Should central banks target nominal GDP?

Notes: Diagram shows the response of monetary policy to an unanticipated inflation shock of +1%.

The benefits of nominal GDP targeting are thought to be both theoretical (see here) and practical, with some pointing to the ease of communicating to the public a monetary policy rule that targets nominal spending (see here). I remain unconvinced by the arguments for nominal GDP targeting. Here are some reasons why.

(1) To the extent that future output is uncertain, nominal GDP targeting does not provide an anchor for inflation expectations. Nominal GDP targeting does provide an anchor for nominal spending. What it does not do, however, is provide a clear focal point for inflation or price level expectations. To see this, note that a nominal GDP target of N* will be met when P.Y = N*, where P is the price level, and Y is real GDP. Hence, the central bank should promise to set P = N*/Y, making the price level countercyclical. Since future output is uncertain, there is no focal point for expectations will gather given a promise to move the price level inversely with output. The same reasoning also applies to inflation, except that expected future inflation will be inversely related to expected output growth under (credible) NGDP targeting. All this matters because having a clear focal point for inflation or price expectations is crucial for keeping actual inflation low and stable. Indeed, there is good evidence from bond yields that inflation targeting has lowered inflation expectations and reduced inflation uncertainty (see here and here). This would be lost under nominal GDP targeting, putting the economy at risk of higher and more variable inflation.

(2) Nominal GDP targeting is not easy to communicate to the general public. Proponents of nominal GDP targeting have argued that it would be easy for central banks to communicate monetary policy in terms of a target for nominal spending. But as I argued in point (1), the problem comes when individuals attempt to forecast the two variables (price level and real GDP) that make up nominal GDP. For instance, the guidance nominal GDP targeting gives about future inflation is minimal, since there are an infinite number of price level and output combinations (or, equivalently, inflation rates and growth rates) which are consistent with any given nominal GDP target. Hence, any inflation rate is desirable, given wild enough swings in output. In such circumstances, central banks would presumably be forced either to deviate from the nominal GDP target (losing credibility) or to specify circumstances in which the target would not apply (big shocks). But then the target itself becomes state-contingent and its simplicity is lost.

(3) Credibility would be strained in the face of demand shocks. Scott Sumner and others have argued that one benefit of nominal GDP targeting is that it provides flexibility in response to supply shocks. It would not require, for example, that the central bank raise interest rates in response to stagflation: a rise in inflation would be permitted, temporarily, while output is weak. But large supply shocks are fairly infrequent. If we look instead at demand shocks, nominal GDP targeting looks less attractive. Suppose, for example, that the nominal GDP target is 100, but that nominal GDP overshoots to 105 due to the price level and output being higher than expected in the face of a positive demand shock. Nominal GDP is now at the level it should be next year, assuming an NGDP target path that rises by 5% per annum. As a result, the central bank now wants nominal GDP to flatline. It is therefore faced with three unattractive choices: keep both inflation and output growth at zero; combine positive inflation with negative output growth; combine positive output growth with deflation. The likely result is that the central bank would not follow through in these circumstances, and its credibility would be eroded. Enough such episodes could reduce credibility to the point where nominal GDP targeting would have to be abandoned.

Once we factor in negative demand shocks and the zero lower bound on nominal interest rates, things look even worse. For instance, suppose that nominal interest rates are near the zero lower bound and nominal GDP undershoots to 95, i.e. 5% below the target of 100, due to a series of negative demand shocks that lower the price level and real GDP. Now, given trend nominal GDP growth of 5%, the central bank would have to promise to raise nominal GDP by 10% next year in order to meet the new target of 105 (=100*1.05). Ten percent – through inflation or output growth, or some combination – when the economy is at the zero lower bound! What central bank can credibly promise that? That would require a massive amount of credibility, probably too much to be plausible in practice.

Of course, the same criticism could be made of price-level targeting. In particular, it will only help us escape rapidly from the zero lower bound if it can credibly promise sufficient inflation to return the price level to its target path, as noted by Eggertsson and Woodford . As Eggertsson and Woodford (p. 185, 199) point out, the key is to build sufficient credibility in normal times that the central bank’s promises will be believed even in extraordinary times (i.e. at the zero lower bound). A problem with nominal GDP targeting is that it would struggle to build credibility in real-world economies that are frequently buffeted by demand shocks. Indeed, as the above examples demonstrate, credibility could be severely strained even by fairly moderate shocks to aggregate demand.

All this is not to say that we cannot improve upon inflation targeting or that researching nominal GDP targeting is without merit. But for the moment at least, I remain unconvinced about nominal GDP targeting and think price-level targeting is a better alternative.

-------------------------------------------------------------------------------------

You can read Scott Sumner's reply to this post here.

Reply by The Incorrigible Market Monetarist.