Stylized facts : real wage behaviour over the business cycle
Date
1990
Authors
Sigurdson, Judith Dawn
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The purpose of this study is to determine some stylized facts about the real wage series over the business cycle. Specifically, the study's goals are to determine if real wages vary systematically over the business cycle and within its own series, as revealed by the series' relationship with business cycle reference variables and its past values, respectively. Any credible business cycle theory should be consistent with the results.
The economic literature since the 1930s has failed to provide consistent evidence to support or refute the procyclical, countercyclical, or acyclical behaviour of the real wage series. Once such information is determined, researchers can test the consistency of their business cycle theory with this stylized fact. Any insight into wages' behaviour over the business cycle is therefore useful.
A thorough examination of the causal relationship between wages and business cycle variables involved various parametrization of bivariate Granger and Sims tests for unidirectional Granger-causality, along with Haugh/Pierce cross correlation analyses. Granger and Sims analyses allow for testing for unidirectional causality between each business cycle variable and real wages. The Haugh/Pierce analysis reflects whether or not real wages and each business cycle reference variable are independent and provides some insight into the lead-lag relationship between each pair of variables. Overall, the Canadian analyses did not suggest causal relationships between wages and the business cycle variables. The exception occurred while analyzing wages' relationship with prices and, to a lesser extent, with housing starts. The overall evidence suggests Canadian wages are acyclical. In the American case there is evidence of a causal relationship between wages and the business cycle variables, particularly housing starts, real output, and employment, suggesting wages exhibit some cyclical behaviour.
For both countries, there is evidence that the real wage series is consistent with the random walk hypothesis. That is to say, the best forecast for next period's wage, using only the information available within the series' past, is the current period's wage (adjusted by its drift parameter).