finance: Roman Frydman & Michael Goldberg on Hayek, Mathematics, Wall Street & Financial Regulations

Economists Roman Frydman and Michael Goldberg presented a paper exploring the significance of the limits of knowledge and mathematical modeling for financial regulations titled “Emerging from the Financial Crisis” (pdf) at Columbia University Friday, Feb. 20, 2009.  From their paper:

In his prescient critique that central planning must fail in principle, Friedrich Hayek emphasized that no mathematical model could mimic exactly what markets do.  Remarkably, prevailing models of asset markets – even those based on behavioral considerations – ignore this key insight about capitalist economies. Instead, contemporary approaches to financial markets and risk presume that an economist can identify precisely the set of factors that determine market outcomes over the shorter and the longer terms.5 These models characterize as routine the way “rational” or “irrational” individuals forecast the future and make decisions, thus representing these decisions with mechanical rules that are specified in advance by an economist. Based on these individual micro-foundations, prevailing models imply that market prices also follow pre- existing mechanical rules.

In fact, as Frank Knight, John Maynard Keynes, and Hayek all emphasized, neither an economist nor the market can get asset prices exactly right in either the shorter or the longer terms.6  This seemingly uncontroversial claim goes a long way toward explaining contemporary models’ gross failure in accounting for asset-price swings and risk. It may also help to explain why policy makers – who, after all, have extensive experience and intuitions about what drives market outcomes – have made scarcely any reference to economists’ models in interpreting the crisis or in thinking about new ways to re-regulate financial markets in response to it.

To remedy this flaw, economic models should reflect modern economies’ key feature: the fact that individuals and companies engage in innovative activities, discovering new ways of using existing physical and human capital and technology, as well as new technologies and new capital in which to invest. Moreover, the institutions and the broader social context within which this entrepreneurial activity takes place also change in novel ways. Innovation in turn influences the future returns from economic activity in ways that no one can fully foresee,7 and thus that do not conform to any rule that can be prespecified in advance.

Indeed, the inherent imperfection of knowledge that Knight, Keynes, and Hayek so clearly recognized is crucial to understanding that financial markets are hard-wired to undergo price swings that revolve around historical benchmark levels. Although some market participants may fall prey to emotions or rely on technical rules, swings in asset prices arise from market participants’ necessarily imperfect knowledge about how these prices will unfold over time. Indeed, once the inherent imperfection of knowledge is recognized, price swings may occur even if all market participants forecast future prospects solely on the basis of fundamentals.

The fact that market participants search for new ways to forecast the future and find profitable avenues to invest their capital neither suggests nor requires that the market does a perfect job in allocating scarce capital. However, placing imperfect knowledge at the center of one’s analysis does suggest that swings in asset prices are an inherent part of how markets function in allocating capital. If swings lie at the heart of what markets do, then eliminating them or transforming them into mere random variations around “true” fundamental values would require the state to close markets down altogether or regulate them so heavily that they virtually cease to play any useful role.


The longer the boom lasted, the more the ratings agencies trumpeted the superiority of structured finance over loans to businesses – and the more investors came to rely on these ratings. Brave new models, which largely ignored the changing structure of the processes driving risk, together with radical deregulation, tempted the mortgage industry into abandoning proven prudential procedures that combined their own judgment and more formal criteria. Instead of lending to “the man who shaved this morning” (to use Albert Camus’ wonderful phrase), they lent mechanically to a FICO score.43 And homebuyers responded by learning how to manipulate their FICO scores. Of course, nobody knew when the reversal would begin. Had the ratings agencies been required to make explicit how their ratings would change under the alternative assumption that housing prices might fall dramatically once the inevitable reversal arrived, projected loss rates on the securities that the investment banks were selling would have been much higher – and their ratings and prices would have been much lower. Instead, by assigning single ratings to assets, the agencies failed to convey the necessarily contingent character of the models and assumptions that underpinned them.

This leads to a simple proposal.44 Ratings agencies should be required to report at least two ratings of securities, along with the methodology used to arrive at each: one assuming that historical patterns will continue, and at least one other assuming reversals in the trends of major variables and the prices of the underlying assets. To be sure, Moody’s, S&P, and Fitch apply “stresses” to their current procedures. But these stresses are hidden in ratings reports, and, as recent events have painfully demonstrated, are woefully inadequate. Furthermore, requiring the agencies to rate securities under one or more pessimistic scenarios would make it harder for them to deliver rosy ratings in return for business from the investment banks.

No single individual or institution can render a definitive judgment on the riskiness of securities. Friedrich Hayek showed that only markets can aggregate knowledge that is not given to anyone in its totality. The new regulatory regime should require rating agencies and issuers of securities have to help markets perform this function

This entry was posted in Limits to Knowledge, Mathematics, Regulations. Bookmark the permalink.