Wage and Salary in the Labour Force Survey

According to Eurostat regulations, collection of data monthly wage of employee is included in Labour Force Survey like optional variable. Eurostat is proposing that collection of data on monthly wages and salaries become obligatory because it’s an important variable for deepening the study of labour market by e.g. comparing pay by different occupational groups or by gender, age, education. With the aim of complying with Eurostat recommendations, in the third and fourth quarters 2004 was conducted the test of variable on monthly wages and salaries in the Labour Force Survey in Italy. We tested different ways to collect this information in order to find the best one. In particular, according to Eurostat guidelines, we tested the impact of asking gross or net salary and the impact of asking exact salary, salary in bands or a double approach (start by asking exact salary and if no answer in bands). So we tested four different formulations: 1) net exact; 2) net in bands; 3) gross exact; 4) gross in bands. With reference to the problem of partial non-response and interview break-offs, the analysis concerns the following issues: the degree of the problem in different waves; the effect of telephone or face to face interviews in CAPI-CATI systems; the impact of proxy interviews and comparisons among waves between a direct and a proxy interview. We are also able to analyse wages for which there is a higher frequency of non-response to this item depending on profession characteristics. Moreover, considering the longitudinal section of the sample, we use the second interview to the same group in the next quarter (the following wave) as a control interview. In this case it is possible to study the characteristics of fours respondent’s profiles: a) response in first and second wave; b) response only in the first wave; c) response only in the second wave; d) non-response in both waves. We describe groups focusing on the similarity/dissimilarity between the groups and on the degree of homogeneity within each group both on the basis of work variables and on the basis of demographic variables. Finally, for studying non-response rate we apply a multivariate analysis technique using other data which is correlated with wages, such as age, educational attainment, profession, economic activit