Wednesday 1 August 2018

The (real) cautionary tale of David Einhorn

Greenlight Capital's David Einhorn was recently the subject of a widely-circulated hit piece by the Wall Street Journal, which came a few weeks in advance of Einhorn reporting a truly disastrous 18% loss in the first half of 2018 (in rising US markets). This result capped a abysmal 30 month stretch of performance, which has seen the former star's cumulative losses mount to some 30% (during a bull market) - a fall from grace that has attracted considerable media attention.

The WSJ article attempted to sketch a narrative of a former high-flyer felled by hubris and decadence, drawing attention to the famed value investor's penchant for afternoon naps; high-stakes poker tournaments; and the occasional glamorous all-nighter on the town, but I'm not convinced. I believe Einhorn has actually been felled by much more prosaic factors - a combination of errors of judgement; the inherent dangers of short selling; and bad luck. The second factor, in conjunction with the third, has resulted in poor investment outcomes disproportionate to the failings of the first factor, and they combine to tell a worthwhile cautionary tale (for those fond of learning vicariously).

So what went wrong?

I am not a fan of short selling, and although the mandate of the fund I run allows it, I hold no naked short positions. The problem I have with shorting is not a shortage of overvalued stocks, but the fact that there are too many ways to lose a significant amount of money shorting even if you are ultimately right on the fundamentals & valuation (which is hard enough to get right in and of itself).

Many hedge funds like to emphasise their low 'net exposure' to the market (for instance, Einhorn is about 95% long and about 75% short, for a net market exposure of only 20%). However, this low 'net' exposure radically understates the level of risk involved for the fund's investors, because in certain market environments, it is entirely possible to lose on both your longs and your shorts at the same time. In Einhorn's case, the reality is that the fund actually has gross active exposure of 170%, and so is therefore more akin to a fund leveraged 1.7x, which will magnify errors of judgement (or simply inopportune macro/style/factor positioning) profoundly.

The putative risk-reduction benefits of long-short positioning implicitly rest on the assumption that stocks across sectors/factors/styles will remain positively correlated at all times, but we have seen both in past market environments (notably the bubble), as well as in recent years, that it is entirely possible for whole segments of the market to go down (especially value/cheap segments) while other segments of the market go up (especially growthy/hyped/bubbly sectors). If you're running a long/short book with 170% gross exposure with significant sectoral mismatches where correlations turn inverse, you're going to get yourself into a lot of trouble quickly.

Furthermore, the potential emergence of this type of market environment should not have been difficult to envisage. As I've discussed in past blog entries, the structural nature of value stocks is that the most likely outcome over the medium term is that the stocks will continue to go down/underperform, while the most likely outcome for hyped growth stocks is that they continue to go up/outperform. This is why value stocks become cheap in the first place and defy the dictates of market efficiency, because the anomaly is difficult for institutional investors to practically exploit.

The outperformance of value vs. growth over time typically comes in small, concentrated doses, where very sharp reversals happen (such as during 2000-02, when value stocks surged while the NASDAQ plummeted almost 80%). Many of the best performing long-only value managers with the best long term track records have spent significant periods of time underperforming the index. Consequently, a long value, short growth book is almost certain to cause you tremendous pain from time to time (and particularly late cycle), and Einhorn should have anticipated this when formulating his strategy. Even if it worked a lot of the time, there were going to be occasional periods where the strategy was at serious risk of blowing up his fund.

Furthermore, shorting actively forgoes some of the most important contributors to the long term success of a well-implemented, long term value strategy: the ability to wait, and at select times, average down. If I purchase a 2% long position in a value stock (at say $100), and it subsequently falls 50%, I can continue to hold and wait. While I am 'wrong' and the stock is falling, the position size is shrinking, so the position becomes a steadily smaller portion of the overall portfolio, and hence the pressure to sell recedes, rather than increases. I can wait. Furthermore, I have the ability to average down. If I invest a further 2% of the portfolio after the stock has fallen 50% (at say $50), I now have a 3% position (4% by cost) at an average cost of just $66.67. If the stock rallies back to just $100 - my original entry - I now have a 50% profit, equal to 2% of the overall fund.

The dynamics of shorting are fundamentally different. If I put a 2% short position on a stock at $100, which I think is worth only $50, and the stock doubles on me to $200,* then I now have a 4% portfolio short position (not a 1% position, as in the case of the long). My position size has doubled rather than halved. This is a serious problem, because unlike with the long position, I cannot sleep comfortably and wait, because risk management considerations will at some point force me into covering, converting what would otherwise be a temporary paper loss (assuming my thesis is ultimately vindicated) into a permanent loss of capital.

Furthermore, I cannot meaningfully average down (or perhaps I should say, 'average up'). If I were to put another 2% portfolio short on at $200, I would now have a 6% short position at an average cost of $133.33. A further doubling in the share price would take this to a 12% short position (vs. a further 50% fall reducing my long position from 3.0% to 1.5%). Risk management considerations would preclude such an action, and create an unacceptable risk of a total blow up.

To succeed as a value investor, it is absolutely essential that you are investing from a position of strength not weakness, and have the ability to 'stay the course' when markets move against you, but the risk of being a forced buyer as a short seller is very real. The above dynamics have likely been a significant contributor to Einhorn's horrific recent returns. Notable is that Einhorn covered his disastrous Netflix short entirely during early-mid 2018, and has doubtless had to reduce his other 'bubble basket' shorts in recent times as well, as they threatened to fell the fund.

This experience betrays a structural problem of a long/short fund focused on valuation. If you're long cheap stocks but have no shorts, you will underperform for a while and it will be painful, but you'll likely live to fight another day and will eventually make back all of your underperformance (and more) when the cycle turns. However, if you're long those stocks, which are going down, while also being short expensive stocks which are going up, with gross leverage of 1.7x, the impact on your returns can be ruinous and permanent - even if you ultimately end up being right.

Einhorn appears to have overlooked this risk, but has been punished disproportionately for this oversight on account of bad luck. During the post-GFC upcycle, and particularly during the last 3-4 years or so, US markets have witnessed record outperformance of growth over value. Einhorn noted in his recent letter that value's underperformance of growth is currently in the 1st percentile of historical experience! By definition, the chance of something this extreme happening was therefore only 1 in a 100!

Some of this divergence has been due to late cycle excesses, but some of it has also been fundamentally justified. The past decade has witnessed a record level of 'disruption' (or perceived disruption) of traditional industries at the hands of Amazon, Netflix, Tesla (perceived), and numerous other technology/software/online businesses, including both listed and VC-funded 'Unicorns'. Many of these companies have raised record amounts of funding, which is being used to run fast-growing, but loss-making businesses that have been taking market share off incumbents.

In more traditional funding markets, they would not have been able to do this to the degree they have for the duration they have (as the lack of earnings would have deprived them of the capital needed to acquire economies of scale), but very permissible tech funding markets have reduced entry barriers to an unprecedented degree and allowed them to plow billions of dollars of losses into rapid growth (aided by the fact that rising tech valuations have meant tech investors have not felt the burden of such operating losses).

'Value' stocks - companies with actual earnings and cash flows - have been on the receiving end of the growth and disruption wrought by these businesses (e.g. traditional retail and media). Some of them have indeed seen their revenues and earnings decline (especially certain traditional retailers and PayTV companies - particularly satellite providers), so disruption fears here are very much real, while others (e.g. traditional auto) have continued to post record earnings, and fears of disruption have so far proven merely hypothetical. Multiples have nevertheless fallen precipitously across these industries. General Motors is a clear example - the company keeps producing record profits and cash flows, but the stock continues to languish on disruption fears, while loss-making 'disruptor' and market darling Tesla continues to scale new heights (until recently at least). Einhorn's long GM short Telsa positioning has therefore hurt him badly, even though he will probably ultimately be right.

This has made the investing environment particularly difficult for Einhorn's style. However, while the failure to anticipate the degree to which the cycle would play out in this direction can be somewhat excused, Einhorn has compounded the damage with unforced errors by shorting scalable, potentially category-killing, and well run-businesses like Netflix and Amazon using inappropriate valuation metrics. As I discussed in my recent Spotify blog post, using near term earnings multiples was a wholly inappropriate way to look at these businesses. I am not criticising his decision to avoid them on the long side (I have avoided them also, although I have been tempted to buy Amazon all the way from $250, but have recently ceased to become tempted), but shorting them was a demonstration of a profound lack of imagination and appreciation for some of the fat-tail upside risks to these stocks, and fat tail risk matters tremendously when shorting.**

Assured Guaranty

More generally, Einhorn's experience with his Amazon and Netflix shorts highlight a more systemic problem with Einhorn's approach where he often appears to get overly caught up in the details (particularly accounting arcania), and often misses the big picture. Einhorn is undeniably an incredibly smart investor, but in markets, it is better to be wise than smart, and for a really smart guy, Einhorn seems to recurrently do some pretty dumb things.

An example is his short thesis for bond insurer Assured Guaranty (AGO US), which he presented at this year's Sohn Conference when the stock was trading about US$35. I was long the stock at the time (and remain long). The stock fell to $33 in early-morning trading post the presentation, but quickly rallied and ended the day up, on a day where the S&P500 traded down. It was a stunning rebuke of Einhorn's analysis. The stock has since traded up to $39.

Einhorn's thesis was that AGO had insured Puerto Rico municipal bonds, which had recently defaulted, and the company had only reserved for 20% of its par exposure (about $1bn vs. $5bn of par exposure), yet the bonds were trading in the secondary market at 25c. AGO was under-reserving its PR exposure, Einhorn argued, and when the market woke up to that fact and/or the company was forced to significantly increase loss reserves, the stock would tank.

The thesis, however, failed to address a fairly obvious point: AGO's book value is currently $60 per share, and the stock was trading at just $35 - a $25 discount. The company has 112m shares outstanding. $25 x 112m shares outstanding =  $2.8bn, and these losses would be tax-deductible, so grossed up for 25% tax, that equals $3.7bn in pre-tax losses. In other words, the market had already taken an additional $3.7bn pre-tax 'reserve' in excess of what was on AGO's books. Add that to its existing reserves of about $1.0bn and you get $4.7bn - i.e. roughly in line with the company's total PR par exposure. The market was already pricing in the worst case scenario and was already well aware of the risk of further PR reserving, and Einhorn didn't tell the market anything it didn't already know - hence why the stock reacted with nonchalance.

Furthermore, there were already clear signs that the PR economy was starting to dramatically outperform expectations (and PR bonds have rallied very significantly since Einhorn's presentation). PR's economy was hit hard by Hurricane Maria in 2017, but insurance and federal government reconstruction funds are now starting to flow, boosting reconstruction activity. I am from NZ, and I can recall NZ's economy booming in the years following the Christchurch earthquake, which destroyed many homes and necessitated a reconstruction boom. Both of these factors - the markets assuming the worst (PR was all over the media and people were universally negative), and the potentially overlooked stimulus reconstruction activity was going to generate, coupled with AGO's long track record and quality management (the company was one of the few bond insurers to survive and thrive during and in the aftermath of the global financial crisis), were the reasons I was long.

Furthermore, AGO has a huge amount of liquidity and cannot be accelerated on its PR guarantees. Instead, interest and principal would have to be paid out slowly over 20-30 years. Consequently, 'event risk' was negligible, rendering a short position that much more difficult to justify (a short that plays out over many years is usually too costly - in terms of the cost of borrow - to carry). Einhorn appears to have missed the forest for the trees here, having been completely caught up in accounting arcania, and having failed to properly consider what was already priced in.

No broken thesis rule

A final gripe I have with Einhorn's approach is his famed 'no broken thesis' rule. This policy insists that each investment position have a clearly articulated rationale (e.g. for AGO, it will be forced to up its PR reserves, which should reduce the share price). If that rationale ceases to be applicable, the position need be closed immediately. The rule is designed to prevent 'thesis drift', where new reasons are invented to justify holding a position when the old ones have ceased to apply. It is seen as synonymous with a disciplined investment process.

The problem is this: Like any flawed methodology in stock markets, it is a rule that pays no heed to price. Steve Johnson of Forager Funds has said that a clear mark of a flawed investment thesis is that it makes no reference to price (and hence the same argument could be made at twice the current price; e.g. arguing that Sydney property is a buy 'because of population growth; limited land; and negative gearing tax benefits', without any reference made to price). I couldn't agree move and believe that to be an excellent way of putting it.

The problem with the no broken thesis rule is that it requires the position be sold regardless of price if the original rationale for holding the stock no longer applies. The approach is problematic because the market nearly always overreacts to near term events - if news comes out that suggests the stock is probably worth 10% less than previously thought, the stock will invariably actually decline 20-30%. It is a recipe for selling out of many positions at extremely disadvantageous times and prices. Personally, I've made a fortune off broken theses stocks in the past, as share prices react disproportionately to negative developments, allowing me to average down at unusually attractive prices. In markets, there is no such thing as a mechanical rule which always works - judgment and rationality in each individual situation are what matter. 

In short, Einhorn has also demonstrated poor judgement on several occasions with many of his positions (both long and short, but especially short), and this has combined with bad luck (market environment) and the inherent dangers of short selling to produce truly atrocious outcomes.


All told, it's a much less exciting read than the WSJ, but I believe these to be the real reasons for Einhorn's spectacular fall from grace.

Will Einhorn be back? Quite possibly. Einhorn is still, without doubt, a significantly above-average investor in my perception, despite the above failings, and I believe it is quite likely his performance will rebound nicely at some point - provided his shorts don't put him out of business before then.

What Einhorn should do is start a long only fund. If he does, I would not be surprised to see it perform reasonably well over time. On the long side, Einhorn's performance has also been mediocre of late but not disastrous. I agree with several of his longs (I also own General Motors and Aircap), but by no means all. I've been unenthused by physical gold for some time (see my recent blog entry on gold), which Einhorn has owned a sizable position in for many years. I was also initially unimpressed with his Brighthouse Financial pick (which I blogged about early this year), as the assumptions underpinning the company's valuation of its liabilities appeared aggressive to me. The stock was trading at $65 at the time. It has since declined to $40-45, and I've changed my tune at this much reduced valuation, and have started to accumulate in the low $40s. Einhorn's long-side record is far from unblemished, but overall I believe him to be a decent long-style value investor.

It is the shorts that have done him in, where some important mistakes have disproportionately cost him. If tech suddenly crashes, his existing long/short fund might also stage a dramatic turnaround, but many of the losses he has already taken on his shorts will likely nevertheless prove permanent. Time will tell, but Einhorn's experiences in recent years are nevertheless a worthwhile cautionary tale.


*The two are inversely equivalent, geometrically. A 50% decline from $100 to $50 requires a 100% gain to get back to $100; and a 100% gain from $100 to $200 requires a 50% decline to get back to $100.

**To be clear, this does not mean I believe Netflix and Amazon to be attractive longs at present (indeed, I can envisage scenarios where Netflix could implode, and although Amazon appears invincible today, it will likely eventually be broken up, and its valuation is currently at nose-bleed levels). It means that I believe them to be dangerous shorts - a belief I have held for many years.