Mohamed A. El-Erian, A. Michael Spence
IntroductionFinancial markets are subject to periodic bouts of systemic risk that affect their functioning and stability, as well as investment returns and volatility. Several elements of investment strategy – including asset allocation, liquidity management, risk mitigation and, in certain cases, even the design of benchmarks and guidelines – can be affected by this reality. Accordingly, and particularly given the ongoing re-alignment of the global economy and markets, it is important to have a handle on the underlying dynamics. To this end, in this paper we analyze those associated with sudden shifts in expectations. It explains how they can morph into particularly disruptive multiple equilibria dynamics, and it points to possible implications for market outcomes, market equilibria and policy responses.
The ContextEarlier work in economics on signaling and screening, including by one of the authors of this article (Mike Spence), sheds important light on issues that are of current interest to markets and investors. Specifically, a significant subset of multiple equilibrium market structures – the ones that are most relevant to financial markets and to their interactions with the real economy – are those in which expectations have two characteristics: First, the expectations are endogenous – i.e., they are determined as part of the process of reaching equilibrium outcomes in the market. Second, they exert a substantial influence on behavior and hence on the market outcomes that are inherently linked with the expectations themselves.
The interactions of these two factors are critical. Indeed, if they are misunderstood, they can easily result in inappropriate analyses, investment decisions and risk management. They also can lead to misguided policy reactions.
If just the first characteristic is present, markets are in equilibrium when expectations are accurate. However, if both are present, then expectations aren’t best described as accurate but rather self-confirming in a serial manner; and the sequence of local equilibria need not lead to a global equilibrium. Consequently, it is this kind of structure that has multiple equilibria and hence the potential sudden shifts in both expectations and equilibrium market outcomes.
For financial markets, balance sheets and the real economy, the endogeneity of expectations is always part of the structure. But shifts in expectations and asset values need not always cause powerful feedback effects on investor behavior and/or the real economy. They do when a slight initial perturbation in expectations is amplified into a rapid, non-mean-reverting shift to a very different set of outcomes.
In assessing systemic risk, we therefore need to look for the conditions where there are substantial feedback effects and where market or policy circuit breakers are either missing, incomplete or uncertain.
Signaling as an Example of a Multiple Equilibrium StructureSignaling occurs in market environments in which there are both informational gaps and informational asymmetries between the buyers and sellers. That is, there is some attribute of the product or service about which one side knows more than the other.
Asymmetrical information conditions are quite common. They occur in labor, financial and insurance markets, and in many more places. As an example, in insurance markets, the observability gap is variation in risk across the entities seeking to buy insurance. This phenomenon is called adverse selection. It leads to sub-optimal outcomes and market failures.
Incomplete information and the inability to distinguish cause a loss of product differentiation. Sellers try to recover the differentiation through signaling. For buyers, signals are visible attributes that sellers can acquire at a cost that, in turn, transmits information to buyers. Signals that survive and contain information in equilibrium have the property that their costs are negatively correlated with the invisible, valued attribute. If that condition is absent, the hypothetical signaling behavior of sellers will not be correlated with the underlying attribute. Subsequent market outcomes will reflect that, and the signal then loses its informational content and falls into disuse and out of the market equilibrium determination mechanisms.
Signaling has the two properties described above: Expectations are endogenous; and they exert a powerful influence on both incentives and choices made by sellers. Signaling models have multiple equilibria, in each of which the expectations are self-fulfilling.
In the current environment of large correlated global macro risks, we naturally tend to think mainly of dangerous equilibrium shifts – away from good ones and toward unfavorable bad ones that are also potentially unstable in that one bad equilibrium gives way to even worse ones in a serial fashion. But, importantly, it can go the other way too.
Yes, you can get stuck in a “bad” equilibrium. For example, in the developing country context in the early stages of development, there is an equilibrium in which there is relatively little growth and suboptimal investment – a bad equilibrium if you like. Any shock, endogenous or exogenous, carries a high risk of tipping the country into an even worse equilibrium. In such circumstances, the leadership and reform challenge is to shift the expectations and hence the underlying investment dynamics to a different and much more positive equilibrium. And it is here that a series of positive equilibriums can impose themselves, with significant implications for investment returns.
Ordinary Market Dynamics: Momentum and Value Investing Financial markets have endogenous expectations. Investors know that expectations and prices can drift away from fundamentals, with a dynamic driven largely by the self-referential nature of the expectations. But there are limits.
In “normal” conditions, the feedback effects of asset price movements on the balance sheets of financial and other institutions (and of households) are not that large; and the wealth effect on activities in the real economy is small as a result. Moreover, arbitrage flows are subject to relatively low liquidity and risk premia.
So while a set of investors may base their choices on momentum and charts, others counter by basing their decisions more on deviations of market prices from some assessment of fundamental values, either short- or long-term. Trading frequency varies, as does the degree of dynamic portfolio reallocations and the premium that can be collected for selling volatility in different market segments.
Generally investors know that at least short term returns are determined in part by other investors’ expectations, but longer-term valuations will revert to levels warranted by the underlying fundamentals. As certain traders cause markets to move away from fundamental values (a mini bubble), the deviations become larger relative to the distribution of underlying value estimates. This entices value investors to move against the trends, thus causing a shift in the market dynamics back toward the central part of the fundamental value distribution.
Depending on the size and responsiveness of assets managed by each class of investor, the deviation from fundamental values can persist for awhile, but it gets pulled back. This can be thought of as a relatively harmless form of fluctuating multiple equilibrium. And it tends to give rise to attractive premiums that investors can capture by selling volatility, either directly in a range of markets (e.g., interest rates, credit, equities, commodities and foreign exchange) or indirectly and less comprehensively through certain asset classes (e.g., mortgages, corporate bonds, and emerging markets’ local rates, credit and equities).
In this rather common dynamic, the deviations are generally mean reverting as expectations shifts do not substantially affect fundamental values. This is not the case when, critically, feedback effects are large and the valuation anchor morphs into a moving target.
For example, in the aftermath of the 2008 crisis, the real economy headed down a highly correlated downward spiral along with the financial markets. Leverage caused a substantial part of the problem, as did pro-cyclical behavior on the part of markets and investors. The result was a significant, across the board repricing of markets, along with “atypical” developments in market correlations and the range of risk premia (including the liquidity premium). As a result, initial changes in asset values that in an unleveraged world would not have produced large real economy effects caused substantial balance sheet damage in the highly levered financial and household sectors. This led to large negative real-economy feedback effects and declining fundamental values. It wasn’t clear what the anchor was, or if there was one. That kind of structure can implode; and it would have were it not for dramatic and unprecedented intervention on the part of policy makers who aggressively substituted public balance sheets for rapidly imploding private ones.
The small feedback condition that we associate with normal stable market conditions was arguably also not what we saw last year in the European sovereign debt markets. Then, a yield run up in Italy or Spain threatened to shift the trio of market, political and policy incentives in such a way that it would have a major effect on credit quality. This had occurred already in Greece. More on this later.
Bank RunsMultiple equilibrium structures are not new. History is full of examples where, absent effective policy circuit breakers, market realities deviated considerably from fundamentals, and did so in a sequential and increasingly disorderly fashion.
Consider the case of the banking sector. The normal levered configuration of banks (in a narrow sense, these are associated primarily with their maturity transformation as short term liquid liabilities fund longer-term and less liquid assets) is key to their role in funding productive investments in an economy. But it also makes them vulnerable to losses of confidence – a situation that was massively accentuated in the run up to the global financial crisis by prop activities and off balance sheet structured investment vehicles (SIVs) and other shadow banking activities.
Sometimes the change in operating paradigm is real, in the sense that the assets suffer some kind of permanent impairment resulting in negative equity. At some point depositors and creditors notice and a race for the exit starts. But as often observed, you do not need such a solvency shock to start a bank run. Just the perception of such a risk will do. Why? Because it is self-fulfilling absent government and central bank intervention. In such situations, solvency itself becomes endogenous.
Extreme bank runs are uncommon these days because governments guarantee deposits and central banks can and do move quickly to monetize the asset side of a solvent bank fast enough to keep it in business, even if the deposit run continues for some time, or if short term private financing in the market is cut off. Eventually the deposit run reverses, borrowing capacity returns, and the original equilibrium is restored.
The Internet Bubble of 2000-2001 The internet bubble of some 10 years ago is an interesting and slightly different case. Valuations of informational technology assets (publicly traded and not) got disconnected from reality, and had some of the characteristics of mini-bubbles described above. But the deviations from fundamental value were quite extreme.
The value investor market correction was delayed by the newness of the technologies and the temporary absence of relevant data to constrain the expectations. In this data-free environment, narratives, unconstrained by relevant experience (there wasn’t any), dominated, usually with a revolutionary, disruptive technology flavor. However, the data-free condition was not permanent. Eventually it became clear that, at least in the short term, forecasts of growth, relevance, revenues and earnings were way too optimistic. The markets experienced a major downward reset, with real economy effects that were large enough to cause a recession.
With hindsight, it seems fairly clear that the market distortions were not caused by an inaccurate specification of the ultimate impact of the technology. Instead, it was the substantial overestimate of the speed with which these new technologies would affect productivity, society and the global economy.
There are a variety of ways to describe this. Essentially, the value-investing anchor was present but much delayed. There were value and activist investors who were skeptics but, in the early stages, either they were less numerous in terms of control over assets or less certain of their views – hence the delay in responding.
The internet bubble was sufficiently large that it did, via the wealth effect, produce a shift in the trajectory of the real economy. For a while this effect was positive and reinforcing. But when expectations shifted and valuations reset, the reverse occurred and the real economy dipped to the point that the central bank opted for low interest rates to cushion the recessionary effect.
Standing back from these specific two cases for a moment, something is clear. In both, there were multiple equilibria affecting market activity, policy anchors and circuit breakers; and they ended up operating with varying speed and certitude.
Sovereign Debt and the Eurozone Olivier Blanchard, the chief economist at the IMF, observed in his 2011 end-of-year remarks: “… post the 2008-09 crisis, the world economy is pregnant with multiple equilibria – self-fulfilling outcomes of pessimism or optimism, with major macroeconomic implications.”
Consider the case of Italy last year. In May, Italian bond yields were relatively stable and well behaved. By August and again in November onward, unprecedented volatility drove them to dangerously high levels – enough to raise legitimate concern about the risk of a debt insolvency sequence.
With yields in the 6% to 7% range and the prospect that they might remain high as maturing low cost debt was increasingly refinanced via high cost debt, there was a material risk of a shift in expectations – not just on the market side, but also on policy incentives. The interactions of these expectations, including through feedback mechanisms, translated into mounting pressures on credit quality, yields, growth and policies. They also had social and political effects. And as the Italian stock market lost a third of its value, the wealth effect kicked in, helping to push Italy’s economy toward negative growth and debt deflation.
No part of this material may be reproduced in any form, or referred to in any other publication, without express written permission. Pacific Investment Management Company LLC, 840 Newport Center Drive, Newport Beach, CA 92660, 800-387-4626. ©2013, PIMCO.
Are you sure you would like to leave?
You are currently running an old version of IE, please upgrade for better performance.