How to Read a Licensing Premium Without Overreading It
The Italian bar-exam study answers the title's apparent contradiction: a license can create a real average earnings return without promising every holder a raise. The final judgment of this series is not that licensing is good or bad in general. It is that a licensing claim becomes intelligible only when we connect the credential to the occupation it opens, the alternatives it replaces, the distribution of outcomes after entry, and the market level at which the rule operates. That framework lets us take the evidence seriously without asking a local result to carry a universal conclusion.
The complete chain is access, practice, distribution, and risk
Start with the object being studied. The Italian evidence concerns a government license to practice law, not education in the abstract and not a portable certificate. Candidates have already completed the law degree and apprenticeship. The first-attempt written score then changes their probability of crossing into legal practice. That sequence is the reason the study can estimate an access return: it compares otherwise nearby candidates whose occupational paths are pushed apart by the 90-point threshold.
The first important connection is between causal credibility and the counterfactual. The study does not compare lawyers with people who never studied law. It compares marginal passers with marginal failers, then follows the alternatives open to the same broad population of law graduates. The non-license path includes public-sector and other work; it is not an empty control condition. The estimated return is therefore the difference between two available occupational trajectories for the local compliers, not a universal value attached to the word “license.”
The second connection is between access and practice. Passing the written exam raises the probability of eventually obtaining a license by about 20 percentage points, but the relationship is imperfect because candidates can retake the exam and because some written passers do not complete the oral stage or enter practice. The license changes the chance of entering legal work; it does not make every later choice disappear. That is why the study's economic object is best described as access to a profession rather than ownership of a document.
The third connection is between the average and the distribution. Over years 2 through 23, the estimated earnings gap for the marginal candidates is 19,825 euros per year, or roughly 50 percent relative to the comparison earnings of non-treated compliers. The effect grows during the early career and then stabilizes at about 20,000 euros. At a 5 percent discount rate, the paper calculates a present discounted value of 248,479 euros over the observed horizon. These are substantial results, but they are summaries of potential career paths, not individual guarantees.
The distribution explains why. Licensed and non-licensed graduates have nearly identical earnings below the median, while the distributions diverge in the upper tail. The larger differences begin around the 75th percentile and become much larger at the 90th. The average is therefore pulled upward by highly successful lawyers. Many people who enter legal practice do not experience the headline premium, and the study does not identify which personal or market characteristics place a practitioner at the top.
The fourth connection is between the upper tail and risk. Lawyers' earnings are substantially more volatile than those of law graduates in alternative work, both across people and across years for the same person. The probability of an earnings decline of at least 5 percent is estimated at about 39 percent for treated individuals, compared with about 10 percent for non-treated compliers.