Moral judgments about economic transactions: Luke Coffman

Lucas (Luke) Coffman defended his dissertation yesterday. He's an eclectic experimenter, and one of his papers looks at the assignment of credit or blame, and how that is influenced by the presence of intermediaries. For example (to pick a Harvard-centric one), is Harvard viewed differently if it hires janitors directly at a low wage than if it contracts with a janitorial services company that employs the janitors?

The baseline condition of one of his experiments is easy to describe: one participant (who you can think of as Harvard) is endowed with $10, which he can divide with a second participant (who you can think of as a janitor), or instead can sell the right to divide the $10 to a third participant (who you can think of as the janitorial services company). Luke then elicits a judgment of the transaction from a fourth party, who is able to punish the first party by reducing his payoff. The results are clear: for a given (low) amount delivered to the “janitor,” punishments are considerably reduced if it is delivered indirectly, through a third party, rather than directly.

Luke designed and conducted many careful controls to better understand what is going on, and rule out plausible alternative hypotheses. (For one thing, choosing to use an intermediary doesn’t seem to fool anyone; people correctly anticipate that using an intermediary will be bad for the lowest paid member of the group, but they nevertheless find it less blameworthy.) One way to think about his results is that they suggest that fairness judgments may be very narrowly framed, and confined more than we had any reason to suspect to very direct interactions, so that intermediated interactions are judged differently than direct interactions.

Luke will be an assistant professor of economics at The Ohio State University next year.

Welcome to the club, Luke.

Networks in markets that unravel, and those that don't: Itay Fainmesser

Itay Fainmesser defended his dissertation yesterday.

One of the papers in his dissertation concerns job markets that have “unraveled” so that a large part of the market consists of early, exploding offers. He develops a network model, motivated by the observation that when many markets unravel (as when medical labor markets start to hire doctors well over a year before they begin employment), hiring also becomes more local (as hospitals start to hire students from local medical schools, etc.). Itay takes this as evidence that when hiring is very early, employers are forced to rely more on their local networks for information. He builds a network model that allows him to investigate which properties of local networks lead to unraveling, and which lead to later hiring. In this model, information about the quality of candidates eventually becomes widely available, but early information about candidate quality can only be reliably transmitted along links of a network. (When Markets Unravel: Social Networks, Information Transmission, and the 'Hiring Frenzy' older version here.)

Another of his papers tackles the question of cooperation in repeated games, where the possible interactions are constrained by a network, and he asks which buyer-seller network structures will support persistent cooperation (where sellers have an opportunity to cooperate by shipping a high quality good, or to defect by shipping a low quality good). It’s a hard problem, and (in a third paper) he and a coauthor invent models and tools to deal with it (in a large-network framework), that allow him to turn some difficult non-monotonic relationships among networks into well behaved statements about the value of links. (Community Structure and Market Outcomes: Towards a Theory of Repeated Games in Networks )

Itay will be an assistant professor of economics at Brown next year.

Welcome to the club, Itay.

Efficient random allocation of discrete goods, Mihai Manea

In this season of Ph.D. dissertation defenses, Mihai Manea just completed the last defense that I'm intimately involved in. His work includes papers on the ex-ante efficiency of allocations involving lotteries in large markets.
This is a topic that comes up in school choice systems, for example, with the random element coming from the fact that a given school may not have the capacity to serve all the students who would like to enroll, so that some random tie breaking procedure has to be employed.

An important paper by Bogolmanaia and Moulin (2001) noted that random serial dictatorship, although ex post efficient, may be ex ante inefficient. They proposed an "ordinally efficient" random mechanism called the probabilistic serial mechanism. An ordinally efficient mechanism chooses a probability matrix (assigning each player some probability of obtaining each object) that cannot be Pareto improved upon by another random assignment that first order stochastically dominates the first.

Random serial dictatorship is the procedure for allocating indivisible goods by putting the claimants in a random order, and then allowing the first to choose the object he prefers, the second to choose the object he prefers from those that remain, etc. This results in a lottery over ex post efficient allocations, which B&M noted may be an inefficient lottery: it induces a stochastic allocation matrix with the property that some claimants would be willing to trade some of their probability of getting one object with one another, to yield a mutually preferable stochastic matrix. One of Mihai's early papers shows that efficient probability-exchange contracts of this sort can be written by agents agreeing to exchange their orders in some realizations: i.e. the agents can write an ordinally efficient ordering-exchange contract that they all prefer to the random serial dictatorship allocation in the sense of first-order stochastic dominance. (Random Serial Dictatorship and Ordinally Efficient Contracts, International Journal of Game Theory 2008.)

One drawback of the probabilistic serial mechanism is that, while it is ordinally efficient (unlike random serial dictatorship) it does not make it a dominant strategy for agents to state their true preferences (again, unlike random serial dictatorship). Mihai and Fuhito Kojima show that this drawback goes away when the number of identical copies of each object is sufficiently large: Strategy-Proofness of the Probabilistic Serial Mechanism in Large Random Assignment Problems . Their result isn't just a limit theorem, rather, they show that there is a finite number such that, when the number of identical copies of each object is larger than that, the psm is strategy proof.

Separately, Mihai and Fuhito study the efficiency of random serial dictatorship in different ways as markets grow large. Mihai shows that , as the market becomes large (in particular, as the number of types of objects grows along with the number of participants), the proportion of preference profiles at which rsd is ordinally inefficient goes to one in the limit. (Asymptotic Ordinal Inefficiency of Random Serial Dictatorship, forthcoming Theoretical Economics.)

Fuhito, together with Yeon-Koo Che, shows, in a market that grows large in a different way (the number of copies of each object type grows along with the number of participants) that random serial dictatorship converges to the probabilistic serial mechanism. (Asymptotic Equivalence of Probabilistic Serial and Random Priority Mechanisms , 2008.)

In general, looking at efficiency ex-ante as well as ex post opens up new avenues for investigation, and how these interact in large markets has been a fruitful way to start thinking about them.

Welcome to the club, Mihai.

Behavioral contract design; Steve Leider

Steve Leider defended his dissertation last week. He's an eclectic experimenter, but among his varied interests is how insights from psychology might change our view of contract design. In particular, if people are nicer than the standard economic model supposes (e.g. more inclined to reciprocate favors, more inclined to keep promises and uphold social norms), how might that change our views of how contracts might function, and therefore how they should be structured?

Among his papers, the following best exemplify that part of his work.

Norms and Contracting (with Judd Kessler) (Job Market Paper) Abstract: We argue that agents create norms specific to their relationships, particularly through the contracts they establish. We build a theory of how the enforceable and unenforceable aspects of a contract determine the norm, and how norms impact behavior. We then demonstrate experimentally that even totally incomplete contracts (i.e. contracts with no enforceable restriction on actions) move behavior substantially towards the first best in a variety of games. A contract with only unenforceable agreements is often more effective than a contract with only enforceable restrictions. Combining enforceable restrictions with an unenforceable agreement is frequently no more effective (and sometimes strictly less effective) than an unenforceable agreement alone. Consistent with our modeling approach that violating the norm creates disutility, many subjects often choose not to make unenforceable agreements, despite earnings substantially higher payoffs with the agreement.

Directed Altruism and Enforced Reciprocity in Social Networks (with Markus M. Mobius, Tanya Rosenblat, and Quoc-Anh Do) [Forthcoming in the Quarterly Journal of Economics] Abstract: We conduct online field experiments in large real-world social networks in order to decompose prosocial giving into three components: (1) baseline altruism towards randomly selected strangers, (2) directed altruism that favors friends over random strangers, and (3) giving motivated by the prospect of future interaction. Directed altruism increases giving to friends by 52 percent relative to random strangers, while future interaction effects increase giving by an additional 24 percent when giving is socially efficient. This finding suggests that future interaction affects giving through a repeated game mechanism where agents can be rewarded for granting efficiency-enhancing favors. We also find that subjects with higher baseline altruism have friends with higher baseline altruism.

Contractual and Organizational Structure with Reciprocal Agents (with Florian Englmaier) [Submitted] Abstract: Empirically, compensation systems generate substantial effort despite weak monetary incentives. We consider reciprocal motivations as a source of incentives. We solve for the optimal contract in the basic principal-agent problem and show that reciprocal motivations and explicit performance-based pay are substitutes. A firm endogenously determines the mix of the two sources of incentives to best induce effort from the agent. Analyzing extended versions of the model allows us to examine how organizational structure impacts the effectiveness of reciprocity and to derive specific empirical predictions. We use the UK-WERS workplace compensation data set to confirm the predictions of our extended model.

Gift Exchange in the Lab and in the Field - It is not (only) how much you give ... (with Florian Englmaier) Abstract: We build on the theoretical results from our companion paper, Englmaier and Leider (2008), that an important aspect in determining the effectiveness of gift exchange relations is the ability of the agent to “repay the gift” to the principal. To test this hypothesis, we conduct a real effort laboratory experiment and a field experiment where we vary the effect of the agent’s effort on the principal’s payoff. Furthermore we collect additional information that allows us to control for the agents’ effort costs and whether they can be classified as reciprocal or not. From our model we derive nontrivial predictions about which is the marginal agent in terms of ability affected by our experimental variation and how different types of individuals, selfish and reciprocal, will react to it. The experimental data lend support to our hypotheses.

Steve's email address will have a "umich.edu" in it starting next semester, when he starts work at the University of Michigan's Ross School of Business.

Welcome to the club, Steve.

Course allocation, by Eric Budish

The problem of allocating courses to students is a famously hard problem of market design. The reasons include the fact that, in most applications, it isn't acceptable to sell the most desirable class places at higher prices to richer students. Also, students take multiple classes, and may have preferences for how their bundle is composed. So the problem is substantially more difficult than how to auction multiple goods, or how to allocate each student a single place, as comes up e.g. in assigning students to schools.

Eric Budish, who defended his Ph.D. dissertation at Harvard this week, has made a substantial, practical dent in the problem. His motivation comes from a detailed study, with Estelle Cantillon, of how classes are assigned to second year MBA students at the Harvard Business School, and how students approach this assignment problem strategically: Strategic Behavior in Multi-Unit Assignment Problems: Theory and Evidence from Course Allocations .

Largely motivated by what they learn about the good and not so good properties of the HBS mechanism, Eric then proposes a new mechanism: The Combinatorial Assignment Problem: Approximate Competitive Equilibrium from Equal Incomes. Eric's work, like market design in general, is eclectic. Among other things, he formulates new notions of what constitutes "fair" outcomes in cases hedged in by the impossibility results that abound when allocating indivisible goods.

Although allocating multiple indivisible items to each student makes the standard economic goals involving efficiency and incentives more difficult to achieve, it gives the designer somewhat more leeway to think about fairness, since although class places are indivisible, the package of classes that each student gets is not. And Eric’s investigations of existing course allocation institutions has convinced him that concerns about avoiding excessive ex-post unfairness are an important constraint on what kinds of mechanisms can be implemented in practice.

Eric's mechanism looks like it has legs, and may be ready for practical implementation in the not so distant future. Perhaps he'll get a chance to have more than the usual impact next year when he brings market design to U. Chicago's Booth School of Business (until recently Chicago GSB).

Welcome to the club, Eric.