Employer Size and On-the-Job Training Decisions

Abstract

The effect of employer size on labor market outcomes have been the subject of extensive, although relatively recent, research by economists. It has been widely demonstrated that large and small firms exhibit a number of differential characteristics with regard to their employment relationships [29; 42]. Employees at large firms have higher wages and fringe benefits [16; 10; 15; 22; 32; 35], higher levels of job mobility within the firm [19], greater job stability [44], a greater tendency to have mandatory overtime rules [23], higher education levels [8], a more structured work environment [20; 26; 46], and higher levels of investments by the firm both in the form of search costs [4; 5] and direct on-the-job training [6; 10]. While numerous theories exist that account for many of these relationships, such as monitoring and on-the-job training models, no single explanation has been able to account definitively for the observed empirical patterns. Clearly size of employer is a significant factor in the dynamics of labor markets that still requires much investigation. The focus of this paper is on one particular aspect of employer size effects-the observed higher training investments in large firms. We propose and test a model in the human capital tradition of viewing on-the-job training (OJT) as an investment [9; 36; 38; 39; 42], which not only explains the greater investment in on-the-job training by larger firms, but additionally generates predictions with regard to the allocation of the firm's resources across investment in people with different demographic characteristics. These latter implications of the model facilitate empirical comparisons to alternative explanations. Our central hypothesis is that large employers exhibit different behavior from small employers with regard to the risk component of the investments in their employees. Other training decision models that have addressed firm behavior under uncertainty have done so in the context of monitoring difficulties introduced by either the higher opportunity cost of managers' time [43] or diseconomies of scale in monitoring [16] in larger firms. Although both of these approaches