For several years journalists at leading publications have reported that enrolling in law school is a bad investment for many and perhaps even most prospective students because of declining starting salaries in the legal industry and data related to the employment status of law grads nine months after graduation.
In our paper, The Economic Value of a Law Degree, we use earnings data from the U.S. Census to investigate the increase in earnings associated with a law degree compared to a bachelor’s degree. Our results suggest that the press mistook a generally weak labor market and high unemployment among the young for a law-specific problem. We find that at current tuition levels and student loan interest rates, the median and even 25th percentile earnings premiums confer advantages that exceed the cost of a law degree by a wide margin. After controlling for observable ability sorting, we find that a law degree is associated with a 60 percent median increase in monthly earnings.
The mean annual earnings premium of a law degree is approximately $53,300 in 2012 dollars. Even toward the lower end, the annual premium is still $17,300 over a bachelor’s degree. For most law school graduates, the net present value of a law degree over a lifetime typically exceeds its cost by hundreds of thousands of dollars.
In a column published on July 25, 2013, Am Law Daily contributor Matt Leichter repeated many misrepresentations of our research that originally appeared on Above the Law—even after Above the Law posted corrections to its article and after we refuted many of these misrepresentations. A second Am Law Daily contributor, Steven Harper, made many of the same mistakes in a column published the following day, and threw in a few disparaging remarks to boot. We’ve discussed many of these errors in a separate piece, but the clearest refutation would be to simply read the paper.
Here are a few—though, by no means all—of Leichter’s misrepresentations. He claims, for instance, that we “assume law school pays off equally for non-lawyers” when we do not make assumptions; rather, we measure the earnings premium regardless of occupation. He claims that we ignore distributional data when we report it. And he largely ignores our extensive discussion of ability sorting in Section II.I. of The Economic Value of a Law Degree, while claiming that we did not consider these issues.
Harper echoes Washington University Professor Brian Tamanaha’s claim— which The Washington Post reported as false—that we only look at means and do not consider different points in the distribution. Harper also incorrectly reports that “a [law] degree returns at most a lifetime average of $687 a month.” In reality, the average (mean) is in fact around $53,000 per year before taxes, while the median is around $32,000 per year in real terms (after taking inflation into account). After taxes, the annual average benefit is greater than $37,000 per year. Harper also incorrectly claims that our findings of a premium depend on specific assumptions about tuition rates or grad school earnings when—as we explicitly note in the paper—our findings are robust and would not substantially change with even large changes in those baseline assumptions. We tried to help Harper by pointing out his errors so that he could correct them, but a few days later he repeated them verbatim.
Harper and Leichter raised some other interesting issues:
Occupation and the Versatile Law Degree
A substantial proportion of law degree holders do not work as lawyers. We discuss this in Section I.A of our paper and online. Leichter and others have claimed that this invalidates our results or that our results are only applicable to lawyers. These critics are incorrect. Across different education fields, many graduates pursue careers that are not directly related to their field of study: engineering graduates do not all become engineers and history graduates do not all become historians. Studies by labor economists have found that increased earnings from education generally extend across multiple occupations, and limiting an earnings premium analysis to a specific occupation is therefore inappropriate. We look at the distribution of earnings for all those who graduate with a law degree, no matter their later career path. If the law degree does not help them earn more in a path other than legal practice, our estimates incorporate this. As we show in the paper, the law degree premium appears to be substantial even for those towards the bottom, and even including many law graduates who do not work as lawyers.
Causation
How can we estimate the causal effects of law school when we can’t observe the counterfactual of how a specific individual would have fared had he or she not gone to law school? As we discuss at some length in our paper in section II.I, (and online), we endeavor to statistically construct an apples-to-apples comparison.
If those who choose to attend law school were more capable than the typical undergraduate, then some of their relatively high earnings after law school might be due to their greater competence before entering law school and not to benefits caused by attending law school. This would bias our estimate up, making it too high.
We took these concerns very seriously from the start. We investigated a variety of ways in which law school graduates might differ from those with just bachelor’s degrees, including differences in college major and GPA, test scores, career, and educational goals, subjective expectations about future earnings, college quality, and parental socioeconomic status. We find that a disproportionate number of law students majored in subject areas with low earning potential, but otherwise have characteristics that predict above average earnings. We estimate the combined effect of these differences on expected earnings and find it to be very small. Although law students have a higher college GPA (about 0.6 standard deviations higher than bachelor’s degree holders), because of their college majors, this predicts only a small (3.3%) increase in earnings. Test scores predict an even smaller boost, and much of that boost is already predicted by GPA.
Any ability biases remaining after our controls may be offset by an equally important source of bias that Harper, Leichter, and many other critics have ignored: less educated, lower income households systematically over-report earnings and more educated, higher income ones under-report. Less-educated survey respondents tend to forget periods of unemployment, while more- educated households tend to forget end-of-year bonuses and SIPP caps maximum reported earnings to preserve confidentiality. This has been documented in SIPP surveys and it biases our results downward, making them too low, since the comparison group of bachelor’s recipients is systematically lower income than the law graduates.
Distribution of Earnings
Harper mistakenly suggests that our research is invalid because it focuses on the mean, but we discuss the distribution of outcomes extensively throughout the paper (starting in Section II.G) and online. Using the same controls we use at the mean, we consider those toward the bottom by computing the earnings premium if one were to earn at the 25th annual earnings percentile throughout one’s life. In the real world, those who are at the 25th percentile of earnings in one year are often having an exceptionally bad year, and will move closer to the median in subsequent years, so what we call the “25th percentile” is actually a more extreme negative outcome for lifetime earnings.
Even then, we find evidence that a law school education yields a substantial earnings premium over the course of a lifetime. We consider the different amounts of tuition one might pay or different discount rates one might apply to the future earnings. We discuss taxes, student loan repayment, prepayment, and forgiveness. We compute premia by decade so that those who retire early, or plan on retiring early, can see how much of the benefit will still apply to them.
Harper also claims that a bimodal distribution of earnings—that is, a distribution with two peaks—would undermine our analysis. This is incorrect. Regardless of the shape of a distribution, we can estimate the earnings premium at the 25th, 50th, and 75th percentiles.
What If This Time Is Different?
Our census data cover the period from 1996 to 2011. Because of the survey design, the youngest cohort in our sample graduated in 2008, in the middle of the most recent recession. As we discuss in section IV of our paper, we look forward to future research investigating whether more recent cohorts of law graduates will see a downturn in their earnings relative to earnings for similar bachelor’s degree holders. Always keep in mind that any future wage effect that applies equally to bachelor’s and law degree holders will not affect the earnings differential between the two. This relative advantage, and not absolute earnings, determines the value of the law degree.
Everyone will experience recessions during their lifetimes. The more important question is if the law degree premium will decline, increase, or remain the same over time. We know from the data that there has not been much of a long-term trend in the 16 years from 1996 to 2011, though the premium does fluctuate from year to year. Some have suggested that the last few years saw an unprecedented collapse in the employment market for law degree holders. And indeed, the broader U.S. labor market has seen declines in recent years. However, in relative terms, the law degree earnings premium, which incorporates unemployment risk, appears to be close to average. That is, many people are suffering, but law degree holders have maintained their relative advantage.
