Labour market regulations aimed at enhancing job-security are dominant in most Western European countries, and Spain is not an exception. These regulations seek to reduce dismissals of workers and fluctuations in employment, and their effectiveness materialize through changes in the costs of adjustment. To understand how these job-security regulations operate it is therefore necessary to learn how these costs affect labour demand and how such regulations modify them. As stressed by Hamermesh and Pfann (1996), knowledge of the structure of adjustment costs is crucial to understand macroeconomic fluctuations in employment.
Most empirical studies (e.g. Nadiri and Rosen, 1969, and Sargent, 1978) presume that the quasi-fixity of labour results from increasing costs of adjustment. Nevertheless, the sources of such costs can be very different depending on whether changes in a firm’s employment are either positive (hiring costs) or negative (firing costs), so that in general adjustment costs will depend on the sign of such adjustment. In fact, the standard assumption of quadratic and symmetric adjustment costs yields an unsatisfactory description of the costs that firms face when adjusting employment. Empirically, the dynamics of labour demand based on symmetric adjustment costs are in general at odds with the data. The rejection is stronger as the level of data disaggregation rises (e.g. from industry to firm). Using data on Dutch manufacturing firms, Pfann and Verspagen (1989) obtain evidence in favour of asymmetric adjustment costs, in which hiring costs exceed firing costs.
Moreover, the assumption of workers’ homogeneity becomes inappropriate and may lead to wrong inferences if the dynamics of adjustment among labour inputs prove to be different. Intuitively, one would expect hiring costs to be larger the higher the skill of workers, since training costs are expected to be lower for unskilled labour. Furthermore, since severance pay depends on the worker’s earnings and they depend on his skill, firing costs will increase with worker’s skill. Empirical findings by Palm and Pfann (1990 and 1993), using aggregate data from the Netherlands and the UK, and Bresson, Kramarz and Sevestre (1991 and 1992) using firm-level data from France, among others, show that the adjustment speed of unskilled workers is generally higher than that of skilled workers. These studies seem to imply that when firms face a shock, they do not necessarily adjust employment uniformly for the different labour inputs.
Recognizing labour heterogeneity requires examining how the costs of changing one type of labour affect the dynamics of the demand for other types. After the occurrence of a shock, the speed of adjustment will not be the same for all labour inputs. A particular labour input will be adjusted more slowly than others if either its variable adjustment costs are more convex or, alternatively, they are simply greater and firms do not know the duration of the shock. Furthermore, the size of the shock that it is needed to adjust a given labour input increases with the fixed costs of adjusting that input. Additionally, stickiness in adjusting one type of labour may affect the speed at which other types are adjusted. In recent work, Rota (1997) finds considerable stickiness in Italian manufacturing employment, as reflected by the significant percentage of firms that do not adjust employment in a particular year. This sort of evidence suggests the existence of a significant lump-sum component in firms’ adjustment costs.
The main purpose of this paper is to evaluate the structure of adjustment costs considering different labour inputs, and allowing for interrelated dynamics among them, and for costs asymmetries between positive and negative employment changes. To do this, I use a Spanish panel of manufacturing firms corresponding to the period 1986-1991. This data set contains annual firm-level information on the number of employees by duration of the labour contract (fixed-term vs. indefinite) and by job (nonproduction or white collar workers vs. production or blue collar workers). I derive and estimate Euler equations for permanent production and nonproduction employees in a standard profit-maximizing framework, using an asymmetric adjustment costs representation. In the sample, there is evidence of a mass point at zero adjustment for each of the labour inputs, which suggests the existence of either fixed costs of adjustment, or indivisibilities in labour inputs, or both. Since the Euler equations are marginal conditions, they are inconsistent with a mass point at zero adjustment, and I therefore will estimate each equation using the subsample of observations for which adjustment is done, yet controlling for endogenous sample selection. I thus will concentrate in the adjustment costs structure that firms face in addition to fixed costs; although my approach recognizes (and controls for) the possibility of such fixed costs, I will not consider their estimation.
Download
PDF Ebook Demand for Labour Inputs and Adjustment Costs: Evidence from Spanish Manufacturing Firms
