Potential output is a crucial benchmark for policymakers because it measures the output level or path an economy can sustain over the medium term. It is typically defined as the level of non-inflationary real gross domestic product (GDP). Since it is a counterfactual object, assumptions are needed to estimate it. This article discusses old methodologies, empirical challenges, and new developments in the estimation of potential output.
Question 1. What is potential output, and why is it different from actual output?
Macroeconomists examine hundreds of statistics, but the most important indicator they look at when they study a given country is output. Output is measured using the real gross domestic product (GDP), i.e., the value of all goods and services produced in an economy, adjusted for changes in prices. Typically, countries want to increase the level and the rate of growth of real GDP because of its close relationship to the level of income and wealth and, ultimately, of welfare. In addition, economists and policymakers may also want to understand whether observed (or actual output) is at the level where it “should be,” given the country’s economic environment. Potential output tries to measure precisely where the economy “should be.” If there is a difference between where the economy “is” (actual output) and where it “should be” (potential output), then economic policy might be able to close the gap between the two (which is the output gap).
A reliable measure of potential output is a critical benchmark for economic policy. Potential output (or GDP) is often defined as the level of output that an economy can sustainably produce over the medium term making normal use of its resources. Yet, this is a concept that is not straightforward to implement in practice because economies rarely operate at normal capacity and undistorted for long. In reality, a barrage of overlapping shocks tends to move real GDP one way or the other, leaving policymakers to decide whether these changes are due to lasting capacity shifts, transitory demand jitters, or simply statistical noise.
Question 2. Why does potential output matter?
Potential output is closely related to inflation, and the definition from the Congressional Budget Office (2001) makes this clear: “… it is a measure of maximum sustainable output—the level of real GDP in a given year that is consistent with a stable rate of inflation.” Central banks manage demand (i.e., actual output) through changes in monetary conditions and interest rates. When actual output exceeds potential output, inflationary pressures emerge. On the other hand, output below the potential level (a negative output gap) implies that there is underemployment (excess supply) of capital and labor, which would motivate a looser policy stance.
But potential output is also important for other areas of policymaking. In the absence of further shocks and when price and wage adjustment is complete, actual output converges to potential output. Hence, the potential growth rate of the economy determines its long-run position. Structural reforms, such as removing inefficiencies in labor and goods markets, might help increase the potential growth rate. The fiscal accounts are also affected by the difference between actual and potential output. For instance, if the economy is growing faster than its potential rate and the output gap is positive, tax revenues would tend to be higher than in normal times because of strong profits, wages, and asset prices. At the same time, expenditures are likely to be lower because of lower unemployment benefits and other social spending. If the fiscal accounts are not corrected by the cycle, policymakers might incorrectly conclude that the fiscal outlook is more favorable and further increase government spending, thereby increasing the debt bias.
Question 3. What is the simplest way to compute potential output?
There are several methods to compute potential output. A simple and popular method is to assume that potential output is a smooth trend around which actual output fluctuates. The widely used Hodrick and Prescott (HP, 1997) filter computes potential output as a two sided moving average of actual output. However, the HP filter has important shortcomings. It can be sensitive to statistical choices (e.g., the degree of smoothing), and it suffers from the problem of reverting to actual GDP at the start and end of the sample. This may lead to close-to-zero estimates of the output gap in real time, which then get revised when new data becomes available. By construction, the HP filter does little to foster our understanding of what actually drives potential output. Finally, the assumption of a smooth trend might be at odds with large shocks, such as a banking or financial crisis, hitting the economy.
Question 4. How is potential output estimated using supply-side measures?
Production function models construct potential output bottom-up from the supply side of GDP based on available labor and capital inputs, as well as measures of total factor productivity and utilization rates of labor and capital. This is the method applied by the European Commission and the Congressional Budget Office, among others. However, the approach requires timely access to micro-level data as well as filtering to eliminate short-term fluctuations from these variables—for example, to determine the “potential” level of labor, capital and total factor productivity available for production—creating problems very similar to the univariate filtering approach.
Question 5. What other macroeconomic variables can be useful to estimate potential output?
Structural multivariate approaches use relationships derived from economic theory to help identify potential output (see Laubach and Williams 2003; and Benes and others 2010). Multivariate filters take advantage of the information contained in selected observable macroeconomic variables such as consumer price index (CPI) inflation, which is related to the output gap through the Phillips curve relationship (see Clarida, Galí, and Gertler 1999). Another variable to consider is unemployment, which is linked to output through the relationship known as Okun’s law (see Ball, Leigh, and Loungani 2013). Including these additional variables adds to the usefulness of estimates of potential output for guiding policy decisions. At the same time, the results are often quite sensitive to the specification and estimation of the underlying partial-equilibrium relationships. In addition, the necessary assumptions about the smoothness of potential output require judgment quite similar to the selection of the smoothness of univariate filters.
Question 6. Do financial variables contain useful information to estimate potential output?
From a more practical perspective, CPI inflation appears to have been a less informative variable to compute the output gap in recent years (see Bayoumi and others 2014; and IMF 2013). Despite major recessions in advanced economies in recent years, deflation was lower than that predicted by Phillips curve models. In addition, in the build up to the recent global financial crisis, CPI inflation was not particularly out of target in most advanced economies. However, inflationary pressures did show up in other price measures outside the definition of the CPI: most notably, in house prices. With hindsight, the severe credit and housing busts that followed suggest actual GDP growth significantly outpaced potential during the boom years.
More generally, the empirical literature on credit booms and busts (Claessens and others 2012) provides a strong case for including financial variables to inform potential output estimates. The multivariate filter approach can be extended to include credit, house, and asset prices. If wide swings in output tend to occur along with wide swings in credit, the approach will ignore the former when determining the level of potential output. In a recent paper, Borio and others (2014) have shown that estimated output gaps that were taking into account financial variables were indeed pointing at more overheating before the crisis, and a more negative gap afterwards, than a conventional HP-filter approach would suggest. This means that once financial variables are included in the analysis, potential output moves more steadily.
Question 7. How useful are fully specified dynamic stochastic general equilibrium models for estimating potential output?
As we have discussed, univariate and multivariate filters aim at extracting potential output as a smooth trend, potentially taking into account information from other relevant sources. However, some of the parameters used in these filters are reduced-form and do not have an economic interpretation, making it difficult to understand what are the channels of transmission. Dynamic stochastic general equilibrium (DSGE) models overcome some of these shortcomings by modeling both the demand and the supply side of the economy. Hence, they can identify GDP fluctuations driven by all shocks that matter for potential output over the longer term. These models also narrow down explicitly the definition of potential output to the output level that would be available if the economy could operate in the absence of price and wage rigidities, but taking into account the reality of real frictions (such as adjustment costs to investment or employment) that demand policies cannot overcome. The latest generation of estimated DSGE models incorporates labor frictions (see Galí, Smets, and Wouters 2011), and financial frictions (see Furlanetto, Gelain, and Taheri Sanjani 2014), with promising results. However, the findings are sensitive to the underlying assumptions and different models can produce different output gaps, so more work is needed to take into account model uncertainty.
BallL. D. Leigh and P.Loungani. 2013. “Okun’s Law: Fit at Fifty?” NBER Working Papers 18668Cambridge, MA: National Bureau of Economic Research, Inc.
BayoumiT.G.Dell’AricciaK.HabermeierT.Mancini Griffoli and F.Valencia. 2014. “Monetary Policy in the New Normal.” IMF Staff Discussion Notes 14/3International Monetary FundWashington DC.
BenesJ.M.JohnsonK.ClintonT.MathesonP.ManchevR.Garcia-Saltos and D.Laxton. 2010. “Estimating Potential Output with a Multivariate Filter.” IMF Working Paper 10/285International Monetary FundWashington DC.
BorioC.P.Disyatat and M.Juselius. 2014. “A Parsimonious Approach to Incorporating Economic Information in Measures of Potential Output.” BIS Working Papers 442Bank for International SettlementsBasel.
ClaessensS.A.Kose and M.Terrones. 2012. “How Do Business and Financial Cycles Interact?,” Journal of International Economics87(1): 178–190.
ClaridaR.J.Galí and M.Gertler. 1999. “The Science of Monetary Policy: A New Keynesian Perspective.” Journal of Economic Literature37 (December): 1661–1707.
Congressional Budget Office. 2001. “CBO’s Method for Estimating Potential Output: An Update,” August2001.
FurlanettoF.P.Gelain and M.Taheri Sanjani. 2014. “Output Gap in Presence of Financial Frictions and Monetary Policy Trade-Offs,” IMF Working Paper 14/128International Monetary FundWashington DC.
GalíJ.F.Smets and R.Wouters2012. “Unemployment in an Estimated New Keynesian Model,” NBER Macroeconomics Annual26(1): 329–360.
HodrickR. J. and E. C.Prescott. 1997. “Postwar US Business Cycles: An Empirical Investigation,” Journal of Money Credit and Banking29(1): 1–16.
International Monetary Fund. 2013. “Chapter 3. The Dog That Didn’t Bark: Has Inflation Been Muzzled or Was It Just Sleeping?,” World Economic OutlookApril2013Washington DC.