Wednesday, December 7, 2022

“Unemployment” and law school rankings (Michael Simkovic)

One area of disagreement among law schools and between U.S. News and some of the law schools boycotting its rankings turns on whether “unemployment” should be defined according to the standard definition used by the U.S. government, the Department of Labor, the Census Bureau, and the Department of Education.  The government defines unemployment narrowly to apply to those who are both not employed and actively seeking employment.  Those who are not employed and not seeking employment are considered to be “not participating in the labor force” rather than “unemployed.” However, U.S. News defines unemployment more broadly to include anyone who is not currently employed, even if they are not working by choice. 

I’ve explained before why I favor the standard definition of unemployment.  Under the standard approach, law school statistics would be comparable to statistics for other educational programs.  Moreover, law schools would not be penalized for admitting students who wish to pursue dual degrees (i.e., an MPP, MD, MBA or PhD) or advanced degrees like a Tax LLM.  Nor would law schools be penalized for admitted wealthy students who do not wish or need to work or people who would prefer to stay home to care for children while their spouse works.

The only real argument against using the standard definition is that law schools will “game” the system by misrepresenting the intent of people who are not working or avoiding contacting those who they suspect are not working.  But a better solution to this problem of trust is to use a third party audit or third party data collection service rather than law school self-report. 

An audit would increase costs, but it might be worthwhile for law schools to invite the ABA to require a random audit that would include direct contact with some graduates who are not working to verify their status (unemployed vs. not participating in the labor force) in exchange for collecting data using the standard definition of unemployment. Alternatively, the ABA could require law schools to outsource the graduate employment data collection function to a neutral third party.  Schools could save money by reducing the number of internal FTE employees handling this data collection function and just paying a fee to the trusted third party.

If these solutions are not worthwhile to law schools or to ranking organizations, it may be that the total amount of fraud is small, notwithstanding a few very salient examples.  Corporate law scholars are familiar with the empirical finding that enhanced auditing under SOX (in response to Enron, Worldcom, etc.) actually reduced shareholder value, presumably because the aggregate amount of fraud was relatively low under the old, less expensive regime.


*. A related, but more technical problem, is how non-response to the survey is handled.  U.S. News assumes that 100% of graduates who do not respond to the employment survey (or whose employment cannot be verified through another source like an employer website) are unemployed.  By contrast, U.S. statistical agencies use the known characteristics of respondents and non-respondents (race, sex, age, location, etc.) to create weights that apply to each respondent.  These weights enable respondents to stand in for the non-respondents who have similar characteristics.  Thus, for example, if non-respondents characteristics suggest higher rates of unemployment than respondents but less than 100% unemployment, then the weights will lead to an unemployment estimate that is higher than using an unweighted average of respondents, but lower than assuming that 100% of non-respondents are unemployed.  In other words, U.S. News's approach to non-response likely leads to higher unemployment rates for law graduates compared to the rates that would be calculated using the standard weighting methods used by U.S. statistical agencies.

Guest Blogger: Michael Simkovic | Permalink