where the charges in Mexico are 60% to 70% lower. Labor and rent are cheaper in Mexico, while the US requirement for malpractice insurance and the longer time taken to train dentists are regulatory factors that affect the cost.

I expect that not all costs are likely to have downwards pressure from the factors I’ve mentioned, but that isn’t necessarily true for all medical costs. Maybe the cure for high prices is high prices. A long time ago I read a claim that the biggest problem in automating diagnosis is patient acceptance. The problem will diminish as medical costs rise, “AI doctors” improve (AI is artificial intelligence), and generations who grew up with smartphones are more used to software giving answers. See “AI found better than doctors at diagnosing, treating patients” Feb 12, 2013 (computerworld.com).

Smartphones can now be used to monitor vital signs and for diagnosis, see “The Future of Medicine Is in Your Smartphone” by ad health, January 19, 2015 (advertisinghealth.co.uk). I recommend finding “I thought I’d seen it all”. A smartphone and sensor can replace a sensor, dedicated computer, display and software, all with a big markup. The gross margins are high, see “Medical equipment and supplies gross margins” where only two out of fourteen margins in the linked list are under 50%, and two (for Medtronic and Zimmer Holdings) are over 70% (all based on the latest quarter). A 70% gross margin is the result of a 233% markup on cost.

In the news recently, Turing Pharmaceuticals raised the price of a HIV drug by 5,000%, prompting a tweet from Hillary Clinton which has been blamed for sending pharma stocks down. Her proposals are in “Hillary Clinton’s Drug Price Plan Would Make The Problem Worse — But That Doesn’t Mean We Should Do Nothing” by Avik Roy, Sept 22, 2015 (forbes.com). The proposal to cap the cost of prescriptions that insurers can pass on to patients, at $250 a month, probably excludes cures. If it didn’t, a single pill that cured a disease would cost a patient no more than $250 in total, while a regimen that lasted for the rest of a patient’s life could cost a patient $3,000 a year times the years of life left. A patient paying the maximum would have to pay much more for the inferior treatment. However, theguardian.com says the cap is for serious or chronic conditions. I expect that means serious AND chronic (because journalists don’t understand logic the way a programmer does), so the cap won’t apply to any condition that is not chronic (i.e. it won’t apply to cures). Googling “what is a chronic disease” confirms that it’s controllable but not curable. Another possible objection to the cap is that there are rare diseases where the R&D cost can’t be spread over as many patients, so the cost per patient will usually be higher.

The average press report has less factual info than the Forbes and Guardian links (assuming they’re right), for example “Clinton takes on ‘profiteering’ drug companies” September 23, 2015 (cnn.com), which does not mention shortening the patent life of new drugs, or the “serious or chronic” qualification to the cap. All the press reports look different to the ‘factsheet’ “Hillary Clinton’s Plan for Lowering Out-of-Pocket Health Care Costs” (hillaryclinton.com), for example she proposes to limit insurance premium increases. Hillary Clinton’s twitter feed is here, but there are many tweets, mostly on other topics. I’m not the first to say that from all the fuss, you’d think she was president.

US deflation stopped around 1950

While I see reasons for costs to fall, it’s a little hard to believe that consumer prices excluding energy will actually fall, at least generally and over several years, simply because that’s been rare (outside of Japan) in the recent era (fiat currencies, policies labelled “Keynesian”, quantitative easing). Wikipedia charts US inflation since about 1655, and there used to be as much deflation as inflation, while there’s only a tiny speck of deflation after 1950. Also, central banks seem to dislike deflation more than ever. Even so, the effect of solar panels on consumers’ own rooftops is likely to be significant by itself, with an average power bill of over $1,320 per year per residence for 2013 (find “Residential average monthly bill” on eia.gov for a PDF).

The many years where prices fell before 1950 did not stop the US from growing to become a superpower.

US price indexes recently

Ycharts give a five year view, which shows inflation dropping to around 0% in 2015. The Consumer Price Index showed negative inflation of -0.1% in August, after seasonal adjustment from +0.2% (from the Bureau of Labor Statistics). There’s also a table with annual inflation figures and monthly CPI figures on “usinflationcalculator.com“.

According to Wikipedia, the CPI and PCEPI inflation indexes are fairly close when averaged over several years, implying divergence when they are not averaged.

More technological advances

Additive manufacturing (which includes 3D printing) offers efficiencies, such as lighter aero engines, and it could reduce the part count in many manufactured items (even if the Star Trek replicator is still a long way off). A reduced parts count means shorter supply chains and less trucking, while rail is used for bulky commodity-like material and is not likely to be very affected. The points are covered in “2015 Commercial Transportation Trends” (strategyand.pwc.com).

When an app can summon a driverless cab, car ownership will become unnecessary for many people. Cars will still be owned for status or because some people just like to own and drive them, and sales of high-performance cars won’t be affected much. The level of ownership could depend on details like how clean the driverless cabs are kept. Like other disruptive innovation, there will be losers, in this case auto makers as there’ll be fewer cars with more use per car. Car park owners will be hit. Faced with falling sales, auto makers could design cheap and attractive “built to own” cars. I think there could be congestion if driverless cars are made to drive slowly in circles to avoid parking charges, but the author of “How Driverless Cars Could Turn Parking Lots into City Parks” believes the opposite, and ran a simulation which supports his view.

There will be an overall benefit through more income to spend on other things, but with temporary disruption to the economy, as happened for oil – “Itemizing The Oil Bust: 75,000 Layoffs And Counting“. Driverless taxis are a long way off, but meanwhile there’s Uber and car sharing companies (acquired by car rental companies) which make it easier to avoid car ownership, at least on average due to more mileage per car per day or more passengers per car.

There are smaller advances, for example IPG Photonics have developed a laser-based paint-stripping system for aircraft which saves labor and avoids using toxic chemicals. That’s just one example and there must be many advances I don’t know about.

For about two months worth of recent advances in bio-tech, see Scientific American’s page. I like “We Transformed Living Pig Cells into Tiny Lasers“. It might sound frivolous, but they can use the technique to uniquely tag a trillion cells, which should help research. Whether it will lead to expensive treatments or contribute to the ‘low cost world’, I can’t say.

One development I’m really not sure about is “Massive open online courses“. Some of the courses are free, and some courses have required text books (sometimes expensive and written by the instructor, it’s been claimed). Wikipedia say the completion rates are under 10%. While MOOCs avoid the expense of conventional eduction, I can’t say how effective they are or will be.

Technology or innovation?

I’ve used the term “technology” rather than “innovation” because innovation depends on technology and can be driven by it. For example the personal computer was one of the innovations produced by Moore’s Law – if you keep making smaller cheaper components, eventually someone will think of making smaller cheaper computers for personal use. I expect fans of disruptive innovation theory will disagree, and it’s not a point I feel strongly about.

Continuing technological advance

Without technology advancing we would still have a stone age economy. I’m not claiming the advance of technology is a new phenomenom, only that it will continue strongly, with a positive effect on discretionary income. Investors’ perception of the effect of technology will swing between under-appreciation and hype. Electrification, radio and prohibition were confidently expected to lead to a “new plateau of prosperity” in the late 1920s. The promised prosperity was real (though not due to prohibition), but it didn’t kick in until after the Wall Street Crash of 1929 and the Great Depression that followed. There’s some similarity to the Dotcom bubble that peaked in 2000, although there was more central bank intervention after the Dotcom bust (find “11 interest-rate cuts” on mises.org, which also lists intervention after previous crises). Recession/depression was avoided/postponed (depending on your point of view). The hype has come true, for example with old media businesses suffering while Apple is the second biggest company by profit (according to “What Are the Biggest Companies in the World?” by Matt DiLallo, Sep 10, 2015 (fool.com)). Currently there are tech stocks and other stocks on high valuations but I’m not seeing the hype or wild optimism that goes with a bubble.

Find where the benefit sticks

When ingenuity produces efficiency, where businesses don’t have a competitive advantage, the benefit will flow either to consumers or to a business in the chain which does have a competitive advantage. Those businesses where the benefit sticks are worth considering for investment, but finding them isn’t easy, for example chemicals companies use energy and raw materials but Eastman Chemical Co. (EMN) doesn’t have much pricing power, even after moving towards specialty chemicals for several years to get better margins (you can confirm the general lack of pricing power by checking the 25 instances of “price” in the Q2 2015 Earnings Call transcript). Another strategy is to invest in businesses likely to benefit from the increased discretionary income that would result from efficiencies flowing to consumers. There’s a little more about that near the end.

Sometimes an economic benefit sticks with land owners. Here’s a neat but unfortunate case from long ago in London. A politician had campaigned for tolls on footbridges over the river Thames to be abolished, because they were high relative to local wages. When he got his way, landlords put the rent up, leaving the workers no better off (find “Some years ago in London”). That was entirely predictable from an old-but-good theory by Ricardo. Local economic successes are likely to inflate rents and real estate, for example around Silicon Valley.

There’s a direct effect for solar PV and batteries – cost-effective batteries for domestic energy storage make solar modules more useful, which should make roofs more valuable, and therefore make buildings worth more. The “Housing Affordability Index (Composite)” (stlouisfed.org) is currently 151.2. Higher means more affordable, and 100 means a median income family has just enough income to qualify for a mortgage on a median-priced home.

There’s a tendency for speculation to force real estate prices up, setting them up for a crash. At the height of the Japanese bubble, some people valued the grounds of the Imperial Palace at more than all the real estate in California. That didn’t last.

Is deflation bad?

Inflation can usually be stopped when the political will exists, for example when Fed Chairman Paul Volcker got inflation down from 14.8% to under 3% in the 1980s. Some extreme cases are exceptions, such as the inflation of Confederate currency, when there was no monetary solution that could save the economy after the Union blockade of Southern ports. Japan has had decades of deflation while QE and low interest rates have not provided a solution. That probably explains why central bankers prefer to risk inflation rather than deflation if there’s any doubt about which to counteract. If Japan’s deflation is cured, the next worry will be that all the money-printing will cause hyperinflation, which is probably the most feared risk, although it’s not likely to last for decades.

Here’s a simple hypothetical case – if prices fall 10%, then even if income falls 5%, those with the income are about 5% better off in real terms (adjusted for inflation). If consumers use their greater real income to increase real spending, then consumer-facing businesses have more real revenue, although the effect on earnings depends on what happened to margins, and given the assumptions it would be harder for stocks to outperform cash. The greater real revenue should feed back up the supply chain. The fall in prices also increases the real value of net cash and net debt. Although that’s hypothetical, I’ll say that debtors’ pain is more newsworthy than the gain from holding cash, and a continual 10% deflation would be at least as bad for investment as a sustained 10% real interest rate – a high risk-free return is a high bar for investments to clear.

The falling price of computing power over decades has not devastated the IT hardware industry, although it hit the incumbent mainframe makers, many of which merged into Unisys which dwindled until it was kicked out of the S&P 500 index. The success of the non-incumbents is in contrast to the current commodities pain, with commodities giant Glencore PLC’s stock down about 75% from the price at its well-timed IPO in 2011. Differences include the cyclical nature of commodities, with companies over-investing in the good times.

Sometimes economists distinguish between monetary deflation and growth deflation. Monetary deflation is the bad sort and it follows a crisis. Hong Kong in the 1990s and Greece lately have had financial crises and not devalued their currencies – Greece can’t without leaving the Euro, and Hong Kong stuck to a dollar peg. The resulting deflation was and is painful. The Japanese bubble saw stocks and real estate rise. The Yen also rose, but as well as market forces there was an international agreement to let the dollar fall and the Yen rise. After crashing in 1990, deflation might have been worsened by the currency recovering too strongly, but it’s fairly complicated and I’ll refer you to Wikipedia.

Growth deflation is exemplified by the information technology industry, where you regularly get more computing power for about the same price. I would apply the term “growth deflation” to energy, as it’s fracking technology that’s driven lower oil and gas prices (although fracking benefitted from OPEC keeping the oil price high for decades), and for solar PV, the falling cost is due to technology and economies of scale, not monetary events. I see the recent collapse in non-oil commodities as an adjustment to over-supply, and would not call it growth or monetary deflation.

Wikipedia‘s deflation page is nearly all about monetary deflation, but they use the term “credit deflation”.

Hedge your bets

Suppose you find the “low cost world” argument credible and invest in companies which you believe will benefit from increased discretionary spending. Now suppose that the benefit of low cost energy, commodities and imports, and efficiencies generally, is not passed on to consumers. That means the benefit has been retained by businesses. The businesses most likely to keep the benefit are those where the low cost impacts and which have a competitive advantage (or “moat” when it’s enduring). If you can find such a stock at a fair price, it should work as a hedge, compensating if the benefit of low costs does not flow to consumers. Strong consumer brands could benefit from either increased discretionary spending or lower costs, and domestic exposure is preferable but may be hard to find or already priced in.

Demand, debt and excess reserves

In “2015 Credit Card Debt Study: Trends & Insights” by Odysseas Papadimitriou, Sep 9, 2015 (cardhub.com), the author warns of credit card debt reaching a tipping point which would be followed by high delinquencies. However, the chart of card debt per household doesn’t look too scary, with the debt slightly closer to the Q1 2011 low than to the 2008 high. Consumer debt is a concern but IMO the data is not strong evidence that overall demand will collapse or be constrained for a long time.

According to the St Louis Fed‘s chart, the ratio of US household debt to GDP has fallen from over 97.5% in 2009 to about 80% now, with the fall bottoming out in 2015. That’s good but it doesn’t necessarily mean the people in debt have seen their incomes grow faster than their debt, or grow at all.

If consumers borrow to spend, is it good for business, or bad because it’s unsustainable? The PDF “Debt overhang and deleveraging in the US household sector: gauging the impact on consumption” from the European Central Bank implies that “in the literature” there’s disagreement over whether more consumer debt leads to more or less consumption. I believe there’s less ambiguity about the effect of lower prices on discretionary income.

Default rates on student debt have doubled since 2011, mostly due to students at for-profit colleges, see “Default Spike in Student Debt Driven by Unconventional Borrowers” by Jeanna Smialek and Janet Lorin, September 10, 2015 (bloomberg.com). Student loans outstanding total over $1 trillion, with more than seven million debtors in default (Wikipedia).

It’s claimed the world is in the wrong part of The Global Credit Supercycle, and consumers have not reduced their debt. I’m not convinced by the chart showing that banks have low reserves if you remove the “transitory” excess reserves. It’s true that the excess reserves could be released if banks wanted to lend, but they are still reserves until that actually happens.

Excess reserves are reserves in excess of the reserve requirement set by the Fed. QE gave banks more money, but instead of increasing lending (and providing for the required reserve), it mostly went into a massive excess reserve. At around $2.6 trillion, the excess reserve is vastly bigger than the required reserve. It means that in principle banks could increase lending massively and quickly, causing inflation (or instability if the reserve requirement is too low). If that happened, in theory the Fed could counteract it by increasing the reserve requirement or raising the rate it pays on excess reserves. Many people don’t trust the Fed and therefore regard the excess reserves as a big problem. For more detail (and more concern) see “The Problem of Excess Reserves” by A Political Junkie, May 14, 2015 (viableopposition.blogspot.co.uk), or “Excess Reserves at the Fed: Lazy Money or Potential Hazard?” by Jim Allen, CFA, 20 February 2014 (blogs.cfainstitute.org). The comments under the second link cover complications which I left out. If you don’t understand all the technicalities, don’t worry, because the experts can’t always predict the effect of policies anyway. Some known effects were discovered accidentally, such as the effect of the Fed buying bonds, which is to increase excess reserves.

One comment says banks are constrained by the opportunities to lend, so the excess reserves are not likely to pour into lending. I agree with the first part, but the result depends on the prudence of borrowers and banks – when imprudent banks meet imprudent borrowers, the opportunities to lend have no obvious limit, as in the financial crisis of 2007-8. I’m not convinced that today’s excess reserves make the old problem much worse, but that could be a minority opinion. The next big financial bust could be bigger due to other factors, such as complicated but inadequate regulation, continued bail-outs by the Fed, and low interest rates for a long time.

Freight and sales

Freight indexes and other freight-related signs are looking bad, especially in Asia, see “Don’t Ignore the Big, Fat Transportation Warning Sign” by Tony Sagami, September 22, 2015 (mauldineconomics.com). One index not included is the U.S. Industrial Transportation Index, which has declined over the past year, but a longer view shows it’s off a peak at the end of 2014, which is not as bad as plunging to new lows. While the signs aren’t good for the short term, in the long run it’s the effect of lower costs on discretionary income versus the long term downward pressure on incomes (next section) that matters. It’s also possible that the U.S. Industrial Transportation Index will tick back up and the recent slide will look like a correction. The bad news about the cancellation of some Asian plane orders in Tony Sagami’s piece has since been offset by this – “Boeing wins $38 billion in orders, commitments from China“. However, an undisclosed amount of orders are not new, instead it’s claimed that buyers had not been found, and now they have. Boeing’s stock fell 1.5% on the news.

Ycharts show five years of US retail sales, with growth that slowed in the past year or two. The monthly figures are revised in later months. July’s double dose of disappointment was the weakness of the June numbers and the downwards revision of May, which had looked good in June. Then August brought joy with good figures for July and revisions-up for May and June. The revisions mean the figures are not reliable for the short term. See “June Retail Sales Drop Unexpectedly, May Sales Revised Lower” by Andrew Soergel (usnews.com) and “US retail sales rise in July, June revised higher” (ft.com).

When May’s retail sales figures came out, economist Joel Naroff thought consumers had been hanging on to the cash they saved from cheaper gasoline, and had started to spend it. That now looks like a premature call on the effect feeding through, and the “big thaw” explanation for May looks credible. See “Shoppers Splurge as Economy Heats Up” by Eric Morath (wsj.com).

The Bureau of Economic Analysis has various tables, including “Table 2.3.3. Real Personal Consumption Expenditures by Major Type of Product, Quantity Indexes” with ten quarters of stats broken down by sector. The total Personal consumption expenditures (PCE) index has moved up from 106.816 in Q1 2013 to 113.400 in Q2 2015, which I calculate is 2.7% pa.

Downward pressure on incomes

Possibly a bigger problem than debt is that some of the factors which reduce costs also keep wages down, reducing discretionary income as a result. See “Corporations Plan for Post-Middle-Class America” by Bernard Starr, 04/06/2012 (huffingtonpost.com). I found that one by searching for “hourglass economy”, a term that does not seem to have been used much since 2012. The thin hourglass waist translates to a small middle income population. For the “gig economy”, here’s a non-believer – “Proof of a ‘Gig Economy’ Revolution Is Hard to Find” by Josh Zumbrun and Anna Louie Sussman, July 26, 2015 (wsj.com). Maybe it’s just too soon to find evidence. In principle, anything that makes employment more flexible and helps the lowest qualified bidder to get the work (subject to the minimum wage), is good for efficiency but bad for wages and consumer confidence (at least immediately – there’s room for argument about the long term effect).

When middle-income jobs are lost, so is the income tax the employees previously paid. Some well paid work seems unlikely to disappear – top management, R&D staff, movie stars etc., so a big chunk of income tax shouldn’t vanish, but middle-income job losses would still be a problem for either government spending or the deficit. About the very high earners, Wikipedia claims nearly one and a half thousand people with income over a million dollars paid zero tax. If ordinary workers lose income while corporations and CEOs benefit from the savings, there’ll be experts looking into how to minimize the tax paid. Governments could see some savings on healthcare from the use of smartphones and (maybe, but not soon) the “AI doctors” I described earlier.

Zero-hour contracts give an employer control over the hours that employees work and are paid for. The author of “5 Reasons Why Zero-Hour Contracts Are the Future of Work” by Maite Baron, July 21, 2014 (entrepreneur.com), believes the contracts will spread from Europe to the US.

Workers compensation insurance shows how efficiency can lose to other factors, as the effect of increasing safety has been offset by higher awards of compensation. Companies have an incentive to avoid regular employment if doing so reduces the insurance expense. Employees without regular employment are likely to be less confident about spending even if their average wage is not lower. Companies that feel pushed to avoid regular employment could see either costs or benefits as a result.

You’ve probably heard of Thomas Piketty (Wikipedia). He argues that financial returns are greater than growth in GDP so inevitably the rich get richer while the poor get poorer. He’s provoked a vast amount of argument both against and in defense. One apparently simple argument against is that no-one can find a Vanderbilt millionaire, and there used to be 13 Rockefellers on the Forbes list of rich people and now there’s only one. IMO you don’t have to believe that inherited wealth persists and grows to see that checkout clerks who’s jobs are automated away are losing while CEOs’ compensation is rising, although I realize that simplifies a complicated issue, and if I was sure it would reduce discretionary consumption I would not have written so much about the factors that might increase it.

The rise of the machines

I’ll start with two old stories. A worker argued that mechanical excavators should be banned because each one put many laborers out of work. The boss said, think of all the jobs you’d create if instead of spades, everyone had to dig with teaspoons.

Henry Ford (allegedly) told a trade union official, “One day I’ll replace all the workers with machines, and they’ll never go on strike.”. The union official replied, “They won’t buy any cars either.”.

Mechanization and other innovations created prosperity rather than persistently high unemployment during the industrial age, as new jobs made up for the jobs that disappeared through mechanization or other labor efficiencies. I don’t claim that progress was painless, but workers in 1950 were better off than in 1850 or 1750.

The same does not necessarily hold in today’s information age. Government jobs may be safe, and the same may be true for academics, top management, R&D staff, movie stars etc., but outside of the safer areas the workforce can’t all do each others’ nails and hair while R&D work on the robotic salon.

This is an interview with the author of a recent book – “Rise of the Machines: The Future has Lots of Robots, Few Jobs for Humans“.

I don’t have much to say about current employment figures or about the minimum wage. I’m concerned with long term possibilities, and if a job would be automated away sooner at $15 per hour, it seems likely that it would be automated away later at $7.25 per hour.

I’ve used this before – “Starwood Introduces Robotic Butlers At Aloft Hotel In Cupertino” by Jordan Crook, Aug 13, 2014 (techcrunch.com)), and now I can add “New Japanese hotel has robot staff and no room keys” by Stu Robarts, July 22, 2015 (gizmag.com). The hotels demonstrate the technical possibility, and I expect they both have novelty value. Eventually, robot-staffed hotels will be cheaper, but it’s uncertain when. Many guests might be prepared to pay a premium for human service, but that’s also uncertain and likely to change over time.

This piece is mostly about the safest jobs – “The Robots Are Coming. Is Your Job Safe?” by Laura Entis (entrepreneur.com), but if your job isn’t mentioned you can try the BBC’s robo-risk search box.

Anyone who wants to invest in robot makers can see a list on “The Minimum-Wage Fight, Chinese Factories, and the Rise of Robots” by Tony Sagami, September 15, 2015 (mauldineconomics.com). I’ll guess the stocks won’t be cheap. The missing robot maker might be Fanuc.

Going into consumer mode

There’s a possibility related to increased discretionary income – developed countries going into consumer mode, rather like “Chimerica“, where China manufactured and exported and the US consumed, and the Great Moderation, characterized by low inflation and moderate business cycles, until the financial crisis around 2007. The links are to Wikipedia. The second piece mentions various factors that contributed to the Great Moderation, but not the role of cheap manufactured imports from China in keeping inflation low, and the recycling of China’s export earnings into US securities (and presumably into some European assets, so European customers didn’t run out of cash). Going into consumer mode depends on the effect of lower costs outweighing the effect of too much debt and the downward pressure on incomes I described above.

Trade imbalances can persist for a long time, creating instability. The last US trade surplus was in 1975. Since then, the dollar and the trade deficit declined until the 1990s when “hot money” that had flowed into emerging markets flowed back to the US in flights to safety, notably from Asian countries in the 1997 Asian Financial Crisis. Lower Asian currencies revived their exports and the US trade gap. (See “The History of the U.S. Balance of Trade” (about.com).) That could be partly repeating, with a downturn in Asia but not a close repeat of the crisis.

In simple terms, the financial crisis of 2007-8 can be described as the result of banks making bad loans and buying poor quality assets. Complex derivatives such as CDOs of CDOs boiled down to bad loans packaged into poor quality assets, with the poor quality obscured by the complexity. Explanations of the crisis can be pitched to blame banks or the government, depending on political preference. It’s possible to view the failure from an international trade angle. In economics there’s an accounting identity:

Savings – Investment = Exports – Imports

For simplicity I’ve ignored the effect of the government defict. By renaming the left and right sides of the identity it can be written as:

Private sector savings = Trade surplus

A trade deficit implies negative private sector savings, which means investment is funded by overall private sector debt. That’s fine so long as the debt is funding productive investment, but the pressure for new loans to sustain the trade deficit means that the investment may be unproductive, for example a home purchase with a mortgage that the borrower can’t repay. In other words, if the banks had not kept lending enough, then the trade deficit could not have been sustained. The level of lending necessary to sustain the trade deficit was independent of the opportunities to lend prudently, and quite possibly exceeded it.

I hope I’ve got that about right. Although the identity looks simple, the well known economist Paul Krugman has accused the Nobel prize winner Joe Stiglitz of getting it wrong, and at least one of them must have erred (see “More On The Exchange Rate And The Trade Balance” April 7, 2010 (krugman.blogs)). For more about the identity see “The Trade Balance” (arnoldkling.com). It starts with Y = C + I + G + X, where Y is GDP, C is consumption, I is investment, G is government savings (taxes – government spending), and X is the trade surplus.

Exporters also have problems. China is a major market for German exports, and concern about China having a “hard landing” shows in the DAX, the German stock index. When the financial crisis and recession hit the US, it was obviously going to affect the Chinese economy, and they reacted with stimulus, probably too much of it causing poor allocation of capital and corruption. (The Economist finds flaws in a widely reported estimate of $6.8 trillion wasted.)

A trade deficit on the current account (the normal trade of goods and services) has to be balanced by an inflow on the capital account. The inflow may be more sustainable when the owner as well as the capital moves, especially if the flight is from justice or injustice, which means that direct investment back into the country of origin would be risky. London has benefitted from incoming Russian oligarchs and their fortunes, although the term “benefit” would be disputed by people who don’t like the high house prices, which rose partly due to a spill-over effect from oligarchs buying expensive property. For the US, see “Fortunes vary for Chinese fugitives in US” May 18, 2015 (scmp.com).

An economist’s view

The economist John Mauldin wrote “Needed at the Fed: An Inverse Volcker” September 12, 2015 (mauldineconomics.com), which is mostly about “good deflation”. As the title suggests, he sees central bankers as a risk. I can recommend the article. There’s some overlap with this piece, and some of the same examples. I’ve aired my thoughts about the prospect of low costs in comments on Seeking Alpha, rather more briefly than here, before John Mauldin’s piece was published.

“Human ingenuity” is an old term. Warren Buffett has used it more than once, sometimes to explain his preference for investing in businesses rather than commodities. There’s nothing new about saying human ingenuity leads to progress, which includes the increased efficiency I’ve mostly concentrated on.

Previous writing on the issues

I posted some thoughts about lower costs leading to more discretionary income in a comment dated Sptember 5 on “Main Street Capital Survived Far Worse. Why Are You Scared Now?” by ColoradoCapitalManagement, Sep. 3, 2015 (seekingalpha.com) (find “About some macro factors”).

Back in April 2014 I commented “If energy gets cheaper and the savings aren’t pocketed by utilities, people will have more discretionary income”. It’s under the “questioned the value of retained projects” link above. You might have to click “Load all comments”.

I said something about the US economy going into consumer mode in a comment dated Aug 26 under “Is A Stronger U.S. Dollar Really Bad News For Emerging Markets? 3 Myths Debunked” (seekingalpha.com) (find “Chimerica”).

In April 2014 I tried writing about factors affecting solar PV like climate and seasons, the relative merits of thermal (non-PV) solar, politics, and the effect on the grid, power utilities, and people who don’t own rooftops. The result was so long and hard to organize I filed it under “Failed blogs” (here). It included the effect of cheap energy on discretionary income, and a theory (not mine) that oil producers would increase output in an attempt to get their oil reserves out of the ground and sold before the oil price dropped.


Main Street Capital is a Business Development Corporation (BDC), they finance small businesses and I believe they are likely to benefit from increased discretionary income, although many of the businesses they finance don’t sell direct to the consumer, and some are in oil & gas or have O&G exposure (find the “Consolidated Schedule Of Investments” in the latest 10-Q.). I expect the small businesses they finance to have high domestic exposure. One problem with MAIN is that their results benefit from issuing new shares at a premium, and if the premium disappears, so does the benefit from it.

Winmark Corporation run franchises specializing in “gently-used” goods, including Plato’s Closet (clothes) and “Once Upon A Child” (clothes, toys etc.). While they should benefit from increased domestic consumption, they aren’t highly levered to it, and when consumers have less spending power or less confidence they want a low price.

Cash America have a chain of pawn shops, and are concentrating on that after regulatory trouble over non-pawn consumer loans. They’ve said that business drops when the price of gasoline falls, and I’m not expecting the gas price to go up, instead I see the stock as a hedge because many stocks would be hit by a fall in discretionary spending.


A key question for anyone who cares about the long term is, will the workforce be better off as a result of lower costs, or will it be worse off as technology reduces the need for their labor. I don’t let the question dictate all my investments. MAIN, WINA and CSH collectively represent a weak bet on lower costs leading to higher discretionary spending, hedged for depressed incomes.

That’s all, thank you for reading this.

DISCLAIMER: Your investment is your responsibility. It is your responsibility to check all material facts before making an investment decision. All investments involve different degrees of risk. You should be aware of your risk tolerance level and financial situations at all times. Furthermore, you should read all transaction confirmations, monthly, and year-end statements. Read any and all prospectuses carefully before making any investment decisions. You are free at all times to accept or reject all investment recommendations made by the author of this blog. All Advice on this blog is subject to market risk and may result in the entire loss of the reader’s investment. Please understand that any losses are attributed to market forces beyond the control or prediction of the author. As you know, a recommendation, which you are free to accept or reject, is not a guarantee for the successful performance of an investment.

Air travel – an overview

Includes US stocks AVAV, BA, CAR, DIS, GE, HTZ, IRDM, JBLU, LMT, MAR, RTN.
Disclosure: I’m long General Electric Company (GE), AeroVironment, Inc. (AVAV).

I’ve listed the headings on a separate page – Headings in “Air travel – an overview”. It might be useful for navigating.


• Annual growth of 4.1% in the global number of air passengers carried is predicted from 2014 to 2034.

• A variety of trends and factors supports the growth, from more Chinese tourists to the “free routing” of airplanes, but a number of risks could affect it.

• Growth in air travel increases revenue in many industries, but due to competition there are few where the money sticks.

• Out of many companies I looked at, I believe only Disney combines pricing power with reasonable safety.

• I have not researched the stocks in depth, and I hope to learn that more of them are safe and have pricing power.

Businesses that should benefit from increased air travel

JetBlue Airways Corp (JBLU) offer air travel with low cost but with good service and amenities. At least they did, according to “A New Era Has Begun for JetBlue, and Travelers Will Hate It” by Brad Tuttle, Nov. 19, 2014 (time.com). According to a more recent article aimed at investors, the service and amenities are as good as ever, see “SWOT Analysis: JetBlue Airways Corporation” by Iason Dalavagas, June 29, 2015 (valueline.com). But – “JetBlue starts charging for all checked bags” by Carl Surran, SA News Editor, Jun 30 2015 (seekingalpha.com).

The policy of never cancelling flights backfired in 2007 when a snowstorm resulted in passengers waiting in a plane for 11 hours, only to have the flight cancelled. JetBlue are not likely to repeat the mistake, instead I see the incident as a general warning that marketing-friendly policies might be stuck to rigidly until experience teaches otherwise. There’s more company history on Wikipedia.

The Export-Import Bank has closed, meaning overseas airlines won’t get new planes subsidized by the U.S. government (find “Boeing’s Bank” below).

JetBlue look cheap on Morningstar with a forward P/E under 11 and Price/Cash flow under 7. Morningstar also show a high quality of fund ownership, giving five funds 5,5,5,4 and 0 stars out of a maximum of 5. Morningstar’s financial history shows that a double digit return on equity and on Invested Capital was only achieved in 2014 and “ttm”. The high leverage is the likely reason why double digit ROE is possible, given the single digit return on assets, which was below 2% from 2005 to 2012. The Operating Cash Flow has been much higher than operating income, but very high capex has resulted in erratic free cash flow, with big negative figures for 2005 to 2008. Although results have ticked up recently, I have not found a convincing reason why the future should follow the recent uptick rather than earlier financial history. If increased demand allowed JBLU to charge for all checked bags without the number of customers falling, sooner or later supply is likely to catch up with demand. My conclusion based on Morningstar’s figures and little research could be wrong, and the high quality fund owners could be right.

I have never researched airlines much because I believe they are likely to compete away any benefit to the industry. Warren Buffett has described himself as an “airoholic”, and said

    “Think airlines. Here a durable competitive advantage has proven elusive ever since the days of the Wright Brothers. Indeed, if a farsighted capitalist had been present at Kitty Hawk, he would have done his successors a huge favor by shooting Orville down.” (Berkshire Hathaway shareholders’ letter 2007)

Airports are more like unavoidable toll-gates. I found two companies domiciled in Spain with some airport ownership, I don’t like them much but I’ve written about them in “Aena SA – Spanish airport operator” and “Ferrovial SA – Spanish infrastructure company with an Airports division“. There are six other stocks in “Airport stocks list – none in the U.S.” (kitchensinkinvestor.wordpress.com).

Airbnb should benefit if air travel increases, but they are not likely to IPO on an attractive valuation, see “Airbnb raising $1 billion in new funding” by Sara Ashley O’Brien, June 18, 2015 (money.cnn.com). Uber might benefit (and might have an IPO) but they face restrictions, e.g. “Uber Drivers Are Getting Into Trouble at DC Airports” by Eric Hal Schwartz, 02/25/15 (streetwise.co). This – “Ride-hailing firm Uber faces big challenges in China” is about the raids, protests and competition in China, which has not deterred Uber from making China their top priority, pushing growth by paying drivers more than the customer pays. See also “Uber Lands Big Chinese Backer“.

The stock of hotel and lodging company Marriott International Inc (MAR) has been volatile, with a sharp drop from $75.31 in May 2006 to a low of $12.63 in March 2009, then rising to $74.80 on July 3, 2015 with a P/E of nearly 28. The low in 2009 was reached without much competition from Airbnb which was founded in 2008. To be interested I’d need to learn a good reason why there will be no repetition of the previous 83% drop (or even half that), or of the negative net income and free cash flow from 2005 to 2008, as they don’t seem to offer the growth that would compensate for the risk.

For Walt Disney Co (NYSE: DIS) (Morningstar, Seeking Alpha) to be relevant here, there would need to be more tourists attracted to theme parks by cheaper flights or quicker journeys than are encouraged to fly abroad as an alternative to visiting a theme park closer to home. I expect that to be true, especially if tourism from Asia expands. A long term share price chart shows a low of $15.83 in March 2009, another low of $30.16 in September 2011, and fast share price appreciation up to $114.97 as at July 3, 2015. The fast rise may suggest waiting for a dip.

Warren Buffett said good things about Disney’s moat in this PDF – “Buffett Lecture at the University of Florida School of Business October 15, 1998” (tilsonfunds.com), mostly about how Disney’s movies are a reliable choice for parents.

The company is excellent at merchandising, and the more characters and franchises they make money off, the more they can show in the parks (I needed this – “What’s the difference between Disneyland, Walt Disney World and the Magic Kingdom?“). The more movies are seen, the more tourists are attracted to the theme parks, and the more they make on rides. A good experience at Disneyland is likely to reinforce brand loyalty.

Apparently, not everyone that gets Disney wants Disney (BTW there was annoying junk on the page) – “Disney Targeted By Cable Bundle Critics In Campaign Urging More Consumer Choice On TV” by Christopher Zara, May 27 2015 (ibtimes.com). They want Disney in China – “World’s largest Disney Store is forced to close just 1 HOUR after grand opening as eager shoppers queue more than a mile to get inside” by Emily Chan and Edward Chow, May 21, 2015 (dailymail.co.uk).

I see aspects of the product as including emotion, experience, utility (keeping kids out of parents’ hair) and digitized information. The mix, the reputation and the cross-selling might be good enough for a moat for a long time, but I don’t see it as guaranteed. Way back when music came on vinyl, copyright must have seemed like good protection – if the product was consumed, profit would be made. It doesn’t work like that now. Disney have not gone the way of record labels, but I can’t be 100% certain that the model won’t be undermined, although that might just be because I don’t know enough or understand enough.

CAR and HTZ are mentioned next, and AVAV, BA, GE, LMT and IRDM are mentioned where they are relevant, further down.

Car rental

Car rental revenue is likely to increase if the number of enplaned passengers increases. Avis Budget Group, Inc. (CAR), Hertz Global Holdings Inc. (HTZ), and Enterprise Rent-A-Car dominate the car rental market and the airport segment in the US. However, when a rental company has an on-airport concession, it has to charge fees with names like Customer Facility Charge or Airport Access Fee and give them to the airport. There are also off-airport concessions, with some disadvantage through inconvenience or dependence on the airport. San Francisco International Airport lists “off-airport rental car shuttle” under “Ground Transportation” here, and they will charge the “ground transportation providers” that provide the service.

Industry consolidation has led to this – “Rental-Car Chains Bask in the Power of Higher Prices” by Justin Bachman, August 06, 2014 (bloomberg.com). Car rental companies compete to some extent with Uber, but some people might reject car ownership in favor of a mix of car-sharing, rental, Uber and conventional taxis.

Car-sharing is like rental but for only a few hours, in urban areas. Avis, Hertz and Enterprise all offer car-sharing, and synergies include demand at different times for rental and sharing. See “Avis to Buy Car-Sharing Service Zipcar” by John Kell, Jan 2, 2013 (wsj.com), and there’s a useful insight into how big rental companies profit on disposals, in “Explaining The Avis Takeover Of Zipcar” by Tim Worstall, 1/02/2013 (forbes.com). Generations used to streaming entertainment instead of buying and owning records or movies may be more open to the idea of car-sharing and other alternatives to ownership.

Hertz are at $17.25 (July 5, 2015) compared to a peak of $31.61, possibly due to this – “Hertz Global discloses another $30M in accounting errors” May 18 2015 (seekingalpha.com), and may be expensive with the forward PE around 20, although Morningstar gives a Price/cash flow ratio not much over two (as at July 3, 2015).

Avis may be a better investment, see “Avis Budget Group Set To Dominate U.S. Rental Cars” by Insider Monkey, May 28, 2015 (seekingalpha.com). Their US airport locations are listed here.

This – “Global Car Rental Market Poised to Hit US$ 106 Bn by 2020” (March 03, 2015) has a reasonable amount of info but they’re hoping you’ll want the full report. It’s positive about car rental at airports but it assumes the global economy will be good.

Given the fees airports collect from car rental, it’s possible that airports may be in a position to grab most of any benefit to car rental from increased enplanement, but that’s not a certainty. In fact it’s an interesting game theory problem – three rivals who compete but who also have the potential to benefit from cooperation are bidding to rent from not just one local monopolist, but a series of them. The ‘series’ makes the problem similar to a repeated game, and cooperation is higher in repeated games because there is more opportunity to punish cheaters. If an airport is squeezed too hard by the rental companies, it might be able to force rental customers to take a taxi or other transport to an off-airport car rental site, and make their money from the transport provider. In future, driverless taxis from Google could increase the competition to the advantage of airports, but domination by Google or further consolidation could follow. I have to stress I’m not accusing anyone of anything illegal, although collusion against airports would not excite much condemnation when the airports are using their monopoly position to add to the rental bill.

Imagine using an app to hail a driverless car. There are regulatory hurdles, but the possibility shows how the players need to be watchful and not let change pass them by. In contrast, there isn’t much car makers and dealerships can do about car sharing and driverless taxis eating into car ownership, and they’re on the wrong side of the discounts which big fleet buyers get, which may explain “Ford Jumps Into the Car-Sharing Pool With Pilot Program” Jun 24, 2015 (go.com).

I mentioned how consumers used to streaming instead of buying might see less need for ownership generally. The change in attitude is likely to spread from the US, but it’s hard to say how long it would take for the desirability of car ownership to fade in China. Maybe more people need to be able to afford cars before ownership becomes a less important aspiration.

The hospitality market

See this 36 page PDF from Ernst & Young – Global hospitality insights – Top thoughts for 2015. “Outbound investment from Asia” starts on page 10, the outbound hotel investment from Asia reached nearly 42% of the global total and it may continue to rise.

Asian tourists are not getting the service they want from western hotels, and that’s especially true for Chinese tourists – “The Services Chinese Tourists Want, But Aren’t Getting, From Western Hotels” by Eliza Ronalds-Hannon, Skift (skift.com). See also “The new chinese traveler is on the move and combines business and pleasure“, which mentions that Chinese tourists abroad have little interest in gambling.

Airport hotels could be next investor targets” by Jan Freitag, June 27 2014 (hotelnewsnow.com) – The chart of ADR (Average Daily Rate, the rental of occupied rooms) shows a rise, but it had not recovered to the peak reached in 2008.

PWC’s near-term forecast for annual growth in revenue per available room (RevPAR) in the U.S. is 7.0% for 2015 and 6.1% in 2016.

This piece about San Francisco, “2015 hospitality forecast: Hotel room shortfall puts damper on tourism boom” by Annie Sciacca, Jan 2, 2015 (bizjournals.com), shows that a shortage can develop which is good for hoteliers but bad for tourism. Investors are put off by low occupancy following the DotCom crash and the financial crisis. Normally I’d say that investors would get over it and invest, or over-invest, but awareness of Airbnb will be very high in San Francisco.

From this piece “PKF Hospitality Research Issues Consistent U.S. Lodging Forecast” June 4, 2015 (hotel-online.com) it’s possible to infer a long-run average growth in US RevPAR of 3.4% to 3.6%, with twice that predicted for 2016 and 2015.

PWC’s “Room for growth: European cities hotel forecast for 2015 and 2016” does not give overall figures for 2015 and 2016, but Dublin is expected to lead with a near 9% increase in RevPAR for 2015. The tone is generally positive, but with a warning that competition is hotting up, and Airbnb is mentioned. One factor is the use of robots and other advanced technology, partly to reduce costs (see “Starwood Introduces Robotic Butlers At Aloft Hotel In Cupertino” by Jordan Crook, Aug 13, 2014 (techcrunch.com)).

Consultants can get it wrong, at least in India – “Bankers reluctant to fund hoteliers as projections go awry” by Anita Bhoir & Varuni Khosla, Dec 10, 2014 (indiatimes.com).

I found no long term projections for the hospitality industry.

Tourist arrivals

UNWTO, the United Nations World Tourism Organisation, has published it’s free booklet “Tourism highlights, 2015 Edition” (PDF). UNWTO use a broad definition of tourists, saying 53% of international tourist arrivals were for leisure, compared to 14% for business (27% were in a broad category including visiting family, pilgramages etc.). They expect global tourist arrivals to grow 3.3% a year between 2010 and 2030, reaching 1.8 billion. I don’t know why a 2015 edition needs to start a projection from five years ago. For emerging markets, the projected growth is 4.4%, double the 2.2% projected for advanced economies. The global figure is lower than IATA’s projection for air passengers carried, although the periods are different as well as the item projected (IATA forecast 4.1% average annual growth from 2014 to 2034).

The 2.2% for arrivals in advanced economies is lower than I’d expect from the pieces I’ve linked to about tourism from China. UNWTO describe China as already being “the world’s top tourism source market”, and their spending abroad increased by 27% in 2014.

International tourist arrivals grew from 25 million in 1950 to 1,133 million in 2014, averaging 6.14% pa, but with major dips. In the same period, the receipts which destinations earned grew from $2 billion to $1,245 billion, averaging 10.58% pa. Assuming that’s not adjusted for inflation, I multiplied the 1950 figure for receipts by 9.80420 to get 19.6084, and the growth in real terms is 6.70% pa. (I used moneychimp and Calculator.net’s Inflation Calculator.)

The trend is for more overnight visitors to travel by air, with 54% by air for 2014, and 39% by road.

The interplay of hospitality, airlines and airports

Increased competition in hospitality is bad for margins in hospitality but good for airlines and airports.

Increased competition in airlines is bad for margins in airlines but good for hospitality and airports.

Other factors such as cheaper fuel costs or less fear of terrorism feed through to benefit all three industries.

It’s harder for airports to compete with each other, possibly because major cities already have airports and it’s not easy to get agreement or permission for another one.

Airport proximity

I found some airports close to each other on “What are the two closest airports to each other in the world?“. There are two airports only 1.7 miles apart in the Orkney Islands, Scotland – Westray Airport and Papa Westray Airport (the links are to Wikipedia). The scheduled flight time between them is two minutes, including taxiing. The two closest airports in the US might be Chignik Fisheries Airport and Chignik Lagoon Airport in Alaska. Those four airports are tiny. Zorg en Hoop Airport (small) and Johan Adolf Pengel International Airport (bigger) in Suriname, the smallest sovereign state in South America, are only 4.6 km apart.

While airports can be close, they don’t proliferate to the point where oversupply makes earnings crash, whereas in many capital-intensive industries, capacity increases until there’s a glut. Bankruptcy is still possible – Ciudad Real Central Airport in Madrid went bust after three years. It’s been alleged that the investors benefitted from construction contracts and never expected the airport to be viable. “14 of the world’s most amazing abandoned airports” has seven U.S. cases, not all civilian or international airports. One airport was attacked by a submarine.

An airport can be a natural monopoly, rather like “one town, one local newspaper” was for towns of the right size before the internet, but that isn’t a requirement for pricing power. London in the UK is served by three airports –

London airport ~ passengers in 2014 ~ distance from London center

Heathrow ~ 73.4 million ~ 14 miles
Gatwick ~ 38.1 million ~ 29.5 miles
London City ~ 3.6 million ~ 6.9 miles (smaller planes only)

Even with three airports, London is still so short of capacity that the mayor backed a proposal to put an airport on an estuary island, which would have cost £70 bn to £100 bn ($111 bn to $159 bn). The project, known as ‘Boris Island‘, has been rejected.

Other airport considerations

Privately owned airports are likely to be restricted by regulation which means they can’t increase landing fees to maximize their local monopolies (e.g. at Heathrow Airport near London). Disputes over landing fees in the U.S. seem to have been partly settled in 2008 – “Federal policy will let airports charge landing fees” 1/15/2008 (usatoday.com), with airports allowed to base fees on time and traffic volume. The piece expressed the hope that eastern airports close enough to be alternatives to each other would compete on landing charges.

When I looked for material about competition between US airports, all I found was this – “The Decline and Fall of the U.S. Airport” by Patrick Smith, January 24, 2015 (askthepilot.com). It’s mostly about long waits caused by the entry procedures for travelers arriving from overseas, even if they’re in transit to a third country. The same author wrote “Cockpit Confidential: Why International Business Travelers Avoid U.S. Airports” in May 28, 2013, opening with a survey by CNN that showed a substantial minority of business travellers to the US would not visit again or recommend visiting, due to the procedures. Monopolies have their limits, and for routes of equal distance travelers prefer to go the other direction around the globe to avoid a stop-off in the US. The list of taxes and fees collected by the government (e.g. “Federal Segment Fee”) indicates that airports don’t keep the monopoly benefits all to themselves, although US airports tend not to be privately owned so all the fees go to one level of government or another.

I’ve only checked reviews for two airports, and picked on Orlando. Orlando Airport scored five out of ten in customer reviews on airlinequality.com, with many complaints about immigration control. The smaller Orlando Sanford Airport scored seven out of ten.

It’s not hard to find material about competition between airports in Europe, but I didn’t find anything very good. This book promo – “Airport Competition: The European Experience” indicates that it’s a complex and subtle issue, and there must be substantial natural monopoly benefits to airports or there would be no need to review the debate in a book, or for a reviewer to asset that competition “can and does exist”.

Non aeronautical sources of revenue have been increasing, and increased traffic results in more revenue from parking, transport (ground based), rental cars, retail and duty-free, see “How Do Airports Generate Money? New Study Shows Nearly Half Comes From Non-Aeronautical Sources” by Christopher Harress, November 18, 2013 (ibtimes.com).

Air freight, FedEx and UPS

The same costs that affect passenger transport generally affect freight, with implications for FedEx Corporation (FDX) and United Parcel Service, Inc. (UPS). Two recent articles are – “FedEx: Down But Not Out” by Daniel Jones, Jun. 22, 2015 (seekingalpha.com) and “United Parcel Service – Will Deliver Returns As The Economy Expands” by Khen Elazar, Jun. 9, 2015 (seekingalpha.com). Investors need to decide if size gives the companies a moat, and if crowdsourced delivery is a threat, see “Amazon eyes crowdsourced delivery service” Jun 16 2015 (seekingalpha.com) and “I Just Tried Uber’s New Delivery Service ‘Rush’ And It Got Me My Package In 20 Minutes” by Alyson Shontell, Oct. 23, 2014 (businessinsider.com). (There’s also UberEATS.) Obviously it’s harder to crowdsource delivery over longer distances, across borders or over seas. The companies ought to benefit from the trend for shopping online.

I have not studied Fedex or UPS in any depth, but they have defined benefit pension schemes and I expect they have employed many staff over the years without very high productivity. Both companies now use “mark-to-market” for the accounting, and maybe I ought to be reassured by this piece – “Five years of mark-to-market pension expense reporting” by Bob Collie, June 10, 2015 (fiduciary-matters.russell.com), which explains that the method is transparent, at the cost of volatility.

Risks to the air travel industry

Air travel is vulnerable to economic downturns, war, terrorism, satellite failure, accidents, strikes, weather and volcanic ash clouds (see “Air travel disruption after the 2010 Eyjafjallajökull eruption“). See also “Cyber-Attack Warning: Could Hackers Bring Down a Plane?” (spiegel.de). This may have been Nato jamming during a military excercise, but they weren’t commenting – “Dozens of aircrafts briefly vanished from Europe radar” Fri, 13 Jun 2014 (sott.net, attributed to “Guardian”). Satellite failure is covered in links under “Satellite communications” below.

I have not said much about recent tragedies because I don’t think they have seriously affected the air travel industry generally. Malaysia Airlines were technically bankrupt and are relaunching, which is mentioned in “Malaysia Airlines plane makes Melbourne emergency landing” June 12, 2015 (bbc.co.uk).

The world economy faces a few problems. In emerging markets, there’s around $9 trillion of dollar-denominated debt which could trigger a ‘flight to safety’ of capital to the US, pushing up the US dollar and making the debt harder to repay. The risk was a theme of mine until I judged the media had caught up enough and the dollar had risen enough to highlight currency risks, although I learned I was way behind the economists Worth Wray and John Mauldin. See “Feeling green – Debt-ridden emerging markets are heading for a nasty dollar hangover” Mar 21st 2015 (economist.com).

IMO Greece is too small to have much effect on the European or world economy, but it may provide the spark for a big drop in stock or bond prices, given the generally high valuations, or it may spark a ‘flight to safety’. For example Greece owes Turkey over $30 billion (from “A Week of Unseen Things”, below). The Greek drama could make markets notice and start a flight of capital from Turkey and then maybe other emerging markets. I’m not saying that’s likely, I’m using it as an example of possibilities that are hard to predict. Some people believe that reduced liquidity in the bond markets increases the risk of a collapse. IMO liquidity is likely to dry up in any crash, as no-one wants to buy the ‘falling knife’. The Fed would probably react with Quantitative Easing, or if interest rates had been raised, lowering them again. With the EU now doing QE, they are likely to do more in extreme circumstances.

Also IMO, if Greece or any other country leaves the Euro (the European common currency), twenty years is long enough for recovery from the shock, and the common currency may do more harm to inefficient members than it benefits the exports of efficient members like Germany. Air travel does not seem to be very sensitive to the world economy, find “only a short-lived slowdown in the growth of passengers carried from 2007 to 2009” below.

Readers have probably heard enough about Russia, but maybe not so much about territorial disputes in Asia. This piece – “Territorial Disputes in Asia Show Sovereignty Still Matters” by Francesco Mancini, July 8, 2013 (theglobalobservatory.org), starts with a fisherman who was killed by Philippine coastguards, leading to outrage in Taiwan, sanctions, and cyber attacks in both directions. The scale is far below Russia’s invasion of parts of Georgia and Ukraine, but it may show a tendency for national pride to cause outrage and a refusal to back down or apologise. I’d guess that escalation is possible but unlikely.

This – “Dormant and Unresolved Border, Land, and Maritime Disputes in Southeast Asia: a New Threat For ASEAN?” is a fairly dull list of academic research abstracts, with only one conflict with casualties (Thailand vs Cambodia), but it claims a “multitude” of dormant and unresolved disputes that could threaten stability. That isn’t reflected at all in Wikipedia’s List of territorial disputes.

China has extensive claims in the South China Sea based on maps with dashed lines, the earliest of which was published in 1947 according to Wikipedia. There are recent articles relevant to the claims on nine-dashed-line (thediplomat.com), and see “Images show Chinese airstrip on man-made Spratly island nearly finished” by Dean Yates, Thu Jul 2, 2015 (reuters.com).

The Japanese occupation of Korea is still a live issue, e.g. “Korean sexual slavery victim to hold protest rally in front of Japan’s embassy in Washington” 2015-06-27, even though South Korea and Japan are both threatened by North Korea. About sites from Japan’s industrial revoluation – “Korea objects to heritage status for Japan’s World War II ‘slave labour’ sites – by Donald Kirk” The Independent, May 26, 2015 (republicofmining.com). Japan and Korea won’t fight a war over their differences, but old issues between East Asian countries could make an effective alliance against Chinese ambitions harder to achieve, or harder to achieve soon enough, which could encourage China to pursue claims more aggressively.

In “Chinese Nationalism: The CCP’s ‘Double-Edged Sword’” by Jessica Chen Weiss, November 25, 2014 (thediplomat.com), the author/interviewee argues that Chinese demonstrations of nationalist anger are genuine and not the result of party manipulation. Japan is a popular target. When Vietnamese protests over an oil rig led to the killing of Chinese workers, China did what it could to keep its population calm, because their strategic objective had been achieved and there was no point in inflaming the losers.

Probably no-one wants a war, given the economic consequences and the danger of escalation to nuclear levels, and I don’t mean to imply that a major war in the region is likely.

Europe isn’t free of internal tension, for example “Serbia asks Russia to veto UN resolution that calls Srebrenica massacre ‘genocide’” July 4, 2015 (globalnews.ca). There’s tension over Greek debt, with occasional reminders about the forgiveness of Germany’s debt, see “Greece and Spain helped postwar Germany recover. Spot the difference” by Nick Dearden). Even so, for most of the countries a border incident or military conflict between them is about as unthinkable as for the US and Canada.

The United States vastly outspends other countries on defense. From Wikipedia, for 2013, in billions –

US $640.2
China $188.4
Russia $87.8

Even so, with threats, trouble or tension around Russia, the Middle East, and East Asia, the US could become overstretched.

air passengers carried after 9-11 terrorism

Air passengers 1990 to 2013

The chart is based on figures taken from or derived from a table provided by the World Bank. I calculated “North America” by adding the figures for the USA, Mexico and Canada, and according to Wikipedia I missed out a lot of countries (see “Spreadsheets” below).

air passengers carried - short
Air passengers 2014 to 2034

The data used for the next chart was taken from the press release “New IATA Passenger Forecast Reveals Fast-Growing Markets of the Future“. IATA forecasts 4.1% average annual growth in air passengers carried globally from 2014 to 2034.

There are problems with the data. It’s imprecise, for example the figure of 2.9 billion passengers for Asia-Pacific in 2034 only gives one decimal place, and corresponds to a range between 2,850 million and 2,950 million (minus one). The full report is likely to be more precise, but it costs $2,500. The problem is compounded because in most cases I had to subtract the projected increase to get the 2014 figure, and for Europe I had to work from only the increase and the annual growth rate. Using imprecise figures in calculations leads to even less precision. The categories don’t look the same as for the historic data in the previous chart, with “Asia-Pacific” instead of “East Asia & Pacific”, “Middle East” instead of “Middle East and North Africa”, and “UAE plus Africa” instead of “Sub-Saharan Africa”. It’s not surprising that the chart is not continuous with the chart up to 2013, but the implied drop in the total from 3,973 million to 3,167 million is 20.3% and may be due to a major difference in methodology.

The figures in the forecast are for flights “to, from and within”, although that’s only stated for Asia-Pacific. It means a flight from one region to another will be counted in both regions (and presumably a return journey will be counted again). Passengers flying from the US to China can be US travellers to China or Chinese tourists returning home. It doesn’t make much difference to plane makers, airports etc but it makes a big difference to businesses in hospitality, attractions, car rental etc., if they are not equally represented in both countries. Increasing Chinese air travel is at least consistent with the “surge in tourists and other visitors to the US” (from China) which I mention later.

IATA is the trade association for the world’s airlines. While readers should consider that IATA may not be impartial, the forecast was made with a subsidiary of Oxford Economics, who claim “Oxford Economics continues to outperform other global forecasters, August 2013“. The reliability of almost any twenty-year forecast is doubtful. The inputs from demographics are probably more reliable than the assumptions about economic growth, which may have been optimistic to get the forecasted 4.1% global growth in air passengers carried. However, my charts show only a short-lived slowdown in the growth of passengers carried from 2007 to 2009, while the financial crisis was severe, and recovery from the recession that followed was slow. There has been no shortage of bad news from Europe since then, also concern about China’s debt and slowing growth, and demand for commodities cooled, affecting many emerging economies, yet the passengers kept flying in increasing numbers since 2009.

I applied a smooth growth rate to get from the 2014 figures I calculated to the 2034 figures.

There is more information in the press release. Under “Explanation of demand drivers” there are paragraphs for “Living standards”, “Population and demographics” (population decline in Japan, Russia, and Ukraine, rapid growth in Africa), and “Price and availability”. There’s no mention of terrorism, war, or the kind of sub-war conflict that could affect flights. Although a reduction in the real cost of air travel is forecast at around 1% to 1.5% pa, it’s my impression that they have probably not fully factored in the potential of “free routing” aircraft across the skies (see below), but find “Price and availability” in the release and see what you think. It’s unlikely they’ve factored in any benefit from holographic radar (see below). There’s also this PDF from the FAA – “FAA Aerospace Forecast Fiscal Years 2014 – 2034“.

air passengers carried 2014 to 2034
Air passengers 1970 to 2034, log chart

A slowdown from 4% growth to 3% growth might not be clear in a normal chart, but when the logarithm is charted it would show up more clearly as the gradient becoming less steep. The top half of the chart shows the discontinuity between the World Bank’s figures which end at 2013, and IATA’s projections which start at 2014. In the bottom half I shifted IATA’s lines up so they joined with the World Bank’s lines. Because I used base 10 (for the logs), an increase of 1.0 represents growth by a factor of ten, e.g. from 7.0 to 8.0 (7.0 means 10,000,000 – with 7 zeros, and 8.0 means 100,000,000 – with 8 zeros). The scale is marked in steps of 0.2, and each increase of 0.2 in the log represents an increase of 58.5% in air passengers carried.

air passengers carried 1970 to 2034 - log
Air passengers, stacked chart

Because the regional figures are for flights “to, from and within” the regions, there will be double counting in the totals represented by the heights of the stacks. The implied totals are good for the percentages in the charts. In other words, without the double counting in the total, I would expect the percentages for each region to add to more than 100%. At least, that’s the theory. My “stacked” total for 2034 is 7.0 billion, but IATA project a total of 7.3 billion for 2034. The difference could be due to the problems with the data I mentioned above, in particular the rounding error of figures to only one decimal place, compounded by further calculation. I have not included every warning about the data in every chart.

air passengers carried 1970 to 2034 - stacked
Air passengers, China, US, India, UK and Brazil

I forgot to say the scale is in millions.

top countries air passengers 1970 to 2034
IATA’s growth figures and the 2034 top 9

Extracted from their press release, average annual growth in passenger flights 2014 to 2034 (to, from and within regions and countries) –

    Total 4.1%
    China 5.5%
    Asia Pacific 4.9%
    Middle East 4.9%
    Africa 4.7%
    Latin America 4.7%
    North America 3.3%
    US 3.2%
    Europe 2.7%
    Japan 1.3%

Predicted top 9 countries for flights in 2034 (by traffic to, from and within) and movement from 2014

    1 China – up from 2
    2 United States – down from 1
    3 India – up from 9
    4 United Kingdom – down from 3
    5 Brazil – up from 10
    6 Indonesia – new in the list
    7 Spain – down from 6
    8 Germany – down from 5
    9 Japan – down from 4

Trends and factors affecting air travel and transport

• Tourism and travel from Asia
• The low oil price and other cost reductions
• Free routing

In more detail –

Tourism and travel from Asia

Increased tourism and travel from Asia and especially China is likely if economic growth continues and discretionary income increases as a result. China’s rebalancing from an investment-led economy towards consumption and the private sector should favor growth, spending and travel in the long term, but the huge amount of debt in China could cuase problems.

The Top Chinese Travel Trends to Watch for in 2015” by Rafat Ali, Skift, Dec 24, 2014, (skift.com). One prediction is a surge in tourists and other visitors to the US.

I found no evidence of a large number of Chinese tourists heading to Las Vegas. My searches turned up “$4 billion Chinese-themed resort coming to Vegas” and Vegas casino operators with casinos in Macau. Macau (Google map) is a region near Hong Kong with an area of 11.6 sq miles and high population density, which could limit the scope for casinos to expand. For comparison Las Vegas is 135.82 sq miles and the metropolitan area is 83 sq miles.

This piece makes five optimistic predictions – “What will 2015 Bring for Chinese Tourism in the United States?” by Avery Booker, 06 January 2015 (chinaluxuryadvisors.com), about the US vs Europe, investment, mobile devices, a mobile ‘chat’ app, and individual tourists.

Value Retail are a private company but they have an interesting business model, selling discounted luxury goods through nine shopping villages in Europe and one in China. They sell to shoppers who want discounted luxury brands and who demand a luxury environment that does not feel like a dumping ground for surplus stock. Shoppers who have travelled thousands of miles particularly value the experience as much as the discount (even if they fund the trip by selling purchases back in China). Brands occupy space in the villages and pay a percentage of sales to Value Retail. The model minimizes the damage that discounting can do to brand image. They followed the principles and predictions of retail guru Marvin Traub, which are in this PDF from 2008.

It looks like Value Retail know how to exploit China’s retail tourism – “Value Retail promotes shopping tourism with Chinese airline partnership” by Staff reports, June 5, 2015 (luxurydaily.com) – it’s a kind of mutual promotion, or cross-selling between different companies. See also “CEO Talk | Scott Malkin, Chairman, Value Retail” June 5, 2013 (businessoffashion.com), which says quite a lot about Chinese shoppers.

China’s retail tourism could be hit by this – “China cuts import taxes to boost consumer demand” 25 May 2015 (bbc.co.uk). Unilateral tariff cuts might seem un-Chinese, but outbound retail tourism costs their economy more than importing the goods, and it’s possible that more tax will be collected in total, even with lower tariffs. The effect will be good for luxury goods makers, and bad for air travel.

Not too worried – “Another Glimpse Into China’s Debt, And Where It’s Heading” by Kenneth Rapoza, 5/29/2015 (forbes.com)

More worried, but about financial zombification rather than a crisis – “The great hole of China” Oct 18, 2014 (economist.com)

There’s an ebook with the snappy title “A Great Leap Forward? Making Sense of China’s Cooling Credit Boom, Technological Transformation, High Stakes Rebalancing, Geopolitical Rise, & Reserve Currency Dream”, where the contributors are mostly pessimistic but not all of them (on Amazon). Here’s a long synopsis by co-editor John Mauldin.

The bank HSBC are considering moving their domicile from the UK to Asia, possibly to Hong Kong. If they aren’t bluffing to influence UK policy, they are likely to have a positive outlook regarding Asia, and with their large business in the region they should be well placed to judge the prospects. The Guardian’s article about it only considers the politics and regulation in the UK.

Because stock prices could conceivably have a wealth-effect on spending, I’ve linked to “A Week of Unseen Things” by John Mauldin, July 4, 2015 (mauldineconomics.com), find “Shanghai Falls to Earth” in it. More Chinese are piling into stocks but as momentum traders rather than value investors, and the likely result over time is stocks going higher but with volatility and sharp pullbacks. Previously I heard about Chinese investors spending less so they could buy into the bubble, which is the opposite of the usual supposed wealth effect, although I can’t substantiate the story. You can get a chart of the Shanghai Composite Index from 1996 on the link by choosing “all data”. It shows that the peak in 2007 has not been surpassed (as at June 30, 2015).

Chinese stock valuations are highly uneven, there’s room for quality stocks to go up while the crazy valuations fall, and as investors take advantage of liberalisation, Hong Kong stocks could rise as Shanghai stocks fall.
See “Opinion: China’s Stock Connect: More boom and bust?” by Craig Stephen, Apr 12, 2015 (marketwatch.com).

A high dollar is not good for incoming tourism. One point to consider is that dollar strength tends to go with low commodity prices, including oil.

One of China’s aims is to make the Remnimbi a reserve currency, challenging the US dollar, but the aim might not be compatible with the growth they want. Reserve currency status seems likely to make the remnimbi higher than otherwise, by increasing demand for the currency. A high remnimbi makes tourism from China more affordable.

It’s been projected that half of the world’s cruise passengers will be from China by 2020, see “CSSC joins race to build China’s first cruise ship” (seatrade-cruise.com). In the US, 49% of cruise passengers were aged 50 or older (in 2011), but Chinese cruises are expected to attract younger passengers and families.

Using rough historic approximations, assuming 300 ocean-going cruise ships averaging 2,000 passengers each, guessing 10 days per cruise (so about 37 cruises per ship per year), I get 300 * 2,000 * 37 = 22,200,000 passengers per year, compared to 3,972,988,179 for air travel in 2013 (from summing the World Bank’s table). That’s about 22 million cruise passengers and 3,973 million air passengers, or 181 air passengers for every cruise passenger. I don’t expect cruise holidays to have much effect on air travel.

The low oil price and other cost reductions

The benefit of the lower oil price is passed on to passengers through competition by airlines, and the low oil price could lead to increased discretionary income and more travel (although some employment is favored by a high oil price). Technology such as lighter, stronger materials and more efficient engines (helped by 3D printing) increases the fuel efficiency of aircraft (although the planes and engines aren’t cheap).

In the U.S., the cost of extracting unconventional hydrocarbons is still coming down. Efficiency has been increasing with experience, which is one of the features of extraction from shale that makes it more like manufacturing than conventional extraction. The next link explains it better than I’ve ever seen before – “The Shale Boom Shifts Into Higher Gear” by Donald L. Luskin and Michael Warren, May 31, 2015 (wsj.com). The fact that shale oil is front-loaded (i.e. a well peaks early) has been seized on by writers who call shale production a “Ponzi scheme”, which seems like a loaded term to apply to unconventional production, although the financing may be another matter.

There is still controversy over how much can be extracted, but there’s also the possibility of applying the techniques in other parts of the world. In “Govt Report Says Shale Oil Will Run Out Soon, Start Drilling Offshore” Mar 26, 2015 (marcellusdrilling.com), the government’s pessimistic view is disputed. Also, “Shale oil and shale gas resources are globally abundant” January 2, 2014 (eia.gov).

The cost or benefit to a company of short term movements in the oil price depends on how much of the impact is hedged.

There’s been much talk of growing algae to make aviation fuel, but Solena Fuels and Velocys plc (VLS.L) (I’m long VLS.L) are pioneering a “biomass-to-liquids” process, see “GreenSky London site selection” about the production of aviation bio-fuel scheduled for 2017. Solena claims to sell fuel at spot prices with no subsidies. The customer British Airways has committed to buy at long term market competitive rates, which may be lower than the suppliers initially expected. It’s possible that the technology could become cheaper once it is established. The partnership lags behind China’s Sinopec and Hainan Airlines – “China commercial flight uses Sinopec bio-jet fuel” 25 Mar 2015 (argusmedia.com).

Sorry, Everyone, Making Fuel Out of Seawater Isn’t Gonna Save Humanity” by Martin Robbins, April 19, 2014 (vice.com) is about the mis-reporting of the Navy’s demonstration of a “fuel from water” process. The author is right that the media is hopeless at reporting such stories accurately, and right in pointing out that the methane by-product is a potent greenhouse gas. There’s actually nothing surprising about being able to turn one form of energy into another (except for ambient temperature heat, see “Maxwell’s Demon” for the physics), and in this case electrical energy was turned into chemical energy – the energy unlocked when fuel is burned. The conversion efficiency matters (and the by-products), but the volume of water required is probably not very important because no-one cares much about the water produced when fuel is burned. If a suitable region has surplus electric power for some periods, it may be economic to use it to make fuel, depending on various factors.

The cost of solar photovoltaic energy is falling, roughly in line with Swanson’s Law, a kind of solar power version of Moore’s law. The installation costs are likely to be stickier. To compete more broadly with oil, intermittent renewables need more affordable energy storage, which might still be some way off, and batteries with higher energy density would help the take-up of electric vehicles.

If you google “solid state lithium ion batteries” (without the quotes), you’ll find big names like Google and Toyota in the research race. I’ve also read that Apple are in the race and General Motors have had some kind of involvement. The U.S. startup Sakti3 seemed to emerge from nowhere, and U.K. appliance-maker Dyson have invested $15 million in them. (I’m long Ilika plc (IKA.L), who are in the race.) Tesla, the high-end electric car-maker, is building a ‘Gigafactory’ to make conventional li-ion batteries for use in vehicles and the Powerwall home battery unit, and Elon Musk (of Tesla) obviously does not believe that new technology was worth waiting for.

The cost of stored renewable energy will put a ceiling on the price of fossil fuels, and the cost will come down, in the same way that the cost of shale oil puts a ceiling on the oil price, and the cost of extracting shale oil is coming down.

The low price of oil means aviation fuel is cheap, but the relation could conceivably change in extreme circumstances. In theory, if demand for the other fractions of crude oil falls, aviation fuel would get more expensive. In the simple case of two products A and B with production of them in a fixed ratio, increasing demand for A will mean more of B is produced and B will usually be cheaper as a result. Similarly, less demand for A makes B more expensive, and more demand for B makes A less expensive. The economics is more complicated for fractions of oil, partly because the fractions produced can be varied by using heavier or lighter oil, and by cracking or coking. This image for the fractions of crude oil is from “Getting gas from Crude” by Heading Out, March 13, 2006 (theoildrum.com), and is available under a Creative Commons license.


This … “General Electric Bids its Fuel Efficient Future on 3D Printing” by Theo Valich, July 15, 2014 (vrworld.com) is about ‘printing’ fuel efficient nozzles. 3D printed parts can also be lighter, because the best design might be lighter and too complicated to make by conventional methods. (Whole jet engines have been 3D printed, but only two copies of an old engine that were put on display.)

While cheaper oil is good for air travel which is good for sales of planes and their engines, the lower oil price reduces the benefit of fuel efficiency, making it harder to sell expensive fuel-efficient aircraft, according to “Lower Jet Fuel Prices Shake Up Aircraft Market” by Robert Wall, Jan. 19, 2015 (wsj.com). General Electric invested in oil & gas at the top of the market, and the WSJ’s “Shake Up” piece implies a risk that the low oil price will hit sales of GE’s efficient engines. However, recent sales news has been good – “GE Aviation tallies $19 billion in engine deals at air show” by Chelsey Levingston (daytondailynews.com). There’s a piece I like about GE’s tech opportunities – “Big Data, The Industrial Internet, And How GE Stands To Profit” by Adam Stockmeister, Jun. 23, 2015 (seekingalpha.com).

Against GE, there’s a history of underperformance with the share price of $26.78 (on July 3, 2015) well below the price of $41.40 on Sept 28, 2007. There’s also a lot of debt, which will be reduced due to the divestment of financial operations, but they’re having problems closing a big acquisition (Alstom, it’s French), while the disposal of Electrolux has hit regulatory opposition. GE’s unions won significant improvements in wages, healthcare and pensions, see “GE Announces Details of Proposed Labor Contracts” June 24, 2015 (businesswire.com). They might expect another good deal in four years time. GE were a top-ten beneficiary of export subsidies that have ended, find “Boeing’s Bank” below.

Boeing Co (BA) contradict the WSJ’s “Shake Up” piece – “Cheaper oil to boost aircraft sales, Boeing official says” March 2, 2015 (thenational.ae), emphasizing the effect of cheaper fuel on increasing air travel.

Boeing’s 787 Dreamliner has attracted some sensational headlines, such as “Boeing 787 software problem could lead planes to fall out of the sky”. Further investigation shows the problem was quickly solved with a software update. I like the headline “Turn your Boeing 787 off and on again, or it will crash”, but for an unsensational overview I suggest “ANALYSIS: After three years in service, how is 787 performing?” by Stephen Trimble, Nov 14, 2014 (flightglobal.com). Apparently, airlines have tolerated various glitches because they like the fuel efficiency. Al Jazeera have been highly critical of the 787, see “Review: Al Jazeera’s 787 Report Misses the Mark” by Vinay Bhaskara, 9/10/2014 (forbes.com). I’m not reassured by the author’s claim that it’s hard to find any major industrial plant without drug use. Boeing claim to have standard drug-testing procedures. I checked the titles of articles on Seeking Alpha back to Sept 11, 2014, they are mostly positive, and the few negative titles I saw were not concerned with safety.

Competitor Airbus Group SE (EPA: AIR) has also had incidents. On Wikipedia, the four incidents involving the A320 family in 2015 can be summarized as: murderous co-pilot (fatal), wind shear (injuries), loss of height (unexplained) (injuries), and landing accident (no-one hurt). About a military Airbus driven by compter-controlled propellors – “Fatal A400M crash linked to data-wipe mistake” by Leo Kelion, 10 June 2015 (bbc.co.uk).

Boeing subcontracted sub-assemblies for the 787 to various suppliers, it was supposed to reduce costs but the cost reductions have been slow. The 787 is expected to break even this year. The company has good cash flow, but the accounting is complicated – “Boeing 787 unit loss declines, but deferred costs rise” by Stephen Trimble, Apr 22, 2015 (flightglobal.com). Those deferred costs were for production and the amortization of tooling, and they rose to over $30 billion. A deferred cost is an expense that will be recognised later, which is accounted for as a notional asset. The notional asset gets wiped when the cost is paid (or recognized as paid, on the books). The $30 billion of deferred costs will hit income as they are recognised. Boeing’s half-a-trillion dollar backlog should keep the cash coming in, but the deferred costs could build up with it. I suggest doing your own research in the area before investing a lot in BA.

“Boeing’s Bank” has shut – “Export-Import Bank Closes: Kill Subsidies To Cut Federal Liabilities, Promote Economic Fairness” by Doug Bandow, 6/30/2015 (forbes.com). The ‘bank’ provided finance at below market rates (for $67 billion of Boeing’s sales over seven years), and took on the risk of non-payment. From the article (page 2) it looks like the financing was not a major factor in closing sales, so if there was any rush to buy Boeing before the bank closed, it was probably not a stampede and not likely to have a massive effect on future sales. Other agencies to promote exports remain.

Boeing and Airbus have been fighting each other over subsidies for years, see “Boeing warns Airbus over reports of state aid” by Russell Hotten, 17 June 2015 (bbc.co.uk).

This is about how China is struggling to make commercial airliners, and the author’s reasons are rooted in the technical complexity of a modern airliner and the organizational complexity of making them – “Can the Chinese Create a Competitive Commercial Aviation Industry?” by Thomas Duesterberg, Jan/Feb 2015 (aspen.us). The Dreamliner seems to have been a struggle for Boeing, which I put down mostly to the size of the challenge rather than any fault of the company.

Airbus in China” (airbus.com) is best read after the previous link. A cynic might say it’s about how Airbus are giving China their technology and know-how, undermining the future of European and US aircraft manufacturing.

Free routing (next) could mean fewer planes are required for the same number of passengers carried.

Free routing

Flight along legacy air corridors means inefficient zigzagging, and there are moves to bring routing into the twenty-first century.

Free flight” Sep 6th 2014 (economist.com) is about the prospect of new technology allowing pilots to “free route” and choose direct routes when possible, with the potential of quicker journeys and lower costs. Direct routes increase the ‘useful miles flown’ per day with no increase in the cost of fuel, aircrew pay, maintenance, or other overheads except for the cost of the new technology. Trajectory-based routes include altitude, and instead of ascent, level, ascent, level, descent, level, descent (etc.), a smoother route in the “height” dimension adds to the effieciency.

One company The Economist mentions is Indra Sistemas SA (IDR:Soc.Bol SIBE), a diverse Spanish consulting and technology multinational which includes developing air traffic management systems (on their “Air traffic” page).

There’s also a promo which summarizes “Free Routing” on “New Space to Fly” Dec 3, 2014 (click “More”), but to download the radio program you probably have to be in the UK (and you’ll need to download the BBC iPlayer).

The US has its “Next Generation Air Transportation System” (Wikipedia) aka “NextGen”, which will also allow the selection of direct flight paths. Implementation started in 2012 and should be complete by 2025.

Security is not mentioned in the Economist piece about free routing. The Wikipedia piece about NextGen in the US only mentions that crucial signals are “unencrypted and unauthenticated”. The signals are ADS-B (Automatic Dependent Surveillance Broadcast). The paths involved are from transponders in airplanes to other airplanes and to air traffic control. The transponders receive a time signal from GPS satellites from which they calculate their position, which is then transmitted.

I’d say that the more automated a system is, generally the more damage can be done by hacking it. A system which manages too much is more likely to have unforeseen glitches, and hopefully the authorities know what change and what pace of change is safe. If a dynamic automated system operates safely in the air, it could strengthen the case for allowing driverless cars and coordination of their operation on the ground.

One problem with global roll-out is that many Air Navigation Service Providers run systems that are behind the capabilities of modern aircraft, and the focus in air traffic management is on encouraging upgrades to allow harmonization and interoperability. That might need to be complete in a region before starting a program of radical change (the world’s nine regions are mapped here).

I’m guessing that another inefficiency exists which technology could fix – when a flight is cancelled, there will be newly available slots for take-off and landing. These are likely to be wasted quite often unless the cancellation is a long time in advance, either because the information system can’t cope, or because people prefer to play safe and stick to the filed flight plan, or stick to the plan to avoid the stress and effort of making a new plan within a tight deadline.

Free routing increases the capacity of airplanes to carry passengers between airports (i.e. more flights per airplane per week), and closer spacing between aircraft increases the capacity of the airspace. If those capacities increase, the effect on an airport depends on various factors (these are fairly abstract and complicated) –

• In principle, an airport with excess capacity which is limited by the capacity of the airspace would benefit the most, but in practice there’s no reason to have provided the excess capacity in the first place.

• An airport at around full capacity which can increase its capacity (e.g. by adding a runway) would have the opportunity to increase passenger volume.

• ‘Second choice’ airports (further from a city or other destination) get the leftovers, and volume depends on the capacity and pricing of their more convenient competitors in relation to total demand for flight to the destination.

• If an airport and its airspace both have excess capacity, there’s no benefit from increasing the capacity of either.

• If the infrastructure is not a constraint, a city can handle a lot of visitors and departures, but a resort will have limited capacity and too much expansion could ruin the experience.

Although fairly complicated, that’s still a simplified version because it ignores time (the difference between peak and off-peak times at airports) and geography (airspace could be congested to the north but not to the south). There are also sub-cases such as where spare capacity becomes a shortfall, and the statements are rough rather than rigorous.

All of those considerations could have some effect on hotels, attractions, conference centers etc. Negative effects only seem likely on a local level, if an airport becomes stranded because a nearby airport can handle more traffic.

The more efficient use of planes will mean lower sales of planes, engines and other parts, unless air travel increases by enough to compensate. Lower costs and increased capacity from free routing and associated upgrades will help to increase air travel, and airlines, airports, car rentals and hospitality will generally see higher volumes than otherwise. Airports may be best placed to capture the cost benefit, but it depends on regulators allowing it.

Rail can sometimes compete with air travel and transport, and while trains obviously can’t “free route”, there is scope for information technology to allow the spacing between trains to be reduced, increasing capacity without needing to lay new track. They’re talking about it in the UK, and working on it in Australia, see “Advanced Train Management System” (lockheedmartin.com). Lockheed Martin Corporation (LMT) are best known for expensive jet fighters, see “Pentagon’s ‘Too Big to Fail’ F-35 Gets Another $10.6 Billion“, but their “What we do” list is diverse enough to include biometrics and robotics.

Existing flight management

Software has already made flight planning easier, see “SkyDemon Flight-Planning Features“. The system is visual, and a route can be made by setting waypoints, and then edited for example by dragging, which will be needed if a warning is generated that the flight goes through restricted space. The feature suggests that at least a degree of free-routing has already been implemented, although NextGen is not due for completion until 2025. Journey time and fuel use are displayed.

Flight plans can still be submitted on forms filled in manually, and you can download a flight plan form (PDF) here.

According to Wikipedia’s “Flight plan” page, for domestic flights a flight plan should be submitted at least one hour before departure, and for international flights, the minimum is up to three hours before departure. Those are recommendations, and it’s possible for a flight plan to be submitted after take off.

The existing systems are already sophisticated, but so far as I can tell, if a change in the weather means that a different route would be more fuel efficient or avoid delay, the pilot is likely to stick with the flight plan that’s been filed, rather than go to the trouble of having a more efficient flight plan calculated and submitted. BTW if severe weather threatens safety, an airplane will usually turn back or be routed around it. In “AirAsia QZ8501: Plane crash blamed on weather“, it’s reported the pilot requested permission to climb to avoid bad weather. Big modern planes have weather radar that usually allows problems to be seen in time, but sometimes (rarely) pilots can under-estimate the risk or make mistakes when conditions get worse quickly.

TRACON stands for “Terminal Radar Approach Control” and a TRACON handles the descent and approach to an airport, usually for more than one airport. A TRACON jurisdiction is divided into sectors with a controller for each sector. The controllers in the TRACON have radar view of all the sectors and can easily communicate with the other controllers, which allows a smooth transition from one sector to another. According to nasa.gov, there is not the same communication between TRACONs, and as a result some routes are longer to preserve safety. Presumably that restriction will be phased out under “NextGen”.

For more background see Wikipedia’s “Air traffic control“.

Holographic radar

The Economist’s piece about Free Flying describes the limitations of existing radar. While radar might never be as accurate as positioning by satellite, using satellite signals depends on systems in the plane as well as the satellite. If there’s an incident or failure on the plane, or if anything such as a solar storm affects the satellites or the ability to pick up the signal, good ground radar is needed. The good news here is that ground radar has just got better.

When Russian jets fly dangerously close to airliners in the Baltic, the Russians don’t respond to radio or give the normal transponder signals that would give position data. The usefulness of ground radar in tracking uncooperative jets depends on how stealthy the jets are. However, it’s been suggested that the military pass on some of their radar surveillance data to civilian air traffic control, which implies that the jets are detectable by radar (see “Free route aspirations face rogue jet threat” below).

Holographic radar has been pioneered by Aveillant, a privately owned spin-off from Cambridge Consultants in the UK. According to the sales blurb – “What makes Holographic Radar different?“, Aveillant’s “static staring” system is on target 100% of the time, compared to under 1% “dwell time” for scanning radars. For decades, engineers have focused on squeezing more info out of that 1%. Previously, radars collected too much information to process effectively, and they operated by electronically filtering out the least useful info. Aveillant’s founders say they realized that advances in digital technology meant the filtering could be avoided.

This – “Aveillant pioneers win industry accolade” by Aimee Turner, November 28 (airtrafficmanagement.net) explains the technology fairly simply, and I’ll quote – “it observes the complete airspace”. It confirms that Aveillant have overcome a key problem – telling a wind turbine from a plane.

There’s more about the technology from suppliers Boston Limited – “NVIDIA GPUs and Supermicro UK servers fuel Aveillant’s Holographic Radar™” January 13, 2014 (sourcewire.com).

This is the most comprehensive piece about the tech – “All-seeing EYE” May 6, 2015 (aerosociety.com).

The radar is good for knowing the position and velocity of drones, which could favor their civilian use.

Aveillant’s CTO is making a 45 minute presentation at the Military Radar Summit 2015 (agenda), on “Holographic RadarTM – Target Centric Surveillance”.

Satellite communications

About satellites and safety – “SB Safety installation marks flight deck comms revolution” 14 August 2014 (inmarsat.com).

About satellites and broadband – “Honeywell partners on advanced in-flight connectivity system” Apr 14 2015 (seekingalpha.com).

Satellite providers angling for new business in air-traffic changes” – More frequent reporting of airplane position will allow them to fly closer, i.e. not so far ahead or behind each other. The increased traffic allowed will be good for airport business, though for a single airport it depends on how often capacity is reached, what limits capacity, would an extra runway be feasible, etc. Demand from aviation for satellite services looks likely to grow in the long term, although it will drop whenever flights or miles flown drop.

The first two links refer to Inmarsat, a UK company which I’ve put on the “Airport stocks list”. A US stock could benefit from the same demand –

Iridium Communications Inc (IRDM).

The forward P/E is under 10 (as at July 3, 2015). Morningstar’s financial history page shows –

Operating cash flow about twice earnings (very roughly).
Capex about twice OCF.
The returns metrics are poor single digit figures.
Income has grown quickly but the rise in the number of shares means that EPS does not show a clear growth trend.

Iridium depends on successful launches for cash flow to cover interest on debt. I have to give credit to slboomer and 2tired2talk for their comments about launches, cash flow and debt under “Betting Against Beta With Iridium” by Celan Bryant, May 4, 2015 (seekingalpha.com). An obvious point is to be sure IRDM have good insurance before investing in them ahead of critical launches. I don’t know if it’s possible to insure every aspect of deployment. Iridium’s big investment could make life tough for competitors. I’ve checked that two European projects won’t compete – the European Data Relay System will relay satellite data to ground stations by laser, and the EU Sentinel satellites will monitor the earth and the data will be freely available.

See “Iridium NEXT — Iridium’s second-generation global satellite constellation“. It’s probably just coincidence that Iridium use “NEXT” and “second-generation” while “NextGen” (above) is the big Air Transportation System upgrade. Iridium talk about “truly global mobile communications on land, at sea and in the skies”.

Getting back to satellites in general –

GPS and the European equivalent GALILEO are provided by governments, but it’s a key part of products and services from Iridium and many other companies.

Why satellites fail?” by Marryl Azriel, March 7, 2013 (spacesafetymagazine.com)

What would happen if all satellites stopped working?” by Richard Hollingham, 10 June 2013 (bbc.com). Air travel is one of the casualties. The author does not seem certain about the far-reaching consequences, and maybe few people if any know for sure.

Satellite insurance” (Wikipedia)

This is of historical interest, it’s a long slow-burner but the tension gets high – “Elon Musk’s Space Dream Almost Killed Tesla” by Ashlee Vance, May 14, 2015 (bloomberg.com).

3D radar

This is just to give some idea of what radar can already do (i.e. non-holographic radar). 3D radar has applications in weather monitoring, air defense, surveillance and the imaging of underground structures. Some links –

3-D Radar of Mystery Object That Hit Louisiana” by AccuWeather.com Staff, 10/16/2012 (accuweather.com)
Acquisition of 3d-Radar – Chemring Group PLC” about mapping underground structures
British fleet’s new radar system can detect a supersonic tennis ball 25 km away” by David Szondy, March 19, 2013 (gizmag.com)

Other staring radar

My amateur guess was that Aveillant’s holographic radar could not do the tennis ball trick (in the previous link), because the best way to pick up a reflection from a small distant fast-moving object is to focus a lot of energy on it, even if it’s only for a short time as the object intercepts a swept focused beam. However, the private South African company Reutech has demonstrated a staring radar array which they claim is good for detecting and tracking small fast-moving projectiles. It’s good at discriminating between projectiles and complex clutter, which seems similar to Aveillant’s radar being able to tell wind turbines from planes. Although developed for military applications, Reutech claim possible applications in automotive, mining and security.

Possible radar losers

It’s possible that staring radar threatens part of some companies’ defense revenue, such as Raytheon Company (RTN). I stress ‘part of’, e.g. Raytheon look well positioned with their expendable drones (which cost about $400,000 each, but as they’re launched from aircraft to jam enemy radar, they wouldn’t be as cheap as surveillance drones).

Thales SA (HO:Euronext Paris) supply air traffic management systems (thalesgroup.com). They combine radar and non-radar for surveillance, which may provide some protection from Aveillant’s radar, and they should benefit from the need to upgrade the CNS infrastructure (communication, navigation and surveillance) to support free routing and associated improvements. See also “Thales launches STAR NG, revolutionary new ATC radar with specific dual civil/military applications“. The name “STAR NG” rather suggests “staring”, as in Aveillant and Reutech’s staring radar. The picture looks like it’s of a directional radar, but there’s no caption to say exactly what is pictured. The radar is an “S-Band Primary Surveillance Radar (PSR)”, a kind Aveillant claim their holographic radar is superior to.

This is better than Thales’ sales spiel at explaining the need for integrated civil/military surveillance – “Thales’ STAR NG heralds primary renaissance by Aimee Turner, June 15, 2015 (airtrafficmanagement.net). The immediate problem is (probably) Russian jets flying close to passenger aircraft over the Baltic Sea, as described in “Free route aspirations face rogue jet threat” by Aimee Turner, June 24, 2015 (airtrafficmanagement.net). I’m not sure how free routing aspirations are threatened, as uncooperative aircraft can fly just as close to planes in an air corridor as they can to free-routing planes. The article says the jets were close to the EU region’s borders, but the European Commission says –

    “Apart from two areas under Russian sovereignty at the far end of the Gulf of Finland and off Kaliningrad, the Baltic Sea is now under the jurisdiction of European Union Member States.”

Keying “russian jets in ” into Google’s searchbox gets a list of prompts with continuations such as English Channel, American airspace, Irish airspace, Alaska, etc. That only shows what people have been searching for, but I checked for the English Channel (between England and France), and found “UK scrambles jets as Russian planes and ships spotted“. IMO using relatively cheap drones to fly dangerously close to Russian jets might be effective (when they are not close to airliners), as it’s clear who would have the most to lose in a collision, but I can’t be sure about feasability and the likely responses from Russia need to be considered. In any case, wherever there are uncooperative jets it will be easier for Thales to sell their integrated civil/military surveillance, but I’m not tempted to buy the stock due to the possible impact of Aveillant’s radar.

AeroVironment, drone maker

AeroVironment, Inc. (AVAV) makes drones and charging stations for electric vehicles. Most of AVAV’s drone business is military, but IMO the opportunity is (or was) in civilian drones.

Data from the EU Sentinel satellites (linked to above) will be free, requiring only registration. For applications where Sentinel’s spatial resolution of 10 m in the visible spectrum is good enough, observation drones might not get the business, although they have the advantage of being able to fly under clouds and as frequently as needed. The satellite’s resolution will not be good enough for the inspection of infrastructure like roads and pipelines.

AVAV have been concentrating on the upmarket “Enterprise” segment of the civilian drone business, but there’s a strong hint in this – “AeroVironment (AVAV) Timothy Conver on Q4 2015 Results – Earnings Call Transcript” Jun. 30, 2015 (seekingalpha.com) that they will enter the consumer drone business (find “consumer appetite” in the transcript), bringing some kind of technological advance to the market.

The consumer market seems like an easy one for anyone to enter and likely to become highly competitive and low margin. AVAV should probably have entered the market earlier rather than later. If they have an advantage that transfers to the consumer market, I can’t see it. “The 7 Best Agricultural Drones on the Market Today” by Andrew Amato, 1, 2014 (dronelife.com) has the “Lancaster” drone from PrecisionHawk at number 3, with LIDAR among other features, and there’s no mention of AVAV. There’ll be more drones on the farm as restrictions ease – “Why 2015 is the year agriculture drones take off” by Clay Dillow, May 18, 2015 (fortune.com).

In “Drones Revolution Means Big Data Cloud Services” by Colin Snow, February 18, 2014 (blogs.sap.com), the president of PrecisionHawk imagines analytics app-stores. That has me thinking, should AVAV be developing an analytics app-store, will they put out their own apps, and will they be late again.

Military drones, Shaving costs” Feb 17th 2014 (economist.com). In short: AeroVironment’s Raven – $76 thousand, University of Virginia’s Razor – $2 thousand. The two drones won’t be exactly comparable, and control through a smart phone with apps does not sound secure.

Management are keeping the cash generated by the business to fund growth, and the cash needed to fund growth depends on how opportunities develop. It’s good that management realise growth is a marathon, not a sprint (my cliche, not management’s).

The financial history on Morningstar is poor. The return metrics have declined since 2006 to pathetic levels, with return on assets, equity and investment all below 1% for fiscal 2015. However, the metrics are based on income, and the operating cash flow and free cash flow have been much higher than operating income for the past three years.

Overflight Fees

One reason for a plane to sometimes avoid a direct route is the overflight fees charged by government agencies (overflight means ‘flies over a country without taking off or landing in it’). The FAA‘s fees look like they go up every October, and they’ve gone up from $43.82 in 2012 to $56.86 in 2014 (per 100 nautical miles). At least it’s simple, unlike the algebra on Eurocontrol.

Every country has the right to deny overflight but there are agreements between countries to allow it, usually under the “Freedoms of the air” (Wikipedia) framework. “Freedom” should not be confused with “no fee”, as the deals to allow overflight also allow a fee to be charged. The fees have to be “reasonable”, which can mean the same fees as for domestic airlines, but the reasonableness of fees can still be disputed.

Russia did not sign up to the “Freedoms of the air”, but they still allow overflights (usually), and earn about €300 million a year from European airlines. They’ve threatened to stop foreign overflights, but it would hurt them more than the airlines according to “Russia’s Still Not Quite Getting These Trade Sanctions Things; Now It’s Overflight Rights” by Tim Worstall, 9/09/2014 (forbes.com). I’m not so sure. Russia doesn’t seem likely to ban overflights by Chinese airlines, or any airline based in a country which has not hit them with sanctions. I’m not sure that European countries could do much if their airlines complained about competitors having the advantage of overflying Russia.

Greece is probably well positioned (literally) to collect overflight fees.

Exotic air passenger transport

IATA’s forecast to 2034 has something missing – “London to Sydney in just two hours? By 2030 travellers will jet the world on a 13,750 mph spaceship” October 2, 2011 (dailymail.co.uk). Unfortunately, in November 2014 Virgin’s spaceplane broke apart after prematurely deploying “feathers” that were meant to slow the descent on reentry, and prevent burning up in the atmosphere. I’d guess if anyone has the will and resources to persist through such an ambitious project, it would be the Chinese government, and success would enable them to leap ahead without the expense of working up to a dreamliner. Their “humans in space” capabilities are covered in Wikipedia’s page about the Tiangong-1 space station.

Passenger airplanes that look like stealth bombers would have lower drag and better fuel efficiency, if they can solve problems like yaw (horizontal rotation) and where to put the windows – “Northrop Grumman’s Futuristic Flying Wing Cribs From its Past” by Jason Paur Gear, 01.25.12 (wired.com).

Electric planes are a long way from being commerical. This plane can fly for up to an hour, and I imagine serious range anxiety.

Solar Impulse have already broken the record for non-stop solo flight with their 118 hour flight from Nagoya in Japan to Hawaii. The solar plane is a sophisticated piece of engineering, but there are practical problems in applying the principles to commercial aircraft. The plane had the wingspan of an Airbus but only carried the pilot, so a purely solar airliner is not feasible. Putting solar cells on airliner wings means less fuel is burned to generate electricity, but the cells and the bonding to the wings would need to be very strong because damage would produce drag and airlines like to keep maintenance needs as low as possible. There’s the question, would and should the minimum safe fuel requirement be reduced to take account of the solar energy generated.

Facebook have test-flown a prototype solar powered drone with an airliner’s wingspan and a car’s weight.

The Autonomous Human Drone Taxi” (lab.moovel.com). The hanging seat is probably too scary for most passengers. Although it’s only ‘proof of concept’, adding a cabin would increase the weight. It might suit applications like supplementing emergency response in remote areas, rather than city use, if the necessary range is achieved.

Links to gurufocus

Altman Z Score (bankruptcy risk)

AVAV safe, BA grey, CAR distress, DIS safe, GE distress, HTZ distress, IRDM distress, JBLU grey, LMT safe, MAR safe, RTN safe

The score indicates the risk of bankruptcy within two years, with a “grey” zone between the distress zone and the safe zone. It ignores cash quantities because they were not usually disclosed when the test was invented, but it has had good statistical support many years after its invention. Professor Altman’s later test is available, for a fee. Like all financial tests that I know of, the Altman Z only considers figures in the main accounts, i.e. nothing off the balance sheet. If your favorite stock is in the distress zone, check the table of history which gurufocus put at the bottom. If it’s been in distress every year since 2005 without bankruptcy or other failure, the indication has been false for at least 8 years.

Beneish M-Score for earnings manipulation


Lockheed Martin (LMT) are reported as being an earnings manipulator, but gurufocus did not complete the calculation so the result should be ignored. The M-Score is unreliable (I don’t mean the calculation by gurufocus) and if a company fails it, further investigation could show there’s no real problem.


These are on Spreadsheets used for “Air travel – an overview”. I have not reproduced the data from the World Bank’s tables. It’s probably easiest and safest to check my work by reproducing it from the sources, find “was taken from the press release” and “taken from or derived from” above.

To get the historic data in similar categories to the forecast data, I had to calculate the air passengers carried for North America. I did that by adding the figures for USA, Canada and Mexico. That missed out a number of countries, mostly Carribean or Central American, at least according to Wikipedia’s list (find “Countries, territories, and dependencies”). As a result, the historic totals are under-reported in my spreadsheet and charts. That’s in the wrong direction to explain the discontinuity between the historic data and the projected data, which shows as a drop from 2013 to 2014.

And finally …

You’ve probably heard of ‘Lawn chair Larry’ (Darwin awards). This YouTube video takes 39 secs before getting to Larry. There’s also “On a high: Adventurer becomes first person to cross Channel in chair attached to helium-filled balloons” by Vanessa Allen for the Daily Mail, 29 May 2010.

Thank you for reading this.

DISCLAIMER: Your investment is your responsibility. It is your responsibility to check all material facts before making an investment decision. All investments involve different degrees of risk. You should be aware of your risk tolerance level and financial situations at all times. Furthermore, you should read all transaction confirmations, monthly, and year-end statements. Read any and all prospectuses carefully before making any investment decisions. You are free at all times to accept or reject all investment recommendations made by the author of this blog. All Advice on this blog is subject to market risk and may result in the entire loss of the reader’s investment. Please understand that any losses are attributed to market forces beyond the control or prediction of the author. As you know, a recommendation, which you are free to accept or reject, is not a guarantee for the successful performance of an investment.