Adjusting for Changes in Labour Composition in Statistics New Zealand's Productivity Series

Abstract

In measuring labour productivity, a composition-adjusted series is generally considered to provide the most representative measure of labour input. The rationale for adjusting for changes in labour composition is that workers are not homogenous, and as such, have different skill levels. Not only should this provide a more accurate measure of labour input, but it can also provide insight into the effects that changes in labour composition have on productivity. In practice, this process is undertaken by cross-classifying labour by proxies for skill, such as educational attainment and experience. The relative skill levels of different groups are estimated via regression analysis, through differences in hourly wages. This paper evaluates the various theories behind compositional adjustment, and presents the results of applying such an adjustment to the Statistics NZ labour productivity series, under various conditions. The ultimate contribution of the paper is an experimental composition-adjusted labour productivity series, for which external feedback is welcomed.