Tuesday, July 29, 2008

Core Inflation and Inflation Targeting

As I mentioned recently, inflation doesn't look nearly as scary if one focuses - as the Fed does - on "core" inflation, which excludes food and energy prices. Some Fed critics - e.g., Willem Buiter - seem to see core inflation as an excuse for going soft. Here is a counter-argument from NYU's Mark Gertler - a co-author of Bernanke's back in his academic days - defending the Fed's focus on "core" inflation in the FT:
Why care about headline inflation versus core inflation? Simply put, a sustained move of headline inflation to the levels of the 1970s is unlikely without an accompanying increase in the core component. The reason is simple: although they can be highly persistent, rapid increases in the relative prices of energy and food cannot go on indefinitely. Once this process dies down, as long as core inflation remains anchored, headline inflation must converge to it.
One criticism of using a core measure is that higher overall (or "headline") inflation feeds into inflationary expectations, which, in turn, influence wage and price setting behavior. Gertler addressed this issue:
Could it be that high headline inflation is unmooring inflation expectations, leading us back to the 1970s through this painful route? Some measures of inflation expectations are edging upwards. This needs to be taken seriously. However, where we should expect the impact of increasing expectations to show up is exactly in the behaviour of core prices and wages.

So far this is not happening. Not only has core inflation remained stable but the growth in nominal unit labour costs, on which most pricing of core items is based, also remains benign. It may very well be that the Fed’s reputation for keeping core inflation stable has kept the expectations relevant for price- and wage-setting in line. Also relevant is that, at least to date, wage- setters appear to understand that, however unfortunate, the relative increase in energy and food prices is something beyond the central bank’s control that they must live with.

Fed governor Frederic Mishkin (also a Bernanke co-author), who is leaving for Columbia University, seems a little more concerned about the expectations issue. In his last speech as a Fed official, Mishkin argued that explicit inflation targets - such as that of the European Central Bank - help stabilize, or "anchor" expectations:

[T]here is substantially greater disagreement in long-run inflation forecasts for the United States than for the euro area. As shown in figure 2, the standard deviation of U.S. inflation forecasts at each survey date is higher than the standard deviation of corresponding euro area inflation forecasts. Moreover, the degree of dispersion in the views of individual forecasters has gradually declined towards negligible levels for the euro area but not for the United States. One obvious interpretation of these patterns is that professional forecasters in the United States are less certain about the Federal Reserve's longer-term inflation goal.

That uncertainty may also explain differences in the behavior of inflation compensation as implied by the gap between nominal and real yields on long-term bonds. Figure 3 depicts far-forward inflation compensation (that is, the one-year-forward rate nine years ahead) for the United States and the euro area. Inflation compensation, sometimes referred to as "breakeven inflation," reflects not only inflation expectations but also a premium that compensates for uncertainty about inflation outcomes at the specified horizon. Evidently, far-forward inflation compensation for the euro area displays much smaller fluctuations than for the United States, consistent with greater stability of inflation expectations and a lower degree of uncertainty about longer-run inflation outcomes. Moreover, regression analysis confirms that U.S. far-ahead forward inflation compensation exhibits statistically significant responses to surprises in macroeconomic data releases--consistent with the view that market participants are continuously revising their views about the longer-run outlook for U.S. inflation. In contrast, euro-area inflation compensation does not respond significantly to economic news.

Mishkin goes on to suggest that inflation targeting does not result in worse performance in terms of stabilizing output:

One concern might be that these benefits in anchoring inflation expectations could come at the expense of the performance of output growth. However, the empirical evidence suggests that central banks with explicit inflation goals do not have worse output performances than central banks, such as the Federal Reserve, that have not specified an explicit numerical goal for inflation.
So far, this is true, but inflation targeting is relatively new, so I think it is too soon to tell (perhaps a good research topic a few years from now...).

Mishkin did not address the issue of which inflation measure to use, though all of the current inflation targeters use headline inflation. His strategy for implementation is for the Fed to incorporate a target into its long-run forecast:
In light of these considerations, I would like to suggest several specific modifications to the Federal Reserve's current communication strategy.
  • First, the horizon for the projections on output growth, unemployment, and inflation should be lengthened. This change might involve simply an announcement of FOMC participants' assessment of where inflation, output growth, and unemployment would converge under appropriate monetary policy in the long run. Alternatively, the horizon for the projections could be extended out further, say to five or more years.
  • Second, FOMC participants should work toward reaching a consensus on the specific numerical value of the mandate-consistent inflation rate, and this consensus value should be reflected in their longer-run projections for inflation.
  • Third, the FOMC should emphasize its intention that this consensus value of the mandate-consistent inflation rate would only be modified for sound economic reasons, such as substantial improvements in the measurement of inflation or marked changes in the structure of the economy.
If the deviations between core and non-core inflation induced by oil price fluctuations, etc., are indeed transitory (i.e. they die out in a shorter time period than the forecast horizon) there is no tension between tolerating headline inflation spikes in the short run and stabilizing it in the long run. However, a commitment to "long run" inflation targeting which allows substantial "short run" deviations seems rather squishy, and I wonder if it really would earn the Fed the additional credibility that true inflation targeters have.


Anonymous said...

Or inflation could be uneven. The Fed makes the very odd assumption that the manifestations of monetary expansionism must happen to certain goods in a certain order for it to be inflation.

Secondly, core inflation mostly includes goods that were industrialized very early and are constantly getting cheaper as a function of technology. It is very likely these sectors would naturally deflate. The computer market, for instance, is notoriously deflationary with products.

Bill C said...

That ties in to the point that Buiter has made that the non-core goods have more flexible prices, so the inflation shows up there first, but in his opinion, it is still a monetary phenomenon.

It seems the argument for the core inflation focus relies on the notion that the non-core inflation shocks are transitory, but it seems that we are in an environment where more rapid energy price increases will be persistent.

On computers, etc., some would argue that the Fed was excessively "loose" in response to the (deflationary) positive technology shocks in the 1990's.

However, overall, services are a larger part of GDP (and therefore the deflator) than goods, so I don't think that is a general problem.