Earlier this year, Career Education Colleges and Universities (CECU), the main for-profit college lobbying association, offered a proposal that would eliminate all accountability metrics from the gainful employment (GE) regulations promulgated in 2015. In place of the current GE rule, CECU suggested that career colleges disclose income estimates for each of the occupations that a program of study claims to prepare students. To meet this requirement, colleges would likely publish data from the Occupational Employment Statistics survey, which is administered twice a year by the Bureau of Labor Statistics (BLS). With the Department of Education launching its renegotiations of the GE regulations today, now is the time to understand why this approach would be such a bad idea.
In general, the current GE rule measures the amount that each graduate of a career-oriented program is earning three years after entering the workforce, relative to the amount of debt each student took on to cover the cost of the program. Given the below average earnings for graduates and the higher than average cost of attending a for-profit college compared to similar credentials offered at public community colleges, it is no surprise that the for-profit college industry wants to eliminate this debt-to-earnings measure wholesale. Aside from the clear issue of no longer disclosing total student debt levels, having colleges only provide national or regional income estimates isn’t going to cut it. BLS data might be helpful for prospective students who haven’t yet decided what to study, but these data will obfuscate wide variation in earnings for graduates of specific degree programs.
Although a general understanding of how BLS data differs from actual graduates’ incomes should be sufficient to dissuade any attempts to implement CECU’s proposal, comparing the data side-by-side confirms just how ill-suited BLS data would be for measuring gainful employment.
But before digging into the numbers, there are several noteworthy reasons that national, state, or local BLS estimates will differ so markedly from GE earnings. First, while the BLS data present a useful resource in other contexts, they are not intended to reflect the amount students can expect to earn after graduating from a particular school with a particular credential. Unlike the GE data, BLS data include earnings for people with a variety of academic and work experiences. For example, a short-term certificate in business administration from a for-profit college will be associated with the same list of occupations as a master’s in business administration from an elite public or nonprofit university. Furthermore, BLS’s Occupational Employment Statistics make no such distinctions between how long someone has worked in a given industry. Unlike the GE data, which capture income three years after graduation, the Occupational Employment Statistics are derived from a sample of all Americans, regardless of career stage. A graduate of a GE program may not be able to find work in her field of study, much less earn the same amount as someone with many years of experience under her belt.
Even if the BLS data were offering an aspirational view to students of their eventual outcomes later in their career, the argument in favor of these estimates is unconvincing. Most of the students enrolled in career-oriented programs, which predominantly offer lower-earning certificates and two-year degrees, are adults looking for an immediate opportunity to boost their income. Adult students who are already in the workforce, living on their own and possibly supporting a family, deserve to know how their investment will pay off in the first few years after they leave school, and whether or not it’s worth the financial and opportunity costs.
With these factors in mind, the disparities between the BLS data and the data gathered under GE would significantly and consistently inflate the amount many graduates could expect to earn in the job market. Overall, median May 2016 national BLS incomes exceed annual median earnings for GE program graduates by an average of $26,000. And out of 8,412 programs* included in the GE data released last year, close to 97 percent had lower earnings than the BLS data would otherwise suggest.
In the chart below, earnings for 10 of the most common programs in the GE data are compared to BLS earnings for the associated field. In these ten fields of study, BLS earnings universally exceed earnings for GE program graduates not only in middling programs (those at the 50th percentile) but also for programs in the 90th percentile of their field. In other words, even graduates from the top ten percent of programs in each field would be led to believe they could earn much more. This trend holds true in 385 of 457 cases -- about 85 percent of the fields of study represented -- and the disparity is large. Overall, median BLS earnings exceed earnings for the 90th percentile of programs in each field of study by an average of $19,915.
Using BLS data would also suppress major differences between programs of study in the same field. For example, in certain areas like business administration, the median national earnings dwarf the $30,393 that the typical graduate from a GE program can expect to earn immediately out of college by over $59,000. But business administration programs have a much larger spread between the top and bottom earnings of graduates. This spread between different colleges would be eliminated from view under any proposal that publishes BLS data in lieu of actual earnings data. Instead, students from a program in the bottom quartile at which they can realistically expect to make less than $25,000 annually might presume they will make up to $90,000 a year -- the amount the average management analyst, construction manager, and other business-related occupations make collectively.
While CECU's proposal may seem sensible on the surface, BLS data in lieu of information about actual graduates' outcomes will hamstring prospective students in their search for a quality program and will prevent efforts to hold colleges accountable for taxpayer dollars.
*In the original GE data, 8,637 distinct programs were included. 225 of these programs either did not match with an occupation, or the occupation was not reflected in the BLS data.
To derive comparable BLS earnings data, we first matched each program’s Classification of Instructional Programs Code (CIP) with its corresponding Standard Occupational Codes (SOC) using a crosswalk published by the Department of Education. Next, we matched the SOC Codes with the May 2016 National Occupational Employment and Wage Estimates published by BLS. While 8,637 programs were subject to the Gainful Employment regulations beginning in 2015, 225 programs were dropped from this analysis since they either do not match with an established occupation code, or the occupation code with which they are associated is missing from the BLS data--a logistical challenge that would require further study if it were adopted for GE programs.
For fields of study that lead to more than one occupation, we use the average BLS estimates for each of these occupational categories to approximate how this disclosure might work in practice. As an example, the national median income for a dental assistant accompanies the median earnings for dental assisting programs. But for some broader areas of study like culinary arts, BLS earnings reflect the average of several different occupations, including private chefs, restaurant chefs, line cooks and others.