Output gaps in practice: Proceed with caution
The ongoing Covid-19 crisis has further highlighted the importance of estimating economies’ growth potential in order to deploy appropriate countercyclical macroeconomic policy in response to temporary shocks. Challenges in measuring the relative falls in demand and supply, as discussed in Brinca et al (2020), coupled with the unusual nature of the pandemic-induced downturn, further complicate estimates of slack which were difficult enough prior to the current crisis. The unknown extent to which the downturn entails temporary versus permanent damage to activity also muddles the task of deciding the appropriate amount of policy support by authorities.
In recent research, we examine this issue by analysing the most common concept for slack, the output gap, which measures an economy’s current growth level relative to that of its potential. The concept is also used in IMF surveillance work (Barkema et al. 2020). While this measure for slack has periodically come under scrutiny, as discussed in Buti et al. (2019), and alternative views have been proposed (e.g. Dupraz et al. forthcoming), it remains a widely used tool in the macroeconomic profession. A relatively accurate measurement of the output gap is thus crucial, as the amount of fiscal and monetary support required to get growth back to full capacity will depend on the estimated amount of underutilized resources in the economy. And even where there is relative certainty on the measurement, alternative means of understanding slack are also crucial to ensure the robustness of the diagnosis.
In our first set of results, we study the historical record of output gaps and find a skewed picture. Using a sample of 197 countries over the period 1995-2018 we find a median real-time output gap of -0.7% and an average of -1%. That is to say, countries in the sample are on average assessed to be operating below their potential growth over this 23-year period. This negative skew is similar to previous studies that look at specific regions (e.g. Kangur et al. 2019 for European countries) and suggests that activity may not move in symmetric cycles above and below a country’s capacity.
The negative real-time output gap, which refers to the gap at time t as estimated at time t, is revised over time, however. In fact, the median of the final output gap declines to -0.1% over time. These revisions could potentially be explained by two factors: revisions in actual output as more data becomes available or, alternatively, revisions to the estimated potential growth of an economy. In practice, we find that the bulk of revisions – roughly two-thirds – is due to the latter. It therefore seems as if the view on the stance of the business cycle changes significantly over time. This suggests that real time estimates of the optimal position of fiscal and monetary policy that are based solely on output gaps need to be assessed with caution.
Further, it is worth considering other indicators of slack that move together with the estimated output gap. Some usual suspects are inflation and unemployment. It is reasonable to expect a positive association between output gap and inflation, as well as a negative one between the output gap and unemployment. A mix of simple correlations, as well as pooled and panel regressions do indeed confirm this. While the coefficients are only partial, the variables are all significant. Furthermore, the co-movement is somewhat improved by moving from real-time to final estimates of the output gap.
How these output gap estimates are interpreted, and the context within which they are placed, is also of crucial importance. Using text analysis techniques as well as manual text analysis, we examine IMF Article IV Surveillance staff reports which are generally produced at an annual basis for the Fund’s member countries. These reports entail a comprehensive review of member countries, including an assessment of each country’s economy and its policies, as well as recommendations. Our sample includes 2536 reports for 195 countries over the period 2000-2019. We first collect basic statistics on the amount of coverage of the output gap and related concepts. As Figure 1 shows, the discussions on the output gap has increased over time, with a notable spike after the Global Crisis. We also see a spike in the use of the term “above potential” towards the end of the sample period, pointing to positive output gap estimates for many countries in the immediate pre-Covid environment (see Coibion et al. 2017 for a contrasting view).
Figure 1 Output gap mentions over time (ratio of staff reports mentioning the relevant term at least once)
Looking at coverage by income group, we find that the frequency of output gap discussions is positively correlated with a country’s income level: 66% of staff reports covering advanced economies mentioned the output gap, versus 29% for emerging markets, and only 5% for low-income countries. In the latter, structural issues are often of greater relevance. Similarly, staff reports for European countries use the term most often and African staff reports use it the least.
Following these basic results, we juxtapose the discussion of the output gap in a given country with estimates of the actual size of the output gap but find no conclusive relationship. That is to say, somewhat surprisingly, output gaps do not seem to be discussed more in countries where the absolute gap is estimated to be relatively larger.
In order to enhance the analysis with other variables, we construct a simple probit model that includes a number of macroeconomic variables, sectoral risk measures, and IMF policy advice (whether to loosen or tighten). The results suggest that output gaps are typically discussed more during periods of low inflation, and generally more in advanced economies.
Finally, in addition to the relationship between the output gap and country characteristics, we examine the policy advice in the IMF staff reports and the level of the output gap. Ex ante, one could reasonably expect a higher output gap to be associated with a recommended tighter stance of policy. Looking at both levels and changes in output gaps and policy advice, we find a a slight positive association between the level of the output gap and the recommended relative tightening of monetary policy but a very limited trend for fiscal policy.
This very partial association may be due to several reasons, including the fact that our measure of policy recommendations is a relative one as it is always compared to the actual policy stance held by country authorities. That is to say, the recommended stance in the report suggests that policy should be looser or tighter than the authorities’ current stance. If policymakers are already responding appropriately to any existing output gap, the advice would not be expected to have a bias in either direction.
Another issue that will weaken the relationship between the output gap and the recommended stance is that the output gap is only one of many inputs that feed into policy advice. For example, debt sustainability issues may prevent a recommendation that the fiscal stance be loosened, and external vulnerabilities may similarly limit the ability to loosen monetary policy. Overall, there is hence a positive but weak relationship between policy recommendations and the level of the output gap.
Figure 2a Monetary policy recommendation vs output gap
Figure 2b Fiscal policy recommendation vs output gap
In sum, the evidence points to a useful but nuanced role for output gap estimation in policymaking and macroeconomic surveillance. This is particularly the case for real-time estimates of output gaps, which seem to be asymmetrically revised upwards as time goes by and thus complicate the assessment of optimal policy in real time. There are no obvious silver bullets that address the shortcomings. However, greater acknowledgement of uncertainty would be a step forward, as argued in Romer (2020). This could involve the use of confidence intervals when presenting output gap results and greater emphasis on both upside and downside risks in policy discussions. This has become particularly pertinent in the current environment of heightened uncertainty following Covid-19 and is likely to remain so throughout and beyond the recovery period.
Authors’ note: The views expressed in this column are those of the author(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management.
Barkema, J, T Gudmundsson and M Mrkaic (2020), “What do we talk about when we talk about output gaps?”, IMF Working Paper No. 2020/259.
Brinca, P, J Duarte and M Castro (2020), “Decomposing Demand and Supply Shocks during COVID-19”, VoxEU.org, 17 June.
Buti, M, N Carnot, A Hristov, K Mc Morrow, W Roeger and V Vandermeulen (2019), “Potential Output and EU Fiscal Surveillance”, VoxEU.org, 23 September.
Coibion, O, Y Gorodnichenko, M Ulate (2017), “The US Economy is not yet back to its Potential”, VoxEU.org, 24 August.
Dupraz, S, E Nakamura and J Steinsson (forthcoming), “A Plucking Model of Business Cycles”, Revise and resubmit at Journal of Political Economy.
Kangur, A, K Kirabaeva, J Natal and S Voigts (2019), “How Informative are Real Time Output Gap Estimates in Europe?”, IMF Working Paper 19/200.
Romer, D (2020), “In Praise of Confidence Intervals”, NBER Working Paper No. 26672.