"More Data" Won't Improve Economists' Lousy Predictions
The Economist has lately been running a series of articles on the shortcomings of the economics profession. Its most recent piece argues the important point that “the [2008-2009] crisis exposed the economic profession’s continued ignorance of the business cycle.” One reason for this ignorance is, they observe, a lack of data: while crises do happen repeatedly, they do not happen often enough for statistical analysis of them to be rigorous. Another reason—underlying the ongoing debate between neoclassical and New Keynesian camps on monetary policy—is represented by the “epistemological woes” of macroeconomics, which the profession “must get to grips with… if it hopes to maintain its influence and limit the damage done by the next crisis.”
However, The Economist is not suggesting an overhaul of the methods and tools of macroeconomics. Far from it. In mapping out the disagreement between schools of thought, they fall clearly on the side of fiscal and monetary intervention, lumping together classical economists with monetarists, non-interventionists, rational expectations economists, and conservatives. In pointing out epistemological woes, they deplore only unsuccessful forecasts, like “underestimating the risks of targeting a low rate of inflation.” In showing avenues for research, they want to bring back to the old discredited Phillips curve.
It is thus not the superficial content of their analysis, but their discussion and approach to these issues which showcase the true shortcomings of the modern, mainstream economics profession.
First, analysis of empirical data remains the fountainhead of economic theory instead of human action. But there will never be enough data to give birth to economic theory, simply because, as Mises explained,
“We grasp the effect of changes in the data by means of our theory” (Mises 2003, 170).
How else would economists know what data to look for, what data to ignore, and why there is not enough data on business cycles? It is through theory that we can disentangle the causal effects of multiple factors on economic variables, not the other way around. Letting data drive theory is like groping in the dark hoping to find not only a light switch, but the very idea of a light switch as well.
On the other hand, and as a result, theory can help us predict the quality and trend—but not the quantity—of the consequences of past changes in the data. In this sense, therefore, macroeconomic models can quantitatively predict the working of the economy much like driving a car with square wheels can predict the results of Formula One racing. There can be no successful forecasts in this sense. As Mises put it,
“All the endeavors that have been and are being devoted to the construction of a quantitative theory of catallactics must, therefore, come to grief. All that can be accomplished in this area is economic history. It can never go beyond the unique and the nonrepeatable; it can never acquire universal validity” (2003, 170).
Moreover, with models that lack insight into the role of time, capital, entrepreneurship, money and economic calculation, qualitative prognostications are equally fruitless. Until we get rid of these methodological woes, modern macroeconomics will continue to lack any power in limiting the damage done by the next crisis.
So if you really need to know about the epistemological problems of economics, don’t listen to The Economist. Read Mises instead.