DK begin by emphasizing the enormous difficulties encountered in trying to answer this increasingly important - and much asked- question:
...obtaining unbiased estimates of the return to college quality is difficult due to unobserved characteristics that affect both a student’s attendance at a highly selective college and their later earnings. In particular, the same characteristics (such as ambition) that lead students to apply to highly selective colleges may also be rewarded in the labor market. Likewise, the attributes that admissions officers are looking for when selecting students for college may be similar to the attributes that employers are seeking when hiring and promoting workers.
First of all, where does one get the data that enables one to look at long-term financial return as a function of college quality? DK use a subset of the 1976 and 1989 cohorts of the Mellon Foundations College and Beyond dataset. The C&B dataset contains information on students from 34 institutions, 27 of which agreed to participate in this current study. Surveys of graduates were carried out to supplement and extend previous surveys of this group. From this, DK were able to extract data describing each individual student's gender, race, high school grade point average, student SAT score, parental education and occupation (used to predict parental income) and whether the student was a college athlete. In addition, they had for each student information on the other colleges to which the student applied, and which of those accepted the student. Data on earnings came from the Social Security Administration, which linked data from its Detailed Earnings Records 1981-2007 to individuals in the C&B Dataset for analysis - this was obviously done by researchers at the SSA, such that DK had no access to individual earnings records. Surrogates used for "quality" of the colleges were average student SAT's, Barron's selectivity ratings, and net tuition.
DK use several variations on a cross-sectional least squares regression model to try to tease out relationships between these data:
We assume the equation relating earnings to the students' attributes is:(1) ln Wi = b0+ b1Qi + b2X1i + b3X2i + ei,where Qi is a measure of the selectivity of the college student i attended, X1 and X2 are two sets of characteristics that affect earnings, and ei is an idiosyncratic error term that is uncorrelated with the other explanatory variables (1). X1 includes variables that are observable to researchers, such as grades and SAT scores, while X2 includes variables that are not observable to researchers, such as student motivation and creativity (that are at least partly revealed to admissions officers through detailed transcript information, essays, interviews, and recommendations). Both X1 and X2 affect the set of colleges that students apply to, whether they are admitted, and possibly which school they attend. The parameter b1 represents the monetary payoff to attending a more selective college.
As examples of the variations considered in this study, the individual earnings Wi that appear in this formula can be taken either as earnings during a specific year (e.g. 2007), or as a sequence of 5 year intervals that allow study of temporal changes in return over the graduate's working life; the quality parameter Qi can be the average SAT of the college attended, the Barron's selectivity rating, or the net tuition.
BK refer to the model in which X2 is set to zero as the "basic model" since it depends on data that is easily obtained, and is the model used in many previous studies. Defining X2 is obviously difficult, since it depends on characteristics not accessible to researchers. However BK attempt to get around this problem by using the students themselves:
we use one of the selection-adjusted models—referred to as the self-revelation model- in Dale and Krueger (2002). This model assumes that students signal their potential ability, motivation and ambition by the choice of schools they apply to. In fact, students may have a better sense of their potential ability than college admissions committees. If students with greater unobserved earnings potential are more likely to apply to more selective colleges, the error term in (the equation above with X2=0) could be modeled as a function of the average SAT score of the schools to which the student applied.
DK tease out a number of very interesting relationships using the various variations of the model, and these are well worth looking at.
However, the main impact of the paper comes in its overall conclusion:
Consistent with the past literature, we find a positive and significant effect of the return to college selectivity during a student’s prime working years in regression models that do not adjust for unobserved student quality for cohorts that entered college in 1976 and 1989 using administrative earnings data from the SSA’s Detailed Earnings Records. Based on these same regression specifications, we also find that the return to selectivity increases over the course of a student’s career. However, after we adjust for unobserved student characteristics, the return to college selectivity falls dramatically. For the 1976 cohort, the return to school-SAT score for the full sample is always indistinguishable from zero when we control for the average SAT score of the colleges that students applied to in order to control for omitted student variables. Similarly, the returns to other college characteristics (the Barron’s Index and net tuition) are substantial in the basic model that controls for commonly observed student characteristics but small and never statistically distinguishable from zero in the self-revelation model, which (partially) controls for unobserved student variables.
In other words, if you simply ignore the unobserved characteristics of the student and use only the observed characteristics, you find there is a strong correlation between the quality of the college attended and earnings, with the benefit increasing with time from graduation. However, if you hypothesize that the students' self evaluation is reflected in the schools to which they apply, and use these data to represent the unobserved student characteristics, you get a radically different picture. This latter calculation tells you that the student's earning power reflects the characteristics of the student himself or herself - the influence of the quality of the college attended on earning power becomes statistically insignificant!
There were some subgroups that provided exceptions to these general results. In particular, black and Hispanic students, and students from families with lower educational attainment showed positive financial returns from attending more selective institutions. DK also consider several possible cases of correlations or lack thereof in the variables that could alter their conclusions.
It is obvious that this will not be the last word on this important subject, but this certainly is a very provocative contribution to the discussion.
You can follow this conversation by subscribing to the comment feed for this post.
I agree with you. This type of projects should be encouraged and I think that these type of projects are the projects for the future. . . . .
Posted by: Financial Advisor Jacksonville | August 19, 2011 at 07:33 AM
I think students that attend more selective colleges are going to pursue higher end jobs that pay more initially. This doesn't mean that a student who attends a lower-end college will not make the same income in later years with motivation and life-long goals that push them to do very well in life. We have to look at the demographic component as well as the quality of the student.
Posted by: Patricia | April 28, 2011 at 08:54 PM