Clearwater: A Paper Company Offering High Returns

May 14, 2024

You may recall from my previous article that I saw opportunities in the packaging company, Greif. But I see now an intriguing opportunity in the packaging-adjacent. Clearwater is a company that makes paper products including bleached paperboard and also tissues for retail consumers, and it offers an attractive free cash flow yield upwards of 15%.

Clearwater recently reported earnings that Wall Street cheered, resulting in its share price going up by 25% (sorry), but even after this advance the company’s earnings power could justify an even greater advance.

Sales in 2023 were $2.083 billion, and operating income was $108 million, and excess depreciation over cash flow produced a free cash flow of $149 million. Set against the current market cap of just under $800 million once excess cash is taken into account, that is a free cash flow yield of nearly 19%. In 2022 sales were $2.080 billion and free cash flow was $124 million, and in 2021, sales were $1.773 billion and free cash flow was $77 million. From 2021 to 2023, paperboard shipments have declined at a single digit percentage while prices increased by over 25%. In the consumer product division, which is roughly of equal size, sales volume increased by 10% and pricing by 15%. This is an encouraging sign that Clearwater is capable of passing on inflation to its customers.

For the first quarter of 2024, sales were $496 million versus $525 million in 2023, operating income was $17 million versus $24 million, and free cash flow was $22 million versus $29 million. Paperboard volumes were unchanged while sale prices declined by 10% as compared to the previous year, and consumer tissues showed a 5% increase in volume while sales prices remained flat. However, the company reported that an extreme weather event affected one of their plants to the effect of $15 million, so that $22 million is a low estimate of earning power. But on the downside, Clearwater also announced that capital expenditures for 2024 would be on the order of $100 million, as opposed to $70 million in 2023.

Now, in terms of free cash flow it is conceptually helpful to distinguish between maintenance capital and growth capital–that is, the amount of capital expenditures necessary to maintain the company’s earnings power as opposed to expenditures designed to increase the company’s productive capacity. The former must definitely be counted as a deduction against free cash flow; the latter does not (subject to the customary skepticism about growth projections, the concept of diminishing marginal return, concerns about management empire-building, etc. etc.). Unfortunately, corporate reporting does not seem to be aware of this distinction, as most managers, including Clearwater’s, do not disclose whether incremental capital spending is maintenance or growth, and analysts seem to be reticent to press them on the matter. Therefore, it is not clear whether the $30 million increase in capital investment in 2024 will be maintenance or growth.

One clue, however, is what the company is doing with the free cash flow it already generates. A company that sees plenty of growth opportunities will deploy free cash flow back into capital assets, while a company that sees fewer growth opportunities will find another use for that money, either returning it to shareholders in the form of dividends or share repurchases, or, in the case of Clearwater, repurchasing debt. Now, Clearwater has only $420 million in long term debt against $840 million in equity, and its interest is covered five and a half times, so I do not see its debt position as particularly worrisome at present, or at least, that lowering its debt further is unlikely to result in a significant multiple expansion. But as the company has chosen the low-return strategy of debt reduction, the implication is that it does not see the possibility of higher returns from investing in additional capacity, and therefore the extra $30 million should at least for the moment be considered maintenance rather than growth capital, unfortunately.

Nonetheless, taking the 2023 free cash flow figure of $149 million as representative of Clearwater’s earnings power, and deducting $30 million for additional capital expenditures still produces a free cash flow yield of 15%, and the company is a reasonable candidate for portfolio inclusion.


DXC Technology: High Cash Flow, but Fundamentally Risky

April 29, 2024

DXC Technology Company is a provider of IT outsourcing services to other corporations, and it has been going through a rough patch over the last few years as shown by a dramatically declining share price. Investors have been holding out for a turnaround from the introduction of new management over this period, and perhaps at the moment they are seeing early signs of it. And in the meantime, the company’s free cash flow yield of 28% has brought it a lot of attention.

DXC offers a suite of offerings including IT outsourcing and consulting, data analysis, etc. The company was formed by the merger of Computer Sciences Corporation with HP Enterprise, and has spent the last few years whittling away at its noncore assets, which has made analyzing its year over year performance trends difficult, as has the fact that the company operates all over the world and apparently doesn’t bother to hedge the bulk of its currency exposure. However, the company has been kind enough to disclose “organic growth” of sales (or to be precise, “organic shrinkage”), and the trends on that appear to be leveling off. Moreover, as a result of this shrinkage, depreciation substantially exceeds new capital investments, making free cash flow significantly higher than earnings.

DXC operates in two sectors: Global Business Services and Global Infrastructure Services. GBS offers analytics and engineering to automate operations and analyze data, writing applications, and run insurance software and business process services in the form of bank cards, payment processing, etc. . GIS offers security, cloud infrastructure and IT outsourcing, and a modern workplace service.

For fiscal year 2023, Global Business services showed organic revenue growth of 2.4%, while Global Infrastructure Services showed organic revenue shrinkage of 7.2%, making a total revenue decline of 5.3% on a comparable basis. As stated before, divestitures and the strong dollar produces a larger overall revenue decline. At any rate, free cash flow from operations for the year came to $1.088 billion (the calculation is somewhat complicated by substantial swings in pension funded status, currency translation effects, and the inclusion of gains on divestments in “other income,” but I think I’ve sorted most of it out). This is an improvement on fiscal year 2022, where sales declined by 9.1% on a comparable basis, affecting both Global Business and Global Infrastructure, while free cash flows came to $490 million. I would also note that restructuring costs declined from $551 million in fiscal year 2021 to “only” $216 million in 2023 and are on course to be about $120 million in fiscal year 2024.

The share price has been declining for some time, but the first quarter of fiscal year 2024 (last August) severely disappointed the markets and drove the price down from $28 per share down to $20, where it lies today after drifting back up to $24 in the interim. In that quarter, organic revenue declined on a year over year basis by 3.6% and free cash flow from operations was $130 million as compared to $203 million the year before. However, $25 million of this difference was due to increases in contract onboarding costs and software (both of which DXC capitalizes somewhat aggressively in my opinion, but as I count them against free cash flow anyway it all comes out in the wash).

However, in the second quarter of fiscal year 2024, free cash flow for the quarter was $233 million versus $252 for the previous year, and in the third quarter, $276 million versus $164 million, making for $629 million year to date as compared to $619 for last year. Projecting this out to the entire year produces $838 million, which based on the company’s market cap and taking excess cash into account, comes to a free cash flow yield of 28% against an adjusted market cap of $3 billion. I should also point out that DXC recently won a lawsuit against Tata Consultancy for trade secret violations, with the jury recommending an award of $210 million, but that would have to be approved by a judge and presumably survive an appeal as well.

So, is that free cash flow yield attractive given DXC’s situation? Well, since sales declines seem to be flattening and restructuring costs are declining, we may say tentatively that DXC is approaching the size and configuration that the management has in mind. If we assume a terminal 5% decline in revenues, and that the company adjusts its operating and capital expenses accordingly, that works out to a multiple of 6 2/3, while a 28% free cash flow yield is a multiple of about 3.75, so there is apparently significant upside.

But on the other hand, there is the possibility that DXC could see a major exodus of clients which will make these past figures an unreliable guide to its earnings power, and it seems clear to me that DXC does not have the competitive moat that Warren Buffett looks for. That, combined with the ongoing declines, suggests that DXC could easily turn into a value trap despite its attractive free cash flow yield. In other words, there is not the asymmetry between the upside and the downside that makes for an attractive value play.


Greif Inc.: Promising but Recent Results are Worrisome

April 22, 2024

Greif is a packaging company that has a global operation to make industrial packaging out of steel and plastic, and a similarly-sized operation to make paper packaging in the United States, and also a line of paper goods. Based on its recent performance it appears to be attractive, but the results of the first quarter of 2024 are concerning.

The packaging industry is pro-cyclical to an extent, meaning that its fortunes rise and fall with the level of GDP, and Greif does cite a weakness in volumes and pricing as the cause of its recent drop in earnings, but this has not stopped the company from making several acquisitions of paper packaging companies in the United States in the last couple of years. If the industry itself is depressed and due for a rebound, this is sound strategy, but if not these acquisitions are highly questionable.

Turning to the figures, Greif has a market cap of approximately $3 billion and $180 million in excess cash on its balance sheet. Sales in 2023 were $5.2 billion as compared to $6.3 billion in 2022 and $5.6 billion in 2021. Gross margin in 2023 improved slightly, but overall operating income was $540 million, fiddling out the gain on disposal of a business. The company was careful to point out that in 2023 Russia represented 4% of its sales, 2% of its assets, and 9% of its operating profits. From the operating income we remove $96 million in interest expense, and applying estimated taxes gives us $355 in operating cash flow. The company also had $37 million in excess depreciation and impairment charges and $20 million in the earnings to noncontrolling interests, producing a final free cash flow of $372 million, representing a hefty yield to equity of 13.3%. The comparable figures in 2022 and 2021 were $519 million and $390 million respectively, again stripping out special or nonrecurring items.

However, in the first quarter of 2024 sales declined by a further 4% compared to the first quarter of 2023, mostly in the American paper packaging sector, but operating profit declined by 30% to $69 million. Interest expense was $24 million, and applying estimated taxes, operating earnings should come to $36 million. (I should point out that the company’s actual tax provision was actually minus $38 million owing to a loss upon repatriation of foreign assets, which I shall speak of later). Taking excess depreciation into account, free cash flow came to $42 million, as compared to $68 million for the same quarter in 2023.

The market seems to have taken this substantial decline in free cash flow in stride, just as it has the annual declines in free cash flow since 2022; the stock has remained roughly at its current level. I suppose it is recognition of the principle that one should not place excessive concern on one quarter’s or even one year’s earnings, particularly as the company and the entire industry, I understand, is already seeing early signs of an improvement. But on the other hand it is questionable whether a business that is subject to such swings is stable or predictable enough to constitute a genuine value investment.

Also, I mentioned earlier that the company’s actual provision for income taxes in the first quarter of 2024 was minus $38 million instead of the $9 million one would expect. This is the result of the company’s “onshoring of certain non-U.S. tangible property,” as stated in the 10-Q. I am tempted to call accounting shenanigans on this move, as the timing of this “onshoring” is totally within the company’s discretion and seems to be timed specifically to make a disappointing quarter look better.

There is one point that concerns me as to Greif’s corporate structure. The company has two classes of shares. The class A shares are entitled to two thirds of the dividends of the class B shares, and are not entitled to a vote unless the dividend has been passed for an entire year. And yet, the class A shares as of this writing trade for $61.36, and the class B shares trade for $62.73, which is barely a year and a half of dividends more. On the one hand, this is further evidence that dividend policy is no longer a major influence on valuation and that the company is seen as being run tolerably well to the point where control is a non-issue. But I still think such a narrow spread between the two share classes is surprising, and naturally I would recommend the class B shares.

So, if one has confidence that the earnings of Greif are temporarily depressed and the company’s perception of improving conditions will return the company to historical performance levels, then the current price is attractive. But as such a view is dependent on potentially optimistic assumptions about the future, I cannot say that Greif is an obvious value play.


Why does everyone hate Tegna stock (apart from the obvious)

April 8, 2024

Tegna, formerly Gannett, is a broadcast TV company (that’s the obvious part), which the market has been incomprehensibly punishing despite its free cash flow yield of 18% or potentially higher. In fact, a recent merger agreement valued the company at $24 (the stock is at $14 as of this writing) failed for regulatory reasons, not for any problems with the deal itself. And of course we have the 2024 election season coming up, which is always a lucrative time for broadcast television.

The reason I say that the stock is hated is that Tegna expended $600 million on share repurchases in the last couple of years (out of a market cap, adjusted for excess cash, of $2.2 billion as of this writing), and yet the stock has not moved. The company’s announcement of $600 million more in buybacks in the next two years have also left it unmoved. It is possible that the post-merger slump has chased away investor interest, but even so, a free cash flow yield this high cannot be ignored forever, as management in its 10-K has committed to returning half its free cash flow at least to the shareholders. The repurchases are a positive sign, as a company that is undervalued should be using its cash flow to repurchase stock, while a company that is fairly valued should be paying dividends, and a company that is overvalued should be focusing on using its stock to acquire other companies.

In fairness, broadcast television is a declining industry, but not necessarily a dying one anytime soon (and don’t forget that cable companies do pay local broadcasters to carry their content). But even a declining company can throw off considerable free cash flow, particularly as depreciation and amortization tends to outweigh new capital expenditures.

Turning to the figures, the election cycle tends to make year over year comparisons difficult, so it would be fairer to compare 2023 to 2021. In 2023, sales of $2.91 billion declined by 3% as compared to 2021 and operating income (not counting the merger termination fee Tegna earned because it is obviously nonrecurring), declined by 25%, which management attributed partially to a more difficult business environment that affected advertising revenue, as well as higher programming costs. Even so, taking excess depreciation into account, free cash flow for 2023 came to $400 million. The comparable figure for 2021 was $634 million, so one can see why there is market pessimism. The non-political advertising situation may improve in 2024 or it may not, but $400 million still represents a yield of 18% on what is essentially an off-year. For the election year of 2022, sales were up 8% compared to 2021, operating income was up by 25%, and free cash flow was $782 million. So even if the decline in free cash flow yields is permanent, political advertising in 2022 added $280 million to Tegna’s sales, and the Olympics also occur in even-numbered years, so 2024 is likely to improve on that $400 million figure to the tune of more than a hundred million easily, particularly as 20% of Tegna’s reachable households live in swing states.

Of course, I wouldn’t be me if I didn’t point out an accounting shenanigan. In the 2023 10-K Tegna reported the free cash flow figures for the two year period ending in December of 2023, and again for 2022. It is fair to present an odd- and even-numbered year together, but a reader who skims might miss the two year figure and conclude that free cash flow was over one billion in a single year, and furthermore might miss the significance of the even-numbered year being counted twice and the smaller, odd-numbered years being counted once each. The company computes that free cash flow in 2023 was $120 million less than in 2021, as their calculation differs somewhat from mine.

And for what it’s worth, the company’s bonds are rated BB+ or just barely sub-investment grade, and according to the St. Louis Fed, the current option-adjusted spread for BB bonds is just under 2% as of this writing, suggesting that new bonds (such as the ones Tegna will have to issue to refinance its existing ones in the next few years), would yield about 6.4% based on the current 10-year Treasury rate, so I do think an 18% cost of capital (or higher given that 2024 is an even year) is unrealistically demanding.

So, even in the absence of the catalyst of a merger I do think Tegna’s substantial free cash flow yield must be recognized by the market eventually, and I recommend it as a candidate for portfolio inclusion.

Disclosure: At the time of this writing I held shares in Tegna.


What exactly is China doing with all our money?

April 2, 2024
Tags: ,

In case you haven’t noticed, China runs a substantial trade surplus with the world and the United States in particular, and its foreign reserves come to $3.3 trillion dollars equivalent, and just under $1 trillion US. Actually, the US holding has declined slightly from its record, partly from diversification and partly from interest rates going up, I speculate. However, there is considerable speculation as to what China intends to do with that money. Certainly not turning around and buying goods and services from the United States; that would defeat the purpose of the trade surplus. Is there a plan to use it in some nefarious plot to destabilize the US economy or the world?

In my opinion, probably not. China’s currency reserves are a side effect of China’s overall economic policy, and may indeed be causing them a bit of a headache. The large bolus of foreign currency is just sitting there, waiting to cause a burst of inflation in China if the country ever decided to something with it.

In order to understand what’s going in China, a good starting place is, of course, postwar Japan, which China studied in order to crib its development plans. Consider first the concept of autarky. Autarky is the notion that an entity, such as a nation, community, etc., should be able to provide all of its needs with resources that it already controls. For a nation the size of China, or a supranational organization like the Warsaw pact, it is theoretically feasible, while of course attempts to try it in North Korea or Pol Pot’s Cambodia ended in disaster. The idea has waxed and waned in popularity over the years and is not entirely associated with Communist regimes.

Prior to World War 2, Japan’s effort to organize the Greater East Asian Co-Prosperity Sphere was itself an autarkic effort, but after the war Japan itself did not have the natural resources to engage in autarky. However, per R. Taggart Murphy’s excellent The Weight of the Yen, it was Japan’s economic policy was instead to switch to financial autarky, by building up its capital base without relying on foreign investment. The key method was to force the Japanese to save a lot of money while keeping interest rates low so that firms would have profits to reinvest and money to pay their large debt balances, and allowing the economy to become cartelized so that the profit margins of the businesses were protected so the high degree of leverage to fund capital investments did not result in bankruptcy. It also required international capital controls so that Japanese firms and citizens had no alternative but to invest in Japanese assets.

I believe that key elements of this policy were followed by China, with the added convenience that China was a command economy and could use heavy-handed regulation and state-owned enterprises rather than trying to force the outcome through a nominally capitalist system.

The economic effects of this policy, if it is effective, is to lower consumption in favor of investment, and to lower interest rates because there is more savings chasing investments relative to what businesses would normally have to offer savers. Now, in a vacuum the chief determinant on demand for capital goods in a free economy, other than interest rates, is the future course of the aggregate wage, which will impact future consumption. If the policy is expected to be unchanged, with consumption being artificially suppressed it is inevitable that there will be excessive industrial capacity and also excessive unemployment since there is no need to employ people to produce goods that no one will buy.

So, what to do? Enter the trade surplus. If the economy can undercut other nations and export to them, the excess capacity and unemployment can both be solved. The excess capacity problem is solved automatically, as the lower required return on assets will inherently outcompete higher priced foreign producers, while the lower interest rates weaken the currency, making your exports cheaper as well. The enforced lower standard of living also lowers production costs, making one’s exports even more competitive. The only downside is an artificially low standard of living, which is a surprising policy goal for a Maoist regime, but here we are. Apparently, China sees this as a worthy price to pay to avoid inflation and instability.

So, the accumulation of dollar-denominated assets is really a side effect of China’s policy, not their intended goal. In fact, by increasing demand for foreign assets the effect is to lower interest rates in China’s trade partners, which in a free market would make them weaker and the yuan stronger. However, China is capable of keeping its currency peg in effect through capital controls. In theory, the lower interest rates facilitated in our country could result in inflation, so the Chinese are exporting inflation as well as importing employment.

One wonders, though, if at some point China will determine that the capital stock of the nation has become adequate and there is scope to reverse this policy. I think it will be a difficult adjustment, in the sense that there vested interests resulting in policy inertia, partly the fear of releasing inflation, and part of it simply that capital is not fully interchangeable, and goods desired in wealthy foreign countries are not suitable for the domestic market. The overhang of uncompetitive state-sponsored enterprises would also have to be dealt with.

Even so, we are seeing signs that this development is taking place. China is now allowing its trade partners to settle its accounts in yuan rather than dollars, and has opened swap facilities with many trading partners, but since the source of the yuan is the Peoples’ Bank at the official Peoples’ Exchange Rate, this seems to be more of a method to avoid having to convert foreign currency into dollars and then into yuan than to actually turn the yuan into an international currency. This same facility is also allowing foreign businesses operating in China to borrow money in yuan as long as the proceeds remain in the country. Moreover, the recent weakness in the yuan is caused not by their policy but by the Federal Reserve and the ECB raising interest rates to tamp down post-COVID inflation. Even so, these steps are in my view not indications that China is ready to allow the yuan to be a fully international currency, or to abandon its trade surplus uber alles policy, and certainly these liberalization steps could even be reversed by administrative fiat, potentially leaving many trading partners in the lurch.


Trump Appellate Bond Reduction Unsupported by Law (in Two Senses)

March 26, 2024

I know that normally I am known for my investing and economic insight here, but I am also a lawyer and this has just come up. The decision by the New York Appellate Division to reduce the bond required to stay enforcement of the judgment against Donald Trump is incomprehensible. I mean that literally. I cannot even say that it is based on faulty legal reasoning; it is, as far as I can tell, based on no reasoning at all.

The New York Appellate Division issued a ruling reducing the required bond to stay enforcement of the state of New York’s judgment against him (called a supersedeas bond) to $175 million from $454 billion as calculated by statute. Notable in this ruling is the lack of any justification or analysis. Perhaps the court will issue a memorandum later, but this order alone contains not one shred of legal reasoning to justify the appellate court’s order. I have not been able to determine whether Letitia James will request leave to appeal the Appellate Division’s order to the Court of Appeals, but if so I can surmise the lack of any reasoning by the Appellate Division may become an issue.

Frankly, if the court had examined Trump’s reply brief, they would have found that he has been playing fast and loose with the decisions cited. Trump’s lawyers did cite certain cases wherein the supersedeas bond was not enforced fully, but all of those cases involved differing factual circumstances that did not apply here.

The first case cited was Texaco Inc. v. Pennzoil, and in fairness the court did reduce a $12 billion bond to $1 billion, but the court’s reasons were that it had been persuaded by evidence that there were not more than $1.5 billion in surety bonds available in the entire world (in 1986), and that the immediate enforcement of a judgment that was likely to be reduced on appeal anyway would cause apocalyptic havoc on the fifth largest company in the nation, affecting tens of thousands of people. Donald Trump has not even represented that he or any of his companies would be forced into insolvency by the enforcement of this judgment.

Trump’s next case was In re Adelphia Communications, which allowed an appellate bond of $1.3 billion against 111 million shares of stock, 9.4 billion tradeable interests, and $7.136 billion in cash. However, the decision under appeal in Adelphia was the final distribution of a Chapter 11 bankruptcy case, under which the shares were to be distributed to 14000 potential shareholders and the other assets to 10,000 interested parties, making it essentially impossible to track them down again if the decision was reversed on appeal. In Trump’s case, however, there is only one recipient of the res of the lawsuit: the State of New York; a key distinction that Trump’s lawyers conveniently leave out.

Trump’s next case is Cayuga Indian Nation v. Pataki, which did waive the bond requirement entirely, but in that case it was the State of New York that requested a waiver of a bond, and the court was persuaded that New York’s taxing power would suffice to provide adequate assurance of the collection of damages, and also there were some constitutional difficulties in a federal court imposing a bond requirement on a sovereign state. Furthermore, the case also contains language to the effect that it is a “well-established principle that quantifiable money damages cannot be deemed irreparable harm [citations omitted] Because the judgment herein is only for ‘quantifiable money damages,’ the State is unable to establish this particular stay element.” Or at least, not without some additional facts adduced by the appellant.

The next case up is International Distribution Centers v. Walsh Trucking, which is again complicated by a bankruptcy. The corporate defendant in this case had declared bankruptcy but the five individual defendants had not, and the court here declined to impose a bond requirement on the non-corporate defendants. This case at least is on point, but the Trump case has not yet been complicated by the bankruptcy filing of any defendant.

The next case Trump cites is TWA v Hughes, which his lawyers characterize as “granting a substantial reduction of the bond amount where “[b]ecause of the unprecedented size of the judgment, the obtaining of a supersedeas bond was impracticable.” However, Trump’s lawyers conveniently leave out that the judgment in TWA was for $145 million and the appellant was allowed to make a $75 million bond and satisfied the balance by stipulating to the condition that his company would maintain a net worth of at least three times the $83 million remaining, as determined by an independent auditor. In other words, the company still provided the additional assurance required, just not in the form of the bond. There is no sign in the instant case that Trump is even offering to put any other assets on the table.

Finally, Trump cites C. Albert Sauter v. Richard S Sauter, which was a federal anti-trust case. The court in Sauter did not require a full undertaking to be posted because “execution is most likely to terminate Richard S. Sauter Co., Inc. as a going concern and eliminate it as a competitor in interstate commerce,” and further required the defendants to escrow what looks like a significant chunk of their financial assets to the court, including all of the shares of Richard S. Sauter, Inc. Donald Trump has faced no such escrow requirements and his appeal does not even represent to the court that the judgment would require the insolvency of the Trump Organization.

In short, then, the cases cited by Trump and his co-defendants in the reply brief simply do not take him where he needed to go in order to meet the legal requirements for a modification of the appellate bond requirement (and frankly, Letitia James should have pointed out Trump’s attorneys’ sloppy research in her surreply brief), and this being the case, the reasoning of the New York Appellate Division is incomprehensible. Which is presumably why the Appellate Division didn’t bother to provide it in the first place.


Deriving a zero yield curve from a par yield curve

January 11, 2024

Have you, like me, ever wondered how you might derive the zero curve from the par yield curve but don’t know where to start? If so, I have good news for you.

The yield curve, or more formally, the term structure of interest rates, is a vital tool for any fixed income analyst, and also of non-trivial interest for anyone who has anything to do with borrowing money. But how to construct the yield curve? What yield to use?

A good place to start in the United States is the yield on Treasury instruments, because Treasuries avoid the issues of credit risk, liquidity risk, and callability that plague other bonds, and those other risks can be assessed later. But even Treasuries offer their choice of yield curves. The first yield formula they teach you at yield curve formula school is the yield to maturity, but that formula assumes that all interim payments are reinvested at the same rate, which is a highly questionable assumption unless the yield curve itself is flat, which it never is. So one would expect the yield to maturity to have fallen by the wayside decades ago.

There is, however, a measure of yield that is completely agnostic as to reinvestment assumptions, and that is the zero curve. The zero curve treats every coupon from the bond as a separate maturity with no interim coupons to reinvest, so it is a much better approximation of the one true yield. One could, of course, call our friendly neighborhood investment banker or fire up the old Bloomberg for quotes on zeroes of various maturities, but there are data quality issues and not all of us have an investment banker on call.

Fortunately, the US Treasury, posts a daily yield curve we can use. Unfortunately, the Treasury publishes the par yield for any coupon bonds, not the zero curve. The par yield is a measure of what coupon rate would cause a bond with the stated maturity to trade at par, which Treasury assures us that they calculates using the finest and rarest of algorithms. The trouble is that as far as I can tell, the par yield can only be constructed from the zero curve, so the US Treasury has just added a step to our process that must be patiently undone. Fortunately analysts are accustomed to doing this anyway when dealing with accountants when examining a company’s financial reports.

Furthering our inconvenience, the Treasury doesn’t publish every rate, or even every yearly rate. It is kind enough to give us the six month and one year par rates, and it is simple algebra to compute the zero rates on a one year bond from these inputs. (I say simple algebra, but the one year rate must be solved numerically, which Excel is perfectly capable of). The method is essentially the same method as bootstrapping a forward rate curve.

However, the next par rate published by the Treasury is the 2 year rate, which is a problem because Treasuries pay semiannually, which means that even if we have the 6 month and 1 year rates, there are two interest rates to solve for here with only one equation, meaning that we have an infinite number of solutions.

Let us suppose, however, that the 18 month rate and the 2 year rate are mathematically related to each other in some manner, such as linearly. In other words, the 1 year rate + r equals the 18 month rate, and the 18 month rate plus r equals the 2 year rate. Brilliant, you think; we have just added a third variable to our problem. But in fact we have defined the 18 month rate and 2 year rate in terms of the one year rate and a single variable, making this an equation that we, and by “we” I mean Excel, can easily solve. And we may continue in this vein with the other stated Treasury par yield dates to fill in the entire yield curve for 30 years.

This will give us a nice chunky polygon which will somewhat resemble the par curve but generally with a slightly exaggerated slope. This is the one prepared using the published Treasury par yield curve of January 2, 2024.

The blue is the par yield curve; the orange is the calculated zero curve. It is usable, but I think there is room for improvement. It makes no sense to me that a yield curve should be discontinuous at each Treasury-published rate, as in my view a yield curve develops organically from the views of market participants and even if some participants are bound to bonds of a particular maturity, other participants are not, and even if the holder of a 2 year bond may not care about the yields on a 25 year bond, they should care about the yields on a 3 year bond, so some smoothing is in order.

So, what can we do? We can enforce continuity by assuming that the outgoing slope of one yield curve section will equal the incoming slope of the next section, and within the section use an approximation other than linear. The Treasury itself uses cubic hermite splines, but after extensive fiddling in Excel I’ve found that at least a sensible-looking graph can be created by using two methods, one by using a natural log function instead of linear, and also allowing the slope of the graph to shift from the incoming slope in each section to the slope implied by the next section (or zero, in the case of the 20-30 year), because I found that a pure backward-looking method was producing distortions because, as stated above, the 2 year and the 3 year interest rates, for example, should have some influence on each other because at least some market participants are capable of choosing between the two when constructing their portfolios.

Unfortunately, the Treasury does not publish rates between the 10, 20, and 30 year rates, and the unusual shape of the yield curve of January 2, 2024, which inverts, uninverts, and then reinverts, makes a wonky shape of the 20-30 year section no matter what reasonable modifications are attempted. I assume the natural log term is useful because, based on my observations, the yield curve tends towards flatness as opposed to more extreme slopes as yields extend into the future, and anyone who claims a genuine insight as to whether the difference between the 28 and 29 year rate should be larger than the 27-28 year rate spread had better have a very good reason for making such a claim.

So, now that we have this cool curve, what are the applications? The benefit of casting the yield curve in sections of functions with defined coefficients that we have solved for is that we can calculate the interpolated rates at any future date, not simply at six month intervals, which will be helpful for the 363 days a year which are not the coupon days. And with these zero curves on any particular date we can apply them to any corporate, as opposed to Treasury bond, and neatly and accurately compute the zero volatility spread offered by that particular bond, in order to search for anomalies that may represent attractive investment opportunities, or, more ambitiously to work out a term structure of zero volatility spreads across several bonds, which may be of interest when constructing a fixed income portfolio, or even to evaluate multiple bonds of the same issuer to see if the market is anticipating some change in the financial situation of one issuer over time.

Also, getting back to yield to maturity, we can also calculate any implied forward rate we like. I said the flaw with the yield to maturity is that we assume that all intermediate cash flows must be reinvested at the same rate. But with this method we can instead assume that intermediate cash flows can be reinvested at the forward rate implied by the yield curve, and discount that future value back to present value, thus possibly identifying attractive opportunities among Treasury instruments or among corporate instruments with the same credit spreads, or even work out a term structure of the gaps between yield to maturity and forward curve implied spreads. I do not recall seeing anyone using this method, but it is a way of sieving out anomalies, and for analysts, anomalies are in theory worth investigating.


Inflation: Does Say’s Law even Apply?

September 14, 2021

I was watching the latest hearing of Jerome Powell before the Senate Banking committee (as one does), and as expected the Democrats were congratulating themselves on unemployment being so low and workers finally having enough bargaining power in the labor market to acquire higher real incomes for the first time in forever. And then both parties added “And this is causing a notable uptick in inflation rates, so we’re hoping you can put an end to these developments as soon as possible.”

Naturally this led to thoughts about the aggregate supply and demand, and whether it is actually possible to get growth without increased prices and whether supply or demand is in the driver’s seat. And naturally my thoughts turned to Say’s law, beloved of supply-siders, which holds that supply creates demand. There is a certain logic here, since one cannot demand a good that is not supplied, although it is equally valid to say that one can be perfectly ready to supply a good, it will not be supplied until it is demanded.

But, consider an ordinary transaction whereby person A buys $1000 in goods and services from person B. Then B can buy $1000 in goods and services from C, and so on in that fashion throughout the entire world, so that one person’s decision to buy $1000 in goods, or $1 in goods, or 1 penny in goods, will echo around the world and produce an infinite GDP. And yet, GDP is obviously not infinite. Why?

Obviously, part of it is that it takes time to figure out what to demand and whom to demand it from, but that isn’t an issue in the time-compressed world of basic economic models (aside: economic models would be more acceptable if they all ended with the word “eventually”). But to Say, I suppose, the more fundamental reason is that the economy’s ability to supply goods beyond what is demanded is physically limited, and so GDP cannot rise beyond the economy’s ability to supply it.

The bigger problem, of course, is the invention of money. Keynes himself criticized Say’s Law as only valid in a barter economy, because with money it is feasible to supply a customer well in advance of demanding anything from him or anyone else. And since the marginal propensity to consume declines with income, obviously some of the money will be diverted to savings, so there’s a perfectly sensible reason why demand will eventually burn itself out.

So, as we all know from Econ 101, the supply curve is positively sloped and the demand curve negatively. However, those constructions do not rule out the possibility that the curves are horizontal or vertical. I do not believe that the aggregate demand curve is horizontal, at least until we have invented the Star Trek replicator, but it is conceivable that the demand curve is vertical at least in the short term, because, per Keynes, demand for consumer goods is a function of income, and demand for capital goods is a function of a lot of things, including the expected course of future income and opportunity costs, and setting aside “animal spirits,” those things do not change unless future developments change them. At any rate, it can be treated that way.

And what is the shape of the aggregate supply curve? Presumably it is not always horizontal; if it were possible to have any level of supply at a given price then we would just as soon have a high supply as a low one and the senators from the banking committee above would not be concerned with inflation. But the question remains, is it vertical?

However, one wonders if we are having Say say more than Say said. It is one thing to say that quantity supplied equals quantity demanded; it is quite another to say that supply = demand. The first statement is true by definition; the second one depends on assumptions that must be examined. Also, I was talking earlier about the relationship between GDP and price level, and if the supply curve is identical to the demand curve, price entirely falls out of the equation.

I would say that this is not the case, or at least not all the time. One of the implications of Say’s law is that ultimately supply cannot exceed demand since supply is what creates demand, but as is said, severe recessions and specifically the Great Depression would seem to refute this view, and if your economic model cannot predict something that has already happened, some revision is in order.

Fortunately, we have wikipedia to come to the rescue. In their view the supply curve is divided into three sections. In the Keynesian section, where severe recessions live, the supply curve is nearly horizontal, and any additional consumption is just drawing up slack in the market. We have seen this in our most recent financial crisis, wherby unemployment was halved, GDP increased for ten years straight, and somehow inflation actually came in below expectations. It would be tempting to think of the Keynesian section as a free lunch, but remember, the concept of the Keynesian section is drawing up the slack in the economy, and the slack had to have been created before the recession it. In other words, the lunch was already paid for but no one wants to eat it.

In the intermediate section, higher GDP does come at the cost of higher prices. I would suggest that this is the “normal” state of the economy, where there is a tradeoff between GDP growth and inflation and where we are living most of the time and the Federal Reserve tries to keep us.

But what is that vertical bar over there on the right side of the graph? It’s the “classical” section, where the supply curve is in fact vertical: the economy is firing on all cylinders and there is no slack to draw up. So, as discussed above, the GDP is in fact independent of the price level; Keynesian stimulus in this situation could increase inflation potentially infinitely without affecting the real GDP. I would argue that this is what was going on in the 70s’ stagflation.So, despite the temptation to throw Say’s Law onto the heap of failed ideas, it is clear that under some theoretical and even real-world circumstances where it would apply, and having made those circumstances clear we are in a better position to use it as a guide to policy.


In a Volatile Market, Step Back and Read

January 12, 2019

The market has been going through some troubling times lately, and the financial sites are full of doomsday articles (although after a few quiet trading days in January, people seem to be in a much better mood). I can readily understand the desire to avoid making bold long moves in the market, as we may be in the last gasps of a bull market (just as we were last year, and the year before that).

Personally, I have not been inordinately worried about the historically high valuations because of the unusually low level of long term interest rates. Most financial articles I’ve read take pains to point out that bond prices and yields go in the opposite direction, but they never bother to point out that stocks and yields work the same way. At least in terms of basic financial theory, stocks are priced the same way as bonds in that expected future cash flows are discounted to present value, and to the extent that recent market movements are the result of expected long-term interest rate increases caused by normalization of Federal Reserve policy, a drop in valuations is only to be expected. My thinking is that we have less of a bubble and more of a balloon that can have the air let out of it without popping.

But, if people aren’t buying aggressively, they certainly need something to fill in their free time, and that makes me think about books. And not just the quality investing/finance/economics books (though I have recommendations in that vein), but in general.

A lifetime ago, when I was in college, it was natural for students to sell their textbooks back at the end of the quarter, and inevitably some books could not be sold back, possibly because a new edition came out, or because of the book’s condition, or (ideally) the course it was for was simply not being offered next quarter and the bookstore had no use for it. There was a large table next to the buyback window where the disappointed students could just dump their unsold books for anyone who was interested. And I, being an enterprising lad, would just scoop up those books by the armload to sell them online. They say that value investors are born, not made, and it seems to me that profiting off of other people’s cast-off books is not so different from profiting off their cast-off stocks. At any rate, having my rent paid was worth having every wall of my apartment taken over by bookcases.

I don’t know that such a method would work nowadays. Textbook publishers use every trick in the book to come up with either frequent new editions or course-customized editions that cannot be resold; campus bookstores are getting better at finding a home for unwanted books anyway, and finally the used books at are dominated by high-volume resellers such as Thriftbooks, Half Price Books, the various Goodwills, etc., who have a script to go through and undercut each other by a few pennies. And in the section where the condition of the book is supposed to go, they just copy-paste boilerplate descriptions or just a blurb about their own company; you often have to scroll through several pages of used offerings to find a single offer that looks like the seller actually had the book in their hands when they wrote it.

Even so, I still accumulate books, for use if not for resale. And my favorite source is library sales, where libraries sell off partly their discards, but mostly the donations they receive, typically at very reasonable prices. As Charlie Munger says, he has never known anyone who is successful in a field requiring broad knowledge who does not read pretty much all the time (an object lesson for more than a few politicians), and in terms of dollars per hour, used books are pretty much at the top of the list of ways to pass the time.

However, at the library sales also see a curious phenomenon of people going through with barcode scanners, feeding the UPCs of books to a mobile app which looks up the prices of books on or other sites, in order to buy and resell them. This activity makes no sense to me for various reasons. Even setting aside the current state of used book sales,  most libraries sort through their donations for books that are worth selling online anyway, and either sell the valuable books themselves or set their sale prices accordingly. Furthermore, these online lookup tools are ubiquitous enough that any book being scanned at a library sale has no doubt been scanned and rescanned by other sellers, and been rejected by them. In other words, the only way for a seller to find a worthwhile book with these scanning tools would be to have lower standards, in terms of profit margin, than any other book scanner in the place, and I would think that low standards are a poor way to justify the time, effort, and storage space required of keeping an inventory of used books.

And then I think back to investing. Isn’t it true that buying a financial instrument is also a battle of who-has-the-lowest-standards, since any stock or bond has been examined by hundreds or thousands of market participants who has declined to pay a higher price? On the whole, though, I think not. The reason that scanning books at book sales is so unattractive is that there are only a few variables involved: the book’s sales rank at amazon and its lowest selling price. But with securities, there is not only the buying and future selling price, but also the cash flows generated by the security in the meantime, which of course influence the buying and selling prices, but it must be conceded that the market is quite capable of being far off the mark in either of them. And it is this characteristic (plus portfolio management issues such as beta, correlation, etc.), that makes the two distinct. It is the same old distinction between an investor, who studies a situation to find a reasonable return on investment with safety of principle, and a speculator who simply hopes that they can find someone who will pay more tomorrow than what they paid today.


Cypress Semiconductor and Value Line: Do Your Own Analysis

March 15, 2017

Warren Buffett, when asked where he gets his investing ideas, was fond of replying “Start with the A’s.” This may seem like unhelpful advice now that so much data requires clicking around and waiting for webpages to load, instead of wading through nice archaic books. However, the Value Line Investment Survey still publishes data in a multiyear format. This allows one to perform visual scans of financial performance, with more room to identifying trends, results, and ratios that are in the “neighborhoodof acceptable for investment purposes with more flexibility and “fuzziness” than could be accomplished by a database query. Although I do tend to start with the Z’s so I can get to them before Buffett does.

Yes, Value Line is generally a useful resource while screening for investments to analyze, but it cannot replace the importance of the analysis itself, particularly if the reported figures must be significantly adjusted. Value Line is partly a statistical service and partly an analyst firm, so they make adjustments to enhance comparability and clear out some confounding outcomes mandated by accounting rules. In most cases this adjustment has its intended effect, but sometimes, when dealing with companies that have substantial equity compensation, mergers, events generally viewed as nonrecurring, and other controversial matters, the effects of their adjustments can be large and distortionate.

Wafer ThinFor example, Cypress Semiconductor (CY) makes a wide array of memory products used in telecom and data communications equipment and the Internet of things. I should point out that I’m not an expert in the semiconductor industry, but it seems to be characterized by intense competition for price and margins and a great deal of merger activity. Cypress has indeed undertaken two mergers in the past year for a combined purchase price of over $2 billion for a company that has a current market cap of $4.66 billion as of this writing.

Dealing with mergers is an endless pain for analysts. Particularly when the mergers are large, sales and margins become incomparable to historical figures and even the capital structure of the company is significantly altered. In Cypress’s case, sales went from $725 million two years ago to $1.92 billion last year. Analysis is further complicated by the fact that companies may acquire other companies as a substitute for capital expenditures, which makes depreciation rates somewhat unreliable. This is a  particular problem for the amount of the purchase price that represents goodwill, which does not amortize but instead is tested for impairment. The effect is to make both depreciation and capital expenditures seem much less smooth and difficult to rely on for projections.

More than that, though, mergers are expensive both in investment banking and legal fees and in the inevitable restructuring that follows, and we analysts have to determine how to handle those costs. On the one hand, most companies are not in the business of selling products, not merging, so it would only be fair to treat these expenses, substantial though they are, as nonrecurring for purposes of calculating the firm’s overall earnings power. On the other hand, some companies make such a habit of merging that these expenses come to resemble a regular part of their operations. Other categories of nonrecurring expenses commonly ignored are impairments and writeoffs of goodwill, the latter of which is especially common in frequent acquirers.

As I stated before, the Value Line Investment Survey attempts to adjust these nonrecurring expenses, but their adjustments are not documented, at least on an individual basis. As a result, Value Line concludes that Cypress earned 21 cents per share in 2015 and estimated that Cypress would earn 48 cents per share in 2016. The actual figures were losses of $1.25 and $2.15, respectively. Furthermore, using Value Line’s earnings per share, capital expenditures, and depreciation figures to estimate free cash flow, one arrives at an estimate of $264 million in free cash flow in 2015 and a forecast of $360 million in 2016. And since Cypress’s current market cap is $4.66 billion, this represents a free cash flow yield of 7.7%, which appears to me to be high for a semiconductor company and also fairly attractive under modest assumptions about growth.

I have performed an estimate of the Cypress’s free cash flow based on the figures in the 10-K in my customary manner. Even giving the company every advantage by treating restructuring, merger-related expenses and inventory writeoffs, and impairments as nonrecurring, I arrived at an estimated free cash flow of $125 million in 2015 and $229 million in 2016, a discrepancy of well over $100 million for each year.

Futhermore, diving into Cypress’s 10-K filings, I discovered what I can only describe as a financial shenanigan. In the books I’ve read about financial shenanigans, such as Financial Shenanigans by Schilit & Perler,  one thing to watch for is a company trying to juice sales growth by accelerating the recognition of sales. The difference between a shenanigan and a fraud, of course, is that the shenanigan is fully disclosed in a footnote in the boring part of the 10-K. And in Cypress’s case, over the course of 2015 and 2016 the company began to recognize some, and eventually all, shipments to their distributors as sales, rather than their previous policy of recognizing sales upon shipment from the distributor to the end customers. The company justifies this treatment by concluding that, with suitable reserves, sales by distributors were sufficiently reliable to allow this treatment. Whether or not Cypress is correct in this view or not, the effect of adopting this treatment was to increase sales by $40 million in 2015 and $59 million in 2016. A financial shenanigan is mostly harmless to the actual figures once discovered and taken into account, but a company adopting them in the face of two large reported losses should tend to raise one’s level of suspicion about the company’s management and financial reporting overall.

Now, from this article one might be left with the impression that I view the Value Line as defective and unreliable. This is not the case; I find Value Line’s figures generally much more reliable, at least as a first approximation, and Value Line’s analyst commentary is generally useful, although I tend to disagree with its timeliness ratings. My point, though, is that investment analysis cannot consist solely of taking numbers from a database and fashioning them into a portfolio. There has to be genuine digging into the financial reports to assess the reliability of the numbers and in the form that an analyst finds most useful.

As for Cypress Semiconductor, I cannot say that it is attractive as a value investment because of its not-outsize cash flow yield and heavy reliance on acquisitions. Value Line and other analysts are optimistic about Cypress’s growth prospects and future developments may justify this optimism, but a value investor would find it very dangerous to indulge growth assumptions, because under sufficiently high growth assumptions, no price is too high.