Experimental Games on Networks: Underpinnings of Behavior and Equilibrium Selection

In this paper, we describe a series of laboratory experiments that implement specific examples of a general network structure. Specifically, actions are either strategic substitutes or strategic complements, and participants have either complete or incomplete information about the structure of a random network. Since economic environments typically have a considerable degree of complementarity or substitutability, this framework applies to a wide variety of settings. We examine behavior and equilibrium selection. The degree of equilibrium play is striking, in particular with incomplete information. Behavior closely resembles the theoretical equilibrium whenever this is unique; when there are multiple equilibria, general features of networks, such as connectivity, clustering, and the degree of the players, help to predict informed behavior in the lab. People appear to be strongly attracted to maximizing aggregate payoffs (social efficiency), but there are forces that moderate this attraction: (1) people seem content with (in the aggregate) capturing only the lion's share of the efficient profits in exchange for reduced exposure to loss, and (2) uncertainty about the network structure makes it considerably more difficult to coordinate on a demanding, but efficient, equilibrium that is typically implemented with complete information.     

Altruism in Networks

We provide the first analysis of altruism in networks. Agents are embedded in a fixed network and care about the well‐being of their network neighbors. Depending on incomes, they may provide financial support to their poorer friends. We study the Nash equilibria of the resulting game of transfers. We show that equilibria maximize a concave potential function. We establish existence, uniqueness of equilibrium consumption, and generic uniqueness of equilibrium transfers. We characterize the geometry of the network of transfers and highlight the key role played by transfer intermediaries. We then study comparative statics. A positive income shock to an individual benefits all. For small changes in incomes, agents in a component of the network of transfers act as if they were organized in an income‐pooling community. A decrease in income inequality or expansion of the altruism network may increase consumption inequality.     

Supply Networks as a Complex Adaptive System: Toward Simulation‐Based Theory Building on Evolutionary Decision Making

In this article, we examine how the firms embedded in supply networks engage in decision making over time. The supply networks as a complex adaptive system are simulated using cellular automata (CA) through a dynamic evolution of cooperation (i.e., “voice” decision) and defection (i.e., “exit” decision) among supply network agents (i.e., firms). Simple local rules of interaction among firms generate complex patterns of cooperation and defection decisions in the supply network. The incentive schemes underlying decision making are derived through different configurations of the payoff‐matrix based on the game theory argument. The prisoner's dilemma game allows capturing the localized decision‐making process by rational agents, and the CA model allows the self‐organizing outcome to emerge. By observing the evolution of decision making by cooperating and defecting agents, we offer testable propositions regarding relationship development and distributed nature of governance mechanisms for managing supply networks.     

Existence of Optimal Mechanisms in Principal‐Agent Problems

We provide general conditions under which principal‐agent problems with either one or multiple agents admit mechanisms that are optimal for the principal. Our results cover as special cases pure moral hazard and pure adverse selection. We allow multidimensional types, actions, and signals, as well as both financial and non‐financial rewards. Our results extend to situations in which there are ex ante or interim restrictions on the mechanism, and allow the principal to have decisions in addition to choosing the agent's contract. Beyond measurability, we require no a priori restrictions on the space of mechanisms. It is not unusual for randomization to be necessary for optimality and so it (should be and) is permitted. Randomization also plays an essential role in our proof. We also provide conditions under which some forms of randomization are unnecessary.     

Berk–Nash Equilibrium: A Framework for Modeling Agents With Misspecified Models

We develop an equilibrium framework that relaxes the standard assumption that people have a correctly specified view of their environment. Each player is characterized by a (possibly misspecified) subjective model, which describes the set of feasible beliefs over payoff‐relevant consequences as a function of actions. We introduce the notion of a Berk–Nash equilibrium: Each player follows a strategy that is optimal given her belief, and her belief is restricted to be the best fit among the set of beliefs she considers possible. The notion of best fit is formalized in terms of minimizing the Kullback–Leibler divergence, which is endogenous and depends on the equilibrium strategy profile. Standard solution concepts such as Nash equilibrium and self‐confirming equilibrium constitute special cases where players have correctly specified models. We provide a learning foundation for Berk–Nash equilibrium by extending and combining results from the statistics literature on misspecified learning and the economics literature on learning in games.     

A Structural Model of Dense Network Formation

This paper proposes an empirical model of network formation, combining strategic and random networks features. Payoffs depend on direct links, but also link externalities. Players meet sequentially at random, myopically updating their links. Under mild assumptions, the network formation process is a potential game and converges to an exponential random graph model (ERGM), generating directed dense networks. I provide new identification results for ERGMs in large networks: if link externalities are nonnegative, the ERGM is asymptotically indistinguishable from an Erdős–Rényi model with independent links. We can identify the parameters only when at least one of the externalities is negative and sufficiently large. However, the standard estimation methods for ERGMs can have exponentially slow convergence, even when the model has asymptotically independent links. I thus estimate parameters using a Bayesian MCMC method. When the parameters are identifiable, I show evidence that the estimation algorithm converges in almost quadratic time.     

Program Evaluation and Causal Inference With High‐Dimensional Data

In this paper, we provide efficient estimators and honest confidence bands for a variety of treatment effects including local average (LATE) and local quantile treatment effects (LQTE) in data‐rich environments. We can handle very many control variables, endogenous receipt of treatment, heterogeneous treatment effects, and function‐valued outcomes. Our framework covers the special case of exogenous receipt of treatment, either conditional on controls or unconditionally as in randomized control trials. In the latter case, our approach produces efficient estimators and honest bands for (functional) average treatment effects (ATE) and quantile treatment effects (QTE). To make informative inference possible, we assume that key reduced‐form predictive relationships are approximately sparse. This assumption allows the use of regularization and selection methods to estimate those relations, and we provide methods for post‐regularization and post‐selection inference that are uniformly valid (honest) across a wide range of models. We show that a key ingredient enabling honest inference is the use of orthogonal or doubly robust moment conditions in estimating certain reduced‐form functional parameters. We illustrate the use of the proposed methods with an application to estimating the effect of 401(k) eligibility and participation on accumulated assets. The results on program evaluation are obtained as a consequence of more general results on honest inference in a general moment‐condition framework, which arises from structural equation models in econometrics. Here, too, the crucial ingredient is the use of orthogonal moment conditions, which can be constructed from the initial moment conditions. We provide results on honest inference for (function‐valued) parameters within this general framework where any high‐quality, machine learning methods (e.g., boosted trees, deep neural networks, random forest, and their aggregated and hybrid versions) can be used to learn the nonparametric/high‐dimensional components of the model. These include a number of supporting auxiliary results that are of major independent interest: namely, we (1) prove uniform validity of a multiplier bootstrap, (2) offer a uniformly valid functional delta method, and (3) provide results for sparsity‐based estimation of regression functions for function‐valued outcomes.     

A Theory of Macroprudential Policies in the Presence of Nominal Rigidities

We propose a theory of monetary policy and macroprudential interventions in financial markets. We focus on economies with nominal rigidities in goods and labor markets and subject to constraints on monetary policy, such as the zero lower bound or fixed exchange rates. We identify an aggregate demand externality that can be corrected by macroprudential interventions in financial markets. Ex post, the distribution of wealth across agents affects aggregate demand and output. Ex ante, however, these effects are not internalized in private financial decisions. We provide a simple formula for the required financial interventions that depends on a small number of measurable sufficient statistics. We also characterize optimal monetary policy. We extend our framework to incorporate pecuniary externalities, providing a unified approach to both externalities. Finally, we provide a number of applications which illustrate the relevance of our theory.     

Investment and Competitive Matching

We study markets in which agents first make investments and are then matched into potentially productive partnerships. Equilibrium investments and the equilibrium matching will be efficient if agents can simultaneously negotiate investments and matches, but we focus on markets in which agents must first sink their investments before matching. Additional equilibria may arise in this sunk‐investment setting, even though our matching market is competitive. These equilibria exhibit inefficiencies that we can interpret as coordination failures. All allocations satisfying a constrained efficiency property are equilibria, and the converse holds if preferences satisfy a separability condition. We identify sufficient conditions (most notably, quasiconcave utilities) for the investments of matched agents to satisfy an exchange efficiency property as well as sufficient conditions (most notably, a single crossing property) for agents to be matched positive assortatively, with these conditions then forming the core of sufficient conditions for the efficiency of equilibrium allocations.     

A Service Provider's Elicitation of Its Customers’ Demand Distributions by a Price Mechanism

In a service environment a service provider needs to determine the amount and kinds of capacity to meet customers’ needs over many periods. To make good decisions, she needs to know the probability distribution of her customers’ demand in each period. We study a situation in which customers’ demand for a given service is random in each period, but inelastic, or modeled well by this assumption, and cannot be delayed to the next period. This article presents a mechanism that allows a service provider to learn the distribution of a customer's demand by offering him a set of contracts through which he can partially prepay for future service for a reduced cost for units of service based on anticipated needs. We describe the form of a set of contracts that will cause the customer to reveal his demand distribution as he minimizes his expected costs. To justify the effort of organizing and offering contracts, we present an application that demonstrates the cost savings to the service provider with better capacity planning using the truthfully elicited distribution.     

Segmentation Opportunities for a Social Planner: Impact of Limited Resources

In this research note, we investigate segmentation opportunities for social planners such as government agencies, nonprofits, and public organizations. These opportunities arise when the potential products are vertically (quality) differentiated and the consumers are heterogeneous in their preferences toward quality. In these cases, whether to offer quality differentiated products and what quality level to choose are important decisions for a social planner. In this research note, we identify the conditions where it is socially optimal to offer either one homogenous or two quality differentiated products. We find that the resource limitations may result in a single product offering and that the quality of the product depends on the maximum surplus per unit resource consumed by the products. We also compare our findings to a profit‐maximizing firm. We find that the resource limitations may cause a profit‐maximizing firm to provide a better service to some consumers than the social planner. Contrary to common wisdom, we also show that the capacity limitations may force the social planner to act like a profit‐maximizing firm in terms of its pricing and product mix choice.     

Implementation With Contingent Contracts

We study dominant strategy incentive compatibility in a mechanism design setting with contingent contracts where the payoff of each agent is observed by the principal and can be contracted upon. Our main focus is on the class of linear contracts (one of the most commonly used contingent contracts) which consist of a transfer and a flat rate of profit sharing. We characterize outcomes implementable by linear contracts and provide a foundation for them by showing that, in finite type spaces, every social choice function that can be implemented using a more general nonlinear contingent contract can also be implemented using a linear contract. We then qualitatively describe the set of implementable outcomes. We show that a general class of social welfare criteria can be implemented. This class contains social choice functions (such as the Rawlsian) which cannot be implemented using (uncontingent) transfers. Under additional conditions, we show that only social choice functions in this class are implementable.     

Usage‐Based Pricing and Demand for Residential Broadband

We estimate demand for residential broadband using high‐frequency data from subscribers facing a three‐part tariff. The three‐part tariff makes data usage during the billing cycle a dynamic problem, thus generating variation in the (shadow) price of usage. We provide evidence that subscribers respond to this variation, and we use their dynamic decisions to estimate a flexible distribution of willingness to pay for different plan characteristics. Using the estimates, we simulate demand under alternative pricing and find that usage‐based pricing eliminates low‐value traffic. Furthermore, we show that the costs associated with investment in fiber‐optic networks are likely recoverable in some markets, but that there is a large gap between social and private incentives to invest.     

Time Horizon and Cooperation in Continuous Time

We study social dilemmas in (quasi‐) continuous‐time experiments, comparing games with different durations and termination rules. We discover a stark qualitative contrast in behavior in continuous time as compared to previously studied behavior in discrete‐time games: cooperation is easier to achieve and sustain with deterministic horizons than with stochastic ones, and end‐game effects emerge, but subjects postpone them with experience. Analysis of individual strategies provides a basis for a simple reinforcement learning model that proves to be consistent with this evidence. An additional treatment lends further support to this explanation.     

Consumption Commitments and Habit Formation

We analyze the implications of household‐level adjustment costs for the dynamics of aggregate consumption. We show that an economy in which agents have “consumption commitments” is approximately equivalent to a habit formation model in which the habit stock is a weighted average of past consumption if idiosyncratic risk is large relative to aggregate risk. Consumption commitments can thus explain the empirical regularity that consumption is excessively sensitive and excessively smooth, findings that are typically attributed to habit formation. Unlike habit formation and other theories, but consistent with empirical evidence, the consumption commitments model also predicts that excess sensitivity and smoothness vanish for large shocks. These results suggest that behavior previously attributed to habit formation may be better explained by adjustment costs. We develop additional testable predictions to further distinguish the commitment and habit models and show that the two models have different welfare implications.     

Utilitarian Preferences With Multiple Priors

This paper proposes a method for aggregating individual preferences in the context of uncertainty. Individuals are assumed to abide by Savage's model of Subjective Expected Utility, in which everyone has his/her own utility and subjective probability. Disagreement on probabilities among individuals gives rise to uncertainty at the societal level, and thus society may entertain a set of probabilities rather than only one. We assume that social preference admits a Maxmin Expected Utility representation. In this context, two Pareto‐type conditions are shown to be equivalent to social utility being a weighted average of individual utilities and the social set of priors containing only weighted averages of individual priors. Thus, society respects consensus among individuals' beliefs and does not add ambiguity beyond disagreement on beliefs. We also deal with the case in which society does not rule out any individual belief.     

Stochastic Choice and Revealed Perturbed Utility

Perturbed utility functions—the sum of expected utility and a nonlinear perturbation function—provide a simple and tractable way to model various sorts of stochastic choice. We provide two easily understood conditions each of which characterizes this representation: One condition generalizes the acyclicity condition used in revealed preference theory, and the other generalizes Luce's IIA condition. We relate the discrimination or selectivity of choice rules to properties of their associated perturbations, both across different agents and across decision problems. We also show that these representations correspond to a form of ambiguity‐averse preferences for an agent who is uncertain about her true utility.     

Networks in Conflict: Theory and Evidence From the Great War of Africa

We study from both a theoretical and an empirical perspective how a network of military alliances and enmities affects the intensity of a conflict. The model combines elements from network theory and from the politico‐economic theory of conflict. We obtain a closed‐form characterization of the Nash equilibrium. Using the equilibrium conditions, we perform an empirical analysis using data on the Second Congo War, a conflict that involves many groups in a complex network of informal alliances and rivalries. The estimates of the fighting externalities are then used to infer the extent to which the conflict intensity can be reduced through (i) dismantling specific fighting groups involved in the conflict; (ii) weapon embargoes; (iii) interventions aimed at pacifying animosity among groups. Finally, with the aid of a random utility model, we study how policy shocks can induce a reshaping of the network structure.     

Rational Inattention Dynamics: Inertia and Delay in Decision‐Making

We solve a general class of dynamic rational inattention problems in which an agent repeatedly acquires costly information about an evolving state and selects actions. The solution resembles the choice rule in a dynamic logit model, but it is biased toward an optimal default rule that is independent of the realized state. The model provides the same fit to choice data as dynamic logit, but, because of the bias, yields different counterfactual predictions. We apply the general solution to the study of (i) the status quo bias; (ii) inertia in actions leading to lagged adjustments to shocks; and (iii) the tradeoff between accuracy and delay in decision‐making.     

Tenable Strategy Blocks and Settled Equilibria

When people interact in familiar settings, social conventions usually develop so that people tend to disregard alternatives outside the convention. For rational players to usually restrict attention to a block of conventional strategies, no player should prefer to deviate from the block when others are likely to act conventionally and rationally inside the block. We explore two set‐valued concepts, coarsely and finely tenable blocks, that formalize this notion for finite normal‐form games. We then identify settled equilibria, which are Nash equilibria with support in minimal tenable blocks. For a generic class of normal‐form games, our coarse and fine concepts are equivalent, and yet they differ from standard solution concepts on open sets of games. We demonstrate the nature and power of the solutions by way of examples. Settled equilibria are closely related to persistent equilibria but are strictly more selective on an open set of games. With fine tenability, we obtain invariance under the insertion of a subgame with a unique totally mixed payoff‐equivalent equilibrium, a property that other related concepts have not satisfied.