There Are Two Okunõs Law Relationships Between Output and Unemployment
There Are Two OkunÕs Law Relationships Between Output and Unemployment
Humberto Barreto and Frank Howland* Wabash College
Abstract This paper corrects a fundamental error in the literature examining the OkunÕs Law relationship between the unemployment rate and the rate of growth of output. Since OkunÕs original work, biased estimates of the Okun Coefficient on Unemployment, output gaps, and potential GNP have been reported by authors who mistakenly assume that unbiased coefficient estimates of the reverse regression are reciprocals of their direct regression analogues. Thus, for example, there is no need to explore a wide variety of alternative answers to the ÒpuzzleÓ of an extremely high Okun coefficient in Japan Ñ the extremely large value is merely a statistical artifact. (JEL Classification Numbers: C1, E3, E6)
1993
Address correspondence to Humberto Barreto, Wabash College, Crawfordsville, IN 47933 USA FAX: (317) 364-4295 E-mail: barretoh@wabash.edu
1.0 INTRODUCTION This paper emphasizes that the research question determines the direction of regression. This insight allows us to correct a fundamental error in the OkunÕs Law literature concerning the magnitude and interpretation of coefficients from the direct and reverse regressions of unemployment on output. In 1962, Arthur Okun found a clever solution to the difficult problem of estimating an economyÕs potential output. Instead of simply measuring the trend of GNP over time and calculating the GNP gap as the deviation of actual GNP from trend, Okun exploited the dependable relationship between the rate of growth of output and the unemployment rate (now commonly known as OkunÕs Law) in order to estimate potential output given actual unemployment. Unfortunately, Okun, and many others following in his footsteps, mistakenly assumed that it is valid to use the reciprocal of the slope of the regression of unemployment on output when making a prediction of output given unemployment. In fact, the
Humberto Barreto and Frank Howland* Wabash College
Abstract This paper corrects a fundamental error in the literature examining the OkunÕs Law relationship between the unemployment rate and the rate of growth of output. Since OkunÕs original work, biased estimates of the Okun Coefficient on Unemployment, output gaps, and potential GNP have been reported by authors who mistakenly assume that unbiased coefficient estimates of the reverse regression are reciprocals of their direct regression analogues. Thus, for example, there is no need to explore a wide variety of alternative answers to the ÒpuzzleÓ of an extremely high Okun coefficient in Japan Ñ the extremely large value is merely a statistical artifact. (JEL Classification Numbers: C1, E3, E6)
1993
Address correspondence to Humberto Barreto, Wabash College, Crawfordsville, IN 47933 USA FAX: (317) 364-4295 E-mail: barretoh@wabash.edu
1.0 INTRODUCTION This paper emphasizes that the research question determines the direction of regression. This insight allows us to correct a fundamental error in the OkunÕs Law literature concerning the magnitude and interpretation of coefficients from the direct and reverse regressions of unemployment on output. In 1962, Arthur Okun found a clever solution to the difficult problem of estimating an economyÕs potential output. Instead of simply measuring the trend of GNP over time and calculating the GNP gap as the deviation of actual GNP from trend, Okun exploited the dependable relationship between the rate of growth of output and the unemployment rate (now commonly known as OkunÕs Law) in order to estimate potential output given actual unemployment. Unfortunately, Okun, and many others following in his footsteps, mistakenly assumed that it is valid to use the reciprocal of the slope of the regression of unemployment on output when making a prediction of output given unemployment. In fact, the