Okun’s law describes the empirical relationship between changes in unemployment and output at the macroeconomic level and has been regarded since its discovery by Arthur Okun (Potential GNP: its measurement and significance. In: Proceedings of the business and economics statistics. American Statistical Association, Washington, DC, p 98–104, 1962) as a building block of traditional macroeconomic models. This article discusses the interpretation of this relationship and summarizes recent developments in the econometric specification of Okun’s law.
KeywordsAggregate demand Demand and supply shocks Okun’s coefficient Okun’s law Output gap Phillips curve Production functions Unemployment
JEL ClassificationsE32 J21
The term ‘Okun’s law’ refers to the empirical relationship between changes in unemployment and output. It is a basic building block of traditional macroeconomic models, where the aggregate supply function is derived from combining Okun’s law with the Phillips curve.
In Okun’s original contribution (Okun 1962), the empirical relationship between unemployment and output is introduced in the context of the quantification of potential output and the measurement of the social costs of unemployment in terms of forgone production.
Okun (1962) presents estimates for the United States which are based on three alternative econometric specifications aimed at quantifying empirically the relationship between unemployment and output growth: (a) regressing (quarterly) changes in the unemployment rate on (quarterly) percentage changes in production (as proxied by Gross National Product, or GNP), (b) regressing the unemployment rate on percentage deviations from potential output, defined as the exponential trend in GNP and (c) regressing the (logarithmized) employment rate on a linear time trend and (logarithmized) GNP. The effect of output changes on unemployment is quantified by the estimated parameter associated with the output variable in each of these regressions, whose inverse is usually known as ‘Okun’s coefficient’. The results in Okun’s contribution indicate that there exists roughly a three-to-one link between unemployment and output changes, in the sense that an increase/decrease of three percentage points in output (or the output gap, depending on the specification) is associated with a decrease/increase of one percentage point in unemployment. This rule of thumb is proposed as a ‘subjectively weighted average’ (Okun 1962, p. 100) of the estimates obtained from the three specifications. Estimations based on data including the post-oil crisis period, which can be found in most modern macroeconomic textbooks, tend to reveal an Okun’s coefficient that is closer to two than to three.
Okun’s arguments (see, for example, Okun 1962, p. 99) suggest that the link found between unemployment and output is not to be understood as a ceteris paribus relationship, but rather as capturing also the effects of simultaneous changes in labour force, hours worked and productivity (see also Friedman and Wachter 1974). Okun argues that a reduction in the unemployment rate would induce an increase in the labour force by persuading discouraged workers to seek work actively, and also presents estimates of the increase in hours worked per employed person caused by rising output. The analysis carried out in Okun’s contribution, based on data for the United States in 1960, assigns approximately 56 per cent of the change in output to the effect of changes in total labour input measured in hours worked, while the rest is attributed to productivity increases. Prachowny (1993) approaches the quantification of the link between unemployment and output by proposing a specification based on a fairly general production function, where the independent effects of changes in unemployment, hours worked, capacity utilization and labour force on output can be estimated separately. In this setting, Okun’s empirical specifications would be appropriate only if certain parameter restrictions on the production function are satisfied. Prachowny (1993) therefore proposes labelling Okun’s law ‘Okun’s theory’ and testing these restrictions directly on the data. The estimates of the direct effect of unemployment on output obtained using this specification are correspondingly smaller than in the original contribution by Okun, although the econometric modelling strategy used (based on estimating the production function in gap form and in first differences) is not without criticism. Attfield and Silverstone (1997) reconsider this approach using cointegration techniques and find estimates that are comparable with the original values in Okun (1962).
Obviously, the relationship observed between changes in output and changes in the unemployment rate is determined by the nature of the shocks hitting the economy. The usual interpretation of the relationship summarized by Okun’s law refers to arguments based on shocks to aggregate demand. Blanchard and Quah (1989) emphasize the importance of identifying demand and supply shocks in order to estimate and interpret Okun’s coefficient. Using a dynamic system formed by the unemployment rate and output growth, Blanchard and Quah (1989) assess the issue by isolating supply and demand shocks and interpreting the responses of these two variables to each type of shock. The results suggest that the implied Okun’s coefficient for demand shocks is slightly above two, while there is no such systematic short-run relationship between unemployment and output changes following a supply shock.
While many macroeconomic textbooks tend to emphasize that the relationship between short-run changes in output and unemployment is a robust and reliable empirical regularity, much of the literature dealing with Okun’s law is aimed at evaluating the robustness of this link across countries (Kaufman 1988; Moosa 1997; Lee 2000), in time (Sheehan and Zahn 1980; Gordon 1984; Evans 1989), across econometric specifications (Weber 1995; Lee 2000) and across states of the business cycle – recessions versus expansions – (Lee 2000; Crespo Cuaresma 2003). The results of this branch of literature point towards the existence of asymmetric, country-specific Okun’s coefficients, with a higher elasticity of unemployment to output changes in recessions than in expansions, and a lower elasticity in continental European countries compared with Canada, the United States and the United Kingdom. The estimates of Okun’s coefficient appear to be sensitive to the specification and de-trending method used for retrieving the cyclical component of output and the unemployment rate. Furthermore, this empirical literature usually reports evidence of structural instability, with a break in Okun’s coefficient taking place in the 1970s.
- Blanchard, O.J., and D. Quah. 1989. The dynamic effects of aggregate demand and supply disturbances. American Economic Review 79: 655–673.Google Scholar
- Okun, A.M. 1962. Potential GNP: Its measurement and significance. In Proceedings of the business and economics statistics, 98–104. Washington, DC: American Statistical Association.Google Scholar