By Mark Thoma
Robert Hall’s speech at the Kansas City Fed’s Symposium in Jackson Hole concerns an important issue in the conduct of monetary policy (links to conference papers). The issue is the measurement and use of output and inflation gaps as a guide to monetary policy (the gaps are the deviations of inflation and output from their target values; I am not going to address whether the appropriate measure is the current or expected future gaps, for this discussion such distinctions are unnecessary; also, ECB president Trichet issues cautions on use of the gap here).
Hall’s paper been covered elsewhere, macroblog is a good place to begin, but the response of the discussants to his paper has not been covered as extensively. One of the discussants is Charles Bean, Chief Economist of the Bank of England. Hall's paper concludes that the output gap is not a relevant or useful measure of economic activity, and strictly interpreted there is no gap at all. Bean disagrees and explains why. Here are Bean's remarks:
Speech by Charlie Bean, Chief Economist, Bank Of England, Comment on Bob Hall’s “Separating the Business Cycle from Other Economic Fluctuations,” Jackson Hole Symposium, Wyoming, Friday 26 August 2005: It is always a pleasure to read a Bob Hall paper ...The kernel of Bob’s argument is that the movements in the main US macroeconomic aggregates, both at cyclical frequencies and longer, are the natural consequence of a well functioning economy responding to shocks to productivity and real spending. Far from following a relatively smooth trend, potential output is stochastic. Indeed in the canonical model, there is no Keynesian-style output gap at all. The claim that business cycle fluctuations are primarily an equilibrium response to real shocks, particularly to productivity, was originally advanced more than twenty years ago by Finn Kydland and Ed Prescott (1982).
Bob provides similar evidence in his paper today. But how convincing is that evidence? Not very, I would argue. …other approaches, such as the structural vector-autoregressions of Blanchard and Quah (1989) and their many imitators, suggest that the contribution of technology to output fluctuations at business cycle frequencies is rather more modest.
Partly because of that evidence, much recent business-cycle research has been directed to incorporating Keynesian-style sticky prices and wages into the workhorse neoclassical growth model. Bob is somewhat dismissive of the importance of sticky prices … One of the least satisfactory features of the early real business cycle models lay in the absence of a compelling explanation for fluctuations in employment and unemployment. ...Various ingenious solutions have been suggested... But none of these solutions are very convincing. Bob …provides an explanation why fluctuations in output may be associated with substantial fluctuations in unemployment, but ... Bob is implicitly of the view that … shocks are transmitted very largely into prices even in the short run. The problem with this conclusion is that it runs against the very large empirical literature which concludes that nominal disturbances in the shape of monetary policy shocks have substantial real effects at horizons out to a year or more (see, for instance, the survey by Christiano, Eichebaum and Evans, 2000). ... So, to me, the evidence still suggests that nominal rigidities, whether in prices or wages or both, are a part of the story of the business cycle....What does this all imply for the conduct of monetary policy? Relatively little is said in the paper about monetary policy, but what there is ...implies that [at] equilibrium …the output gap presumably should always be identically zero. As already indicated, I do not buy the argument that nominal rigidities are irrelevant, in which case … one could in principle define an output gap. But even then, is it a useful construct?
...Taylor’s rule was originally meant as a rough reduced form empirical description, not a prescription for how policy should be set. …we form a view of the outlook for inflation and output over the medium term together with the attendant risks, and then make a judgement on the appropriate policy stance in the light of that outlook. An assessment of the slack in the economy is an essential ingredient in forming that view, but it is not made merely by fitting a smooth trend through GDP. Instead our staffs employ models of varying complexity in which potential output and the natural rate of unemployment evolve stochastically in the way modern macroeconomic thinking suggests and also rely heavily on other sources of information, including a whole range of official data and business surveys. And we are acutely aware of the uncertainties, treating statistical estimates of the output gap with a very considerable degree of scepticism.
To conclude, Bob is certainly right to highlight the fact that potential output is not a smooth trend and that the natural rate of unemployment and the neutral rate of interest are not constants. But I do not believe we can dispense with the concepts altogether…
References
- Ball, Laurence and David Romer, “Real Rigidities and the Non-Neutrality of Money”, Review of Economic Studies, Vol. 57, No. 2. (Apr., 1990), pp. 183-203.
- Blanchard, Olivier J. and Danny Quah, “The Dynamic Effects of Aggregate Demand and Supply Disturbances”, The American Economic Review, Vol. 79, No. 4. (Sep., 1989), pp. 655-673.
- Christiano, Larry J., Martin Eichenbaum and Charles Evans, “The Effects of Monetary Policy Shocks. What Have We Learned and to What End?”, in John Taylor and Michael Woodford, Handbook of Macroeconomics, (2000).
- Kydland, Finn E. and Edward C. Prescott, “Time to Build and Aggregate Fluctuations”, Econometrica, Vol. 50, No. 6 (Nov., 1982), pp. 1345-1370.
- Mortensen, Dale and Christopher A. Pissarides, “Job Creation and Job Destruction in the Theory of Unemployment”, Review of Economic Studies, Vol. 61(3), (July 1994), pp.397-415.
My sentiments are with Charles Bean's point of view. To me, there are two questions. First, can the gap be measured? The answer to this question is that we are not able to separate GDP and other series into trend and cycle components without making assumptions, and the specific assumptions needed for the decomposition are not provided by theory. Thus, given current technology, there is always some degree of uncertainty in our gap measure even if it is a theoretically sound principle.
The second question is whether we should measure the gap, i.e. if theory supports employing the gap in monetary policy decisions. It depends upon the theory used as noted in Bean's remarks. Using an RBC model as a guide, the answer is no. Using a New Keynesian model, the answer is yes. However, given that model uncertainty exists, i.e. given that central banks are not certain which theoretical structure applies, robustness and risk management dictate that policymakers calculate outcomes under various versions of both RBC and New Keynesian models (and perhaps others too). To complete such an exercise, a measure of the gap is needed and would be calculated under the assumptions of the New Keynesian model.
"Instead our staffs employ models of varying complexity in which potential output and the natural rate of unemployment evolve stochastically in the way modern macroeconomic thinking suggests and also rely heavily on other sources of information, including a whole range of official data and business surveys."
Well quite, and here I entirely agree with Bean. You don't simply follow a mechanical rule, you have a committee, you consult a range of estimates and data sources, people try to give intuitive weights, and then you vote. This is why Greenspan at his best is irreplaceable, and why, at the end of the day, there is much more of art than science involved in good central banking.
"Instead our staffs employ models of varying complexity.."
This is the alchemy bit, maybe they chant a strange mantra while doing it :).
"Bob is certainly right to highlight the fact that potential output is not a smooth trend and that the natural rate of unemployment and the neutral rate of interest are not constants"
Doesn't Hall imply that more than being not constants, they are fictions, convenient (Bean) or inconvenient (Hall) ones?
Posted by: Edward Hugh | Thursday, September 01, 2005 at 12:05 PM