Labour Supply in a High Unemployment Economy: Labour Supply Function
In almost all such studies, participation is treated as synonymous with being in employment, reflecting the North American norm of low unemployment. As we argue below, equating labour supply to employment may not be appropriate in high unemployment economies. Despite advancements in the estimation procedures, there are still some conceptual problems in estimating a labour supply function. In all these approaches, it is assumed that workers are free to choose their hours of work and the observed hours of work reflect a labour supply decision. This implies that the desired/optimal hours of work (hd) that maximise the utility (given the budget constraint) are assumed to be equal to the actual hours of work (ha).
This assumption is too restrictive for the following reasons. First, even when there is a range of jobs offering different hours of work, workers are unlikely to be mobile due to their skills not being transferable across jobs. Second, information asymmetry usually characterises the labour market (say due to search costs). Third, there are demand-side restrictions regarding the choice of hours of work within jobs. Finally, particularly in high unemployment economies including Kosova, there are demand-side restrictions regarding the availability of jobs at the market wage.
A number of studies are now available supporting the view that at least part of unemployment does not represent leisure, but rather a constraint on labour supply due to the lack of demand for labour. The observed high inactivity (such as in the Kosovan labour market) implies that joblessness may not be attributed to the reservation/asking wage exceeding the offered wage (although this cannot be ruled out completely), but it is rather a matter of jobs not being available at the market wage. This implies that for a significant number of unemployed individuals, the actual hours of work being zero does not reflect their labour supply decisions.
There is no evidence available for transition or developing economies regarding the gap between the actual and desired hours of work. For Canada, Kahn and Lang find a considerable difference between actual and desired hours of work (a third of the sample of female workers want to work more than they actually work, while some 17 percent want to work less). For the Netherlands, Euwals and van Soest find that on average the desired hours of work per week for men and women are 11.5 percent and 10.5 percent higher than their actual hours of work respectively. Brown and Sessions find for Britain that more than a third of the employed sample would like to work less, while only 4 percent would like to work more than they actually work.
Given the discussion above, when estimating the labour supply function, it may be incorrect to specify other variables constraining an individual’s behaviour to be the same when ha=hd as when ha<hd, since the estimates might be misleading as far as policy implications are concerned. This problem was emphasised in the early literature of labour supply determinants (Pencavel, 1986), but very few studies acknowledge this or make this distinction in the empirical estimation. The limited empirical evidence suggests that the estimated coefficients from the sample where ha=hd are significantly different compared to estimates from the sample where ha<hd.