Guest Post by Thomas Hanes: Plaintiffside Litigation

Eli’s note: This is a repsonse by Thomas Hanes to the post on plaintiffside litigation financing that I wrote a few weeks ago.

In Short

There is not a principal-agent problem with respect to plaintiffside firms that is analogous to the principal-agent problem arising from stockholders running corporations during bankruptcy.

The key mistake made by my esteemed, illustrious, and even handsome colleague is his failure to see that costs of litigation do fall on plaintiffside lawyers: those lawyers have to do the painful work of bringing the litigation itself. For that reason, the litigation system should function well: so long as the rules of evidence are not unreasonably weighted to favor one side, the deadweight loss of litigation is split pretty equally between the plaintiffs (who must do work to prove their claim) and defendants (who must do work to disprove it).

So long as costs are split evenly, then most claims should end in settlement: the defendant should make a payment to the plaintiffs of the expected value of the plaintiff’s claim, offered in exchange for peace in the valley. That expected value should be, generally speaking, the value that encourages efficient behavior by deterring negligent actions. There is no problem here.

My Esteemed Colleague’s key error is this: he thinks that the payment to the plaintiffs is deadweight loss, but it isn’t. It’s a transfer. The only deadweight loss is the expense/hassle of the litigation itself, and plaintiffside firms (and litigation financiers) do pay that deadweight loss - the litigation financier has to pay the plaintiff lawyers to spend hours of work preparing their claim.

So, the tort system should work to promote efficiency. At least, any inefficiency would be a failure of the underlying law, not the financing structure of plaintiffside litigation.

If the procedural law is not unfairly weighted to one side, and the underlying substantive law is sensible, then the transfer of the expected value of one’s claim should be a good thing: the law had given everyone a package of rights that incentivized them toward good behavior (or upheld justice or whatever), and thus the transfer mandated by the court (or agreed to in settlement) upholds a system that makes life go along swimmingly. The transfer is not deadweight loss.

So the bad thing that we should worry about is not payments to plaintiffs, but litigation expenses. But for litigation expenses, the “capped downside, unlimited downside” problem is absent. Plaintiffside lawyers do not have a capped downside with respect to litigation expenses.

Therefore, the litigation-finance system should not be over-incentivizing plaintiffside litigation. In fact, it under-incentivizes it - for reasons I will explain below.

The Learned Hand Rule and its Discontentments

To see why the plaintiffside litigation system under-incentivizes litigation, let’s examine how things ideally work under the Learned Hand rule: a stylization/idealization/approximation of the concept of negligence that everyone is taught in their one-l torts class.

Under Learned Hand, an activity is negligent if, on the margin, it creates ex ante risks of harm that exceed its ex ante prospects of benefits. Thus, whenever someone is harmed by negligent action, they can sue for the damages resulting from that activity, and thus any negative-externalities of actors are internalized, and actors will only undergo actions that have a positive expected value. The tort system under Learned Hand imposes a general Pigouvian tax (a tax on all negative externalities in the amount of the negative externality), and everyone acts efficiently. That’s the idea of Learned Hand.

Now, things become more complicated once you introduce legal uncertainty and litigation costs.

Suppose there is an activity that has some benefits and some costs. There’s no black-letter rule as to which of those costs and benefits are legally cognizable.¹ Thus a court might hold that the activity is negligent (and thus wish to deter it) while it might not.

Because of legal uncertainty, some activities might be deterred even if the court wouldn’t actually want to deter it. Assuming plausible risk preferences (risk aversion, but not of the rawlsian kind), legal uncertainty is likely to deter activity somewhat more than the activity would be deterred by the law people thought would probably solidify eventually. People avoid risks.

So, generally speaking, legal uncertainty means that activity will be deterred more than it should be deterred in order to encourage economic efficiency. The Learned Hand rule can go wrong.

Furthermore, there is another overdeterring bias baked into the system, even when law is certain. The bias results from factual uncertainty.

Note that, once the cognizeable harms and benefits are established under law, factfinders (typically juries) have to assess the magnitudes of those benefits and costs. Juries have famous biases - in particular, they tend to punish defendants for being big corporations, and they do not like entities that engaged in a formal cost-benefit analysis that accept a certain degree of… mortality risk.

But even if juries are unbiased, there is a bias worked into the Learned Hand rule itself - the Learned Hand rule biases actors against actions with big impacts.

Even if juries are reasonable, there will always be a fuzzy line between negligent behavior and responsible commercial action. Thus, there will always be risks of litigation. And, while the Learned Hand Rule internalizes negative externalities (harms to others) it does not internalize positive externalities (benefits to others). Thus, if you invent a new business practice that lowers congestion, that helps various people all through the world, but you cannot sue them for that benefit. Whereas, if you invent a new practice that worsens congestion (or facilitates crimes on the other side of the world), you can be sued. So, under Learned Hand, your negative externalities are internalized, but your positive externalities are not, which means it’s better to engage in a practice that affects no one than a practice that affects everyone (with if those effects are an even mix of positive and negative impacts).

So, the Learned Hand rule encourages small, conservative actions, and discourages large, impactful ones.

Now, the Learned Hand rules tries to cabin this bias by only allowing suit for actions that are on net bad for society (i.e, negligent). But factfinders are a fickle bunch - you can never be quite certain which actions they will disapprove of. So, it’s best to err on the side of conservatism in one’s business practices.

Thus, through several biases (legal uncertainty & and the Learned Hand Rule’s bias toward small-impacts) the basic structure of the tort system can overly deter activity. There are other biases that push in the opposite direction (most importantly: the requirement that damages be sufficiently concrete and certain) but limitations on damages are only so powerful, given juries’ tendency to ignore them.

The Further Quirk of Settlement

Beyond biases inherent to Learned Hand, the quirks of settlement negotiations might divorce the realities of tort litigation from the economist’s ideal.

What is settlement? Why does it happen? Owen Fiss would tell you settlement is cowardice. That is a bit silly. Litigation is like war; settlement is diplomacy. Trials should only happen when negotiations break down. Litigation costs are deadweight loss, and successful negotiations will avoid it.

Think about it this way: a legal claim of uncertain validity is a big pot of money. There are two claimants to that money. Thus, there is some expected value which is the ex ante predicted amount that the plaintiff would get by going to court. And there is some expected amount of the pot that defendants would, on average, retain.

So the two claimants can either split up that pot and go their separate ways (this is called settlement) or one of them can go to court to try to get the whole thing. I will call the predicted amount that the plaintiff will get (ex ante) is EV(P): the size of the pot multiplied by the probability that the plaintiff win.

But here’s the thing - if the plaintiff pursues their claim in court, the plaintiff doesn’t really get EV(P). They get EV(P) minus the cost of hiring lawyers (and minus court fees, the cost of hiring experts, etc.).

So, it would be better for all parties concerned if the defendant handed the plaintiff EV(P) before any litigation costs were incurred, and then the two parties signed a settlement agreement and called it a day.

But that won’t quite happen, because the litigation costs introduce their own pot of money that can then be fought over.

You see:

EV(P) is the amount the plaintiff could expect, ex ante, to win in court.
EV(P) - {expected plaintiff litigation costs} is the net amount they would expect, considering litigation costs.
EV(D) is the amount the defendant would expect to get in court.
EV(D) - {expected defense litigation costs} is the net amount they would expect.

So, a plaintiff is better off getting a settlement of any amount greater than [EV(P) - {expected plaintiff litigation costs}] rather than going to court.

But the defendant would also be better off settling for any amount greater than [EV(D) - {expected defense litigation costs}] than going to court.

So what? Well, that means that {expected defense litigation costs} + {expected plaintiff litigation costs} is a pot that has to be split in negotiation. As long as each party gets more than the net-expected benefit of litigation, they would prefer to settle rather than litigate. So… who gets more than that?

Something something negotiation. The way that negotiation splits pots is more a matter of psychology than economics; there’s no way for a mild autist like myself to predict in advance how pots will be split.² Negotiation is a crypstic labyrinth of passions where logic cannot tread.

So, we cannot say conclusively that the dynamics of settlement negotiations favor of defendants or harm them.

But we can be certain of something: the dynamics of negotiation bias companies toward not engaging in activity that might lead to litigation later. To get that second pot, firms want to credibly commit to an aggressive negotiating position. Credibly committing to a position has costs.³ So, the dynamics of litigation mean that {defense litigation costs} cannot be reduced to zero. There are pre-litigation costs (the costs of credibly committing to one’s position in later settlement negotiations) that one must pay in order to succeed in settlement negotiations later. Thus, there is an unavoidable cost that companies must pay when they enter a litigation-risky line of business, and those costs go beyond the payments that go to plaintiffs: the existence of the tort system creates deadweight loss even when parties settle.⁴ Thus, the tort system biases corporations toward lines of business that do not introduce litigation risks beyond what is economically efficient.

The Counteracting Bias: The Weakness of the Plaintiffside Litigation System

So, several biases leave the Learned Hand rule (despite its intuitive appeal) a less than perfect law, to overdeterring behavior by overencouraging litigation. First, legal uncertainty means that the tort system overdeters potentially negligent behavior (so long as actors are risk-averse). Second, uncertainty over factfinders overdeters big, impactful activities in contrast to those that minimally impact anyone. Third, the costs of credibly-committing to positions (and negotiating settlements) bias corporations away from lines of business that pose litigation-risks.

But then there is a great counteracting bias: the litigation finance system biases the system toward under-deterrence.

Let me draw a quick analogy to agriculture. Because… I want to. The typical form of tenancy in the modern world is rent. Under the “rent” system, the landowner gives control of a parcel to a tenant, the tenant makes decisions about management and gets the full returns from their efforts (since the rent is fixed) and so the maximally efficient decisions are made. Contrast that to the sharecropper system. In sharecropping,⁵ the benefits of the tenant’s labors are split between the tenant and the landlord. If the tenant works particularly hard, the immediate result is that more crops are grown. Then, some of these crops go to the tenant while the remainder goes to the landlord. Thus, a sharecropper will put in less effort than is optimal. Any efforts that just barely break even in terms of total productivity won’t be taken by sharecroppers, because the sharecropper only gets a portion of the benefit.

Why did I tell you this? Because litigation finance operates on a sharecropping system. Plaintiffside lawyers are paid by getting a percentage of the benefits of litigation: a “contingency fee.” Thus, like a sharecropper, they are underincentivized to bring claims. Some claims with positive expected values for plaintiffs will not be pursued, because they don’t have positive expected value for plaintiffside lawyers. Like a sharecropper, plaintiffside lawyers won’t pursue a claim if it’s only marginally worthwhile.

Why are they paid this way? Well, there are probably a number of reasons that plaintiffside lawyers are paid this way, but one is the doctrine of “champerty:” In most jurisdictions, it is illegal to sell your legal claim to someone else (the way a landlord “sells” the potential produce of a land by renting out the land to a tenant.)

Thus, the funding system for plaintiffside litigation underincentivizes litigation, relative to what the Learned Hand Rule contemplates.

To the extent that the United States is overly litigious, that is not because of litigation finance, it is because other aspects of the tort system overdeter activity or overly redistribute to plaintiffs.

Note that the difference between “legally cognizeable harms and benefits” and those which are not cognizeable constitutes economic efficiency, in a real sense. Alternatively one could constitute economic efficiency from willingness to pay (that would be the classical interpretation, but is hard to apply to phenomena that cannot be bought or sold). In reality these two conceptions tend to align, since market-prices usually determine damages awards. ↩
One caveat: economists can tell you that repeat players have an advantage over one-shot players. If you’re a repeat player, you can credibly commit that you will only pay plaintiffs EV(P) - {litigation costs} + $1. So long as your behavior is public, you will benefit from your reputation, so plaintiffs will believe that you will stick to your commitments. They, as one-shot players, cannot benefit from their reputation, so they might as well give in rather than go to court. At first that might seem to advantage defendants (corporations) over plaintiffs (random people). However, oftentimes the plaintiffside firms themselves have a reputation to sustain (especially big names like Susman Godfrey or Edelson) and so it’s not obvious whether the dynamics of negotiation actually favor defendants. ↩
I will not bother to argue for this position. You should just believe it. ↩
Furthermore, even settlement involves some litigation expenses - you have to pay the lawyers who perform your settlement negotiations for you. ↩
Many people have strong moral intuitions about rent vs. sharecropping, presumably because of the association between sharecropping and the post-bellum South. These people seem to believe that sharecropping is particularly landlord-friendly or tyrannical. They are idiots. Sharecropping has existed all over the world. The main reasons to have sharecropping are that it shifts risk toward the landowner (which might be preferable) and that it allows proper pricing of the land when the landowner can not identify (and credibly signal) its true productivity/value. Another reason might be if the landlord can take actions during the tenancy. ↩

Capitalism sucks when you haven’t just blown up the entire world’s capital stock

My Response to Hanes’ Guest Post