ZEW Discussion Papers
Maintaining (Locus of) Control? Assessing the Impact of Locus of Control on Education Decisions and Wages
Piatek, Rémi and Pia Pinger (2010), Maintaining (Locus of) Control? Assessing the Impact of Locus of Control on Education Decisions and Wages, ZEW Discussion Paper No. 10-093, Mannheim. Download
This paper investigates the impact of an individual’s level of locus of control, a concept commonly used in social psychology (Rotter, 1966), on educational choices and wages. We establish that more internal individuals, i.e., who believe that reinforcement in life comes from their own actions instead of being determined by luck or destiny, earn higher wages. However, the positive effect of a more internal locus of control only translates into labor income via the channel of education: once schooling is controlled for, the impact of locus of control on wages vanishes. Using data from the German Socioeconomic Panel (GSOEP), we address the problem of measurement error by extracting a latent factor reflecting locus of control. In addition, we account for the problem of reverse causality and truncated life-cycle data in that we combine information on both young individuals, who have not yet entered the labor market, and on older, working-age individuals. Our estimation approach follows the work by Heckman et al. (2006b), Hansen et al. (2004) and Carneiro et al. (2003) in that we use Markov chain Monte Carlo methods to simulate the parameters of the model. Specifically, we use a Gibbs sampler with flat priors that sequentially draws the parameters of interest from their respective conditional distributions. Furthermore, we build on a strategy developed in Cunha et al. (2005), which allows us to retrieve the distribution of locus of control from a sample of young individuals, and to estimate its impact on outcomes in a sample of working-age individuals. Producing identification of different parts of the likelihood using different samples, we are able to correct for potential biases that arise due to reverse causality and spurious correlation, and to measure the impact of premarket locus of control on later outcomes.
We find that locus of control is an important predictor of the decision to obtain higher education. Furthermore, we find that premarket locus of control, defined as locus of control measured at the time of schooling—before the individual enters the labor market-does not significantly affect later wages after controlling for education decisions. In light of the existing literature, which finds mostly positive effects of contemporaneous locus of control measures on wages, this indicates that it is important to distinguish between premarket skills and those that are already influenced by labor market experience and age. Last, simulation of our model shows that moving individuals from the first to the last decile of the locus of control distribution significantly shifts the distribution of schooling choices, thus indirectly affecting later wages.
Keywords: locus of control, wages, latent factor model, data set combination