Leamer and the state of macro

(Pardon the long post, but this is a topic that I love.) Adding another post to a topic that Jason and I have discussed often both on and off this blog (post 1, post 2), I wanted to post a link to a podcast interview with Ed Leamer on EconTalk. Leamer is a renowned economist in international trade and econometrics. Russ Roberts' interview with Leamer is interesting and insightful, and I recommend it as a good listen. But I finished the podcast feeling very confused. Leamer argues both that current macroeconomics does a terrible job at explaining the data and at having a story to explain the data. This is not an inditement, but rather a good indirect description of the two current ways of approaching macroeconomics.
The fundamental premise of Leamer's arguments in the interview, as well as that of his new book, is that macroeconomics should be based on data and based on a story. I think "story" is just another world for "model"--that is, some framework for thinking about the data and interpreting it. Yet, early in the interview, Leamer reveals his disdain for the more model-heavy (or story-heavy) segment of macroeconomists, those "real business cycle characters."

He goes on to make what I think is a false acusation, that macroeconomics does not pay enough attention to the data. Macroeconomics is always checking itself against the data. In a previous post, I included a diagram entitled "the cycle of economics," which shows the interplay of theory, data, and policy in moving the field forward. All three are necessary, and have been present, in the bumpy evolution of the field of macroeconomics since the 1930s.

In fact, macroeconomics came to prominence during the Great Depression--a period of dramatic data and policy that economists tried to make sense of and interpret through theory. Before the 1930s, the prevailing macroeconomic doctrine was called classical macroeconomics and highlighted the forces of competition and free markets from Adam Smith and David Ricardo.

But with the Great Depression, Keynesian theory came to prominence. John Maynard Keynes laid out a framework that implied that the Great Depression was primarily a shock to demand, and that government policy (spending) could mitigate the effects of negative demand shocks in the short run. Keynesian theory was based the idea that markets had imperfections and frictions, an assumption that was missing in classical theory. Keynesian theory also implied a tradeoff between inflation and unemployment--the Phillips curve.

Keynesian theory prevailed and informed U.S. policy well into the 1970s because it fit the data so well. See... theory, data, and policy. But then the data changed. In the 1970s, the U.S. saw stagflation--the phenomenon of increasing inflation and unemployment rates. Keynesian theory had no explanation. This gave rise to neoclassical economics and then real business cycle (RBC) economics. RBC theory explained short-run fluctuations with shocks to productivity that moved aggregate supply around. However, the RBC models could not explain much of the data. A Keynesian revival filled the void (New Keynesian economics).

Many articles have been written recently about the current state of macroeconomics (see Jason's post from a few weeks ago for good references). The punch line is that the New Keynesian and RBC branches of macroeconomics are truly converging. The New Keynesian's have adopted more rigorous microfoundations (this is the model or story part) and do a better job of matching the data. But their models still lack a degree of specification (see Chari, Kehoe, and McGrattan, 2009, AEJ Macroeconomics) that is important for making policy recommendations. The RBC advocates have adopted many of the frictions of the New Keynesians (like imperfect competition and sticky prices) and have strong microfoundations, but still are not as good as the New Keynesian models at matching the data.

So now, after my long description of the history of macroeconomics, back to Leamer's premise. Has macroeconomics gotten away from the data? Not a chance. We are just iterating through the cycle of economics, which is the social science version of the scientific method. I am convinced that good work is currently being done by both New Keynesians and RBC economists who are working on the respective weaknesses of each branch of the field. Both branches are making their models better--better at matching the data, better at allowing interpretation of the data, and better at informing policy.

Leamer seems to be arguing that the models need to be more informed by the data and need to have a story. My argument here is that this has been the objective of the field of macroeconomics since the 1930s. We're looking for the right story (model) that matches the data and allows us to interpret it.

Lastly, both Roberts and Leamer discuss in the interview the chances of being able to evaluate whether or not the current government stimulus program in the U.S. was effective. I think that this will be the main question of macroeconomics for the forseeable future. The answer to the question is that it is impossible to judge the efficacy of policy without a model. You need a specific story about how the economy works in order to explain why aggregate variables move together and what was the policy effect.

Leamer's discussion of the data was very insightful, and I think he is right about the need for a story. I just think that he is missing the fact that the current efforts in macroeconomics to improve our models are simply an effort at improving our stories.