Studies of labor market returns to postsecondary credentials have typically relied on Mincerian, difference-in-difference, or individual fixed-effects approaches, each of which assess post-degree wage returns at a single point in time, rather than assessing dynamic growth across time. However, when estimating returns to sub-baccalaureate credentials (such as certificate and associate degrees), it may be particularly important to understand the shape of students’ individual wage trajectories across time, given that most students are already employed before they begin college, and many continue to work while enrolled.

In this paper, we examine the wage trajectories of Virginia community college students, using unit-record transcript data from all 23 community colleges matched with quarterly Unemployment Insurance records from several states. Each student is tracked for at least seven years, including pre-enrollment, during-college, and post-credential wage observations. To explore wage trajectories as a growth process, we employ a multilevel growth curve modeling approach (MGCM). Under MGCM, individual students’ repeated measures of wage are modeled as a function of time, allowing for individual heterogeneity in growth trajectories and an examination of  how these trajectories differ according to student characteristics (such as gender, race, age, or type of credential). To examine how trajectories shift as students enter and exit college, we employ piecewise MGCM, which allows each student’s growth estimate to vary across pre-enrollment, during-enrollment, and post-credential periods. We compare the wage trajectories of students who earned short-term certificates, long-term certificates, associate degrees, or transferred to earn a bachelor’s degree, compared to a base group of students who enrolled but later dropped out.

Consistent with previous research, we find stronger returns to longer-term degrees. However, the growth curve analysis approach reveals several important methodological considerations. First, credentials vary in the extent of their immediate wage jump, as well as in the extent to which they bend the wage growth curve upward.  Thus, conclusions regarding which credentials have higher returns will vary according to a researcher’s selection of follow-up timeframe. Second, different types of students vary strongly in both their pre-college and during-college wage curves, indicating not only that students experience heterogeneous opportunity costs for college attendance, but also that the key “parallel growth” assumption underlying the individual fixed effects model may not hold. Third, the pre-college earnings of younger students seem to underrepresent their actual earning potential; as a result, an individual fixed effects approach would tend to exaggerate the returns to schooling for these students. Applying these insights, we discuss how standard methodological approaches to the estimation of labor market returns could be adjusted to allow for more accurate estimates.