[Dear new readers: Please click here if you'd like to see the rest of this blog, which focuses on energy, security, and climate.]
Do you think that it’s straightforward to figure out whether high oil prices cause recessions? Many people apparently do. The 2011 Nobel Prize in Economics, awarded today to Thomas Sargent and Christopher Sims for “empirical research on cause and effect in the macroeconomy”, should make them reconsider. The basic reason is simple: if it was easy to separate cause from effect and thus measure the relationship between stimulus and response, they wouldn’t be awarding Nobel Prizes for related progress.
Indeed the difficulty of distinguishing economic cause from effect is a big reason for economists’ longstanding interest in oil price shocks, particularly those of the 1970s. Those price hikes were clearly spurred by events outside the economic system, in 1973 by the Arab oil embargo, and in 1979 by the Iranian revolution (though even on these points there is some dissent). Identification should thus be simple: the oil shock is the cause, and any macroeconomic change is a consequence. This provides an unusually clean laboratory in which to study economics.
Alas, there is a problem. The oil price shocks of the 1970s prompted big interest rate hikes in consuming countries, as policymakers tried to stem inflation. One now must ask: were the economic slowdowns that followed the oil price shocks the result of the shocks themselves, or consequences of the monetary policy reaction? The difference matters, because one leads to an energy policy solution, while the other points to better monetary policy as the right response.
In a seminal 1997 paper, Bernanke, Gertler, and Watson (BGW) came down strongly on the latter side. They drew on methodology developed by Sims (one of the new Nobel laureates) and Tao Zha to fit a model that incorporates both the direct response of the economy to oil price rises and a monetary policy component. Then they simulated the model with one change: instead of having interest rates rise in immediate response to increased inflation, those rates remained constant. The result? Most of the economic slowdown triggered by the oil price rises disappeared. The conclusion is clear: it is the monetary policy response, not the oil price shocks themselves, which slowed growth and triggered recessions. This has clear policy implications: it may be more effective to fight oil price shocks with good monetary policy than with energy policy itself.
I hadn’t noticed before this morning that the original BGW paper was accompanied by an illuminating critique by one Christopher Sims. Sims offers two related criticisms. The first is that the alternative policy modeled by Bernanke et al is implausible. Superficially, the alternative policy (where interest rates remain unchanged) permits high inflation, which Sims sees as unsustainable. More fundamentally, Sims has trouble seeing a way to get from the pre-shock monetary policy to the post-shock one. He argues in essence that if your starting point is a monetary policy that responds strongly to oil price shocks (or, more generally, inflationary pressures), it’s basically impossible to credibly switch to one that ignores them. If your new policy isn’t credible, though, the response to it won’t be the one that’s predicted by the model. Sims is basically arguing that, at best, policymakers will eventually be forced to return to their old (tighter) monetary policy rule, that markets will anticipate that, and that the benefits of looser money in the near-term will thus be blunted.
Sims’s second criticism, based on the so-called Lucas critique, is closely related. He essentially argues that a fundamentally new policy rule, even if it is successfully implemented, might not have the modeled effects. That’s because some of the parameters identified in specifying the original model might implicitly reflect the old policy rule. If that’s true, and a policy change thus effects deeper changes in the economy, the model projections of the new policy can’t be trusted; in turn, that makes it impossible to say that it was monetary policy, not high oil prices, that did the real economic damage.
How have these two competing views stood the test of time? One can argue, at least superficially, that U.S. policymakers have succeeded in credibly and sustainably changing the policy rule. Back in the 1970s, monetary policy responded to headline inflation; today, it responds only to core inflation, which excludes energy and food prices. To be sure, this only gets you part way to the hypothetical strategy described by BGW and dismissed as not credible by Sims, since changes in oil prices ultimately pass through in part to core inflation, to which the Fed still responds. Moreover, European policymakers have been far less swayed by the case for ignoring headline inflation. Still, the historical track record suggests that on this particular matter, Bernanke et al were, at least in principle, on largely defensible ground.
The other critique offered by Sims is tougher to rebut, since it cuts at the heart of the modeling enterprise. Back in the 1970s, market participants pretty much knew that rising oil prices would prompt monetary tightening. That presumably affected their behavior more broadly. Today, they can be similarly (though not entirely) confident that the Fed won’t react to increasing oil prices by cranking up interest rates; once again, that might change their broader behavior. Extracting lessons for today’s economy from patterns displayed in the 1970s may thus be a fraught exercise.
And that, to return to where we started, is the real bottom line. Anyone who claims to have looked at the historical data and established a strong, immutable relationship between oil prices and macroeconomics is being overconfident. Cause and effect are tough to disentangle in the macroeconomy. Changes in monetary policy since the 1970s make extracting lessons from the first two big oil shocks exceedingly difficult. Policymakers are stuck with the inevitable uncertainty that remains.