Employment participation, unemployment and non market work: Composition models of the United States labour force

Abstract

Data on the proportions in each of the labour force status categories sum to one and form a composition. Using the log ratios of the parts and a transform to orthogonal coordinates in a lower dimension provides a consistent way of handling this constraint. This paper emphasizes that a composition is a vector of values not a single number and shows a coordinate for employment participation is directly associated with a coordinate defining the level of search by those without jobs.The model is extended to deal with heterogeneity due to gender and current status and to generate a new analysis of the gross flows data. The associations observed provide evidence that micro decisions can be strongly modified by the macro situation and a Keynesian demand driven framework fits very well.