General and Specific Training.

Evidence and Implications

ABSTRACT

Using data from the Employer Opportunity Pilot Project (EOPP) survey and the National Longitudinal Survey of Youth (NLSY), we explicitly document the specificity and generality of employer-provided training, and we analyze how wage growth and mobility are influenced by our direct measures of specific and general training. In spite of the emphasis that labor economists have placed on specific training, we find that employers in the EOPP and workers in the NLSY indicate that most of the skills learned in training are useful elsewhere. Our results are consistent with several recent models that predict that employers will often extract some of the returns to the general training they provide.

I. Introduction

In his seminal article on investment in human capital, Becker (1962) defined completely general training as a human capital investment that raises a worker's productivity at other employers to the same extent as at the employer that provides the training. Similarly, completely specific training is defined as a human capital investment that increases productivity only at the employer that provides the training. Recent empirical work in the human capital literature has attempted to proxy for specific versus general training by looking at the different effects of on-the-job versus off-the-job training (Lynch 1991, 1992), or by looking at the different effects of company versus school training (Loewenstein and Spletzer 1997, 1998a). In this paper, we document the degree to which employers and workers believe that the skills learned in employer-provided training are useful at other firms, we analyze the extent to which employers are able to substitute relevant experience elsewhere for their own training, and we e xamine how wage growth and mobility are influenced by direct measures of the generality of training.

The empirical work in this paper significantly expands on our earlier work on this topic (Loewenstein and Spletzer 1998a). In our earlier work, we used only one data set and were forced to use a proxy for the generality of training. In contrast, this paper uses data from both the Employer Opportunity Pilot Project survey (EOPP) and the 1993 National Longitudinal Survey of Youth (NLSY). Both of these data sets contain direct measures of the generality of training. A strength of the EOPP data is that they can be used to ascertain whether employers are able to substitute previous training elsewhere for their own training, and a strength of the NLSY data is that they are longitudinal and can therefore be used to analyze job mobility and wage growth both within and across jobs. We find, perhaps somewhat surprisingly in light of the emphasis that labor economists have placed on specific training, that the majority of employer-provided training is general.

In Section II we briefly review Becker's on-the-job training model and then discuss how certain modifications of the model can alter the prediction that workers incur the entire cost and realize the entire return to general training. We next summarize the empirical evidence on the effects of training; given the importance of the human capital model and the recent plethora of empirical work regarding on-the job training, the paucity of empirical evidence on the generality of training is quite striking. In Section III we describe the information from the EOPP data and the 1993 NLSY data regarding the generality of workers' training investments. The data suggest quite strongly that most training is general--more than half of the employers and the workers believe that all of the skills learned in training are useful at other employers, and roughly three-quarters indicate that most if not all of the skills learned are useful at other employers. However, these simple descriptive statistics do not by themselves mea n that most employer-provided training is general.

For training to be general, alternative employers must in fact recognize that a worker's training at a previous employer raises his productivity elsewhere. In Section IV we use a unique measure of relevant experience in the EOPP data to examine whether there is any evidence that informational asymmetries in the labor market effectively transform general training into specific training. Our results provide strong evidence that employers not only frequently recognize the value of skills that workers have learned from previous training, but also substitute previous training for their own training when the skills required on the current job are general.

We then return to the predictions of the human capital model and ask how the wage and mobility effects of training depend on the degree to which the training is specific or general. Wage equations estimated from both the EOPP and the NLSY data indicate that skills learned in past jobs are rewarded by a worker's current employer; our mobility equations are also consistent with the hypothesis that much on-the-job training is quite general and provide confirmation of the results in our wage equations. Interestingly, we are unable to find any systematic difference in the wage returns to specific and general training provided by the current employer. We believe that this finding can be explained not only by a combination of measurement error in the data and questions that do not measure precisely the economic concepts of specificity and generality, but also by the fact that employers often share the costs and returns to general as well as specific training.

II. Previous Research on Specific and General Training

Because general training raises a worker's future productivity at other employers, Becker (1962) argues that the employer providing completely general training will not be able to capture any of the future returns. Therefore, the employer will not be willing to pay any of the costs. In contrast, an employer can obtain the returns to specific training and is thus willing to share the cost. Indeed, Becker notes that many divisions of the returns and costs to specific training are possible and that the optimal division will depend on the wage responsiveness of the quit and layoff rates. Naturally, since the returns to specific training are lost when an employer-worker match dissolves, specific training should unambiguously be associated with lower turnover.

Recently, several authors have questioned Becker's conclusion that workers receive all the returns and incur all the costs to general training. Bishop and Kang (1996) have shown that the existence of a liquidity constraint on the part of the worker can lead to the employer sharing the costs and returns to general training. [1] In our earlier work (Loewenstein and Spletzer 1998a), we argued that contract enforcement considerations can lead to employers sharing the returns and costs to purely general training. [2] These recent models provide a helpful perspective from which to interpret the empirical results in this paper: if employers do in fact share the costs and returns to general training, then the within-job wage profiles of workers receiving general training may not look very different from the within-job wage profiles of workers receiving specific training. In this case, it may be necessary to infer training's generality by looking at whether the training hinders mobility and whether alternative employ ers value the skills learned from the training.

As a result of the paucity of data, researchers in the empirical training literature have been forced to proxy for the differences between general and specific training when analyzing wage growth and mobility. Lynch (1992) finds that on-the-job training raises wages at the current employer but not at future employers, whereas off-the-job training raises wages at future employers but not at the current employer. In a similar vein, we find in our earlier work (Loewenstein and Spletzer 1998a) that training raises future wages more for workers who switch employers than for workers who remain with the employer initially providing the training. This differential return is especially high for training received at a vocational institute or business school or in the form of seminars outside of work. We note that these findings are consistent with employers sharing in the returns to general training. [3]

With regard to mobility, Lynch (1991) finds that individuals with on-the-job training are less likely to leave their current employer, while individuals with off-the-job training are more likely to leave their current employer. Loewenstein and Spletzer (1997) find that individuals with company training are less likely to leave their job, whereas individuals with school training have mobility patterns similar to those with no training. If on-the-job training is more specific than off-the-job training as Lynch presumes, and if company training is more specific than school training as Loewenstein and Spletzer presume, then the job mobility results found by these authors are quite supportive of the human capital model.

Instead of relying on proxies, it would be helpful to have explicit information concerning the generality of the training. The recently released 1993 National Longitudinal Survey of Youth data provide us with such a measure. The Employer Opportunity Pilot Project also includes data from a question inquiring about the generality of the training. [4]

III. Is Training General or Specific?

A. Evidence from the NLSY

Data from the National Longitudinal Survey of Youth (NLSY) are perhaps the major source of our current knowledge about formal training. This survey focuses on young persons at the start of their labor market careers, and the longitudinal nature of the survey enables an individual's pattern of training receipt, job mobility, and wage growth over time to be observed. Our sample from the NLSY consists of 9,362 person-year observations from which we can analyze the effects of specific and general training on wages and mobility. [5] The measure of training incidence comes from a worker's response to the question: "Since [date of the last interview], did you attend any training program or any on-the-job training designed to improve job skills, help people find a job, or learn a new job?" In our 1993 sample of 4,814 individuals, approximately 18 percent received training during the previous year. [6]

In 1993 for the first and only time, the NLSY included the question: "How many of the skills that you learned in this training program do you think could be useful in doing the same kind of work for an employer DIFFERENT than [current employer]?" There were five possible responses to this question: all or almost all of the skills, more than half of the skills, about half of the skills, less than half of the skills, none or almost none of the skills. Sixty-three percent of workers receiving employer-provided formal training responded that all or almost all of the skills they learned are useful in doing the same kind of work for a different employer. Another 14 percent of individuals receiving training indicated that more than half of the skills they learned are useful at a different employer, and about 12 percent of the workers indicated that about half the skills they learned are useful at doing the same kind of work at another employer. Only 11 percent of the workers indicated that either less than half or none of the skills are useful at another employer.

When one decomposes the generality of training by the type of training the worker receives, the results are broadly consistent with one's intuition, thereby increasing one's confidence that individuals are able to give reasonable responses to the question about the generality of training. [7] We find that individuals report company training to be the most specific type of training and school training to be the most general, with inside and outside seminars somewhere in between. Our longer working paper contains tables and a more detailed analysis of these crosstabs.

Although the responses to the specific-general training question (and the responses to the type of training question) may well contain a sizable amount of measurement error, the fact that 63 percent of individuals receiving employer-provided training report that all or almost all the skills they are learning are portable across employers suggests quite strongly that a great deal of training has a large general component. This conclusion is reinforced by data from the "informal training" section of the NLSY questionnaire. [8 ]Specifically, in the 1993 sample, 51 …