The ABA’s just-released consolidated dataset on law school employment outcomes presents nice opportunities for data analysis. Unlike Bernie Burk, I’m not particularly interested in the relationship between bridge positions and USNWR rank: that seemed overdetermined to me. (Bernie is also, unfortunately, using a noisy measure for resources. Why not simply use the ABA-data on expenditures per student to predict bridge positions? Here’s a hypothesis: the correlation will be much higher than USNWR rank.) In any event, I imagine that everyone will be using these data to look at school-specific outcomes. Let’s do something different.
Much more interesting are columns BK-BQ, which detail where (geographically) students are placed. What can we learn?
Almost all American Law Schools are Basically Homers. The ABA asks schools to identify the state where the highest number of their graduates are employed on graduation. By my hasty count, there are eight schools in the country where that first state is not the state where the school sits. Duke (graduates go to D.C. or NYC); Harvard (graduates to go D.C. or N.Y.C.); Michigan (graduates go to N.Y.C. and California), UVA (graduates go to D.C. or N.Y.C.); Western New-England (graduates go to CT); Widener-Delaware (graduates go to Pennsylvania); Penn (graduates go to N.Y.C.); and Yale (graduates go to N.Y.C. or D.C.). That is, deciding where to go to law school goes a long way to picking the State where you will live after graduation. What schools (apart from those just listed) produce the least number of graduates employed at home? Vermont, Appalachian, Notre Dame, Vandy, Ave Maria, New Hampshire, Washington University, and Cooley. Note how this list mixes schools with very bad job outcomes (i.e., a small percent of their class is employed in the home state because a small percent of the class is employed) and those with very good outcomes (i.e., a small percent of their class is employed in the home state because many are employed elsewhere).
Which States Are the Most Popular Runner-Ups? Schools are also asked to identify the second and third most common destinations for their students. New York, D.C., California, Illinois and New Jersey top the runner-up list. New Jersey and D.C. are impressive, as they are the second choice as almost as many students as they are the first choice of others. Or to put it differently: D.C. and New Jersey receive disproportionately more law students than other states, as a percentage of the national employment market. The third-runner-up market is similar, though Virginia and Massachusetts make an appearance in the list. (As does Alaska, which is the primary and secondary destination category of exactly zero American law schools, but the tertiary destination of two.)
Which States Are the Most “Oversupplied” With Law Graduates? Given these data, I assume that for most law students, job seeking begins at home. That enables me to ask: which states have the worst environments for incoming lawyers. I estimate this answer by dividing total graduates per state by total jobs in each state (itself a product of adding together the first, second and third “choices” that law schools provide). I know that there are problems with this calculation, even assuming audited data. For example:
- It ignores missing data on location, and schools where there a large number of graduates going to the “fourth” largest state destination;
- It assumes that schools have correctly coded location data;
- It assigns schools like Yale to a State (CT) where they do not in fact send the majority of their graduates.
We know that this missing and skewed data makes a difference. Schools reported that ~37K of ~43.5K, or 86%, of law students were employed in some capacity in this dataset. But my location-based analysis finds state-specific jobs for only ~29K law students, or 66%. This isn’t malfeasance by schools, and it isn’t evidence of conspiracy. Schools are required to report employment status for all graduates, but employment location for only the top three states. (Tracking students in this way is, after all, expensive.) Nonetheless, we can learn something interesting from these data in aggregate. Below the jump, I’m going to discuss the distribution of these geographically identified jobs.