Advertisement

Op Eds

Volcanic Ash Allowing

Think about the perfectly-timed choreography: the battle over the emergency exit row seat, the subprime dining selection (assuming there is still choice by the time they get to you), the Twitter flight status updates, the arrival dinner, the planned drinks date, and the conference or company or checking account that is paying for the jet-setting. Now insert a volcanic eruption somewhere—say, Iceland—and rethink all of the above.

Suddenly the choreography turns into sauve qui peut: Airlines are in absolute disarray, expense accounts bill thousand-euro Eurostar tickets, the German chancellor buses from Rome back home, and, at the ever-so-fancy Dorchester Hotel tea promenade in London, one can be offered a private jet seat back to New York for only 10 thousand euros.

The paradox is clear enough: Even though skies were clear over London and almost all civilian test flights seemed to have gone swimmingly, European skies remained closed for days on end, stranding thousands and disrupting all plans. Why? Because the models run by European regulators said it wasn’t safe.

Economists would tell you that everything happening is perfectly rational: The efficient market hypothesis of ash cloud flying calls for airline financial ruin, outrageous train prices, and a market for jets at the Dorchester. The market would have predicted it all, if only our models had incorporated Eyjafjallajökull data. As if.

But people’s behavior, clearly enough, is neither rational nor remotely predictable. Neither was the volcanic eruption. Having just attended the inaugural conference of George Soros’s Institute for New Economic Thinking, one thing is clear: The ash cloud over the continent is the perfect metaphor for the state of economics as a field. Both in economics amidst the global financial crisis as well as, it turns out, in volcano ash predictions, we are relying too much on models that have proven fallible.

Advertisement

There are those who still insist on rational expectations theories and EMH as the best predictors of human and market behavior. Put simply, their point is that mistakes may be made, but economics as a field can only be useful with mathematical modeling as the central pillar of analysis. Based on growing amounts of data, models may be incomplete, yet they are also perfectible. Such trust is essential, since it assumes that fully determined outcomes are possible and that more data can overcome any shortcoming.

Admittedly, it is hard to argue with such a positivist position, particularly when financiers and policymakers are willing and eager to rely on the appearance of mathematical certainty they provide. Certainty is attractive: Investors like companies with steady returns, just like individuals purchase insurance to hedge against the unknown; even marriage, if you will, is a quest for certainty.

Ironically, the challenge to this position is anything but new. It goes back to the German historical school of famed authors like Arnold Toynbee and Joseph A. Schumpeter. As John Maynard Keynes himself wrote in the preface to his masterpiece, “The difficulty lies, not in new ideas, but in escaping from the old ones, which ramify, for those brought up as most of us have been, into every corner of our minds.” Fully determined models, for ash clouds or financial markets, may be resilient, but they are intrinsically incomplete, for they account for neither the inherent unpredictability of human action nor the intractability of certain data.

The first is closely related to a philosophical idea popularized by Soros himself: reflexivity. Originally developed as a sociological concept, reflexivity incorporates the essential feedback loop between cause and effect into analysis. This is best illustrated through self-fulfilling prophecies; think about a bank run à la Northern Rock: If rumors abound about a bank being illiquid, depositors will self-fulfill the prophecy by queuing to clear their accounts. There is a ‘tipping point’ for bubbles or bursts alike, which explains where market performance is anything but normally distributed. As Nassim Taleb made a career out of showing, we end up having “once in a century” events far more often than that. What economists call “fundamentals” are little more than a mirage, subject to constant positive (bubbles) or negative (bursts) feedback loops. Make no mistake: This is a world where animal spirits rule supreme.

The second idea further complicates the ideological position of traditional econometricians by borrowing heavily from physicist Werner Heisenberg’s principle of uncertainty. If the observer adds an element of uncertainty by the mere fact of observing, then fully determined prediction is not a matter of how much data one can gather. Rather, it is “computationally intractable,” meaning that if there were an answer, the amount of data required to compute it is beyond not only our current methods, but anything we could ever achieve—we just cannot expect computers to model the very effect of their modeling.

Considering these essential shortcomings, what is most surprising is that we still trust our models so much. The fact that the mirage of certainty has crept from economics into other disciplines is obvious enough: A famous Harvard government professor often complains that no data-heavy, model-driven graduate student gets a good job in political science these days. Perhaps more importantly, such flawed assumptions have grounded endless flights this week. Just yesterday, EU regulators acknowledged they should have conducted more actual tests rather than outright banning European flights for five days. The EU director general for transport, Matthias Ruete, admitted as much when referring to the models that had led to the ban: “[They are] a black box in certain areas.”

In some Latin-derived languages such as Spanish the term for “principle” is the same as that  for “beginning.” Hence “principio” can convey either idea, which suggests that perhaps the conceptual gulf in languages like English should not be quite so steep. The abandonment of fully-determined mathematical models that cannot deliver the certainty they promise and the consequent embrace of uncertainty may thus be, for economics as well as volcanic ash predictions, a good new principle, but perhaps an even better new beginning.

Pierpaolo Barbieri ’09, a former Crimson associate editorial chair, is now Lt. Charles Henry Fiske III Harvard-Cambridge scholar at Trinity College, Cambridge.

Tags

Advertisement