The Economic Value of a Law Degree (part 1 of about 5)

In the classic film It’s a Wonderful Life, George Bailey suffers financial hardship, becomes depressed, and wishes he had never been born. As Bailey attempts suicide, a Guardian Angel, Clarence, intervenes. Clarence magically shows Bailey an alternate universe in which Bailey never existed. Clarence helps Bailey realize that although his life may be hard, a world without Bailey would be far worse for those Bailey cares about.

In an ideal world, we could do for law students what Clarence does for Bailey: run the universe twice. In the first version, the law student attends law school. In the second version, he or she follows another path. With perfect knowledge of long-term outcomes, the student could decide which choice leads to the better life.

In the real world, the closest we can come to this ideal is to compare past outcomes for two groups of individuals who are similar to our prospective law student and were substantially similar to each other, until one group obtained law degrees while the other group did not.

This is the approach that Frank McIntyre and I take in The Economic Value of a Law Degree.  Using large samples and detailed earnings data from the U.S. Census Bureau’s Survey of Income and Program Participation, we measure differences in annual earnings, hourly wages, and work hours between those with law degrees and those who end their education with a bachelor’s degree. Because we include those who are unemployed or disabled, our analysis incorporates differences in risk of unemployment.

Our focus on law degree holders’ lifetime earnings premiums is a major advance over previous studies. Previous studies have generally either focused on “lawyers” (which excludes many law degree holders), on generic professional degree holders (which includes many individuals who did not attend law school), or on starting salaries, which typically represent less than 3 percent of lifetime earnings.

We control for many demographic, academic, and socio-economic characteristics other than law school attendance that predict earnings. In a supplemental analysis using data from the National Education Longitudinal Study, we incorporate additional control variables and tests for ability sorting and selection.

We find that after controlling for observable differences, law graduates earn much more than similar bachelor degree holders across the distribution of outcomes.

The figure below shows law degree earnings and wage premiums in percentage terms at different points in the distribution. At the mean, law degree holders typically earn 70 percent more than similar bachelor degree holders, at the median, the difference is 60 percent, and at the 25th percentile, it is 55 percent.

Figure 1: Law degree earnings and hourly wage premiums are high even at the low end of the distribution.

Slide 17 Blog Post 1

Most of the increase in earnings is from higher wages per hour, not longer hours.  In dollar terms, the annual earnings differences are substantial—more than $53,000 at the mean, $32,000 at the median, and $17,000 at the 25th percentile.

Figure 2: The annual earnings premium is large even at the low end of the distribution

Slide 20 Blog Post 1


For those who would like to quickly flip through our results, a PowerPoint presentation is available here.

Part 2 of this series of blog posts will discuss the timing of increased earnings and the implications of financing costs for the lifetime value of the law degree.

Part 3 will discuss the use of historic data to project future earnings, and whether trends suggest that a “structural shift” has eroded the earnings advantage of a law degree in recent years.

Part 4 will discuss the contributions of legal education to the federal budget, and the implications for programs like Income Based Repayment that help spread the risks associated with investment in higher education.

Intermittent posts and subsequent posts will address remaining substantive questions and comments about data and methods.


You may also like...

6 Responses

  1. Derek Tokaz says:

    “We also cannot determine the earnings premium associated with attending a specific law school.”

    Is this not a serious limitation of the study that deserves more attention? Prospective students are not faced with the general prospects of an at large JD — they are concerned with the prospects at a specific school and with a specific cost.

    I fear that some will try a bit of logical sleight of hand by saying “A JD is a good financial decision; this is a JD; therefor this is a good financial decision.”

  2. Ben Barros says:

    Re: Derek’s question, I agree that per school information would be useful, but it would be very complicated to do. You’d not only need to track earnings over time by law school, but the baseline earnings for the candidate pool for each school. This is one of the many things Elie missed in his rant – sure, a Yale J.D. is probably more valuable over time than a Western Michigan J.D., but the baseline earnings potential of a person who gets into Yale is going to be higher than the earnings potential of a person who gets into Western Michigan.

  3. There is likely a reasonable working correlation between lifetime earnings and school rank (i.e., a smaller percentage of JD grads will earn BigLaw salaries at lower ranked schools). So the study does in that sense track this because it looks at income at different percentiles. And it still demonstrates that for a significant majority of JDs law school is a positive NPV proposition.

  4. Lawrence Cunningham says:


    Valuable addition to the discussion! Keep up the good work.

    And welcome to Co-Op; hope you enjoy your visit. Looking forward to further posts.


  5. Did we really need a study to show that law school was a good investment for people who graduated prior to 2008? Was that ever up for debate? Isn’t the debate actually about the relative value of a law degree specifically for people who graduated, oh, 2006/7 and after, and specifically for people considering law school now? If so (and yes, that’s so), shouldn’t it be important for the authors of this paper to insert something like “PAST PERFORMANCE DOES NOT GUARANTEE FUTURE PERFORMANCE” in their analysis? I’d think so, because these statistics (even if they were applicable in a post-2008 legal industry, which they aren’t) have no predictive value at all. Also, the caveat about “unobservable” characteristics is an exception that swallows the whale. (I mixed that metaphor myself.) It is absolutely impossible to control for variables between populations with similar undergrad courses of study but different post-bacc paths.

  6. Tyler,

    If you review the paper and power points I think you will agree that the authors take your issues into account to the extent the data is available.

    The substantial positive earnings premium they find is over the lifetime of a JD holder. Further it is cyclical. The cycle went down – at a rate less steep than in the wake of the dot com crash – and has now turned back up, as it has over the past two decades.

    That upturn in the cycle is not likely to be materially impacted by lower earnings of recent graduates even if those are lower than has historically been the case. The weight of their lower earnings in the first few years after graduation is not enough to turn that positive up turn negative, although I would defer to the authors to confirm this conclusion.

    The other issue you raise is the question of the predictive power of past experience. It is true for example in the financial markets that hedge funds have blown themselves up hoping for a reversion to the mean when in fact something new is happening. However, again deferring to the authors to weigh in, my understanding of their analysis of this issue is that the legal market has shown remarkable stability and has features that are very resistant to change when it comes to human capital.

    I study the economic institutions of financial markets, particularly market microstructure, where assumptions are being made about the impact of CEO statements, for example, on stock prices at the millisecond level. There, indeed, there is real danger in making the kinds of assumptions about the past as prologue. But I think the claim is that the legal market is far stickier.