The purpose of the study is to theorise on participation in recurrent training and to estimate a statistical model. The main focus is the links between latent variables, such as observed skills based on literacy proficiency, firm size, literacy practices at work and at home, and other latent correlates, such as educational attainment, labour force status, experience, occupational status and earnings. Three models are specified: A model predicting literacy proficiency; a model predicting earnings; and a model predicting participation in recurrent training. The two first models represent intermediate steps in the construction of the final model.

The data set employed in this study is derived from the International Adult Literacy Survey. It combines tools from educational assessment and household survey, in order to collect background information about the participants and their labour force experience, and makes use of an innovative measure of human capital stock, namely proficiency on a functional literacy assessment. The analysis also includes countries from different parts of the world: Eastern Europe; Northern Europe; Southern Europe; and North America. Structural Equation Modeling has been used.

The results show that all three models confirm the importance of educational attainment as one of the main predictors of literacy proficiency, earnings and recurrent training, with the effects of educational attainment most often mediated by other variables. Hence literacy proficiency and educational attainment are both important for modeling earnings or participation in recurrent training. The magnitude of educational attainment is seen to provide justification for signalling theory, job matching theory, and principal agent theory, as earnings and occurrence of recurrent training are related more to employees' educational attainment than to their actual levels of knowledge and skills. Work related learning and home related learning are strong predictors of the outcome variable in all three models. The main policy implication of this finding is that no form of education should be favoured, but rather each should be considered valuable and complementary to the others.

The total effect of parents' education suggests that parents continue to have an influence on the life careers of their children a long time after the completion of initial, formal education. In other words, the kind of start a child has in life is a very strong predictor of his or her actual level of literacy proficiency, earnings, or whether or not he or she will take part in further training, lending support to social capital theory, as well as to resource conversion theory.

The importance of the long arm of the job is confirmed for the prediction of earnings, as well as for the prediction of recurrent training. Labour force status, firm size and, to a lesser extent occupational status, are important indicators.

No particular relationship is observed between men and women as regard to models predicting literacy proficiency and earnings. Conversely, gender differences in recurrent training are evident, with, on the one hand, determining factors for women, for example, earnings, and, on the other hand, determining factors for men, such as firm size and literacy practice at work. Hence, the study suggests that women's participation in recurrent training is determined by different factors than that of men.