Is there compelling evidence against increasing returns to matching in the labour market?

Matching models of the labour market have been of particular interest in macroeconomics where the notion of 'thick-market' externalities can lead to multiple equilibria. This has led to some recent interest in constructing empirical estimates of labour-market matching functions. This paper argues that existing estimates do not provide compelling evidence against the hypothesis of increasing returns in matching. The assumption made in several studies, that the relevant pool of job searchers is proportional to the stock of unemployment, is a potentially important source of downward bias in returns-to-scale estimates. We show the source of this bias theoretically and illustrate its magnitude by estimating Canadian aggregate and regional labour-market matching functions over the period 1978-88. This evidence suggests significant increasing returns to labour-market matching.