What Happened to the Financial Markets in May?

Many global financial markets went on a wild ride in May. What changes caused the big falls in prices in some asset classes? According to many commentators, they were due to a sudden "increase in risk aversion." What does this mean? Is that what really happened? Did people really become less willing to take the risk of crossing the street? As always, the truth of the matter is more complicated than many media commentators can squeeze into a thirty second spot. To understand what happened in May, we must first start with a model of how the price of an asset is determined.

So-called "fundamental" asset valuation models usually determine the price of an asset by discounting future cash flows to their present value. The formula for doing this is [cash flow/(1+discount rate)^Year], where "year" is the number of years from now that the cash flow occurs. These present models use one of two basic methodologies to take risk into account. The first does this by adjusting the discount rate (the denominator of the present value calculation), and the second takes risk into account by adjusting the the cash flows.

The first is by far the most common approach. It forecasts the "most likely" future cash flows an asset will generate, and then discounts them to their present value using a "risk-adjusted rate of return." The latter is established by starting with the risk free rate (e.g., the current yield on inflation indexed government bonds) and adding to it an additional premium than reflects the riskiness of the cash flows being discounted. To put this slightly differently, this approach takes the riskiness of the future cash flows into account by adjusting the denominator of the valuation calculation.

The alternative approach takes risk into account in the numerator, by adjusting the projected cash flows, which are then discounted to their present value at the risk free rate. Here is a quick and easy example that illustrates this approach. To make it easier, we'll leave inflation out of it, and assume all payments are in inflation-adjusted (real) terms. Say I offered to sell you a Bolivian government bond that promised contractual payments of U.S. $100 per year for five years. Like most people, you probably don't walk around with an appropriate risk premium for Bolivian risk at the top of your mind. However, you can easily answer this question: if I could exchange my five risky Bolivian payments for five annual payments from the U.S. government (or whatever home government you choose), how large would these latter payments have to be to induce me to make the exchange? If you believe that Bolivian government risk and U.S. government risk are equivalent, then your answer would be $100 per year. However, if you believe the Bolivian payments to be riskier than the U.S. government payments, you would accept payments of less than $100 per year from the latter. For example, you might accept five $70 payments from the U.S. government in exchange for the five $100 payments from the Bolivian government. Moreover, once you had decided on these amounts, you could easily discount them to their present value at the U.S. government rate for a bond of the same duration (to simplify, let's just call this the five year real government bond rate). Assuming this rate is 2.25%, the present value of the five $70 payments would be about $335. This is the maximum price you would logically pay for the Bolivian government bond I am offering you.

Now that you understand the basic concept of taking risk into account in the cash flows rather than the discount rate, let's expand it a bit more. Logically, the amount you would accept from the U.S. government equals the expected value of the Bolivian payments. In turn, this expected value logically reflects the sum of the different amounts you expect to receive under different scenarios, times the probability of those scenarios. The following example helps to make this clear:

As you can see, the 8.7 expected cash flow in Year 1 equals the forecast cash flow under the Most Likely Scenario (10) times its probability (50%) or 6.0, plus the probability adjusted cash flows under the Upside (12 x 10%) and Downside (5 x 30%) Scenarios, or (8.7 = 6.0 + 1.2 + 1.5). When the five annual expected cash flows (shaded) are discounted at the risk free rate of 2.25%, their present value is 64.12.

It is also easy to convert this calculation into the more traditional approach that takes risk into account in the denominator (discount rate) rather than the numerator (cash flows). You do this by determining the "risk adjusted" discount rate that makes the present value of the Most Likely Scenario's cash flows equal to 64.12. In this case, that discount rate is 7.11%. This discount rate is composed of two parts. The first is the 2.25% risk free rate. The second is an Implied Risk Premium of 4.86%. This premium compensates investors for bearing the risk inherent in the asset's cash flows.

So far, so good. We'll now extend this example a bit further to generate some insight into May's events. Let's first examine what happens in our example if something in the economy changes, and causes forecast cash flows under all scenarios to be reduced by 10%. This result is shown in the following table:

As you can see, while the reduction in forecast cash flows affects their expected present value, it has no affect on the Implied Risk Premium. To put it slightly differently, an economic change that reduces expected cash flows across all scenarios can reduce the price of an asset without implying any increase in the premium an investor requires to hold the asset. This is one explanation for the falls in some asset class values in May. For example, the announcement by the World Health Organization in mid-May that there was increasing evidence of easier human-to-human transmission of H5N1 influenza might have caused some investors to reduce their cash flow forecasts under all scenarios.

Now let's look at what happens if, instead of the forecast cash flows under all scenarios being affected, there is instead a change in the probabilities attached to different scenarios. More specifically, let's look at what happens if the probability assigned to the Downside Scenario over the next three years increases. The affect of this change is shown in the next table:

As you can see, in this case, not only has the expected present value declined, but the Implied Risk Premium has also increased. The increased probability that the Downside Scenario will occur means that the asset is riskier; hence a higher premium is required to compensate an investor for holding it. This is another explanation for the falls in asset class prices we saw in May. For example, news about growing weakness in the United States housing market (which supports consumer spending, and, indirectly global economic demand) could have changed investors' estimate of the likelihood of a significant downturn in the global economy.

Undoubtedly, some of you reading this will say that what we have written thus far is a far too rational explanation of the way financial markets work. And of course you are right. In reality, there are an infinite number of possible scenarios and associated subjective beliefs about the probability they will occur (e.g., see our March Economic Update for examples). More importantly, not all investors in a market make their buy and sell decisions on the basis of some notion of fundamental (i.e., present) value. Rather, there is a second broad group of investors who use a different forecasting approach, and they too can have a large impact on prices. These investors forecasts the future price for an asset based not on its underlying economic fundamentals, but rather on the basis of the expected actions of other investors. In many cases, the forecasted behavior of other investors is derived from an analysis of past prices for the asset in question. This investing strategy is known by many names, including "momentum", "charting", and "trend-following." It is not necessarily an irrational approach. The more difficult it is to accurately forecast the fundamental (i.e., present) value of an asset, the more its price is likely to be affected by investor psychology and market price trends. In other words, the weaker the anchor provided by fundamental valuation, the more emotion (or, as John Maynard Keynes called them, "the market's animal spirits") will determine the price of an asset. (For more on this, see "Valuation Uncertainty and Behavioral Biases" by Alok Kumar, "Information Uncertainty and Stock Returns" by Frank Zhang, "Technological Revolutions and Stock Prices" by Pastor and Veronesi, and "Investor Sentiment and the Cross-Section of Stock Returns" by Baker and Wurgler). However, these conditions are inherently less stable than those in which there is a greater consensus about fundamental values. As a result, prices in markets dominated by trend-following investors tend to be more volatile, and prone to large changes over short periods of time.

The most interesting question is what causes these sharp changes to occur. There is no easy answer here, because of the underlying "infinite regress" problem involved, of the "I expect that you expect that I expect..." variety. (For more on this see "Everything That I Have to Say Has Already Crossed Your Mind" by Koppl and Rosser and "A Cognitive Hierarchy Model of Games" by Camerer, Ho and Chong). Moreover, accurately forecasting the behavior of other investors is made more difficult by the awkward fact that some of them are not perfectly rational (though that cannot be said of model-driven trading programs). There is ample evidence that not every investor has the same amount of knowledge, nor the same access to information, nor the same ability to learn from experience. They also suffer, to varying degrees, from different cognitive biases (e.g., excessive optimism and a reluctance to change one's opinions, even in the face of conflicting evidence), pay attention to different news items, and have differing needs for social conformity (i.e., to make the same trades as their peers). The presence of such "boundedly rational" players in a market sometimes gives rise to herding, fads and similar behavior patterns (for two interesting non-financial examples, see "Information Cascades in Multi-Agent Models" by Devang and Lee, which looks at the movie industry, and "Management Fads, Pedagogies, and Other Soft Technologies" by Bendar, Huberman and Wu). In the context of investing, this can lead to bubbles and crashes, in which asset prices can depart, sometimes for long periods and by significant amounts, from some notion of their underlying fundamental values before a sharp correction occurs.

In a market characterized by a substantial percentage of trend-following investors, large price swings occur when an accumulation of changes, both bits of news and communications between investors, reaches a "critical" or "tipping" point. Sometimes this tipping point is catalyzed by a highly visible story of the "emperor has no clothes" variety. More often, however, there is no highly visible cause; for example, people still argue about just what triggered the October 1987 equity market collapse. These cases are typically characterized by a growing feeling that "the market" is changing, that culminates in a decision to act. When many investors reach this point at about the same time, a significant order imbalance can result (e.g., many more sellers than buyers), causing market liquidity to dry up and a large price change to occur.

Could we have seen just this process occur in May? One could certainly argue that (as we did in our March Economic Update) there is a high degree of uncertainty in the global economy today, which translates into similar uncertainty about the fundamental value of many asset classes. And there are plenty of commentators who have stated their belief that trend-based trading has increased. Under these circumstances, a highly visible piece of news like Philip Coggan's May 13/14 Financial Times "Something's Gotta Give" article, or trades out of certain assets by highly regarded fundamental investors could certainly have catalyzed a significant change in investor emotion and behavior. Moreover, in a market characterized by the presence of many leveraged investors (like hedge funds), even moderate asset price changes can trigger substantial selling volumes. So a change in trend-followers' forecast for the most likely behavior of other investors - triggering a decision to get out first - is a third theory that explains the asset price falls we saw in many markets in May 2006.

In point of fact, the most likely case is that all three explanations played a role in last month's events. Yet do any of them indicate a widespread rise in investor "risk aversion?" There is a difference between an overall change the compensation investors require to bear a given amount of risk and a change in an investor's perception of the riskiness of a given asset. The former is truly a change in "risk aversion", while the latter is a forecast change. When a change in risk aversion occurs, one would expect to see not only declines in the price of more risky assets, but also increases in the price of less risky assets. While May saw significant falls in prices for risky asset classes such as emerging markets equities, it did not see significant increases in prices for low risk assets like short term U.S. Treasuries. While this may yet occur, it did not happen last month. Moreover, if an increase in risk aversion does occur, it seems likely to be a temporary rather than permanent change.

Given the size of their long-term financial goals compared to their current savings, many investors have recognized (no doubt reluctantly) that, unless they are willing to sharply reduce current consumption (and few are), they must bear a high level of investment risk. Absent this, they have no chance of earning the relatively high returns they need to achieve their objectives. While at some point in the future widespread debt forgiveness and/or a cultural shift away from conspicuous consumption may cause this situation to change, today these still seem unlikely to occur. Thus, any increase in investor risk aversion that happens seems likely to reverse.

None of this should be interpreted to mean that we don't expect further declines in some asset prices in the months ahead. As described at length in our March Economic Update, we expect investors to continue to increase the probability they give to downside scenarios, with additional negative consequences for prices in some asset classes, including equities of all types, riskier bonds, commodities (especially products that track energy heavy indexes like the GSCI), and commercial property (though like commodities, its price fall should be limited by its attractiveness as an inflation hedge). At some point, these falls will most likely trigger a true increase in risk aversion, which will be marked by rising prices for government bonds (both real and nominal return), foreign currency bonds (especially if you are a U.S. Dollar based investor), timber, and probably gold and silver. Our guess is that this will occur when asset price declines spread to domestic equity or commercial property markets. However, contrary to assertions by some market commentators last month, we have not yet reached this point.

| 2006-2007 Benchmark Portfolios | This Month's Issue: Key Points | Retail Financial Services Trends and Opportunities | Product and Strategy Notes: Technology and Active Management; Hedge Funds Update; and Update on H5N1 | Global Asset Class Returns | Asset Class Valuation Update | What Happened to the Financial Markets in May? | This Month's Letters to the Editor: Convergent/Divergent Investment Strategy; EMN Funds vs Long/Short Equity Funds; Equity Market Neutral in Portfolios; YTD PCRDX Performance; Japan Overvalued?; Forecasting Volatility; Portfolios During Economic Downturn; Timber and Residential Property; Concentration of Swiss Equities in Swiss Portfolios; and Changing Asset Allocations |