RDP 9809: Estimating Output Gaps 4. Conclusion

Policy-makers are interested in the output gap for two reasons. First, excess capacity indicates that the economy could produce more and could generate more jobs. Second, excess capacity is an important determinant of price and wage outcomes, and so contains key information for monetary policy. This paper reviewed five techniques used to estimate the output gap – linear time trends, Hodrick-Prescott (HP) filter trends, multivariate HP filter trends, unobservable components models and a production function model. It shows that the estimates of the output gap depend on the estimation method selected, assumptions made in estimation, and the sample period, making it difficult to determine the absolute size of the output gap at a particular point in time. This difficulty is made worse when the output gap is estimated over periods which include a disinflation (as occurred in Australia in the early 1990s). Over such periods, the output gap should be negative on average, but univariate statistical techniques force the gap to average zero. Output gap methods which rely on this assumption yield positively biased estimates of the current output gap. This is not necessarily the case with more complex techniques, since they allow estimates of the gap to be conditioned on this information. But the way the conditioning relationships which underlay these models are structured is of critical importance to the estimate of the output gap, which underscores the need for careful analysis and judgment.

The profiles of the output gap produced by these different techniques, however, are broadly similar. This suggests that the relative size of the gap can be determined by comparing the gap at the date of interest with the past profile of the gap. It also indicates that the gap measure can be useful for model-based analysis, since the differences between gap measures will be captured by adjustment to the constant and slope coefficients. For example, using a simple mark-up model of inflation, all of the output gap measures examined improve the fit of the inflation equation and substantially reduce ex post prediction error. The output gap performs this task better than output growth, indicating that the degree of excess capacity in the economy is an important determinant of inflation.