Human capital increases through training, be that implicit on-the-job training (as captured by tenure), or explicit classroom-type formal training leading to human capital accumulation. Our focus in this paper is mostly on the latter: we are interested in how a firm’s explicit decision to train depends on aggregate and sectoral output fluctuations.
It is not ex-ante obvious whether investments in human capital are counter cyclical, pro-cyclical, or a-cyclical. Since the opportunity cost to train workers is lower in downturns, a negative productivity shock should be associated with increased training. This channel is highlighted by deJong and Ingram (2001) who find that training activities “are distinctively countercyclical” and by Devereux (2000) who argues that during downturns firms hoard labor by assigning high-skill workers to lower-production activities such as training, thus avoiding layoffs and the fixed costs associated with firing and re-hiring workers. On the worker side, the literature documents that college enrollment is counter-cyclical (e.g. Dellas and Sakellaris (2003)); typically, enrollment in universities increases when the economy is not doing well and good jobs are harder to find.
Nevertheless, a different adjustment is also possible. A positive shock may be related to the adoption of new technologies which not only require training but also can provide increased returns to skill. This is the channel identified by King and Sweetman (2002). Using administrative Canadian data, they find that “re-tooling” (measured as quits from work to school) is pro-cyclical, consistent with a model where the outside option of high-skill jobs goes up during episodes of higher output, increasing the return to skill and therefore the value of training.
This paper has two main contributions. First, using the Canadian Workplace and Employee Survey (WES), we find that (i) training moves counter-cyclically with aggregate output fluctuations (more training in downturns), while at the same time (ii) the relative position of sectoral output has a positive impact on training decisions (more training in a sector doing relatively better). Second, in order to illustrate the mechanisms at work, we incorporate training decisions into a standard Mortensen-Pissarides model.
In order to measure the quantitative impact of aggregate and sectoral output fluctuations on training incidence, we use the Canadian WES dataset, an eight-year panel of firms from 1999 to 2006, representative of all industries except for agriculture and public administration. The unit of analysis is the establishment. Throughout the paper, we refer to it interchangeably as either “establishment,” “firm,” or “workplace.” The WES is a very appealing data set because response rates are consistently high across all panel years, and sample sizes are relatively generous, especially compared to other firm-level data. Most importantly, the panel nature of the WES allows us to remove the unobserved firm-specific fixed effects in the empirical analysis.
Download
PDF Ebook The Impact of Aggregate and Sectoral Fluctuations on Training Decisions
