Models of Labour Services and Estimates of Australian Productivity

Abstract

This paper examines the manner in which labour services are modelled in the aggregate production function, concentrating on the specification of the relationship between the number of persons employed and average hours worked. We argue that, given the presence of quasi-fixed costs of employment, hours of work and the number of employees cannot be perfect substitutes. We then show that estimates using total hours worked as the measure of labour input implicitly assumes that they are perfect substitutes and this false assumption results, inter alia, in biased estimates of the rate of labour and multifactor productivity growth in Australia.