Econ 522 – Lecture 21 (Nov 27 2007) Hope you all had a nice Thanksgiving. Reminder: last homework due next Tuesday. Thursday: class evaluations – please bring a pencil! Over the last two months or so, we’ve developed theories of property and nuisance law, contract law, and tort law. We’ve looked at how rules of legal liability create incentives, and how these rules can be chosen to achieve efficient, or close to efficient, results. With damages in nuisance law, with expectation damages in contract law, and with compensatory damages in tort law, we assumed it was possible to make one party’s liability for damages exactly match the harm he caused to the other party, so that he would internalize this harm and therefore make efficient decisions. Implicitly, we made two big assumptions: the legal system works flawlessly the legal system is costless The first assumption we made explicitly – by assuming we could set damages precisely in relationship to actual harm – and, during tort law, we considered the effect on incentives when it is violated. The second assumption we made implicitly – by ignoring the costs of the legal system in figuring efficiency, and also by ignoring the private costs of litigation when considering the parties’ incentives. The next three lectures, we will relax these two assumptions, and explicitly consider the details of the legal system and the incentives it creates. We’ll start with an example from the Polinsky book. I hit you with my car and did $10,000 worth of damage. (Sorry.) You and I both know that I was negligent, but we also both know that courts aren’t perfect – if we go to trial, there’s an 80% chance I’ll be held liable, and a 20% chance I won’t. If I am held liable, damages will be correctly set at $10,000; so if we go to trial, you expect to recover (on average) 80% X $10,000 = $8,000. However, if we go to trial, we’ll both have to hire lawyers, and lawyers are expensive. Suppose going to trial will cost each of us $3,000. So now your expected net gain from going to trial is $8,000 – $3,000 = $5,000. Similarly, my expected cost if we go to trial is $8,000 + $3,000 = $11,000. Of course, since a trial will (in expectation) cost me $11,000 and earn you $5,000, it’s possible we can agree to settle without going to court. Any settlement between $5,000 and $11,000 makes both of us better off. So perhaps this will happen. However, it’s also possible we disagree about the likely outcome of a trial. You probably have some private information about the degree of your injuries. I probably have some private information about how recklessly I was driving. First, suppose I’m more pessimistic about my chances at trial than you. That is, you think I’m 80% likely to be found liable, but I think it’s more like 90%. So you perceive your expected gain from trial to be $5,000; but I perceive my expected cost to be 90% X $10,000 + $3,000 = $12,000. This makes the range of possible settlements we’d both agree to wider, and makes settling more likely. On the other hand, suppose I’m optimistic about my chances. You still think I’m 80% likely to be held liable, but I think it’s more like 10%. You expected gain from trial is still $5,000; my expected cost (given my beliefs) is $4,000. So now we’re very unlikely to settle. Finally, even if our beliefs are compatible, that is, there is a range of settlements which would make us both better off than going to trial, the private information we both have might lead to a failure to settle. Recall from before, that if each of our threat points are private information, we might fail to reach an agreement because one of us tries to hold out for too big a share. So even if we both had the same beliefs about the likely outcome of a trial, private information could lead us to fail to settle. This leads us to a few quick observations: With litigation costs, if we agree on the likely outcome of a trial, there will always be gains from settling out of court, and a range of settlements we would both prefer to trial If the two sides are relatively pessimistic – the injurer perceives his expected liability to be higher than the victim – settlement is even more likely If the two sides are relatively optimistic – the injurer perceives his expected liability to be lower than the victim – settlement may be impossible Even if the two sides have the same beliefs or are relatively pessimistic, private information may lead to failures in bargaining But then there’s another thing as well. Recall that under a strict liability rule, or under a strict liability with contributory negligence rule, the injurer bore the cost of accidents, and therefore internalized them and took efficient precaution. But that assumed the cost of being sued was equal to the damage done. With unpredictable courts and litigation costs, the private cost of being sued for damages can be either greater or less than the actual cost of the accident; so this could lead to either too much or too little precaution. But it’s trickier than that as well. If we assume that settlement talks are likely to break down, and most cases will end up going to trial, then the total social cost of an accident includes the resources expended during a trial. That is, rather than $10,000, the cost of an accident is really $16,000 – the harm done, plus the cost of a trial. If accidents do more harm, this means more precaution is cost-justified – the optimal level of precaution is higher than before. We’ve already spent a lot of time looking at how incentives respond to the private cost of accidents, so we’ll put that question aside. However, in the next couple of lectures, we’ll go into greater detail about the legal process itself – how these costs are incurred, and the effects this has. Cooter and Ulen point out that the legal process has a large number of steps: Once an injury has occurred…. The victim can decide to sue or let it go The victim and injurer can immediately settle out of court, or else begin the process of preparing for trial o This consists of exchanging information relevant to the case – more on this shortly Once information has been exchanged, the two sides can again bargain over an out-of-court settlement, and can either settle or go to trial At trial, the victim (now the plaintiff) can win or lose The loser at trial can choose to appeal or not We’ll look more closely at each of these. First, however, it will help to have in mind what the theoretical goal of the legal process should be. Recall that the economic essence of tort law was to minimize the total social cost of accidents. Similarly, in economic terms, the goal of the legal process is to minimize its total social costs. These costs come in two varieties: direct (administrative) costs, and error costs. Administrative costs are obvious. If a legal process is going to require judges, you have to hire judges. If it’s going to require courtrooms, you have to build a courthouse. The more complex the process is, the more it is likely to cost. Error costs are less obvious. Any legal process will be imperfect – some defendants will not be found liable when they should be, damages will be set incorrectly, and so on. We can think of an error as an any judgment that differs from the theoretically perfect judgment, that is, the judgment that the court would impose if it were infinitely wise and had perfect information. An error in, say, computing damages after the fact affects distribution but not efficiency. However, anticipated errors also affect the costs that each side perceives as stemming from their actions, and therefore changes incentives and may lead to actions which are not efficient. Error costs are the costs of any distortions in actions (precaution, activity levels, etc.) due to imperfect incentives caused by flaws in the legal system. Theoretically, then, we can see the goal of a legal process as minimizing the sum of these two costs: the direct costs of administering a legal process, plus the economic effects of errors due to that process. The next several sections of Cooter and Ulen consider in depth each of the different stages we already mentioned: the decision to pursue a legal claim or not the decision to settle immediately or exchange information the decision to settle then or go to trial the trial itself appeals We begin with the question of whether or not to sue. In a rational world, this comes down to calculating the amount you expect to gain from suing, and comparing it to the cost. Looking at the problem from the victim’s point of view, we can turn all the questions above into a decision tree, assign values and probabilities to the different outcomes, and calculate the overall expected value of a legal claim. SUE? Don’t File File SETTLE IMMEDIATELY OR EXCHANGE INFO? Settle “Discovery” SETTLE OR GO TO TRIAL? Settle Trial WIN OR LOSE AT TRIAL? Win Lose APPEAL? Win Lose (In the U.S., before going to trial, the two sides in a lawsuit go through “pre-trial discovery” – basically, the two sides exchange information they have that is relevant to the case. This could be doctor’s reports about the extent of injury, police reports about the accident, lists of witnesses each side intends to call, and so on. In many European countries, there is no pre-trial discovery, and instead, the first part of the trial itself involves a “giving of proofs,” in which the two sides offer evidence to support the basic facts of their claim.) To keep the arithmetic simple, Cooter and Ulen make the numbers unrealistically small – you can think of these as hundreds of dollars. They assume that the damage done was $100. Starting at the bottom of the tree, suppose that if you lose at trial, an appeal costs $20, and will be successful 10% of the time. That is, 10% of the time you’ll win the appeal and get a judgment of $100; the rest of the time, you’ll get nothing. Either way, you pay the costs of $20. So an appeal has an expected value of 10% X $100 + 90% X 0 – $20 = –$10 so the victim expects not to appeal if he loses at trial. Now go back a step. Suppose going to trial costs $20, and you’ll win with probability ½. So with probability ½, you’ll win a judgment of $100, and with probability ½, you’ll win nothing. So the expected value of going to trial, knowing that you won’t appeal if you lose, is ½ (100) + ½ (0) – 20 = $30 Now go back a stage, and consider bargaining for a settlement. Suppose that it’s out of your hands whether bargaining will be successful or not; 70% of the time, you’ll reach a settlement, for an average of $50, and incur costs of $1. The other 30% of the time, talks will break down, and you’ll go to trial. So now the expected value of reaching this stage is 70% * ($50 - $1) + 30% * ($30) = $43.30 So this is the expected benefit of reaching the stage where you’ve already initiated a suit, failed to settle initially, and already gone through discovery. Now go back another stage, to the initial decision to settle immediately or go through discovery. Again, Cooter and Ulen assume this is not a decision but a random chance – with 70% probability, you’ll reach a settlement (averaging $50, and costing $1 in legal costs), and with 30% probability, negotiations will fail. They assume the discovery process costs $3.30. So the expected value of getting to this stage is 70% X ($50 – 1) + 30% * ($43.30 – 3.30) = $46.30 Finally, they assume it costs $10 to file a lawsuit; in this case, the expected gain, $46.30 (with all the later costs built in), outweighs the filing cost, $10, so you would expect the victim to file. Obviously, the exact numbers are arbitrary, but the general idea seems reasonable. For each injury – that is, each accident, or breached contract, or invasion of property, or nuisance – that occurs, we expect a claim to be filed if the expected value of the claim exceeds the filing cost. So there are three things that seem to directly influence the number of claims: the number of injuries the cost of filing a complaint the expected value of a claim Obviously, more injuries should lead to more claims; the per injury, lower filing costs, or higher expected value of claims, should lead to more claims as well. They give a cool of example of how things can get more complicated, however. Think about the number of legal complaints filed, as a function of the level of damages typically awarded by the court. Obviously, when damages are very low, defendants have no reason to agree to generous settlements, and so the expected value of a claim will be low – since all the “upside” of a claim comes from either court-imposed damages or a settlement. As damages rise, the expected value of a claim rises, and so the number of lawsuits should rise. However, as damages continue to rise, the cost of accidents to injurers rises as well, and this may start to lead to greater precaution (or to less breach of contracts, or fewer nuisance activities). When damages get very high, injurers will go to great length to either prevent accidents or avoid liability, and so the number of complaints should be low again. Therefore, as a function of damages, the number of claims filed might have an inverted-U shape. Obviously, different injuries in different situations are different, so the expected value of claims should vary widely. Whatever this distribution, filing costs basically divide this distribution into those where the victim will find it worthwhile to sue, and those where he won’t. Higher filing costs mean fewer actions (at least on a per-injury basis). In the U.S., courts do charge fees for filing a claim and for subsequent stages of the process, but these fees are much less than the actual cost to the state; that is, the state bears much of the cost of the legal process, but does charge some fees. In some civil law countries, the state charges no fees at all for using the civil courts. Come back to our earlier point: economically, the ideal legal system is one that minimizes the sum of administrative costs and error costs. Higher filing fees mean fewer actions, and therefore lower administrative costs. But higher filing fees mean a greater number of injuries will go unpunished, leading to a greater distortion in incentives and therefore greater error costs. The filing fee is set optimally when these two exactly balance on the margin: that is, when the administrative cost of an additional complaint is equal to the error cost of providing no remedy in the marginal case, that is, the case which is right on the border between justifying a lawsuit and not justifying a lawsuit. The magnitude of error costs depends primarily on how strongly peoples’ behavior responds to the incentives caused by liability. The textbook puts this another way: the social value of reducing errors depends upon whether the errors affect production or merely distribution. In some situations, failing to provide a remedy for a harm will impact distribution – the victim will be worse off, the injurer better off – but will not change anyone’s behavior. This might be the case with lawsuits involving hunting accidents: my reasons for not wanting to shoot my friend probably have very little to do with having to pay his widow if I kill him. So failing to “punish” the marginal hunting accident will be bad for widows and good for hunters, but may not have much impact on the number of accidents, so the social cost of these errors is very small. On the other hand, consider a contract setting where I paid up front for you to provide a valuable service. Whether or not you want to live up to your end of the deal might depend very much on whether you think I’ll bother to sue you if you walk away. So failing to “punish” the marginal breach might have a substantial impact on production – whether you actually perform the service, or even whether I agree to the contract in the first place. So we conclude that when errors have large incentive effects, filing fees should be kept low; when errors have small incentive effects, efficiency requires higher filing fees. The textbook makes some obvious, and uninteresting, points about the effect of the number of lawyers on the supply of legal services, and therefore on price. Feel free to read about it if you like. As long as there are filing fees or other litigation costs, there will be some situations where the harm done to each victim is below the threshold to justify a complaint. One solution when the harm is small to each individual but large overall is a class action lawsuit. This is where one or more plaintiffs bring a lawsuit on behalf of a large group of people harmed in a similar way. The book gives the example of a California man who sued his bank over a $6 fee for bouncing a check. $6 obviously exceeds the costs of pursuing the claim, so he sued on behalf of all the bank’s customers who were charged the same fee. In order for a class-action suit to proceed, the court must decide to “certify” the class. This has to be done thoughtfully, since participation in a class-action suit extinguishes each victim’s right to sue later. (Some class action suits allow individuals to choose whether to participate or opt out, preserving their right to sue on their own.) If a classaction suit succeeds – if it leads to either a settlement or a judgment at trial – the court must then approve the plaintiff’s proposal for distributing the award to the other members of the class. Economics suggests class-action suits are appropriate where individual harms are very small but aggregate harms are large, especially if the avoidance of liability will have strong effects on incentives. However, there’s also a view that class-action suits come with a danger: that when a class is large enough, losing at trial would be so catastrophic for the defendant that even when the claim is very dubious, the defendant can’t take the risk at trial and is basically forced to settle. (These have been referred to as “blackmail settlements.”) (There’s also a view that some class-action suits are motivated more by lawyers looking for a plaintiff rather than the other way around.) This last point brings us to the next one: the agency problem between lawyer and client. Ideally, the client wants the lawyer to work on the case until the marginal cost of more work (the opportunity cost of the lawyer’s time) equals the marginal benefit (in increased expected value of a settlement or judgment). However, this is very hard to achieve via a contract. A lawyer being paid by the hour has an incentive to do too much work. A lawyer paid for each individual service has an incentive to do them quickly and sloppily. A lawyer working on contingency – for example, a lawyer who receives 30% of the eventual judgment or settlement – internalizes some but not all of the benefit of working, but all of the cost, and so has an incentive to work too little. (One solution to this problem would be for lawyers to work on 100% commission. That is, they pay the client some up-front amount, and then get to keep whatever settlement or judgment they get – functionally, the client sells their legal claim to the lawyer. The lawyer would then internalize the full cost and full benefit of additional effort, so they would work the optimal amount. Of course, this creates a different problem – the client, who is probably a key witness, now has no incentive to testify or assist in the case. Also, it’s illegal pretty much everywhere.) Since the lawyer tends to know more about the law than the client, the client can’t always tell when he’s getting good advice, or whether more (or less) effort would be optimal. In addition, there is some randomness to the legal process, so a lawyer’s effort level can’t always be judged by the outcome of the case. Given these problems, people often choose lawyers based on reputation and long-run relationships. Established firms have an incentive to maintain their reputation by hiring good lawyers and getting them to do good work. Given that, clients are willing to pay a premium to hire a firm with an established reputation. Exchange of information. Trials are costly for both parties. Clearly, if the two parties can anticipate the likely outcome of a trial, and come to a settlement with similar terms, they both end up doing better. One of the obvious ways that this can fail to happen is if the two sides disagree about the likely outcome of a trial. The parties might disagree about the chance the defendant is held liable at all, or about the likely size of damages. In our usual risk-neutral world, all that matters is the expected value of the judgment each side expects. As we said before, when the plaintiff’s view of the expected judgment is higher than the defendant’s view, the two sides are relatively optimistic – each one has beliefs favoring their own interests. When the two sides are relatively optimistic, the range of settlements that both of them would agree to is smaller than if they shared the same beliefs, and may not exist – that is, there may be no settlement which both the defendant and the plaintiff would agree to over a trial. (On the other hand, when the two sides are relatively pessimistic, there is a wider range of settlements that both sides would prefer to trial.) Once a suit has been filed but before it goes to trial, the parties have the opportunity to negotiate a settlement, as well as to exchange information relevant to the trial. In addition to voluntary sharing of information, there is some required sharing. In the U.S., this is the “discovery” process: each side must supply the other side with the evidence they plan to use, and answer questions about the case. (So named because each party has the right to “discover” facts the other party has about the case.) In Europe, there is no discovery pre-trial, but the first stage of the trial involves a similar sharing of information in front of the judge. Thus, Cooter and Ulen ask two distinct questions about the effect of information exchange: does voluntary pooling of information promote settlements out of court? does involuntary pooling of information promote settlements out of court? First, we consider voluntary pooling – that is, information one of the parties willingly shares without being required to. In general, parties tend to disclose information that corrects the other side’s relative optimism. For example, suppose I hit you with my car, and we both know I’ll be held liable. I am relatively optimistic: I think you likely sustained only minor injuries, and that damages might be $1,500. You know that you sustained more serious injuries, and have x-rays and doctors’ reports to prove it; given this information, damages will likely instead be set at $15,000. Suppose going to trial would cost us each $3,000. As it is, since I expect a trial to cost me $4,500, I have no reason to offer a settlement larger than that. Since you expect to gain $12,000 from a trial, you have no reason to accept a settlement smaller than that. Settlement looks unlikely. But you’re more than happy to show me the x-rays and doctors’ reports. Once I see this evidence, I might agree that damages will be close to $15,000. Given that, I might be willing to offer a settlement close to that level, saving us both the cost of going to trial. So when the information you have corrects my relative optimism, you’re happy to share it with me; and this encourages settlement. On the other hand, suppose we started our relatively pessimistic. I heard something go “crunch” when I hit you and am worried I broke your hip, and expect damages to be $15,000. You know the crunch was your iPod, and your actual injuries were minor. Given this, I might be willing to offer a fairly high settlement; you might be willing to accept a reasonably low one; a settlement seems likely. You have no reason to voluntarily inform me your injuries were minor, as this would likely make me lower my settlement offer. In general, parties tend to withhold information that would correct the other side’s relative pessimism. And since relative pessimism makes settlement likely, this withholding of information likely encourages settlement. Cooter and Ulen explain these results in the following way: Trials occur when the parties are relatively optimistic about their outcome, so that each side prefers a trial rather than settlement on terms acceptable to the other side. When the parties are relatively optimistic, at least one of them is uninformed. Pooling of information before trial that reduces relative optimism promotes settlement. Furthermore, by revealing private information to correct the other side’s false optimism, the party making the disclosure increases the probability of settling on more favorable terms. The same intuition holds in reverse for information which corrects false pessimism – revealing the information is bad for the discloser, and discourages settlement. So voluntary disclosure will tend to share information that corrects false optimist but not false pessimism, promoting settlement. So that’s voluntary disclosure. What about involuntary (forced) disclosure? Involuntary disclosure will tend to reveal the information that the parties initially chose to withhold – that is, information that corrects relative pessimism. In this way, it may make settlement less likely. On the other hand, involuntary disclosure reduces uncertainty, and makes the two sides’ threat points more clear. In this way, it may make reaching a settlement easier. So the overall effect is unclear. (One other thing, of course, is that the existence of a disclosure rule may some make parties less willing to settle before disclosure, since they know they won’t be forced to disclose any harmful information, and want to see what the other side is forced to reveal. So involuntary disclosure may delay settlement until after disclosure occurs.) The disclosure rule in the U.S. is very extensive: well before trial, both parties reveal the basic arguments it will make, the evidence that supports them, the names of witness, and the nature of each witness’s testimony. Witnesses or evidence that are not disclosed ahead of time may not be allowed at trial. Each side can inspect the other side’s evidence and question its witnesses. Most European countries have little or no pre-trial discovery. Part of this difference may stem from the constitutional right in the U.S. to request a trial by jury. Jury trials are costly – the jurors are taken away from their jobs – so there is more value in having the trial itself proceed quickly and without interruption. Pre-trial discovery makes this more likely. In Europe, juries are rarely used in civil cases. So delays and interruptions are less costly, and much more common. In addition, in European civil law, judges take a more active role in developing arguments and exploring evidence. In the U.S., the judge plays a more passive role, serving as a referee between the two opposing sides. Finally, we can look at the effects of voluntary and involuntary disclosure on the two types of costs we considered earlier – administrative costs and error costs. Voluntary disclosure, we said, encourages settlements, and therefore decreases the number of trials. In addition, having more information out in the open ahead of time should ideally simplify and quicken the trial itself. So we expect voluntary disclosure to reduce administrative costs. In addition, as information is shared, the parties get closer to agreeing on the likely outcome of a trial, so the terms of a settlement likely get closer to the result of a trial. If we believe that the judgment at trial would likely be correct, this reduces the size of errors in outcomes, and therefore reduces error costs. So voluntary disclosure should reduce both administrative and error costs, reducing the overall social costs of the legal process. Involuntary disclosure, we said, does not predictably encourage or discourage settlement – it’s ambiguous whether it will lead to more or fewer trials. Pre-trial discovery is expected to lead to simpler, shorter trials. However, discovery is also a costly process. It’s unclear whether the overall effect on administrative costs is positive or negative. Even more than voluntary disclosure, though, involuntary disclosure pools much of the information that would come out at trial; so when settlements occur, we expect them to deviate less from the likely outcome at trial. So we expect involuntary disclosure to reduce error costs.