March 17th, 2015
June 7th, 2014
I spent 13 years in Silicon Valley software businesses, knew venture capitalists, and pitched to them dozens of times. Here is my impression as a value investor.
Venture investors are usually brilliant at operating aspects of new businesses — competitive strategy and marketing — but ironically not always at investment analysis. Many, of course, have a deep understanding of value. But some, through optimism, ignorance or self-interest, may persuade themselves that alternative metrics will eventually turn into traditional value. They sometimes do not grasp what is meant by traditional value (meaning either net assets or earning power), nor why it would be relevant to their own work.
This view has worked for them for decades, so who am I to say it is wrong? All I can say is that it doesn’t make sense in the context of business history.
Historically, more often than not, the best startups are profitable almost immediately, on relatively low investment. This is even generally true of businesses you would think of as high-capex today: Ford and Chrysler went profitable early, on low investment.
In the modern era, this has remained true. Apple, Dell, Microsoft, Cisco, and even Google to a lesser extent, were profitable almost from the start, requiring little capital up front.
Tech investors themselves seem not to notice the disconnect: today’s tech titans, whose huge returns made modern VC investing possible, were profitable almost from inception, and hence do not resemble modern VC investments, which suffer years of losses before selling still-profitless companies on to strategic buyers or public markets.
The widespread perception of startup investing — that a normal profile involves years of losses, eventually leading to high long-run returns — is mythical. It’s the exception, not the rule.
The isolated exceptions include Fedex, where years of big initial losses eventually paid off. More commonly though, big upfront business investments in an unproven business usually end up impaired, or do not soon translate to cash flows that are large relative to total investment. Think 1860s railroads, 1930s airlines, AOL, Amazon, LinkedIn, and so on. A sound and fury of endless capex or customer acquisition.
This approach has worked for venture investors who are not value investors, at least since the mid-1990s, when interest rates fell low enough to create huge incentives for speculation.
History would suggest that, when interest rates rise to normal levels, investors will demand higher returns for speculative ventures, reducing exit potential for unprofitable companies. This would force traditional value principles onto the industry, and probably make it smaller.
October 29th, 2012
In 1930, John Maynard Keynes wrote an essay entitled “Economic Possibilities for Our Grandchildren” (free PDF from the Yale economics department). In general, it presents an optimistic long-term economic outlook, but in his short-term prognosis — presented at the very beginning of the Great Depression — Keynes writes of “technological unemployment,” which he defines as “unemployment due to our discovery of means of economising the use of labour outrunning the pace at which we can find new uses for labour.” This is interesting, because Keynes reveals that even he, and even then, saw the Depression resulting from non-monetary causes. When productivity rises faster than people can adapt, mayhem ensues.
Perhaps it is this force, rather than any “debt supercycle,” policy decision or monetary force, that is the ultimate cause of sustained global economic slump. Occasionally, labor productivity rises faster than organizations can adapt and faster than people can retrain, creating years of industrial discontinuity. The entire economy must refactor itself around a new reality — during which large numbers of people remain unemployed for lack of relevant skills. Let’s call such conditions not a Great Recession, but rather a Great Refactoring.
Keynes saw this only through the lens of the 1920s, during which farm automation and factory electrification triggered a huge increase in labor productivity. At the end of that tunnel was a golden age of jet planes, tailfinned Cadillacs and swaggering Rat Packs — but first came 15 years of depression, revolution and war.
The Great Refactoring hypothesis also goes a long way to explain the developed world’s recent long-term slump, and its resistance to policy decisions. Starting in the 1990s, cheap communications and computing revolutionized the industrialized world, very much as factory electrification had done 70 years earlier. Suddenly, unskilled labor could be placed in developing countries at much lower cost, and mid-skill back-office labor could be offshored or automated out of existence. From the first-world perspective, this brought an abrupt and huge increase in labor productivity, triggering an economic restructuring that continues to this day.
The “technological employment” perspective might argue that Japan’s lost decades are not a different event from today’s Europe and Anglosphere, but rather an especially sensitive canary in the same coal mine. Because Japan began the 1990s with among the highest domestic manufacturing employment, and among the lowest flexibility in its industrial organization and labor force, it was among the most exposed and least responsive, and hence first to decline and last to recover.
The Great Refactoring hypothesis would explain why few if any countries have ever ended a sustained slump through monetary, currency or fiscal policy, despite many and varied attempts. It argues instead that debt buildups and financial bubbles are merely effects, not causes — symptomatic malinvestment, by firms, individuals or central banks, in the early stages of a Great Refactoring.
The hypothesis would predict that countries that embrace refactoring, such as Denmark and Sweden, and to a lesser extent Germany and the United States, should recover faster than the rest of the world, even if they follow monetary and fiscal policies that don’t closely resemble each other. We’ll know in twenty years.
From a policy perspective, this sounds fatalistic, implying there exists no single, centralized dial one can twist, such as interest rates, that will shorten such a slump. At best, the Fed may blunt symptoms — worthwhile if it prevents war or revolution — but at the cost of distorting the economic signals that guide decisions, thus slowing the refactoring that brings real recovery.
On the bright side, this view implies that, at every level, from individual scale up to nation scale, aggressive adaptation wins the day: to recover, learn new forms of organization, and new skills to fit that new organization, as fast as possible. That’s actionable.
June 14th, 2012
In an earlier post, I pointed out that a company’s price/earnings ratio is mathematically identical to its present value, if growth is assumed to be zero.
So why would we want to assume growth is zero? One reason is it makes for a simple calculation. But another, deeper reason is that the present value equation is highly sensitive to small estimation errors in the growth and discount rates. You can easily fool yourself with present value, especially when bond yields are low.
Let’s take an example. Seasoned Aaa corporate bonds currently yield 3.5%. This is a ready alternative to stocks, with some risk premium built in (unlike Treasuries), so one could argue that this is a reasonable discount rate in the current rate environment (I would call this discount rate crazy, but one could make the argument, so bear with me for a couple of paragraphs).
Suppose we have a cash flow of $100, with a terminal growth rate of 3%, very close to our chosen discount rate. The present value of that cash flow will be large: $20,600, or 206 times this year’s earnings.
Now suppose your growth estimate is just slightly wrong — in fact, terminal growth will turn out to be half a percent lower, 2.5%. The present value falls not by a little, but by half, by 103 times earnings.
Or suppose your growth estimate was exactly right, but bond yields drift up just slightly, say 50 basis points, toward their long-run average of 7.5%. Again, present value falls by half, on an estimation error of just half a percent.
I chose an extreme example to make the point, but this is every bit as true away from the extremes: present value is a nonlinear function that can lead you wildly astray when its growth and discount rate estimates are too close to one another, or too close to zero.
The Graham-style approach — buy at a P/E below a fixed number, say 10 — essentially assumes a growth rate of zero and a discount rate of 10%. This will cause the buyer to miss many, many potential opportunities. But the few that pass such a screen (and also have stable earnings and strong balance sheet) are resistant to human error: they will tend to do well even if your predictions of future rates are wrong, even if management makes a boo-boo, even if financial crisis intervenes, even if the world enters a sustained low-growth deleveraging period.
May 14th, 2012
If the eurozone is the Beatles, then Greece is Ringo — great personality, fabulous mustache, but neither technically proficient nor irreplaceable.
Ringo never left the Beatles. Quite the reverse. Why? Like Greece, Spain and Portugal, he didn’t have terrific independent career options.
Instead, depending on who you ask, Lennon or McCartney left first. They were the band’s Germany, more indispensable than personable.
Like Lennon & McCartney in 1970, Germany has a solo career ahead. Germany writes exportable hits. Germany plays multiple instruments. Germany’s name can fill stadiums (to a fault, twentieth-century historians might argue). Germany has a reputation to defend, and may wish to leave the band before it is tarnished by a bad album.
Of course, EMU without Germany is like the Beatles without Lennon & McCartney: a subject of retrospectives.
April 23rd, 2012
Buried in the baseball book Moneyball is a deep investment insight: one can gain an investment edge not only from better analysis, but also from better fundamental data. Few investors seem to appreciate this as much as Joel Greenblatt’s Gotham Asset Management.
According to Santangel’s Review (an excellent newsletter that reviews the performance and methods of successful professional investors — we have no affiliation), Gotham maintains its own in-house fundamental data feed for 3000 US and 1200 foreign stocks. Like the heroes of Moneyball, Gotham tries not only to interpret data better, but also to interpret better data.
To see why this is so great, contrast it with what everyone else is doing.
Nearly all US investors today, from the big hedge funds all the way down to hobbyists working from home, rely on the same few sources for all their financial data about US companies: Compustat/CapitalIQ (both owned by Standard & Poor’s), Bloomberg, and Thomson Reuters. Do you use Yahoo Finance? They simply republish CapitalIQ. Google Finance or MSN Money? They both use Thomson Reuters.
With such concentration in the fundamental data sources, it’s difficult to avoid herd behavior: everyone is madly competing for an analytical edge, even as they all use nearly identical data and tools.
Moreover, if you spend a lot of time looking at public company financial statements, as I have, you will start to notice that these sources are sometimes wrong: there may be delays before data is reported accurately, or a company’s balance sheet may have unusual line items that are not easily reflected in the standardized formats used by the data feeds.
By compiling their own basic financial data about companies — optimized to maximize accuracy for the types of investments they like to make — Gotham creates a different, and presumably better for their investment purposes, investment universe. They are then better able to march to their own drummer.
Most investors lack the scale to do this with the whole investment universe. But even individuals can do it with a limited subset of all possible stocks.
Ask yourself: what do I know about this company that cannot easily be extracted from the fundamental data feeds?
April 4th, 2012
Lloyds says its fund management unit, Scottish Widows Investment Partnership, will fire over half its staff in a move to computer models for investment decisions. This should be interesting.
My concern is not Luddism — I’ve been a computer programmer for over 30 years, and write code for investment purposes. What I question is the most likely trend: a widespread move by pension and insurance funds into mechanical long/short relative-value investing. I suspect that not only will this approach perform poorly when done en masse, but also may actually further decouple markets from intrinsic value.
Mechanical long/short relative value investing just means that you employ a computer to buy the cheapest stocks on the market, and sell short the most expensive. This is reasonably logical, backtests very well, and has worked for hedge funds for years.
“Relative value” sounds like value investing, but strictly speaking, it’s not, because it implicitly uses the general market price level as an input. If the market’s average P/E ratio were 50, the computer might eagerly buy stocks at 30 times earnings. If the market’s average price to book ratio were 10, then the computer might eagerly buy stocks at 5 times book value. “Relative value” investing is thus detached from intrinsic value.
Absolute value investing means that you insist on buying at a discount to intrinsic value — for example, you insist on paying no more than 80% of tangible book value. If the market goes to 1000% of tangible book value, and you can’t find anything at 80%, then you stay in cash.
Absolute value strategies force you to buy high intrinsic yield; as a result, they tend to force you out of an expensive stock market, and to force you into a cheap one.
Relative value doesn’t do that: when it buys at 30 times earnings, as in the example above, the reason is certainly not intrinsic yield (a miserable 3.3% in this example). Instead, it holds out the hope that its price ratio will rise to the average, which in that example was 50.
Therein lies the fallacy in relative value: it presumes regression to the wrong mean. The market’s average price ratio will not always be 50. The market average itself regresses to a historical mean: about 16.5 times earnings over the past 125 years. Thus, if you’re paying 30 times earnings, you should be thinking not about the current average, up at 50, but the long-term average, far below at 16.5.
Why mention this now? Because a dogpile has begun. The more widows and orphans (i.e. pension fund managers) that pile into backtested relative-value strategies, the more the market as a whole becomes free to decouple from its intrinsic value. Relative value worked in past years precisely because few other were doing it. When the whale funds get in, the party is over.
At this point, one might argue that none of this matters for a neutral long/short investor. If you have equal amounts long and short, then when an overpriced market crashes, you make more from the short side than you lose on the long side.
Again, this was true for hedge funds that used relative value over the past couple of decades. But not everyone can do it at once — the more money thrown at this strategy, the less well it works. Widely shorted stocks are expensive to short. Worse, they may become impossible to short in a panic, causing that downside protection to vanish just when it is needed most.
So what keeps shares connected to their underlying intrinsic values? Trading decisions by actors that care about intrinsic. Much of the money in the market today is in index funds, which care not at all about intrinsic (or even relative) value. Now we may see traditionally large, conservative institutions moving to mechanical relative value, another approach that cares not at all about intrinsic value. Who is left to compete price toward value? Only absolute-value investors (whether human or electronic).
In coming years, this should bring more opportunity to absolute-value investors, and increasing consternation to relative-value funds.
March 26th, 2012
The following is adapted from the December 2 issue of Spinoff & Reorg Profiles, and probably even more true now than then…
Central bank interventions distort equity markets more today than they did before 2007. This looks likely to result in a bumpy unwind someday.
Yields compete. When one asset class earns a given return, another asset class of similar exposure (however you want to define exposure) will tend to earn something similar. Even where this is not completely true, asset class competition acts like gravity, pulling risk-adjusted yields toward one another.
One such relationship, for equity investors, is between equity earnings yields and long-term bond yields, variously codified in the NPV terminal value equation, the Graham & Dodd P/E Matrix, et al. As a result of that relationship, when bond yields generally fell from 1981 to 2007, the result was a long bull market.
Traditionally, central banks manage only short-term interest rates, letting the market set long-term rates. As a result, long-term rates are usually part of a market ecosystem, both signaling, and responding to signals from, other asset classes. Since 2007, though, central banks are not just setting short-term rates, but also intervening directly in longer-duration bonds, including sovereigns and mortgage-backed securities. The motives are understandable, but the result is a paradox in equities: equity yields still move in response to the “gravitational force” of bond yields, but those yields are increasingly set by centralized command, not free market forces.
The historical record for command interventions in free markets is, to put it mildly, poor. Market forces can at best be delayed, so the usual result is distortion, then failure. More worryingly, the failures are frequenty abrupt, as the cost of distortion overwhelms the benefit of intervention, or as political priorities change. Currency interventions, for example, often end with an abandoned peg and abrupt devaluation. Subsidies are suddenly abandoned; price controls are suddenly relaxed. We suspect something similar may eventually occur in long-duration bonds, with only the specifics yet to be resolved — an abrupt rise in yields, or an abrupt devaluation, or more likely both.
We’re quite cautious about an abrupt rise in equities, in response to an artificially induced decline in long-term yields, as occurred in recent months. To decide whether the S&P index is currently cheap in general, we are curious to know what longer-duration bond yields would be, were they freely priced.
February 15th, 2012
Though they have never had the high success rate of spinoff investments, post-bankruptcy issues were for many years a good place to look for bargains.
We argue this will be less true in coming years, for two reasons: first, interest rates cannot fall any lower, and second, revenue growth will likely be low for a few more years. The flip side of this is that distressed debt may become more interesting.
When a company goes broke, not all its stakeholders are equal. Trade accounts get first dibs on repayment, then secured creditors, then senior unsecured, then other bonds, then preferred stock, and finally common stock. More often than not, by the time senior creditors are repaid, there is nothing left for stockholders, and often not enough for junior bondholders.
When that happens, the common stock is canceled, and the company issues new common stock to junior bondholders as partial compensation for default on the bonds.
(There are additional complexities, as a majority of any class of stakeholder can block a bankruptcy workout to get better terms — but we’ll save those opportunities for a future post.)
The cultural gulf between equity and bond investors can be wide (bond guys are from Mars, stock guys are from Venus), so when a bondholder receives common stock, he often will dump it at the first opportunity. Like a spinoff, this can create selling pressure as the company emerges from bankruptcy, depressing prices just after its new shares list.
Ironically, just as prices are depressed by such selling pressure, the underlying business is often in its best shape in years, since much of its debt has just been offloaded in bankruptcy. Newly profitable, the company can, one hopes, slowly repay remaining debt and deliver a solid investment return to shareholders.
The success rate here is necessarily lower than for spinoffs, because companies usually go broke for a good reason. The canonical cause is a cyclicality that regularly renders the company unable to pay the high fixed cost of debt; cutting debt reduces, but does not eliminate, the possibility that this will happen again, and so one sees a high recidivism rate among bankrupt firms.
But if post-bankruptcy success was difficult before, it’s even tougher now, because debt cannot get any cheaper. For the past 20 years, excluding the occasional debt panic (2008), corporate bond rates have generally fallen, making new debt cheaper than old. Thus a post-bankruptcy firm that is still just barely able to pay its remaining coupon could count on eventually rolling over into cheaper debt, and thus higher profit. This looks almost certain to reverse over the next 10 to 20 years.
Not only will debt be more expensive, but growth will likely be slower, at least in the US. Together, these two conditions make post-bankruptcy issues more of a long shot than they have been in decades — for equity investors. By contrast, distressed debt investors may have a field day, as they pick and choose through companies struggling to roll debt over in a rising-yield environment.
September 2nd, 2011
Microsoft is is now so much smaller than Apple that it will have trouble competing in new markets — yet it makes a better investment for most. I use Apple products, but own Microsoft stock. Why?
Investors over 30 may find it hard even to accept the objective fact that Microsoft, the biggest tech titan of the 1990s, the one business schools use as the textbook example of “unregulated natural monopoly,” the one that returned more than 100x to investors in a decade, is now a pipsqueak compared to Apple. Apple, which was at that time one of the saddest stories in the same industry, and nearly insolvent at least twice in the late 1990s.
How conditions have changed. Apple’s net profit exceeded Microsoft’s for the past 5 quarters — and was about double in the most recent quarter. MSFT still has some network effect, but now lacks the scale to compete with Apple on resources alone.
It’s surprising this hasn’t garnered more attention. Several quarters ago, financial media fluttered to the top of the cage when AAPL market capitalization surpassed MSFT. Yet market cap rank totally doesn’t matter. By contrast, earnings and cash flow rank are competitively substantive, yet here go unreported.
The usual business case for Microsoft depends on the idea that “Microsoft always eventually wins” — it has a long, if recently corroded, reputation as a brutal competitor, based on the following linchpins:
- “Network effects — everyone needs Microsoft to interoperate with each other.” This is still absolutely true among businesses and governments. But in consumer devices, Apple now has the network effect, through iTunes. Apple’s tightly integrated multi-device platform for managing and viewing content creates ever-stronger lock-in as users develop a content collection managed through iTunes.
- “Microsoft has more resources.” This is no longer true. Apple now wins any pure resource contest, other things equal.
- “Windows PCs are cheaper.” This is no longer true. Though Apple still has small market share, it also has a very narrow product line, resulting in huge scale economies on a per-product basis. The result: Apple is more profitable at any price point than traditional PC firms. This is particularly clear in newer product categories like tablets, where Apple is generally recognized as the low-cost producer.
Conclusion: buy MSFT. Huh?
The reason is that MSFT is 40% cheaper than Apple on an earnings multiple basis, and can very likely ride out at least a few more years of gains on its business/government customers alone. It has been a tremendous bargain in recent years, delivering double-digit EPS growth while trading at 8 to 12 times earnings — because of this, I bought MSFT in November 2008 and again in December 2011, still own it, have done well with it, and am not selling yet.
Plenty of people have made piles of money with AAPL in recent years, but because of the higher P/E multiple (currently 17), purchasers are constantly exposed to a product flop. Unlike business buyers, consumers, even if locked in, can simply stop buying. This is why I happily own only Apple hardware, but only Microsoft stock.
Of course, unless Apple stumbles soon, it will shortly be several times the size of Microsoft, and will be able simply to buy its way into the enterprise. To put its scale in perspective: with just 3 years’ earnings, Apple could buy SAP outright, and instantly be the biggest thing in enterprise software. They probably won’t do it that way, but there will come a time, perhaps soon, when Apple has simply exhausted every consumer market, and has no choice but to go after enterprise. When it does so, it will bring resources many times greater than any potential competitor. When that finally happens, MSFT holders, including this writer, should beware.
Shell-shocked investors naturally jump at sudden noises, and the euro blowup will be a loud one. (Is the blowup inevitable? I argue yes.)
The longer-term economic consequences of a currency breakup affecting 600m people will likely be bigger than that of a few failed investment banks. Yet, when the day finally comes that we wake to learn France and Germany have let the Mediterranean nations swing, we Americans may face no threat of US financial collapse on the scale of 2008, for the following reasons.
- US firms are better prepared now. The financial system and indeed most public companies at least appear better capitalized than in 2008, and exposure to the euro is likely well anticipated. Our case for inevitable breakup (previous post) is not simple, but certainly not rocket science, and the collapse is happening slowly and in the open. So we suspect there has been adequate time for financial firms (outside Europe, at least) to limit the damage. In contrast, Lehman came as a surprise, because all banks had presumed the U.S. Treasury understood their interlocking exposure well enough never to permit an outright bankruptcy; anticipating rescue instead of bankruptcy, other banks never guarded against the prospect of cascading failure.
- Thanks to our newfound collective sensitivity to loud noises, the breakup is also better anticipated by investors, and thus better incorporated into price. Shares are already much cheaper relative to earnings than they were then.
- Because the euro is not the global reserve currency, a crisis may be more geographically contained.
- The US Treasury and the Fed have been on full alert now for three years. Again, Lehman was more of a surprise, requiring more time to put together a policy response. Not so here. The Fed stands ready, firehose of cash in hand, ready to extinguish the faintest wisp of smoke. Presumably it will step in instantly to rescue systemically important domestic firms that failed to manage euro risk. Short-term credit then remains intact, and we settle into an orderly global recession.
None of this is to say that it won’t be bad; but not, perhaps, as bad as we think.