This paper presents a micro-econometric model of the female labour market in Italy. The model is derived from a theoretical framework in which the labour market is assumed to be segmented, i.e. firms offer different labour contracts to different types of workers. In this context, female workers decide whether or not to participate in the labour market according to the contract offered, to their individual characteristics, to the regional features of the labour market, and to the economic performance of the sector in which they want to be employed. This is why the econometric model is estimated using a pseudo panel of data in which the sectional dimension is defined by the number of sectors, regions and types of female workers (age classes). The time dimension goes from 1970 to 1990. The econometric model is composed by four equations which endogenize female workers' participation decision, firms' labour demand and the features of the optimal contracts, i.e. wages and numbers of hours. The estimation results, obtained by explicitly accounting for the aggregation of individual equations, confirm the main implications of the theoretical model. In particular, hours and participation decisions crucially depend on the worker's characteristics, rather than on labour market conditions, thus leaving a minor role to the "discouraged« and "added« worker hypotheses. Moreover, the results stress the importance of quantity adjustment costs in the Italian labour market.NON TECHNICAL SUMMARYThis work explores the recent developments in the female labour market in Italy with a general approach, integrating demand and supply decisions in a regionally and industrially segmented model. In order to achieve this, labour demand is defined both in terms of hours and employed workers, while only the participation decision is modelled on the supply side. Solving both the worker and the firm maximisation problem, four equations have been specified for each type of female labour: hours, wages (which together define the contract which is offered by the firm), participation and demand for workers. Each type of worker differs from the others with respect both to productivity and aversion to work. The latter variable, which can be thought mainly as disutility of work as long as working hours increase, is known to the firm. This knowledge enables the firm to present an optimal contract to the worker, i.e. a contract with the minimum wage that will induce the worker to work that number of hours. The profit maximisation level of labour demand and of hours for each type of worker, and hence the wages, are determined for a given product demand. If one type of worker is more efficient than all the others, he/she will be fully employed. Demand for all the other types will depend on the former employment level. Without adjustment costs, the number of hours and the total wages will not depend on the production level: otherwise, the econometrically estimated equation of hours will contain it. Parameter estimation has been achieved using a completely new data base on women participation, hours and wages, disaggregated by age, region and sector. The four equations have been properly aggregated in order to match the available data. Consistent estimates of the model parameters have been subsequently obtained applying a pseudo-panel technique. The results are broadly consistent with our expectations, as far as the variables which enter the various equations are concerned. Adjustment costs appear to be relevant, since value added is significant both in the hours and wage equations; moreover, single degree holders in the region, the variable that proxies the most efficient group of workers, does not seem to have a statistically significant impact on labour demand. Hours, wages, and also participation seem to depend on the worker's characteristics, while in the labour demand equation, the only significant variables are wages and value added. From the theoretical premises, and from the empirical results, the existence of some involuntary unemployment is to be expected. Firms follow an efficiency wage mechanism in wage determination, and have no incentive to lower it, because it would induce an efficiency loss. Demographic variables seem to have a declining influence on wages and participation decisions. Educational attainments influence participation directly, with no effect on hours and wages. In addition to wages, also labour market conditions do not have a significant effect on participation, either proxied by value added or by the employment/active ratio. The negative trend in the hours equation, as well as the positive one in the wage equation, are consistent with the historical experience of all industrialized countries. Coefficients have proven to be stable through time and across groups. Fixed effects are significant in the first and third equations. Differences among them are not large and assume the expected sign. Therefore we can conclude that workers' heterogeneity is captured by the model variables, and its effect is largely consistent with theoretical predictions.
