Compensation and employment impact of a full-time MBA program

Human capital theory suggests that investment in human capital positively affects earnings through higher productivity. Additionally, human capital investment positively impacts employability. There is a great degree of controversy in the literature on whether the MBA degree is of economic value.  Evidence brought forth by prior research on this subject is either mixed or suggests that there is little evidence that obtaining the MBA degree positively impacts salary or career attainment. These results are of concern in today’s competitive market place where MBA graduates are expected to experience frequent job changes or even career shifts.  Additionally, given the considerable direct and indirect costs of an MBA degree and the recent economic slowdown, a better understanding of factors and their influence on the rate of return to MBA education is in order.

This paper examines the returns to an MBA degree for a sample of full time two-year MBA students who completed their degree at a small, private MBA program.  We merge three confidential datasets which contain a robust set of personal characteristics. Our paper makes several unique contributions to both the topics under study and the methodology. Instead of the survey of registrants for the Graduate Management Admission Test often used in the literature, we use the confidential data of MBA graduates which includes details that are unobservable in other data sets. This information enables us to correct omitted variable bias to a greater extent than previous studies. In addition to wage rates, we analyze wage and bonus which gives a more complete and fair representation of compensation for MBA graduates.

We use a two-part model (2PM) to predict how different factors influence MBA compensation. Because of career change and unemployment for some graduates and the skewed distribution of compensation, 2PM is an appropriate analytical framework. In the first part, we use a multivariate probit model to predict the probability of employment with or without career change. In the second part, we use a Gamma GLM with log link to estimate the amount of compensation. We use Park tests and Hosmer–Lemeshow tests to confirm our choice of models is appropriate. Even with rich personal information, our analysis can still be affected by omitted variable bias, selection bias and measurement errors. To address these issues, we use nonlinear instrumental variables techniques in both parts of the model.