With the wolves at the door of the Fed (to be as dramatic about it as possible), it’s a good time to think about the evidence on the effects of central bank independence. The textbook answer seems to be that there is sound econometric evidence that central bank independence causes price stability. Reading some recent papers that criticize the currently accepted evidence, I’m not sure this evidence is actually very persuasive. But more importantly, I don’t think it matters. In fact, I suspect that the economic profession’s apparent consensus around the econometric results is more based on their theoretical plausibility than on the actual soundness of the methodology. Anyone who finds the existing econometric results convincing should ask themselves this: if the question were “what are the effects of having a Wal-Mart?”, would you accept the same caliber of statistical evidence as meaningful?
There are several criticisms to the existng empirical evidence, including the fact that the most often cited paper that also contained a popular index was never actually published, but I will only list two important methodological challenges here;
1) There is a serious endogeneity problem because countries that get central bank independence have more economic freedom, a political system capable of establishing laws that future citizens will want to appeal but cannot, and other positive characteristics such that the establishment of a central bank is certainly a highly non-random event.
2) There are statistical issues with the indexes used to measure central bank independence, including a much cited index co-created by Larry Summers. These rankings usually are a weighted or unweighted average of either “Yes/No” answers or “rank on a scale of 1 to X” types of questions relating to specific measures of independence. For example, “does the legislative branch have final authority on policy issues?”. These ordinal measures can be useful in creating typologies, critics argue, but they should not be treated as a continuous variable with a regression coeffient that can be interpreted as such.
Although I find the criticisms somewhat compelling, I’m not really going to get into the legitimacy of the methodological criticisms here; you don’t want to read it, and I don’t want to write it. The reason I raise them is simply to ask, if the critics are correct, and we have no serious econometric evidenence that central bank independence mattered, would it really matter? Would anyone believe that central bank independence was any less necessary to control inflation?
It seems to me that theoretical and common sense justifications for central bank independence are actually sufficient. I find the simple time inconsistency problem. i.e. that we would always want to commit to low inflation now but break that commitment later, pretty darn compelling. Also, the voting public would never simply “ride out” recessions, but would always insist on seeing that someone “in charge” is “doing something”. When voters insist that something must be done, getting rid of discretionary monetary tools just leaves us with more fiscal policy. And I prefer monetary policy*.
In a famous and influential paper on the issue Larry Summers and his coathor, Alberto Alesino, wrote that “the impact of central bank independence on economic performance is ultimately an empirical question.” I guess my question is, does it need to be?
*To avoid a beaten with a liquidity trap cudgel and a DeLong smackdown, I must add the disclaimer that I prefer monetary policy in most circumstances, not all.