The last word on the wage curve?

Since 1990, there has been extensive international research on the responsiveness of wages of individuals to changing local labour market conditions. For many countries, an inverse relationship between wages and local unemployment rates has been found. In their book, The Wage Curve, Blanchflower and Oswald argued that the unemployment elasticity of pay is around -0.1 in most countries. In a 1995 literature survey, Card referred to this striking empirical regularity as being close to an 'empirical law of economics'. Nonetheless, reported elasticities do vary, even excluding outliers, between about -0.5 and +0.1. There is also considerable heterogeneity among wage curve studies in terms of data and model specification. This paper carries out meta-analytic techniques on a sample of 208 elasticities derived from the literature to uncover the reasons for the differences in empirical results across studies. Several causes of variation are identified. There is also clear evidence of downward publication bias. In addition, many reported t-statistics are biased upwards due to the use of aggregate unemployment rates. A maximum likelihood method and a trimming procedure are used to correct for these biases. Both methods give similar results for our sample. An unbiased estimate of the wage curve elasticity at the means of study characteristics is about -0.07.