A workshop on Knightian uncertainty
September 12, 2007
An inspiring event just took place at Columbia. On September 7-8, David Stark and I organized a workshop on Knightian uncertainty. It brought together distinguished economists, including Nobel laureate Douglas North, as well as sociologists and management scholars, to talk about Knightian uncertainty.
Why uncertainty? It is not an exaggeration to argue that most post-war advances in social sciences have focussed on risk. From Black Scholes to Porter’s Five forces, Nash equilibrium to the efficient market hypotheses, the social sciences have modeled with great success situations in which the future can be predicted. But in a world of innovation, hedge funds, financial bubbles, climate change and terrorism, Knightian uncertainty — not risk — sits at the center of every decision-makers’ agenda. The workshop prompted reflection on these issues. What does uncertainty imply for policy-makers? for managers? For arbitrageurs? For strategists?
Here’s a few of the issues that we touched on.
First, the sheer importance of uncertainty. This was cogently articulated by Douglas North. Moving past the intellectual concern with institutions that won him general acclaim, North has turned to the problem of radical uncertainty, mental models and economic growth. North takes a really long perspective –five or ten centuries– and observes that some countries have risen from poverty to wealth, whereas others remain in misery. Why such wide differences? None of the orthodox economic explanations, he argues, account for it. To North, the explanation lies in the differences in mental models that guide the creation of institutions which, in turn, promote economic growth (or the lack of it). On the same session, Joe Porac expressed his disagreement with one of main premises of the workshop — that uncertainty is behind much of the activity that we see in the economy. Porac pointed to possible psychological mechanisms that may be having a similar effect.
Uncertainty is also central to the corporate world. For example, how should managers confront uncertainty? One of the most counterintuitive aspects of Knightian uncertainty is that the recipes that work best in a predictable world of risk can create major disasters under a scenario of uncertainty. This difficulty was emphasized by Nassim Taleb with the concept of rare events or “Black Swans.” Taleb distinguished between two realms of action: the real of non-rare events or “mediocristan,” and “extremistan.” The former describes phenomena that adhere to a normal Gaussian distribution (size of mountains, weight of people), while the latter are described non-normal distributions (wealth of Bill Gates, stock price movement in 1987). In extremistan, actors have to prepare for the potential occurrence of or rare but disruptive events. They face extraordinary success… and failure. Extremistan is where the key problems, the bankruptcies, the defaults, the shocks arise. It is particularly dangerous for decision-makers take mediocristan for extremistan, and assume they know more than they really do.
How, then, should companies operate in an extremistan-like world of uncertainty? As noted, it requires a very different organization. Firms that operate in this environment, according Anna Grandori, should espouse an “epistemic rationality”. Traditional concerns with optimality and saving resources should be replaced by efforts to discover the environment. The notion of heuristic, Grandori argued, also needs to be redefined: under uncertainty, the challenge no longer lies in recognizing familiar patterns but in discovering new models, novel relationships, etc. Search should no longer stops when performance is good enough, but when the firms’ insights make accurate predictions. The entire organization, in short, needs to be engaged in active research of one kind or another.
A different answer was provided by David Stark’s concept of heterarchy. Uncertainty poses managerial problems that go far beyond profitability. In extreme cases such as transition economies, the definition of success is itself in flux. Profits may not be the key measure of success; employment, sales, production — even location — may be the central reason why a company gains resources. How do you organize for such flux? Stark called for a different internal logic of organization — one that allows companies to adhere to competing conceptions of worth. To accomplish this, companies need to avoid hierarchy and bureaucracy. Instead, firms should pursue a dense network of horizontal interactions and distributed power relations.
The managerial problem, and more specifically the problem of strategy implementation, centered the presentation by Sarah Kaplan. Strategy involves the future. But consider the conundrum faced by an optical fiber company immeditely after the burst of the Internet bubble. A company that had been virtually living in the future for several years was suddenly confronted with a collapse in their market, as well as in their own confidence to predict the future. Kaplan identified the mechanisms whereby the company managed to restore sense and bridge their past, present and future.
How do capital markets confront uncertainty? The question goes at the heart of the conversations in the workshop, as well as at the core of Wall Street’s present headaches. Indeed, the cleavage between risk and uncertainty is at is widest in finance. From the CAPM model to the Black-Scholes equation, orthodox finance has provided Wall Street practitioners with the tools to engage in a new financial strategy, quantitative finance. But in contexts of uncertainty, those models seem to stop working and sometimes even turn against their users. In this sense, Emanuel Derman described quantitative finance in a lucid and succinct manner. Mathematical formulae, he argued, provide a mind-broadening ability to associate the value of a stock to the value of some other, seemingly different one. Unlike physics, which works with axioms, modern finance operates by analogy. But these analogies may break down in contexts of uncertainty. Examples of these include the 1987 crisis, the 1998 crisis and… the current subprime debacle (which was in the mind of all participants). What do to then?
In some ways, my own presentation addressed this very problem, and pointed to financial tools as a solution. I examined how merger arbitrageurs used a specific visualization device, known as the “spread plot,” to confront their own interpretations with those of their competitors. The spread plot provides arbitrageurs with information about what their rivals think. By alternating between their own estimates of merger probability and the probability implied by the spread, arbitrageurs stay alert to their own misinterpretations and are prompted into search. As arbitrageurs adopt financial positions, their beliefs feed back into the spread. Over time, the social use of the spread plot among the arbitrage community leads to gradual convergence in probability estimates. The arbitrageur’s solution to uncertainty, I suggest, is a financial tool: the spread plot.
Harrison Hong widened the discussion of financial uncertainty with a generalized model of disagreement in markets. His discussion centered on discrepancies between the beliefs of market actors and the ways in which these influence traded volume and price volatility. Behavioral models, Hong argued, provide some explanation for either of these, but no single model can explain extraordinary movements in both volume and prices. And yet, this is the central trait that distinguishes situation of uncertainty such as the dot-com bubble. In many ways, the concern echoes the problem raised by Beunza. Namely, that because arbitrageurs draw on all the social clues at their disposal, perception and action, prices and trading, fluctuate together. Hong presented several models in which differences in interpretation, constraints on shorting and differences in media coverage combine to yield the co-movement in prices and volume that is observed. In short, then, disagreement and differences in interpretation is crucial. Without them, one cannot account for bubbles or other periods of uncertainty.
The role of disagreement on Wall Street was further studied by Raghu Garud. A key group in this setting is Wall Street’s securities analysts. They play central role in articulating, using and debating different perspectives about company value. Garud described the role of analysts in advancing different opinions about Amazon.com during the emergence of the Internet. Analysis even interpreted the same piece of news in radically different manners. Eventually, however, their differences in interpretation collapsed when the milestones and rhythms promised by the dot-com optimists were not confirmed by events. But the process took three years.
Uncertainty and strategic interaction
What does “strategy” mean when the opponent is unknown? Traditionally, strategic interaction has been studied in game theory. But according to Adam Brandenburger, game theory can sometimes be unhelpful; specifically, orthodox game theory typically assumes that both players in the game know how the other person thinks, her interests and her payoffs (the so-called assumption of “common knowledge”). Bringing uncertainty into our understanding of games radically shifts the problem from an abstract exercise of calculative anticipation into something much different. In the so-called school of “epistemic game theory,” strategic interaction is seen instead as the way in which the players discover each other. What does the other person think? What does he or she stand to gain and lose?
In understanding “the other,” it is helpful to keep in mind the logic of justification that defines him or her. According to Laurent Thevenot, it is not the same to operate in a domestic regime (one rooted in tradition) than in a market setting (determined by competition) or a civic setting (shaped by the general interest). These different regimes not only bring with them different geographical locations, goals and metrics for success, but even differences in values. Such multiplicity of worth has been studied in detail by the French school of “economics of convention.” Conventions, according to them, are the solution to the problem of coordination in games.
But how does a convention arise? Such was the question addressed by Olivier Favereau. The heterodox economist explored the process whereby a collective of people who do not share a language are able to espouse a convention, underscoring the role of intersubjectivity in bringing about common knowledge.
All in all, an extraordinary event. The description above only begins to scratch the surface…. very soon, I’ll also be reporting on some very provocative commentaries by Bruce Kogut, Harrison White, Yuval Millo, and Fiona Murray.