Has Law Become Stagnant?

My last post provided an overview of my draft article The Cost of Justice at the Dawn of AI and explained the basic logic of Baumol's cost disease for the practice of law. Just as in any other market, if the productivity of lawyers increases at a slower rate than the rest of the economy, legal services will become more expensive. And if a technology like artificial intelligence leads legal productivity to increase at a faster rate than the rest of the economy, then legal services will become cheaper.

In this post, I address the question whether law has stagnated – that is, whether legal services have become relatively more expensive as a result of low productivity increases. Legal actors and the law review literature usually ignore the cost disease. More broadly, legislators and other actors have almost entirely ignored my primary inquiry—how changes in legal productivity and thus costs over time might make the legal system more or less efficient and thus have implications for a wide range of legal doctrines and practices. But a few commentators have considered whether the law suffers from the cost disease, and all unanimously have concluded that it does. I tend to agree, but that the case may be more difficult to prove than they allow.

Bill Henderson, for example, carefully documents that indices of the costs of legal services have risen faster than the consumer price index. Similarly, Emery Lee and John Brooks have argued that Bureau of Labor Statistics data support the conclusion that the legal sector has stagnated. Eric Helland and Alex Tabarrok have argued that the cost disease has afflicted law, as well as "other labor-intensive, high-skill services." Various bloggers have concluded that law suffers cost disease without detailed evidence. With all this agreement, any believer in the Condorcet Jury Theorem would need to assign a fairly high probability to the proposition that law has stagnated.

The problem is that it is very difficult to measure changes in the quality of legal services over time, and quality changes that have occurred thus might not be reflected in the Bureau of Legal Statistics data. Even if billable hour rates have risen more than general inflation, if an hour of legal services accomplishes much more today than in the past, law could be a productive sector. Indeed, over the decades, lawyers have benefited from labor-saving technology such as word processing, online legal databases, and e-discovery software.

Moreover, the legal system itself develops ever more precedent. Maybe the accumulation of precedent means that a higher and higher proportion of problems that arise can be resolved easily. If so, the legal system as a whole is becoming more productive. But it's possible that new legal precedents generally create more issues for lawyers to argue. And even if precedents tend to clarify things, the accumulation of precedent might not keep up with the ever increasing complexity of our legal world.

What are the stakes in the question of whether law is stagnating? Stagnation implies that the machinery of the legal system is becoming ever less capable of affordably making relevant legal distinctions. An increasing number of cases will depend on the dynamics of negotiations in the shadow of the law, and while increasing costs will cause more cases to settle, bargaining outcomes may stray quite far from what a hypothetical inexpensive legal system would require. Meritorious defendants may pay nuisance settlements, meritorious plaintiffs may accept a small amount on the dollar, and variables such as relative risk aversion and status as litigation repeat players may greatly influence amounts paid.

These changes aren't easily measured. Data on how much cases settle for is not always readily available, and even if it were, one could not know what the results of such cases would be at trial. Some have imagined a system of random selection of some cases for trial, but in the absence of such a system, we cannot easily observe how much the high cost of litigation is introducing noise or error into adjudication. And any gradual increase in this noise could easily go unnoticed. In the first installment of this series, I noted that no previous scholar has even linked the "vanishing trial" to the cost disease. This highlights that the small amount of commentary on law and the cost disease has been siloed and that the possibility of stagnation has failed to penetrate the general legal consciousness. In my view, the vanishing trial is the strongest piece of empirical evidence for the cost disease, unless one believes that the law is rapidly becoming much clearer and easier to apply.

The legal system still features some trials, but their shadow is becoming ever larger and ill defined. Continued stagnation would have implications for class actions, arbitration, trial procedure, and the choice between rules and standards. But past trends toward stagnation do not establish that the law will continue to stagnate. In the next installment, I'll consider whether we should expect AI to reverse stagnation, if indeed the law has been stagnating, and after that, I will turn to the question of how the legal system can prepare for the possibility that productivity may move in one direction or the other.