How the Transparency Movement Reinflated the Law School Bubble

Of course I’m click-baiting you! But in place of the vicious criticism you were expecting, you shall receive bitter irony instead. Frankly, I think you’re coming out ahead, so be thankful, you ingrate!

So why did I flag you down?

It appears the Bureau of Labor Statistics is changing its employment projections methodology, specifically its measure of how many workers will be replaced in occupations in its 10-year projection periods—as opposed to the number of positions that the economy will create. Apparently this is a project the BLS has been engaged in for a while, and the comment period is over, so why I didn’t know about it before now escapes and saddens me.

The BLS’s employment projections have long been a go-to source for law school critics. The ~24,000 projected annual lawyer job growth rates they showed every two years contrasted excellently with the ~40,000 law graduates each year (and the even greater number of bar admits). No longer.

Background: Developed in the early 1990s, the BLS’s occupational replacement methodology uses a simple age cohort analysis. For instance if there are fewer employed lawyers in the 55-59 cohort today than there were in the 50-54 cohort five years ago, then you have a rough number of how many people in that age group left the occupation. Do that for all the adjacent cohorts and add together all the negative net changes, and you have the replacement rate. The math behind it is a little bit more complicated and there are some exceptions, like if the occupation is projected to decline overall, but that’s the basic concept.

But the BLS isn’t satisfied with this methodology any more. It suffers from sample bias for occupations with small numbers, and it leans on the assumption that it’s mostly young workers who replace older ones. The bureau is interested in finding the “actual” replacement rate, i.e. one that includes workers transferring to other occupations or leaving the workforce altogether who are concurrently replaced, not just retirees, whom the current methodology tends to capture. This way the projections will include everyone who switches jobs, e.g. fast-food workers for retailer clerks and vice versa, when such changes would otherwise net out under the current methodology. The new methodology is based on Current Population Survey data and regression analysis (which always turns out well) rather than historical trends.

As evidence that the new methodology achieves its purpose of finding more replaced workers where the current one does not, the BLS points to … lawyers because there are external data on employment rates. I’m totally not kidding. It writes (and I editorialize in brackets):

Not all law school graduates become lawyers, but the American Bar Association (ABA) conducts a census of employment outcomes for all law school graduates in order to count the number who find employment in positions that require bar passage (effectively, lawyers). Since ABA began collecting this data in 2011 [Not correct, see below], the number of graduates finding employment in such positions has averaged 29,000 per year. Because some graduates who don’t immediately find such positions may become lawyers later in their career (for example, many graduate become law clerks, a position that does not require bar passage, for a few years before becoming lawyers) [Citation please?], this number [the 29,000 graduates—it’s unclear] should be less than the total number of new entrants into the occupation.

Under the current method, BLS projects an average of 19,650 job openings per year, while the new method projects 41,460 openings per year [!!!!]. Again, no direct comparison between the ABA number and the BLS numbers is possible due to conceptual differences [which, of course, does not rule out indirect comparisons], but the results under the current method are significantly below the actual number of new graduates finding work in the occupation [!]. The new method projects a higher number of openings, which allows for additional entrants not immediately after completion of a law degree.

Okay, data on law graduate unemployment has actually been around for many years, e.g. the NALP and the Official Guide, crude though it was. I’ve written about the strong correlation between falling proportions of graduates finding bar-passage-required jobs and graduates taking JD-advantage jobs or not finding any work. This is evidence of a saturated lawyer market, even if it’s caused in part by slack aggregate demand.

The BLS could also look at lawyer-licensing rates courtesy of the National Conference of Bar Examiners, which it probably should be doing instead of law graduates. So when the BLS says the data are only now available, it’s not doing its homework.

However, the irony—and this is really incredible—is that all those demands for transparency in the employment data, after accusations of misrepresentation and deceit, have perversely led the government to (indirectly) compare the number of graduates in bar-passage-required jobs to its current estimates and use it as evidence that those graduates must be finding bar-passage-required jobs long after graduation.

As arguments from incredulity go, this is a pretty good one. As usual, there are other fallacies.

For one, the BLS is assuming that all occupation changes are positive sum. Everyone who leaves law practice is making the best choice among alternatives (ultimately), so too does everyone who chooses to become lawyers. Thus, departing lawyers need to be replaced. The new methodology automatically rejects the possibility that new entrants force out existing ones and that if more people chose better alternative occupations to law, then fewer lawyers would exit, and everyone would benefit. (Except law professors.) Now, whenever someone leaves the law, there is by definition a shortage, a misallocation of human capital that can only be met by sending more people to law school.

…Especially in light of the eye-exploding 41,460 annual job growth rate, courtesy of the BLS’s new, inscrutable regression approach. It’s certain that some number of lawyers enter practice long after graduation, but assuming 29,000 grads get bar-passage-required jobs, that leaves another 12,500 lawyer jobs each year that must go to earlier graduates despite the swelling numbers of JD-advantaged, unemployed, and other grads who aren’t absorbed earlier.

This leads to an unbelievable replacement rate under the new methodology: If 834,700 projected lawyer jobs in 2022 less 759,800 lawyer positions in 2012 yields 74,900 jobs due to growth, then the cumulative replacement rate (74,900 – (41,460*10 years)) is 339,700 lawyer positions that will “need” replacement over the next decade. If the legal profession has been going through a 44 percent 10-year replacement, then there should never have been a backlog to begin with, and it’s something we should have heard about by now. By contrast the current model shows only a 16 percent 10-year replacement rate.

There’re a few other reasons why the methodology change isn’t a good idea, like aging lawyers, but this post isn’t about that. Rather, it’s a rebuke to everyone who crusaded for transparent employment data based on the rational, debt-guzzling law student assumption. Thanks to them the law schools will soon be saying that the graduate-to-annual-job ratio is (indirectly) in equilibrium right now. The demand for lawyers is there, they’ll say, just after an undetermined period of crippling malemployment … and at a time in their careers when no one is measuring it … except for those After the JD people who found that 24 percent of bar-passers weren’t practicing after 12 years.

Cheekiness aside, it’s likely the BLS (and state governments) will change their projections methodologies accordingly despite law being an unrepresentative occupation with substantial early-career turnover. Be prepared for the dark age of lawyer employment projections.