Can Science Help Solve the Economic Crisis?

A very high power group of financial experts and scientists anwered yes in a December 11, 2008 paper with this title on edge.org, but only after launching an "economic Manhatten Project," proposed by Eric Weinstein "to develop a new paradigm for economic theory and modeling markets."  This is in stark contrast to the March 9, 2009 New York Times article on how the misuse of quantitative analysis on Wall Street led to the financial crisis.  The edge.org authors present an excellent critique of the elements of neoclassical economics that have failed us and an ambitious plan for a independent nonpartisan global commission of economists to develop the best open source models and present policymakers with the likelihood estimates of whether their policy proposals would have their intended effects.  What better evidence is there that such efforts can go badly astray than the financial crisis of the past year?

As the New York Times article explains, some very smart and very powerful financiers on Wall Street believed so much in the quantitative models they developed over the past 20 years that they really believed they could make money no matter what the market did.  Now they're learning otherwise.  Somewhere along the line, a lot of risk got buried in all of those complex derivatives transactions, much of which still hasn't fully come to light.  It's like my sister-in-law's Thai soups, they taste great until your eyes start watering and you start struggling to breathe as you reach for the nearest milk product.

Having spent many years working with individual income tax model and other policy models on Capitol Hill, I have learned some hard lessons:

1. Quantitative models can be very seductive.  Policymakers see the point estimate, "This amendment will cost $X billion," but they don't see the underlying assumptions or the quality of the data.  If it came out of a computer, it must be right was how most policymakers responded.  Some things can be estimated quite accurately, and some can't.  I don't ever remember a policymaker being told that his or her amendment couldn't be accurately estimated.
2. Quantitative models reach rigorously logical conclusions based upon the theory, assumptions, and data built into them.  Computer models are very short sighted.  Computer models will draw a straight line to infinity, when a curve might fit reality better.  How to bend that curve is the hard part.
3. Quantitative models don't turn corners, i.e. predict an economy falling as sharply as ours has in the last six months.  They assume continuity with historical data.  They assume continuous functions.  They assume economic relationships hold together throughout the forecast period.  I haven't heard of any model that, even ex post, could predict the economic path we're on now.
4. Economic policymaking is path dependent and is heavily dependent upon expectations; economic models aren't there yet.  Prior attitudes about the future matter, and they are notoriously difficult to build into models.  When expectations suddenly change, everything changes very suddenly.  The data lags.  I vividly remember going home for the holidays in mid-December, 1974 with one of President Gerry Ford's "Whip Inflation Now" buttons as a souvenir and coming back to Capitol Hill three weeks later to find the economy in free fall.  No model predicted that ex ante or ex post either.
5. Confronted with model failure, it's tempting to drop economic theories that lack predictive power and to replace them with inductive methods.  Just crunch all the data you can find that might be relevant, and the patterns will emerge.  Then you can back into a theory.  Then you have no idea what to do when the real world heads down another path.  Developing better theories is very hard work, but, in my opinion, it's still the best way to go.

I'd like to think that human beings are smart enough to learn from their mistakes before they become catastrophic.  Unfortunately, in the past year, we've learned that if you throw large enough bonuses and large enough promised rates of return at smart people, they will make irrational financial decisions.  Very few were smart enough to say no to creating increasingly risky mortgage backed securities, and very few were smart enough to stear clear of Bernie Madoff.  If only these humans wouldn't mess up the models!