RDP 9809: Estimating Output Gaps 1. Introduction

The output gap, defined as actual less potential output, is an important variable in economic and policy systems. It is important in its own right, since there are fewer jobs and lower profits than would otherwise be the case when actual output is below potential. Moreover, the output gap has a well-documented role in the theoretical and applied literature in explaining price and wage inflation. Since potential output is not directly observable, however, neither is the output gap. This paper reviews various methods of estimating output gaps with the aim of showing some of the pitfalls and complexities of the different techniques, rather than providing a definitive measure of the output gap. Indeed, the analysis in this paper underscores that estimating output gaps is fraught with uncertainty and requires considerable judgment.

Section 2 traces through some of the techniques used to estimate output gaps. It starts by setting out some definitions and reviewing the output data. The methods used to estimate the output gap include linear time trends, Hodrick-Prescott (HP) filter trends, multivariate HP filter trends, unobservable components models and a production function model. Some of these methods are univariate techniques, using particular assumptions about the time-series properties of output to identify potential output. Other methods combine these techniques with economic information about the output gap. Estimates of the output gap are shown to be sensitive to the general model used, the particular specification adopted for the model being used, and the sample period.

Section 3 assesses these output gap measures. Output gaps are estimated for two purposes. The first is to provide information about excess capacity in the economy at a particular point in time. From the perspective of monetary policy, the output gap over the forecast horizon is of most interest. The second purpose is to use a time series of the output gap in modelling exercises. For example, given that excess demand pressures are a key cause of rising inflation, the output gap can be included in price or wage inflation equations to obtain a more precisely estimated equation and more accurate forecasts.

The analysis in Section 2 shows that, on the one hand, estimates of the output gap are imprecise and can vary considerably across estimation methods at particular points in time. On the other hand, however, there is considerable similarity in the broad time profile of the various gap estimates, suggesting that the gap relative to its past contains useful information. While it is difficult to identify the absolute size of the output gap, it is possible to infer its relative size. This underscores the need for careful analysis and practical and considered judgment. Moreover, the similarity of time profiles suggests that most gap measures may have similar explanatory power in econometric models. This is shown by comparing the estimates in a standard mark-up model of inflation for various gap measures. Including any of the gap measures canvassed in Section 2 in an inflation equation improves the explanatory power of the equation and substantially reduces prediction error. Output gap measures also tend to explain inflation in estimating equations better than output growth rates do.