Empirical Economics Isn’t Yet as Smart as Dentistry
In macroeconomics, ideas are more powerful than facts. That’s a shame, but it isn’t a criticism. Even with the best will in the world (a condition, admittedly, that’s rarely satisfied) economists who debate fiscal and monetary policy couldn’t settle their disagreements by appealing to facts. Their discipline is too difficult, or too far from being a proper science, for that to happen. Ideas trump data time and again.
The thought is prompted by two burning topics being discussed this week by the world’s top economic officials who are gathered in Washington to attend the semi-annual meetings of the International Monetary Fund and the World Bank. The first is a paper trashing a study by Carmen Reinhart and Kenneth Rogoff on the relationship between public debt and economic growth. The other is new work from the IMF on the relationship between inflation and unemployment.
The Reinhart-Rogoff paper looked at historical data and found that when the ratio of public debt to gross domestic product exceeds a “threshold” (their word) of about 90 percent, economic growth drops sharply. These are two of the world’s most distinguished researchers -- Rogoff is a former chief economist of the IMF -- so their findings were taken seriously. Now three economists from the University of Massachusetts at Amherst contend the paper is “riddled with faults” and its headline finding is wrong.
The critique raised a storm of gleeful outrage mainly because it caught Reinhart and Rogoff in one (and only one) cringe-making error. As they admitted, they bungled the coding on a spreadsheet and excluded several countries’ data from an important calculation. The critics also accuse them of incorrectly weighting observations and of “selective exclusion” of data -- but the first, as Reinhart and Rogoff have said, is a matter of opinion and the second, with its implication of deliberate falsification, is wrong (the missing data became available too late for inclusion).
Quantitatively, it’s the difference of opinion on weightings, not the spreadsheet error, that matters. Qualitatively, though, it’s the embarrassment over the error that leaves Reinhart and Rogoff defenseless. They can point out until they’re blue in the face that the critics ultimately agreed with them that high debts are associated with (somewhat) slower growth. It won’t stop the smirking.
You could draw several lessons from this episode, but here’s the one I’m interested in. The Reinhart-Rogoff finding, everyone agrees, was influential -- and now it turns out to have been marred by an error. Does that mean the professors are to blame for misguided fiscal policy all over the world? Many seem to think so. But did the Reinhart-Rogoff paper change anybody’s mind? I wonder. Conservatives embraced it for supporting a view they already held (high levels of public debt are dangerous). Progressives rejected it for opposing a view they already held (fiscal stimulus is necessary).
Progressives, to be sure, had a solid rationale for their position, because the Reinhart-Rogoff numbers didn’t address causation -- does high debt cause slow growth or vice versa? Trouble is, that question is hard to settle. Even harder to say is whether some general finding of that kind would apply to any case in particular -- such as, should the U.S. budget deficit be bigger or smaller next year? Good empirical analysis can hope to inform such discussions. Don’t expect it to settle them.
The IMF’s new study on inflation and unemployment is another instance of the limits of the discipline. The fund’s economists show that since 1995 inflation has been less sensitive to unemployment than in earlier years. Previously, lower unemployment tended to mean higher inflation. Lately, inflation has stayed put regardless of whether unemployment is high or low. The researchers attribute this to the firmer anchoring of inflation expectations that comes from having independent central banks dedicated to managing those expectations with a policy regime called “flexible inflation targeting.”
It won’t be long before we’re calling this study influential -- because it has something for everyone. It can support either of the two main prescriptions in contention. Supporters of aggressive fiscal or monetary stimulus can say: “With inflation so well anchored, we can push more decisively for higher employment. Central banks should be bolder.” Those who take the opposite view can say: “The anchoring of expectations has helped us to avoid deflation and an even worse recession. Don’t put that in jeopardy with aggressive stimulus that works the anchor loose. Central banks should be more cautious.”
This argument would be moot if we knew how much of today’s unemployment is cyclical as opposed to structural. But, again, we don’t. If it’s cyclical, and there’s spare capacity, higher demand would increase employment without creating inflation. If it’s structural, perhaps because of a skills mismatch or other impediment, higher demand would be inflationary. In the U.S., the Federal Reserve seems convinced that cyclical unemployment is still high, leaving room for continued stimulus. But the evidence isn’t conclusive, and the new IMF findings don’t shed much light on this one way or the other. In short, choosing a policy still involves a lot of ignorance and a complex balance of risks.
Empirical work can push macroeconomics forward inch by inch. It’s a noble endeavor, and I salute the practitioners -- especially when they code their spreadsheets properly. But let’s not delude ourselves that policy in 2013 is founded on it, or even should be founded on it. Ideas, good and bad, still prevail over facts. John Maynard Keynes famously said he hoped that economists would one day be as humble and competent as dentists. Brilliant as many of them undoubtedly are, they haven’t yet advanced to that point.
(Clive Crook is a Bloomberg View columnist. The opinions expressed are his own.)
To contact the writer of this article: Clive Crook at email@example.com.
To contact the editor responsible for this article: Max Berley at firstname.lastname@example.org.