Analysis of composition effects on wage behaviour

In the Spanish labour market, characterised by high labour market inflows and outflows,1 aggregate wage behaviour may be significantly affected by changes in the composition of employment. This effect may have been particularly important in the period dating from 2008, in which the marked job destruction has been concentrated in lower skilled and in less experienced workers whose average wage is below the Spanish average. It is therefore important to isolate the mechanical changes which this phenomenon may have caused in wages, in order to determine accurately how wages have responded to economic developments. Table 1 shows that the changes in the composition of employment since 2008 have been extremely sharp and generally skewed towards a significant loss in the relative weight of lower-paid workers. For example, the share of younger workers in total employment fell by nearly 9 percentage points (pp) between 2008 and 2012, and that of foreign workers dropped by nearly 2 pp in the same period. Similarly, the fall in employment was particularly pronounced in workers with a lower educational level, the proportion of which dropped from 14.8% in 2008 to 10.2% in 2012. Finally, in terms of work experience, the share of workers with fewer years of service in the firm decreased sharply, reflecting the greater impact of the crisis on temporary jobs. Specifically, the share of workers with more than three years experience expressed as a percentage of total employment increased from 61.7% in 2008 to 73% in 2012. In aggregate wage statistics (those normally used for economic analysis), such large changes in the composition of employment may give rise to relatively significant statistical effects which have to be taken into account when attempting to make a precise diagnosis on the behaviour of labour costs and their relationship with the cyclical situation. However, estimating these effects requires disaggregated information on the wage received by each worker, as well as the characteristics of the incumbent and of the job held. This article estimates these composition effects in the private sector of the economy using data from the social security administrative labour records (Muestra Continua de Vidas Laborales, hereafter “MCVL” by its Spanish abbreviation), which provides such information up to 2012 for a representative sample of social security records. The rest of the article is organised as follows. The second section briefly summarises the methodology used to estimate the aforementioned composition effects. The third section sets out the main results of the exercise and analyses the relative contribution of each of the employment characteristics considered. Finally, the fourth section sets out some brief conclusions.