Bernanke and Woodford on Targeting the Forecast

Ben Bernanke and Michael Woodford wrote “Inflation Forecasts and Monetary Policy” in 1997. The thrust of their paper is a critique of having the central bank “target the forecast.” Most market monetarists have adopted Scott Sumner's insistence that the Fed target the forecast. Sumner credits this view to Svennson.

Bernanke and Woodford describe a scenario where the central bank targets the consensus of private forecasters, which comes right out of Hall and Mankiw’s 1994 paper, “Nominal Income Targeting.” However, they cite Dowd (1994) and Sumner (1995) which propose pegging the price of a CPI futures contract. Interestingly, Bernanke and Woodford have little complaint with Svensson’s (1997) proposal.

They see his proposal as:

“the central bank’s internal forecast is prepared with the use of a structural model, but that the model and data on the current state of the economy are used to determine the policy action, that according to the model, should result in a forecast equal to the target.”

And what of private sector forecasts? They also write:

“a central bank can choose a policy that has the effect of keeping private inflation forecasts equal to the targeting without having the form of a “forecast targeting” rule, and it is desirable that it do so.”

Incredibly, they even believe that the central bank can use the private sector’s forecast of inflation shocks and potential output shocks to plug into their structural model, and so keep the private sector’s expectation of inflation on target (pp. 681-2.) They describe this in the context where the private sector is mistaken so that the central bank is sacrificing that element of inflation stabilization that it could provide by ignoring the private sector’s forecast and just using its own better estimate of inflation and potential output shocks.

They go on to explain (p.682):

“A version of the proposal might reduce the extent to which the central bank is required to stabilize incorrect private forecasts rather than inflation would be to simply require the central bank to give public testimony about the motivation of its policy stance, that might well include discussion of its own inflation forecast. Private sector forecasts that disagree with that of the central bank might well be matters that would require comment on the part of the central bank, but one could accept an explanation on the part of the central bank of how its own forecasts are made as sufficient demonstration of a good faith effort to achieve the inflation target.”

The Federal Reserve doesn’t exactly target inflation, but rather inflation and unemployment. After its most recent meeting, the Fed gave its internal forecasts of both, and explained that both would remain below target for an extended period of time. Perhaps Bernanke should review what he wrote four years ago.

Like most market monetarists, I reject inflation and unemployment targeting and instead favor targeting the growth path of nominal GDP. Replacing inflation with the target for nominal GDP, what Bernanke and Woodford wrote would be appropriate. The Fed should set policy instruments so that its internal forecast is on target. However, it should also be just a little bit less arrogant than Bernanke and Woodford imagine, and be willing to make some adjustments based upon what the “wrong” private forecasts suggest.

Is it really true that if the private forecasters plug in their values for inflation and potential output shocks into a “true” model of inflation and report that, the result is that the eigen value is within the unit circle and the inflation rate is indeterminate? (p.671)? I don’t think so.

The fundamental problem with Bernanke and Woodford’s approach is that they assume that all forecasters have one true model and true information. If so, the central bank should just pay one of the forecasters for this information. Or why not hire one of them in place of their current staff?

The reality is that no one, not the central bank’s internal forecasters or any of the private forecasters knows the true model or has all of the needed information. They disagree. The goal should be to somehow combine both the central bank’s own forecast with those of the private forecasters to generate a market expectation that can be utilized to adjust current market conditions such that the market expectation is that nominal GDP will be on a slow, steady growth path.