The Problem of Cooperation Costs
Potential cooperation as an explanation for underdevelopment, quantity-based responses to externalities, and punishments carried out by our criminal justice system
The UCLA approach to price theory tends to put a lot of emphasis on the nature of exchange. This focus tends to influence the type of questions asked. By focusing on the nature of exchange, one tends to think a lot about scenarios in which there are gains from trade to be had, but that go unrealized. A lot of these types of questions have been influenced by Ronald Coase and thus tend to focus on transaction costs.
Transaction costs can be somewhat tautological. If it appears that there are foregone gains from trade, the explanation can always be framed as due to some type of transaction cost. Simply blaming transaction costs, however, doesn’t tell the whole story. We need a precise definition of transaction costs. Furthermore, depending on how broadly we define the term, we need to make sure to differentiate between different types of transaction costs. Failure to do so will provide us with an explanation, but little gained understanding of the world.
With that in mind, this week I wanted to write about a type of cost emphasized by Earl Thompson. In an unpublished working paper, Thompson argued for the important role of what he called “cooperation costs.” Broadly, what he meant by cooperation costs are costs associated with communication, coordination, and commitment among economic actors. His argument was that cooperation costs were important for thinking about sequential decision-making. In particular, he argued that when people engage in strategic, sequential decision-making, cooperation costs become really important.
He thus coined the term “potential cooperation” as a contrast to the idea of “potential competition” more familiar to price theorists and students. The idea behind potential competition is that even a firm with market power might behave more like a competitive firm when there is the threat of entry by a potential competitor. Whereas potential competition might push society toward greater gains from trade, Thompson’s idea of potential cooperation tends to limit gains from trade or lead to wasteful allocations of resources.
The basic idea is as follows. In a world of zero cooperation costs, all sequential decision-making would maximize the gains from trade. This is because anyone could communicate what economists would call a reaction function, “if you do Y, I will do X.” If the cost of communicating, coordinating, and committing are all zero, and assuming both Y and X would expand the gains from trade, these actions would be taken and make everyone better off. However, when these actions are costly, sequential decision-making can lead to foregone gains from trade. The reason is that, given a fixed cost of cooperation, it might be difficult at the outset of any sequential decision-making for everyone to agree to cooperate. Yet, once beneficial sequential actions start taking place, it becomes possible for a subset of decision-makers to cooperate and expropriate the gains from the people who have already taken action. Because they not only get the private gains they would have gotten otherwise plus the gains to other decision-makers, this subset is able to overcome the fixed cost of cooperation. Of course, if any of the early decision-makers know of this possible future cooperation, they will be unlikely to take beneficial actions because they expect their personal gains to be expropriated.
He argues that this concept can explain why certain places seem to end up in development traps, why sometimes our policies to deal with externalities seem inconsistent with lessons from introductory economics, as well as why the criminal justice system is structured in the way that it is.
Let’s dig in and think about this idea in a little bit more depth by applying it to the examples that Thompson provides.
Underdevelopment
Thompson starts with a simple example. Suppose that there is a site where one could open a mine and a site where someone could open a port. If both the mine and the port were opened, then one would need a road to connect the mine with the port so that the production from the mine could be transported to the port. An important question that we might think about is why some places seem to have no problem getting these investments made and why other places seem to get stuck in development traps in which these investments are not made.
To think about this question, Thompson first goes through conventional arguments for coordination failure. For example, if the decision to invest in the road, the mine, and the port were all made simultaneously, it is easy to show that you could get two Nash equilibria. In one equilibrium, all of the investments get made. In the other equilibrium, none of the investments get made. In other words, since the productivity and the value of any one of these investments is dependent on whether the investments get made, whether all three investments are made depend on whether each investor believes that the other investors will make their investments.
As Thompson correctly points out, this might be an interesting technical argument for development traps, but it doesn’t survive close scrutiny. Investments need not be undertaken all at the same time. The investments could occur sequentially. In fact, pecuniary externalities would seem to imply that sequential decision-making would lead to investment and prevent development traps. For example, suppose that someone decides to open the port. Subsequent to this, there is greater value in building a road that connects the port to the possible site of the mine. Furthermore, once the port and the road already exist, now the value of opening the mine is substantially greater than it was prior to any of these investments.
If cooperation costs are zero, this is precisely what we would expect to happen. One investment will lead to a subsequent investment.
However, suppose that cooperation costs are large enough to prevent these investors from being able to agree to these sequential investments, but less than the total surplus of the investments. Now, consider the same example. An investor decides to open the port in anticipation that the others will invest. It is possible that the road will subsequently get built. Yet, once the port and the road are built, the investor in the mine might refuse to construct the mine unless he can extract all of the surplus of the first two investments. Alternatively, investors in the road and the mine might cooperate to extract all of the surplus from the port investor. If the investor who opens the port knows that this is a possibility, he never opens the port in the first place because he doesn’t stand to earn any of the returns on the investment. Hence, the development trap.
Thompson argued that states that escape such development traps are those that are able to find solutions to these investment problems and create institutions that prevents the expropriation of gains from trade from early investors. He also argued that this is one reason why states that have substantial internal social conflict might be more susceptible to development traps: the cost of initial cooperation and ex post expropriation is substantial.
Externalities
Another application of his logic is to think about externalities. The standard textbook argument is that negative externalities should be taxed at a rate equal to the marginal external cost. Despite that standard argument, the real world is filled with quantity regulations and zoning rules to deal with pollution. Perhaps that is because an insufficient number of policymakers have successfully learned introductory economics. Thompson disagreed. He argued that policymakers clearly understood something that is ignored by the introductory textbook: cooperation costs.
As I have written about before in my post “Externalities and the Numbers Problem,” how we deal with externalities seems to depend on the number of victims and the number of imposers. Thompson argues that this is because of the costs of communication, coordination, and commitment.
For example, suppose that the state puts a Pigouvian tax on pollution that is equal to the marginal external cost of pollution. This is what the introductory textbook tells you to do. Now imagine that a factory that produces a lot of air pollution decides to locate in a large neighborhood. The residents of the neighborhood do not want to live by the factory and its pollution. Moving to another neighborhood will be costly, not only because moving is costly, but also because the value of the house and property is now lower due to the proximity of the factory. Knowing this, the factory could purchase the land and commit itself to building the factory. It could then offer the residents of the neighborhood a chance to buy the land back from the investor so that he cannot build in the neighborhood.
Like the underdevelopment example, the potential factory builder is able to expropriate some of the value of the housing in the neighborhood simply by committing to build if the transfer is not made. To prevent this sort of unproductive behavior from going on, governments tend to impose zoning rules that prevent factories from building in a neighborhood.
Crime and Punishment
A final issue that can be addressed in this context is the system of crime and punishment. Brian wrote about this recently in reference to David Friedman’s work on punishment. Brian frames the issues as follows:
As I said, Friedman’s paper is about punishment. He asks why we use prison instead of fines. From your textbook economics perspective, fines are much better. The defendant loses money, the state gains money, and no resources are destroyed in the process. Prison is horrible from this perspective. The defendant loses years of freedom, and the defendant’s potential economic output is wasted. Unlike fines, where the state at least gets some money, the state here pays the costs. That’s the opposite of getting money.
There’s a somewhat standard law and economics answer that fines cannot adequately punish poor defendants. If someone commits a crime worth $100,000 in damages but only has $10,000, a fine can’t deter them. Prison fills the gap. There’s definitely something to that.
Thompson poses a similar question in his paper. Economic theory would seem to suggest that fines would be more efficient forms of punishment than things like imprisonment. Furthermore, the issue of solvency referenced in Brian’s framing doesn’t seem to be an adequate answer. For example, if someone commits a crime with a social cost of $1 million, but only has $10,000 to their name, clearly they won’t be able to pay the appropriate fine. Nonetheless, the criminal could be made to pay some of the costs via a fine and the remainder in prison. That is often not the case. Also, we do allow for both criminal and civil procedures. If solvency is the problem, these seem to occur in the wrong order. The criminal case is prosecuted first and the civil case follows. Knowing this, the wealthy criminal has an incentive to spend as much as possible on defense to try to assure his freedom, leaving little residual wealth to be claimed by the victim(s) in a subsequent civil trial. If this was about solvency of the criminal, wouldn’t it be the opposite? Or wouldn’t they be combined into one process?
Instead, anticipating David Friedman’s argument by about a decade and a half, Thompson argues that the use of imprisonment is designed to deal with his idea of cooperation costs. His argument is that it only appears that imprisonment is the socially wasteful punishment because it assumes the objectivity of all judges, juries, and witnesses and ignores financial incentives to law enforcement and prosecutors in the alternative system. Punishments like imprisonment are used instead of fines because otherwise the criminal justice system would have an incentive to excessively enforce crimes and even bring dubious cases against people. They might also have an incentive to frame people for crimes they didn’t commit.
He also argues that this explains other elements of our criminal justice system. For example, since imprisonment is socially costly, one way to recoup some of the cost would be to lease the labor of prisoners to the private sector. However, this practice was prohibited by statute in the twentieth century. This was done to limit the incentive of the state and its private sector partners to use the criminal justice system to expropriate the human capital of prisoners.
As further evidence of this claim, Thompson notes that the elimination of the “murder fine” in England and its replacement with imprisonment in the twelfth century also coincided with the use of juries. He argues that both imprisonment and the use of juries fix the same problem. Just as fines would tend to lead to over-enforcement and potential extortion, it is much easier for a judge to collect a bribe from a defendant in exchange for a lesser or zero punishment than it would be for a jury to all cooperate to extract a similar bribe. Furthermore, the bribe is likely to be diluted by the fact that it must be shared by members of the jury. Even just the possibility of extracting a bribe makes it more likely for police to make an arrest and prosecutors to pursue a case if the police and the prosecutor are able to cooperate with the judge who collects the bribe. The jury not only makes it more difficult to extract the bribe, it would also be more difficult and thus more costly for a jury to cooperate in such a bribe scheme with others in the criminal justice. Unlike judges, jury members are unlikely to have repeated dealings with police and prosecutors thereby making cooperation costs much larger.
In short, potential cooperation creates the conditions for members of the criminal justice system to steal real capital from the wealthy and human capital from those who lack wealth. By using imprisonment rather than fines as a punishment, jury trials instead of single judge presiding, and prohibiting the lease of prison labor to the private sector, the system increases the costs of potential cooperation and therefore limits the corresponding expropriation.
Conclusion
Thompson’s idea of cooperation costs is an example of the sort of analysis that UCLA price theorists were known for. By focusing on particular types of costs and how those costs influence the nature of exchange, he was able to formulate hypotheses about why countries end up in development traps, why quantity regulations are sometimes used to deal with externalities, and why the state uses imprisonment rather than fines as punishment for crime. He has other examples as well. He argued that these types of costs might explain elements of anti-trust law and were at the root of the public provision of certain services, like firefighting and flood insurance. This argument is also indicative of Thompson’s style, both provocative and thought-provoking.