Weaknesses of the output gap in forecasting inflation
Peter Warburton - 2 July 2020
If inflation forecasting were easy, almost everyone would succeed in this endeavour. The record shows otherwise. Simple models of the inflationary process are essential teaching tools but have serious deficiencies when deployed as forecasting tools. Neither money supply growth (however defined) nor the Keynesian output gap is a reliable guide to the inflationary outlook amidst monetary regime change. The output gap framework would have failed miserably in its predictions of inflation in the 1930s and the 1970s. When seismic societal and policy changes are underway, there is no substitute for a detailed consideration of all the forces at work.
Over the past decade, Economic Perspectives has published more than 30 full length Global Inflation Perspectives and countless slide presentations. My colleague, Tom Traill, compiles our Global Inflation Heat Maps every month and Global GDP heat maps every quarter, containing inflation measured by the GDP deflator. Our thesis is that inflation is a complex, multi-faceted phenomenon, that encompasses (among other things) geopolitics, political economy, inter-generational dynamics, sociology, and macro-financial economics. During periods of institutional and political stability, it is possible to model the inflationary process with reasonable accuracy. However, these intervals of stability are punctuated by radical upheaval, making inflation forecasting especially difficult. This is such a time.
The notion of an output gap is a helpful heuristic device in a teaching context. However, its usefulness as a real-time forecasting tool was demolished more than 20 years ago by Athanasios Orphanides and Simon van Norden. They demonstrated that “subsequent revisions of the output gap are of the same order of magnitude as the output gap itself, that ex-post revisions are highly persistent and that real-time estimates tend to be severely biased around business cycle turning points, when the cost of policy-induced errors due to incorrect measurement is at its greatest.” The output gap rests on monumental assumptions and is subject to serious measurement problems.
It is highly inadvisable to take output gap proxies, such as that shown in figure 1, as providing a definitive guide to the inflation outlook. If estimates of the output gap had been available in the early 1930s, they would have been just as cavernous (see figure 2) as today. Nicholas Crafts and Peter Fearon estimate that there was still a US output gap of 25 per cent in 1933, assuming trend growth at the pre-1929 rate. Yet inflation rebounded to 5 per cent in 1934. We currently have no idea what is the sustainable trend rate of US GDP growth nor the steady-state rate of US unemployment. What we do know is that extraordinary measures have been taken to shore up nominal demand, that central banks have committed to absorb vast quantities of government debt, and that further efforts are being made to forestall bankruptcies and defaults. The big unknown is the damage to the capability of private sector businesses to produce and distribute goods and services, conditional on an adequate rate of profitability. Supply-side dynamics will play a key role in the evolution of inflation.
We are holding an inflationary webinar (Zoom call) at 3pm on Tuesday 7 July entitled “The Wolf of Main Street”. If you would be interested to join this call, please send an email to email@example.com
Figure 1: US industrial capacity utilisation minus the unemployment rate (per cent)
Data source: Eikon Reuters
Source: The Econtrarian, 1 June 2009
Share on LinkedIn: