Yesterday, we introduced Coase, and the result known as the Coase Theorem or Coase Conjecture: that when the cost of private negotiations are zero, they will lead to efficient outcomes, regardless of initial allocations of property rights or liability. Or, to put it in a market context, that when the costs of transacting are zero, that externalities can be overcome (and efficiency restored) by making property rights more complete.
However, our statement of the Coase Conjecture immediately suggests the converse: when private negotiations are not costless, or transaction costs are not zero, that initial allocations will matter for efficiency. This happens for two reasons: first, when transaction costs are high, they will prevent certain trades that would have been beneficial; and second, any resources actually spent overcoming the transaction costs are, in a sense, wasted.
Coase again: “If market transactions were costless, all that matters (questions of equity apart) is that the rights of the various parties should be well-defined and the results of legal actions easy to forecast. But as we have seen, the situation is quite different when market transactions are so costly as to make it difficult to change the arrangement of rights established by the law. In such cases, the courts directly influence economic activity. It would therefore seem desirable that the courts should understand the economic consequences of their decisions and should, insofar as this is possible without creating too much uncertainty about the legal position itself, take these consequences into account when making their decisions. Even when it is possible to change the legal delimitation of rights through market transactions, it is obviously desirable to reduce the need for such transactions and thus reduce the employment of resources in carrying them out.”
Coase offers two examples of institutions that may emerge in response to high transaction costs: firms, and government regulation. Going back to his example of a farmer and a rancher, if for whatever reason it is very difficult or costly for them to come to an agreement among themselves, one solution is for the ranch and the farmland to be both be purchased and operated by the same firm, so that the firm balances the costs and benefits of both activities and maximizes the total value of production. The second example, government regulation, is the same idea, since he imagines the government as a sort of “super-firm” which considers the costs and benefits of each activity to everyone.
Transaction Costs So, what are transaction costs? Transaction costs are anything that makes it difficult or expensive for two parties to achieve a mutually beneficial trade. Cooter and Ulen divide them into three categories:
search costs – difficulty in finding a trading partner
bargaining costs – difficulty in reaching an agreement on the terms of the trade
Search costs are self-explanatory. When we think of common, standardized goods, there are likely lots of buyers and lots of sellers, so it shouldn’t be hard for them to find each other. When we think of rare or exotic goods, search costs may be very significant. eBay
Bargaining costs are a little less obvious. We used the example before of my car being worth $3000 to me and $4000 to you. Once we find each other, all we have to do is haggle over price and agree on something in the middle – doesn’t sound so hard. However, this assumed that both of us knew exactly how we valued the car, and knew each others’ threat points. When these assumptions fail, things get more complicated.
First of all, there’s the possibility that I know something about the car that you don’t. I’ve been driving it a while, I might know that the transmission is about to fail, or that it needs new brake pads, or that it doesn’t start well on cold mornings. So you might worry that if I’m willing to sell it to you for $3500, maybe it’s because there’s something wrong with the car. So one aspect of bargaining costs might include taking it to a mechanic to verify its condition and try to get an objective measure of the value of the car. Famous paper by Akerloff, “The Market for Lemons,” dealing with this problem (adverse selection), showing that under some conditions, it can cause the market to fail completely.
Next, even if we agree on the physical condition of the car, I might not be sure exactly how badly you want it. Suppose the car is worth $3000 to me, and $3100 to you, but I don’t know what it’s worth to you – all I know is that you value the car at somewhere between $3000 and $5000.
Since you value it at more than $3000, there are definitely gains from trade – it’s definitely efficient (and Pareto-improving) for me to sell you the car. But now you try to convince me that the car’s only worth $3100 to you, and that I should therefore sell it to you for $3050. But here’s the problem: anything that you say to try to convince me, you could also say if the car was worth $4000 to you, and I have no way of knowing whether you’re telling the truth or lying to get a better deal. So if I give in and sell you the car for less than $3100, I can’t escape the possibility that I’d also sell you the car at that price if you valued it at $4000. So I end up saying, “Maybe you’re telling the truth, and maybe you’re lying, but I won’t sell it for less than $3500.” Which means that some of the time, even though you value the car more than me, we don’t reach a deal.
There’s a famous paper by Roger Myerson and Mark Satterthwaite, “Efficient Mechanisms for Bilateral Trade,” which shows that when there’s private information of this sort, there is no way to guarantee that the efficient outcome will always be reached – there’s always some probability that an inefficient outcome (in this case, no trade) occurs.
In this case, bargaining fails because we don’t agree on what each of our threat points are, that is, our threat points are not common knowledge. There have been a number of papers on bargaining, both theoretical and experimental, that reinforce the fact that people are more likely to come to agreements when they know each others’ threat points. One interpretation of threat points being private information is simply that tastes are subjective – I don’t know how much you like the color of my car, for instance. But another source of uncertainty about threat points is when property rights themselves are ambiguous. If we consider again the problem of the rancher and the farmer, and suppose that the efficient outcome is for the farmer to build a fence to protect his crops. But now suppose that the property rule (or the liability rule) is ambiguous – whether or not the rancher is liable for his crop’s damage is open to interpretation, or depends on the exact details of the situation, so the court’s decision is unpredictable. In that case, the rancher and the farmer might not agree on what would happen if no fence was built; and so each one might be uncertain about the other one’s threat point, and therefore it might be very hard for them to come to an agreement. This is one of the arguments for clear, simple, well-defined, unambiguous property rights – that they make negotiations easier, that is, effectively lowering transaction costs.
There’s another way in which bargaining can be difficult, or even impossible, which is when instead of a single buyer and a single seller, there are many parties to the deal. This is one of the situations discussed in the paper by Calabresi and Melamed.
Suppose there’s a developer, who wants to build a shopping mall, and he values the land he wants to build on at $1,000,000. Now suppose there are currently 10 houses on that land right now, and each homeowner values his property at $80,000. Clearly, there are gains from trade: the combined value of the plots is $1,000,000 to the developer, and $800,000 to their current owners.
But now think about one of the homeowners. He thinks, “If we all sell our land to the developer, this creates $200,000 of surplus. I don’t mind if all my neighbors sell out cheaply, but I want a piece of that!” And he figures that he’s the only one smart enough to ask for more money, so he asks for $120,000, figuring that still leaves the developer with a big enough surplus. And now some of his neighbors do the same calculation, and ask for more money for their land. And since it’s very hard to negotiate with 10 people at the same time, negotiations may fail.
(One of the papers I’m working on is a game-theory model of many-to-one bargaining. The idea is this. Everyone accepts that if 9 of the homeowners have sold out to the developer, the last guy is in a pretty strong bargaining position, so he can probably get a pretty high price for his property. But since everyone knows that, nobody wants to be the first to sell out – they’d rather wait for their neighbors to sell, and then be the last, so they get a better price. So even when cooperation, or trade, is efficient, and might occur eventually, there can be huge delays before the trade happens, due to everyone waiting around hoping to be last.)
And the same thing can happen when there are many buyers instead of many sellers. Suppose that instead of a shopping mall, the land was being bought up to be turned into a park, that would benefit 10,000 people in the community, and each of them would receive benefits worth $100 from the joy of having the park. Even if the homeowners were all willing to sell for $80,000, or $800,000 total, it might be impossible to raise that much through voluntary contributions, since each citizen might think, “We only need to raise $800,000, and all my neighbors should be willing to pay $100, so even if I don’t pay anything, the park should get built!” This is the problem of freeriders – once the park is built, its use won’t be limited to the people who paid for it, so people may try to avoid paying, preferring to get the benefits for free.
So when negotiations need to take place between lots of people, rather than just one buyer and one seller, there is a risk of holdout – individual sellers holding out for high prices – and freeriding – individual buyers trying to get the good for free. Either of these could cause negotiations to fail, or to take a long time to conclude. So these can be thought of as bargaining costs.
(One final source of bargaining costs is hostility. I’ll come back to this at the end of the lecture. Many divorce agreements end up being settled by litigation, which is more costly than negotiation, not because the parties disagree about their threat points or for any other rational reason, but because the parties are angry with each other and don’t want to come to a rational agreement.)
The final type of bargaining cost that Cooter and Ulen consider is enforcement costs. Obviously, if I’m just buying an apple from a fruit stand, there are no enforcement costs – I give him my money, he hands me an apple, and the deal is done. But think about our example from before, of a rancher and a farmer. Suppose that even though the rancher is not liable for his herd’s damage, it’s cheaper for him to fence in his herd, so the farmer pays him to build a fence. But now the farmer has to make sure that he actually builds it, and maintains it – the deal is part of an ongoing relationship, and has long-term consequences. In the case of pollution rights, if a factory pays for the right to pollute a certain amount, or neighbors pay a factory not to pollute in excess, someone has to monitor the factory and make sure they abide by the agreement. This involves ongoing costs.
So now, let’s recap what we know. Coase tells us that when there are no transaction costs, the initial allocation of property rights (or liability) doesn’t matter for efficiency, since people will trade until efficiency is reached. On the other hand, Coase tells us that when transaction costs are high, the initial allocation is important, since trade may not be feasible. This leads to two different notions of what the goal of property rights should be:
Structure the law to minimize transaction costs.
Cooter and Ulen phrase this as, “structure the law so as to remove the impediments to private agreements,” and refer to this as the Normative Coase Theorem. If the law is able to reduce transaction costs, then voluntary exchange will be more likely to lead to efficiency. Cooter and Ulen refer to this as “lubricating” bargaining – making it easier for bargaining to proceed without costs.
We said before that one source of bargaining costs is uncertainty about threat points; this suggests that bargaining costs are reduced when the law is simple and unambiguous, so that everyone is clear about everyone’s rights. This seems to favor rules like fast fish/loose fish and allocating the fox to Pierson, the guy who actually killed it – these are simple rules, there is little to dispute once the rule is established, and this should make both sides’ threat points clear and encourage trade to occur when it is efficient.
However, this is not the only possible goal of the law. Like we said, when transaction costs are high, the initial allocation matters for efficiency; so a different conception of the goal of the law could be,
Structure the law so as to minimize the harm caused by failures in private agreements.
Or really, structure the law so as the make the allocation more efficient to begin with, so that fewer negotiations are required and their failure is less costly.
This goal was put forward by Hobbes, who felt that people could not be counted on to be rational enough to cooperate. This view of the law is the Normative Hobbes Theorem, and suggests that the law should aim to allocate property rights to whoever values them the most, so that transaction costs become irrelevant and efficiency is achieved.
This seems to favor rules like iron-holds-the-whale and giving the fox to Post, who was chasing it, as a means of providing better incentives, leading to efficiency without bargaining.
So now we have two possible guidelines for what property law should aim to accomplish – one, lubricate private transactions, and two, allocate rights to whoever values them more. So now we have to ask, when is one of these aims appropriate and when is the other? We can answer this by thinking about the cost of each guideline.
When transaction costs are reduced, they are still unlikely to be eliminated – that is, lubrication works to a point, but there will still be some transaction costs remaining. When these are low, efficiency will nearly be achieved; when they are still high, the outcome may still be very inefficient.
On the other hand, in order to start out at an efficient allocation, lawmakers must figure out who values a right more highly. This is not always obvious. So we can imagine the lawmakers must face some sort of Information Costs to come to the correct conclusion. (This can be thought of either as costs they actually incur in researching the situation, or as the costs of being wrong some of the time.)
Which brings us to the principle reached by Cooter and Ulen:
When transaction costs are low and information costs are high, structure the law so as to minimize transaction costs
When transaction costs are high and information costs are low, structure the law to allocate property rights to whoever values them the most
Next, we’ll take on the question of how property rights are enforced, that is, what remedies are used when rights are violated. This is what is dealt with in the paper by Calabresi and Melamed, “Property Rules, Liability Rules, and Inalienability: One View of the Cathedral.” Calabresi and Melamed state that their goal is to treat both property and liability law under a common framework, rather than keeping them as distinct topics. We’ve already been doing this: when we think about the rancher-farmer question, we can pose it in terms of liability – is the rancher liable for the damage his herd does? – or in terms of property – does the farmer’s right to his property include the right to be free from trespassing cows? or does the rancher’s right to his herd include the right to not be punished when they stray? Calabresi and Melamed consider both cases to be cases of “entitlements” – is the farmer entitled to land without trespassing animals, or is the rancher entitled to be free from herd-damage liability? Similarly, am I entitled to have a noisy party, or is my neighbor entitled to be undisturbed by my noise?
Calabresi and Melamed discuss three possible enforcement rules for protecting an entitlement.
Property Rules, or Injunctive Relief
These are when you are legally barred from violating my entitlement without my prior agreement. This is the usual rule for protecting private property – if it’s mine, you simply can’t take it unless I give it to you. (And if you do choose to take it, you’ve committed a crime – you don’t just owe me the value of what you took, you may go to jail, you may face other consequences.)
An injunction is basically a court order clarifying someone’s rights and specifically barring someone from violating them. For example, if a factory is polluting and the neighbors object and take it to court, the court might issue an injunction, which would bar the factory from further pollution. The factory could still negotiate with the neighbors and reach some bargain where the neighbors agree not to enforce the injunction, but this would be completely at the discretion of the neighbors.
Liability Rules, or Damages
These are when you can violate my entitlement without my agreement, but must compensate me after the fact for whatever damages I incur.
Under a liability or damages rule, the factory could go on polluting, and the neighbors would sue them for damages; the court would then have to calculate an objective value of the damage done. The neighbors can argue that the damage incurred was high, but they are only awarded what is considered fair value, not what they argue they would have held out for in the beginning.
This is the type of rule behind eminent domain. If the government wants to build a school, or an army base, or a town dump, on land that I own, they can force me to sell; and they can force me to sell at what is considered fair market value, not to hold out for whatever amount I want based on my sentimental attachment to the house I grew up in or anything else. Liability rules obviously work better in settings where prior negotiation is impossible – clearly, I have an entitlement to not be hit by someone’s car when I’m walking, but it’s hard to imagine them approaching me beforehand and bargaining for the right to hit me.
(Damages are backward-looking – they compensate for harm already done, while injunctions and property rules are forward-looking – they specifically forbid future harms from occurring.)
A property rule will always be more favorable toward the injuree (the person whose entitlement is to be violated), and liability rules are more favorable toward the injurer. This is because the punishment for violating an injunction without the other side’s permission is much harsher than damages, since it may involve criminal trespass or violating a court order – so when the two sides bargain, the injurer will have a much lower threat point when facing an injunction, so the injuree will end up with a higher payoff if they do choose to cooperate. (We’ll see an example.)
In a world without transaction costs, either rule should be sufficient to allow private bargaining to lead to efficiency; however, since the two rules change the two sides’ threat points, they change the payoff achieved by each side during negotiations. In a world with transaction costs, of course, they may lead to different results. Consider the example in Cooter and Ulen.
There is an electric company E that emits smoke, which dirties the laundry at a laundromat L next door. The electric company earns profits of 1000, and could stop emitting smoke by installing scrubbers in its smokestack, at a cost of 500.
Without smoke, the laundromat earns profits of 300. Smoke does $200 of damage (reducing profits to 100). The laundromat could also avoid the damage by installing filters on its ventilation system, at a cost of 100.
Let’s look at the noncooperative results under different legal rules:
Polluter’s Rights: E earns 1000, L installs filters and earns 200
Pollutee has right to Damages: E earns 800 after paying damages; L earns 300
Pollutee has an Injunction: E installs scrubbers, earns 500; L earns 300
Clearly, the efficient outcome is for L to install filters, leading to combined profits of 1200. (This avoids $200 in damage at a cost of $100, and is the cheapest way to avoid the damage.) This is the outcome that occurs in the Polluter’s Rights case. In the other two cases, the noncooperative outcome is inefficient; but if there are not transaction costs, the two sides could still negotiate an agreement to achieve efficiency. But the threat points would be very different, so the division of surplus will be different under the two rules.
Under Damages, the threat points are 800 and 300, and there is an additional 100 to be gained by cooperating (avoiding $200 in damage by installing filters for $100). If we assume these gains from cooperation are split evenly, the Damages rule, without transaction costs, leads to profits of 850 and 350.
Under an Injunction, the threat points are 500 and 300, and there is an additional 400 to be gained by cooperating (avoiding the damage using $100 filters instead of $500 scrubbers). If this additional surplus is split evenly, it leads to profits of 700 and 500.
Under Normative Hobbes, the initial allocation should be chosen to be efficient; in this case, this is Polluter’s Rights. (There’s no reason for this to be true more generally, just in this example.) Under Normative Coase, if transaction costs are lowered sufficiently, efficiency will be reached under any of the rules; the law should be designed in a way that makes transaction costs as small as possible. (We’ll come back to this in a bit.)
Which rule is more efficient in general? An injunction is cheaper for a court to implement – it simply clarifies the property right, but does not have to calculate the amount of damage that was done after the fact. Damages are more difficult for a court to implement, since they must assess the monetary value of damage that was done. (We’ll return to the question of how damages are computed when we get to tort law; but for now, just realize this requires going to court, expert testimony, and a judgment.) However, in this example, if private bargaining fails, damages lead to a more efficient result than an injunction, and this turns out to be true more generally. This leads Calabresi and Melamed to the following conclusion:
When transaction costs are high (or there are impediments to private negotiations), a liability rule (damages) is more efficient
When transaction costs are low (or private negotiations can be expected to succeed), a property rule (or injunctive relief) is more efficient
(Why liability is more efficient when bargaining is likely to fail: if no agreement, the injurer can still choose to injure and pay damages, which in this case is cheaper than preventing the harm himself; with an injunction, he has to prevent the harm himself, even if this is more expensive.)
Cooter and Ulen use this to explain how things are typically done in certain types of disputes. “Private bargaining is unlikely to succeed in disputes involving a large number of geographically dispersed strangers because communication costs are high, monitoring is costly, and strategic behavior is likely to occur. Large numbers of land owners are typically affected by nuisances, such as air pollution or the stench from a feedlot. In these cases, damages are the preferred remedy.” On the other hand, property disputes generally involve a small number of parties who live near each other and can monitor each others’ behavior easily after reaching a deal; so injunctive relief is usually used in these cases.
Cooter and Ulen also point out, though, that in the first case – where transaction costs are high, so bargaining is likely to fail – a liability rule is only efficient when the court is able to correctly calculate the amount of damages. On the other hand, injunctive relief is efficient any time the court can determine who values the right more, regardless of its absolute level. This leads them to a different interpretation of efficient remedies in the high-transaction-costs case:
“When transaction costs preclude bargaining, the court should protect a right by an injunctive remedy if it knows which party values the right relatively more and it does not know how much either party values it absolutely. Conversely, the court should protect a right by a damages remedy if it knows how much one of the parties values the right absolutely and it does not know which party values it relatively more.”
And in the latter case, injunctive relief is assumed to be more efficient because it tends to be simpler and clearer and therefore more likely to encourage negotiations. However, as we discussed before, successful negotiations generally require clarity over each side’s threat point. In the case of the electric company polluting near the laundry, the cost to the electric company of complying with an injunction might not be known to the laundry, and this could lead to difficulty in bargaining. On the other hand, efficiency of a damages remedy would require the electric company to know how much damage the laundry would sustain, so it will know the laundry’s threat point in order to negotiate to have the laundry install filters. This leads Cooter and Ulen to restate the second rule:
“Where there are few obstacles to cooperation (i.e., low transaction costs), the more efficient remedy is the award of an injunction when the plaintiff can estimate the defendant’s compliance costs more readily than the defendant can estimate the plaintiff’s damages.”
Calabresi and Melamed defend injunctions as being optimal by assuming that the parties will privately negotiate after the court rules. Cooter and Ulen (on their website) mention a paper by Ward Farnsworth, examining whether this occurs.1 Quoting:
Farnsworth "examines twenty nuisance cases and finds no bargaining after judgment in any of them; nor did the parties’ lawyers believe that bargaining would have occurred if judgment had been given to the loser. The lawyers said that the possibility of such bargaining was foreclosed not by the sorts of transaction costs that usually are the subject of economic models, but by animosity between the parties and by their distaste for cash bargaining over the rights at issue."
Naturally, Professor Farnsworth asked the lawyers why they though that no bargaining occurred after judgment. The lawyers cited two impediments to post-judgment bargaining. "First, in almost every case the lawyers said that acrimony between the parties was an important obstacle to bargaining. The parties in these cases often thought that their adversaries were behaving in ways that were unreasonable, discourteous, and unneighborly. Frequently the parties were not on speaking terms by the time the case was over (sometimes much earlier). ... The second recurring obstacle involves the parties’ disinclination to think of the rights at stake in these cases as readily commensurable with cash."
Calabresi and Melamed also talk about a third type of protection for entitlements: inalienability. This is when an entitlement is not transferable or sellable. For example, I an entitled to not be a slave, and I am not allowed to give away or sell that right. This could be defended in some cases by invoking externalities; in others, it seems to be more a case of paternalism, that is, the government thinking it knows better than you do what is good for you. (We can think of this as the case of owning a historical landmark: it is your property, but it may be unalienable, in that you cannot sell it to someone who would put it to a different use.)
(ENDED HERE) They also point out that there is a fourth option, relative to the three we showed in the example above with the electric company and the laundry. Recall we considered the case where the electric company was free to pollute unless they reached a bargain with the laundry not to; where the electric company could pollute and pay damages; and where the electric company could only pollute if they negotiated an agreement with the laundry (injunction). However, we did not consider the case where the laundry could stop the electric company from polluting, not by bargaining with them, but by paying objective (court-decided) damages. That is, we would start in a world of polluter rights, but give the pollutee the right to veto the pollution and pay the electric company for whatever losses this caused them.
(In the example we did above, this would not be efficient; but Calabresi and Melamed point out some instances in which it might be.)
1 Ward Farnsworth, “Do Parties to Nuisance Cases Bargain After Judgment? A Glimpse Inside the Cathedral,” 66 U. Chi. L. Rev. 373 (1999).