The response of US college enrollment to unexpected changes in macroeconomic activity

Abstract

This paper estimates the extent and magnitude of US college and university enrollment responses to unanticipated changes in macroeconomic activity. In particular, we consider the relationship between enrollment, economic growth, and inflation. A time series analysis known as a vector autoregression is estimated and impulse response functions are calculated to measure the enrollment response to these economic shocks. The results suggest that enrollment patterns differ by gender and are consistent with economic theory.

Keywords:

• college enrollment
• macroeconomics
• time series
• economic growth
• inflation

Acknowledgement

An earlier version of this paper was presented at the 2007 ASHE conference in Louisville, KY. We thank session participants and attendees, as well as two anonymous reviewers, for useful comments.

Notes

1. Data are reported for institutions that were accredited by an agency or association that was recognized by the U.S. Department of Education or recognized directly by the Secretary of Education. While the use of annual data is common in the study of macroeconomics, our analysis is limited to annual observations (e.g., as opposed to semester) due to the availability of statistics on enrollment.

2. Specifically, the change in enrollment is given by the first‐difference of the natural log of enrollment and thus may be referred to as the growth in enrollment.

3. No evidence of a significant time trend in the enrollment growth rates was detected using standard augmented Dickey‐Fuller (ADF) tests with constant and trend, although the growth rates series were found to be stationary. Further, a series of rolling Chow breakpoint tests failed to show any evidence of structural change in the enrollment growth rate series.

4. For example, expected future real income is future nominal income less the expected inflation rate, where educational attainment enhances nominal income. In a simple model, one can illustrate that individuals will choose the level of education to obtain based upon their notion of future purchasing power. Since real future income falls with unexpected inflation, then individuals may choose to acquire additional education to compensate for this risk.

5. Tables 2 and 3 show the results from the estimation of the two VARs. Adjusted R‐squares are 0.63 and 0.79 for male and female enrollment equations in the two VARs, respectively. No evidence of remaining serial correlation was detected using the LM test, additionally, the residuals in the VARs appeared to be normal as indicated by the Jarques‐Berra statistic, kurtosis and skewness. Thus, no more information is contained in the regression residuals that could be used to improve model performance (i.e., the VAR models are correctly specified). Table 4 (Panels A and B) provides the corresponding χ ² test statistics for block exogeneity tests. The null hypothesis being tested is that the lagged values of the dependent variable (i.e., GROWTH or INF) in the enrollment growth equation are jointly insignificant. Rejection of the null (i.e., Prob. value < 0.05) indicates that the lagged values help to predict the current value of the independent variable under consideration (i.e., enrollment growth). Rejection of the null is said to constitute ‘Granger‐causality.’ When considered jointly (i.e., as a block), evidence of Granger‐causality of macroeconomic factors to enrollment is found for both males and females.