Chapter 11

Risk

The lucky person passes for a genius.

—Euripedes

Crises have a way of inviting reassessment. Ex post, we ask ourselves what went wrong, how we failed to anticipate events, and how we might refocus our efforts so as to manage events better in the future. Sometimes this requires newly innovated methodologies and sometimes it simply means incorporating things we had known all along but had somehow de-emphasized.

One such de-emphasis has been the distinction between uncertainty and risk. The University of Chicago economist Frank Knight first distinguished between these two terms as far back as 1921. And yet, as events began to unfold at the onset of the 2008 credit crisis, we still talked about risk for the most part as if it were a set of “known unknowns.” In this case, hubris had a rather high cost.

In general, risk involves choices we make in a world in which outcomes are random but their probabilities are known in advance. Gambles are examples—they involve risk but there is no uncertainty. Uncertainty, rather, deals with unknown risks. To extend our gambling analogy, uncertainty is a poker game in which the risk, which can be modeled precisely in a deck of cards, is elevated because of the behavior of other players (for example, bluffing). Risk, therefore, is amplified by uncertainty. Still, despite knowing this, the great moderation of the Greenspan years nevertheless lulled us into complacency until we found that standard risk models didn't work well in a world complicated by increasing uncertainty.

Uncertainty tends to operate in a feedback loop during crises in which peoples’ decisions are not self-regulating. As a result, we get extreme outcomes (Taleb's Black Swans) that are generated by increased uncertainty and hyperelevated risk. Short selling, for example, can produce unstable results through self-reinforcing feedback. Some crises, then, are reminiscent of regimes in which risk becomes endogenous to the system. Moving from exogenous to endogenous risk is a very difficult transition to make in modeling—one from a world of known unknown risks to one in which risk itself is a product of the system.

Exogenous risks are always present—they are the roll of the die, so to speak, and the forces of nature independent of our own behavior that affect outcomes. Endogenous risks arise because of uncertainty and are intimately related to how our behavior is affected by the behavior of others. Manias, bubbles, panics, and cascades all occur because of the dynamics of human behavior. As events of the credit crisis unfolded, we realized quite clearly that we had moved beyond the capacity of standard risk models to manage the challenges that confronted us.

“But this time is different” (Reinhart 2009). Is it? The history of speculation can be traced at least as far back as the Roman Empire. Relatively recent episodes include the South Sea Bubble of 1720, the railroad mania of the 1840s, the crash of 1929, the Japanese bubble of the 1980s, the tech bubble in the 1990s, and the housing bubble that ended abruptly in the summer of 2006. The answer to the question why we did not seem to learn from these experiences is not entirely obvious. One possibility, however, is that there always seem to be different players with no existing collective social memory of previous crises, and therefore the mentality that, indeed, this time is different. Regardless, the relevant question is what have we learned from the current crisis? Without question, we learned that risks can be endogenous, that volatility is but one aspect of risk, that other risk sources such as liquidity, leverage, counterparty, and systemic risks (to name a few) can be just as important, and that complexity, though wonderfully elegant, often harbors hidden risks.