Air-Pockets, Free-Falls, and Crashes

“Abrupt market weakness is generally the result of low risk premiums being pressed higher. There need not be any collapse in earnings for a deep market decline to occur. The stock market dropped by half in 1973-74 even while S&P 500 earnings grew  by over 50%. The 1987 crash was associated with no loss in earnings. Fundamentals don't have to change overnight. There is in fact zero correlation between year-over-year changes in earnings and year-over-year changes in the S&P 500. Rather, low and expanding risk premiums are at the root of nearly every abrupt market loss.

“One of the best indications of the speculative willingness of investors is the ‘uniformity’ of positive market action across a broad range of internals… I've noted over the years that substantial market declines are often preceded by a combination of internal dispersion, where the market simultaneously registers a relatively large number of new highs and new lows among individual stocks, and a leadership reversal, where the statistics shift from a majority of new highs to a majority of new lows within a small number of trading sessions.

“This is much like what happens when a substance goes through a ‘phase transition,’ for example, from a gas to a liquid or vice versa. Portions of the material begin to act distinctly, as if the particles are choosing between the two phases, and as the transition approaches its ‘critical point,’ you start to observe larger clusters as one phase takes precedence and the particles that have ‘made a choice’ affect their neighbors. You also observe fast oscillations between order and disorder in the remaining particles. So a phase transition features internal dispersion followed by leadership reversal. My impression is that this analogy also extends to the market's tendency to experience increasing volatility at 5-10 minute intervals prior to major declines.”

Market Internals Go Negative, Hussman Weekly Market Comment, July 30, 2007

“We've started to see the pattern of abrupt jumps and declines at 10-minute intervals that is often a hallmark of nervous markets. My continued concern is that numerous market plunges have been indifferent to both interest rate trends and even valuations, with the main warning flag being deterioration in the quality of market internals, as we observe at present. Both in the U.S. and internationally, ‘singular events’ tend to occur well after internal market action has turned unfavorable, and prices are well off their highs.

“Though I don't want to put too much emphasis on intra-day behavior, if you examine tick data or daily ranges before major declines both in the U.S. and elsewhere, you'll generally see price movements become chaotic at increasingly short intervals even before the event itself. One way to describe it without mathematics is to spin a quarter on the table and watch (and listen to it) closely - you'll observe a similar dynamic at the abrupt point that the coin moves from an even spin to an irregular one, and again just before it stops. If you imagine a pen drawing out its movements, you would see it tracing out faster and faster circles as it moves from stability to instability.”

Broadening Instability, Hussman Weekly Market Comment, January 28, 2008

Over the years, I’ve frequently observed that the most favorable market return/risk profile we identify is associated with a material retreat in valuations, coupled with an early improvement in market action across a broad range of market internals. That combination does not require a retreat in valuations to historical norms, and whether that shift occurs at still-elevated valuations, depressed valuations, or in sporadic intervals, we fully anticipate such opportunities over the completion of the current market cycle and those ahead. It’s important to begin on that optimistic note; to recognize that there will be what we expect to be identifiable opportunities as conditions change, even though now is the precise antithesis of such an opportunity.

I’ve done all I can to clarify that my “permabear” reputation is an artifact related to my necessary but admittedly ill-timed 2009 insistence on stress-testing our methods against Depression-era data, as employment losses, economic contraction, credit strains, and market pressure became uncharacteristic and “out of sample” compared with the post-war data on which our methods were based. The ensemble methods that solved that “two data sets” problem improved on our pre-2009 methods, but they did lack various trend-sensitive overlays that we introduced in the late-1990’s in response to the tech bubble. Aggressive yield-seeking speculation provoked by the Federal Reserve’s zero interest rate policy led us to reintroduce variants of those overlays, in order to narrow our defensive response to persistent and uncorrected overvalued, overbought, overbullish extremes. All of that made for a terribly awkward transition from our pre-2009 methods of classifying market return/risk profiles to our present methods. Still, understanding that narrative (see Setting the Record Straight) is particularly important at present, in order to disabuse the belief that present conditions can be dismissed simply because we encountered those challenges in the first half of this wholly uncompleted market cycle.  

With regard to that 2009 stress-testing decision, I’ve frequently observed that our most historically reliable measures of valuation (which indicate that current valuations are about double their pre-bubble norms), were actually quite reasonable by late-2008 and early-2009 (see in particular my October 20, 2008 piece, Why Warren Buffett is Right and Why Nobody Cares). The problem is that the similar valuations, during the Depression, were followed by a further loss of two-thirds of the market’s value. It might be useful to recall that the legendary value investor Benjamin Graham was hedged going into the 1929 crash, but once valuations got down to what appeared to be reasonable valuation, he removed most of those hedges. Graham and his Graham-Newman partnership went on to suffer a 60% loss by 1932. One of the key lessons from our 2009-2010 “two data sets” challenge was that Depression-era data, as well as the credit crisis, imposed more demanding requirements on measures of market action (particularly measures of “early improvement”) than were typically needed in other post-war cycles.

Neither our stress-testing against Depression-era data, nor the adaptations we’ve made in response extreme yield-seeking speculation, do anything to diminish our conviction that historically reliable valuation measures are of immense importance to investors. Rather, the lessons to be drawn have to do with the criteria that distinguish periods where valuations have little near-term impact from periods where they suddenly matter with a vengeance.

The effect of valuations on subsequent market returns is conditional. While depressed valuations are a good indication of strong prospective long-term returns, depressed valuations don’t prevent further – sometimes massive – losses in the near-term. A retreat in valuation becomes reliably favorable mainly when it is joined with an early improvement in market internals. All of history (not just the Depression-era and the 2008-2009 collapse) imposes demanding requirements; not least that internals aren’t collapsing and credit spreads aren’t shooting higher, as they are today. Conversely, overvalued, overbought, overbullish extremes are associated with total market returns below risk-free interest rates, on average, but that average features an unpleasant skew: most of the week-to-week returns are actually positive, but the average is harmed by large, abrupt losses. Such extremes become reliably dangerous when they are joined by deterioration in market internals.

Present conditions create an urgency to examine all risk exposures. Once overvalued, overbought, overbullish extremes are joined by deterioration in market internals and trend-uniformity, one finds a narrow set comprising less than 5% of history that contains little but abrupt air-pockets, free-falls, and crashes.

In recent weeks, the market has transitioned to the most hostile return/risk profile we identify: the pairing of overvalued, overbought, overbullish conditions with deterioration in market internals and price cointegration – what we call “trend uniformity” – across a wide range of stocks, sectors, and security types (see my September 29, 2014 comment Ingredients of a Market Crash). As in 2007 and 2000, we’re observing characteristic features of that shift. One of those features is that early selling from overvalued bull market peaks tends to be indiscriminate, as deterioration in market internals and the “average stock” often precedes substantial losses in the major indices. As of Friday, only 28% of NYSE stocks are above their respective 200-day moving averages.

In the current cycle, both the Russell 2000 small-cap index, and the capitalization-weighted NYSE Composite set their recent highs on July 3, 2014, failing to confirm the later high in the S&P 500 on September 18, 2014. Through Friday, the NYSE Composite is down -7.3% from its July 3rd peak, and the Russell 2000 is down -12.8%, while the S&P 500 is down only -4.0% over the same period. What’s happening here is that selling is being partitioned in secondary stocks, and more recently high-beta stocks (those with greatest sensitivity to market fluctuations). Market action is narrowing in a classic pattern that reflects the effort of investors to reduce risk around the edges of their portfolios, in what typically proves an ill-founded belief that a falling tide will not lower all ships.

Abrupt market losses are typically not responses to obvious “catalysts” but instead reflect a shift in investor preferences toward risk aversion, at a point where risk premiums are quite thin and prone to an upward spike to normalize them. That’s essentially what’s captured by the combination of overvalued, overbought, overbullish coupled with deteriorating internals. Another characteristic of these shifts is increasing volatility at short intervals – what I described at the 2007 peak and in early-2008 by analogy to “phase transitions” in particle physics. The extreme daily and intra-day market volatility in recent sessions is typical of that dynamic.

Fed policy and risk premiums

Recall that very early into the 2000-2002 and 2007-2009 bear markets, the Federal Reserve began to aggressively ease monetary policy, but that did not prevent stock prices from going on to lose half or more of their value (see Following the Fed to 50% Flops). The short-term responses of the market were certainly positive, but those responses turned out in hindsight to be exit points. As I’ve noted before, if one is going to invest by aphorism, history teaches that “don’t fight the trend” strongly outperforms “don’t fight the Fed.” With respect to our own experience in the half-cycle since 2009, the primary lesson to be drawn is not that Fed policy trumps all other considerations. Rather, the lessons to be drawn relate to the criteria that distinguish periods where monetary easing is supportive to the markets from periods where policy shifts are irrelevant or even contribute to the loss of investor confidence.

Again, that distinction has a great deal to do with market internals and trend uniformity, because those measures of market action provide a signal about the tolerance or aversion of investors toward risk. In effect, Fed easing is effective provided that risk-free cash is considered an inferior holding. Fed easing is useless if investors actually prefer to hold risk-free cash as a safe haven.

There’s certainly a feedback circle to this: the purely psychological belief that Fed liquidity is a magical risk-removing fairy dust can certainly support increased risk tolerance, but that tolerance should still be read directly out of market internals and trend uniformity. When investor preferences shift toward risk aversion, more liquidity doesn’t support stock prices. Yield-seeking speculation fails to emerge because low or zero interest rates on cash are preferred to the prospect of steeply negative returns. As the market collapses of 2000-2002 and 2007-2009 demonstrate, aggressive Fed easing does not prevent extraordinary market losses once investors have the risk-aversion bit in their teeth.

The Fed certainly has a legitimate and often helpful role in crises when it is needed to act as a “lender of the last resort” by lending to solvent but liquidity-constrained financial institutions. Good public policy acts to responsibly relieve legitimate constraints on the economy that are actually binding. At present, however, the financial system is already drowning in trillions of dollars of idle cash reserves, which don’t need to “go” anywhere, because once a dollar of base money (currency or bank reserves) is created, it remains in existence, in the form of base money, until it is retired by the Federal Reserve. In other words, zero-interest sideline cash is zero-interest sideline cash and will remain zero-interest sideline cash until it is retired, and the only thing that $4 trillion of base money does for the economy is to change hands as a hot-potato that nobody wants to hold so long as risky assets appear to offer better returns than zero.

No doubt – this pile of zero-interest hot potatoes has helped to compress risk premiums across the entire range of risky assets toward zero (and we estimate, in some cases, below zero). But understand that the bulk of the advance in financial assets in recent years has not been a reasonable response to the level of interest rates, but instead reflects a dangerous compression of risk premiums.

The effect of zero-interest rates is measurable, and the arithmetic is straightforward. The expectation of another 3-4 years of zero interest rates (versus normal short-term interest rates of say, 4%) implies that risky long-term assets could reasonably be priced 12-16% above where they would be priced in a normal interest rate environment. That premium would reduce the prospective returns of those risky assets by that same 4% for 3-4 years, but would preserve normal risk premiums. But on valuation measures that are reliable across a century of history, including recent years, the valuation of the S&P 500 is now more than double its pre-bubble historical norms (and only looks more tolerable because investors do what they always do at cycle peaks, which is to capitalize peak cycle earnings as if they are fully representative of the entire stream of future long-term cash flows).

In short, every 3-month period of additional zero-interest rate policy promised by the Fed is worth about a 1% premium over historical valuation norms. Another year would be worth a premium about 4% over historical norms. But with the market more than double historical norms on reliable measures, the Fed would have to promise a quarter of a century of zero interest rate policy before current stock valuations would reflect a “reasonable” response to interest rates. No – stocks are not elevated because low interest rates “justify” these prices. They are elevated because the risk premium for holding stocks has been driven to zero. We presently estimate negative total returns for the S&P 500 on every horizon shorter than 8 years.

At present, prospective market return/risk should not be read from Fed policy. It should be read from valuations and the quality of market internals and trend uniformity, which we view as the best way to infer investor risk tolerances, the level of risk premiums, and the pressure on them. If these measures improve, a fresh easing of Fed policy would allow for further yield-seeking speculation. But in the context of extremely compressed risk premiums that are being pressed higher; in the context of an overvalued, overbought, overbullish market that has been joined by deteriorating market internals, broadening dispersion, and a loss of trend uniformity – all bets on the Fed are off, as they were in 2000-2002 and 2007-2009, until we observe a favorable shift in those measures of investor risk preferences.

Warning: Examine all risk exposures

All of that said, there’s no assurance that the present instance will match historical experience. As I noted at what in hindsight turned out to be the market peak in October 2007, in a piece that bears the same title as this section (see Warning: Examine All Risk Exposures):

“There is one particular syndrome of conditions after which stocks have reliably suffered major, generally abrupt losses, without any historical counter-examples. This syndrome features a combination of overvalued, overbought, overbullish conditions in an environment of upward pressure on yields or risk spreads. The negative outcomes are robust to alternative definitions, provided that they capture that general syndrome. We can't rule out the possibility that investors will adopt a fresh willingness to speculate (which we would observe through an improvement in market internals). Such speculation might prolong the current advance modestly, but even this would not substantially alter the risks that have ultimately been associated with overvalued, overbought, overbullish conditions.”

Though we should allow for a potential improvement in market conditions, I do believe that now is a particularly bad time to rely on the idea that “this time is different” with money you cannot afford to lose. This does not require forecasts about market direction – only proper consideration of market risk. Make sure that the portfolio of risks you do hold is the portfolio that you want to hold over the completion of the market cycle, understand the risk profile and actual losses that various asset classes have experienced over prior market cycles, take account of the prospective returns that are embedded into current valuations, and insist on historically reliable measures of valuation that demonstrate a strong association with actual subsequent returns over numerous market cycles across history.

My view is that even passive buy-and-hold investors should primarily focus on ensuring that the effective duration of their portfolio is not significantly longer than the horizon over which they expect to spend the funds. In other words, the duration of the assets should be matched with the anticipated horizon of spending needs (or liabilities). The estimated duration of the S&P 500 Index is roughly 50 years, 10-year Treasury bonds presently carry a duration of about 9 years, and cash has zero duration, so a passive investor expecting the average date of spending to be about 15 years in the future might match that with an asset portfolio of similar duration. Examples would include a 20%-55%-25% mix of stocks, bonds, and cash, respectively, or perhaps a 24%-33%-43% mix, but in any case not more than about 30% in equities.

The challenge here is that we associate each of those 15-year duration portfolio mixes with expected nominal total returns of less than 2% annually over the coming decade. Based on historically reliable valuation measures, we presently estimate prospective 10-year S&P 500 nominal total returns of just under 2% annually here, so increasing the equity portion does not improve the expected portfolio return. Investors should understand that “prices and valuations are high” is another way of saying “future returns have already been realized, leaving little to be gained for quite some time.”

Fortunately, as valuations retreat, durations shorten. For example, at the 1982 low, the dividend yield of the S&P 500 reached 6.7%, bringing the duration of the index down to 15 years, so from a duration-matching standpoint, even an investor with an expected spending horizon averaging 15 years could have been comfortable with 100% of assets in equities. At the 2009 low, the yield was a more moderate 3.8%, but that still implied a 26-year duration, making a 60% equity allocation quite reasonable even for a passive investor expecting to spend the assets, on average, 15 years hence.

Alternative investments are a bit trickier, as their exposure to market risk can vary. Since the potential for portfolio loss is a significant consideration, one approach might be to gauge relative risk by comparing maximum losses on a compound (log) basis. For example, if the worst historical drawdown of A has been 33% over several market cycles, and the worst drawdown of the market has been 55%, the relative risk of A might be estimated as log(1-.33)/log(1-.55) = 50% of that of the market. That odd-looking compounding arithmetic essentially captures the fact that it takes two back-to-back 33% losses to produce one 55% loss. Similarly, if worst historical drawdown of B has been 18%, the relative risk of B might be estimated as log(1-.18)/log(1-.55) = 25% of the market itself, as it takes four back-to-back 18% losses to produce one 55% loss like the S&P 500 experienced in 2007-2009.

An active investor would typically consider allocations not only from the standpoint of duration, but also broader conditions that affect returns over shorter portions of the market cycle. Presently, we don’t believe that active investors should expect a positive return from unhedged equities at all here, given the combination of rich valuations and deteriorating internals, which suggests skewing holdings toward cash or hedged alternatives until more favorable conditions emerge. Again, we view the strongest market return/risk profiles, and the best opportunities for unhedged investment, as coupling a material retreat in valuations with an early improvement in market internals. Now is the antithesis of those conditions.

The foregoing comments represent the general investment analysis and economic views of the Advisor, and are provided solely for the purpose of information, instruction and discourse. Only comments in the Fund Notes section relate specifically to the Hussman Funds and the investment positions of the Funds.

Fund Notes

The Hussman Funds remain defensive toward equities, and modestly constructive toward Treasury bonds and precious metals shares. With market conditions in the same hostile configuration as we observed last week, the same considerations about day-to-day fluctuations apply. In Strategic Growth Fund, the recent decline has brought the major indices close to or below the strike prices of the long index put option side of our hedges. Though those put options currently represent a small percentage portfolio value (and could lose that amount if the market was to advance between now and their late-year expirations if we were to hold them throughout such an advance), they are also likely to change in value if the market declines significantly. In that event, we would expect good amount of “give-and-take” depending on whether the market advances or declines on a given day. So in the event of a significant market loss from here, these positions will tend to account a significant portion of day-to-day changes in net asset value in both directions.

We’ve long observed that deterioration in market internals and weakness in the “average stock” often precedes weakness in major indices (which are the most liquid vehicles for hedging). As a result, equity portfolios are likely to be somewhat more susceptible than usual to “tracking” differences between the stocks we hold long and the indices we use to hedge. With the broad NYSE Composite down 7.3% since its July 3rd peak, and the Russell 2000 down 12.8%, compared with a loss of only 4% in the S&P 500 over the same period, there is something of a headwind to the selection of individual stocks at the moment. Indiscriminate selling may produce further periodic day-to-day headwinds until investors begin picking the wheat from the chaff a bit more deliberately. Day-to-day changes in net asset value will also be affected to some extent as this process – common to the early declines following cyclical market peaks – takes its course.


© Hussman Funds


© Hussman Funds

Read more commentaries by Hussman Funds