DP5801 Drift and Breaks in Labour Productivity
|Publication Date:||August 2006|
|Keyword(s):||bootstrapping, frequency domain, median-unbiased estimation, Monte Carlo integration, structural break tests, time-varying parameters, variance ratio|
|Programme Areas:||International Macroeconomics|
|Link to this Page:||cepr.org/active/publications/discussion_papers/dp.php?dpno=5801|
We use tests for multiple breaks at unknown points in the sample, and the Stock-Watson (1996, 1998) time-varying parameters median-unbiased estimation methodology, to investigate changes in the equilibrium rate of growth of labor productivity?both per hour and per worker?in the United States, the Eurozone Australia, and Japan over the post-WWII era. Results for the U.S. well capture the 'conventional wisdom? of a golden era of high productivity growth, the 1950s and 1960s; a marked deceleration starting from the beginning of the 1970s; and a strong growth resurgence starting from mid-1990s. Interestingly, evidence suggests the 1990s? productivity acceleration to have reached a plateau over the last few years. Results for the Eurozone point towards a marked deceleration since the beginning of the 1980s, with the equilibrium rate of growth of output per hour falling to 0.9% in 2004:4. Results based on Cochrane?s variance ratio estimator suggest a non-negligible fraction of the quarter-on-quarter change in labor productivity growth to be permanent. From a technical point of view, we propose a new method for constructing confidence intervals for variance ratio estimates based on spectral bootstrapping. Preliminary Monte Carlo evidence suggests such a method to possess good coverage properties.