from Energy, Security, and Climate and Energy Security and Climate Change Program

Can Monetary Policy Mitigate The Effects of Oil Shocks? A Follow-Up

October 17, 2011

Blog Post

More on:

Monetary Policy

Fossil Fuels


I wrote a week ago about an old paper by Bernanke, Gertler, and Watson (BGW) that attributed much of the impact of past oil price shocks to the response of monetary policy. Jim Hamilton followed up this weekend:

Michael Levi (hat tip:Marginal Revolution) and Jeremy Kahn are among those who recently rediscovered some earlier research by Ben Bernanke and others that concluded that the economic downturns that followed historical oil price shocks could have been avoided if the Fed had followed a more expansionary monetary policy at the time. Here I call attention to some subsequent research that took another look at their evidence and reached a different conclusion.”

Hamilton goes on to describe a followup paper that he published with Ana Herrera in 2004. They criticize BGW on two fronts. First, they argue that the monetary policy alternative proposed by BGW isn’t credible, since it’s a huge departure from the historical pattern. I thought that that was precisely the issue I’d discussed in my original post. Indeed, as I argued, one can put this matter to an empirical test. During the 2008 oil price spike, headline inflation rose to over 5 percent, but core inflation maxed out at about 2.5 percent. Had this been 1973, policymakers would have responded to headline inflation and jacked up rates, in turn slowing the economy; as it happened, they focused on core inflation, and didn’t. As I noted in my original post, however, there are limits to this dynamic, since energy price rises eventually pass through to core inflation, forcing a response; thus, I suggested, good monetary policy has limits in what it can accomplish. (As a side note, Hamilton claims that, in 1973, policymakers would have had to hold constant a federal funds rate that actually rose by 900 basis points in order to implement the BGW policy. That’s not true: only part of that 900 point rise was a response to the oil price shock; it’s only that part that would have had to be curtailed to match the BGW policy.)

The trickier critique offered by Hamilton and Herrera (HH) focuses on the fact that BGW estimate the relationship between oil prices and macroeconomic variables using a model that includes seven monthly lags. What this basically means is that if oil price shocks have big consequences for the economy more than seven months after they happen, BGW won’t notice that. Indeed HH point out that most studies of the oil-economy relationship show that the biggest impacts come after twelve months. HH redo the BGW analysis, but with a twelve month horizon rather than a seven month one. They find a bigger impact for oil price shocks – and a much smaller role for monetary policy.

What Hamilton doesn’t mention in his post is that Bernanke, Gertler, and Watson responded to the HH critiques. To address the question of whether their counterfactual monetary policy is credible, they simulate another counterfactual where rates are held flat for a year before responding to inflation. In response to the criticism that using seven monthly lags misses much of the picture, they make two points. First, they argue, shifting to twelve monthly lags has its own problems, namely that it introduces 245 new parameters to the model, increasing uncertainty. Thus, they note, “it is possible that the weaker effects of systematic monetary policy found by HH in models with more lags are the result of additional sampling uncertainty rather than a structural feature”. Instead, they offer a compromise, simulating a model that uses four quarterly lags; this allows them to looks at effects that take as much as a year to materialize without introducing too much uncertainty (since it keeps the number of parameters down). They report the results for a model with four quarterly lags and only temporary monetary policy restraint. The result is that monetary policy still explains about half of the GDP impact – less than in their original paper, but more than in HH. The same result holds for a model with six quarterly lags.

I should probably close by noting that I don’t want to suggest that this is the final word, just like I didn’t mean to suggest in my last post that BGW was the last word either. (The point of that post was to use some comments from a new Nobel laureate to highlight how hard it is to figure out whether and how oil prices affect the economy.) There are plenty of other authors who disagree with both of the papers I’ve discussed here. This only reinforces the point I made in the previous post: policymakers are inevitably going to need to operate in the presence of limited knowledge. That, I suspect, is something that no academic paper will change.