Research Area C2

Information economics

PIs: Moldovano, Rady

Contributions by Dilmé,Hoffmann,Krähmer,Lauermann,Pollrich,Szalay,Tran

 

 

Topic and goals

Information shapes the way individuals and groups make decisions. Modern economic theory recognizes information as a scarce good that is dispersed among many agents: time and resources have to be spent to acquire information, and individuals typically differ in the information generating technology they can deploy. The performance of institutions that govern collective decision-making crucially depends on the incentives they provide for individuals to acquire information, on their ability to aggregate information, and on the information to which they themselves give access. The research area thus focuses on three themes: information acquisition, information aggregation, and information design.

 

 

State of the art, our expertise

Information acquisition. In many situations, decision-relevant information must be acquired at a cost. A mechanism then not only serves to elicit information ex post but, as emphasized by Bergemann and Välimäki, must also provide appropriate incentives for its acquisition ex ante. This may lead to additional distortions beyond those that are caused by agents’ private information. As a case in point, Krähmer–Strausz [KS11] considered a procurement problem where the supplier chooses how much information to acquire about his production costs. Depending on the supplier’s ex ante information, the optimal contract systematically induces inefficiently low or inefficiently high levels of information acquisition. Also considering a procurement setting, Szalay [Sza09] allowed for general information acquisition technologies. In [Sza05], he determined optimal decision procedures and information acquisition in a communication game with commitment. In ongoing work, by contrast, Deimen and Szalay assume no commitment to decision rules. When information can only be acquired by costly search or experimentation, an intertemporal trade-off arises between the immediate payoff today and the option value of a possibly higher payoff tomorrow. Klein–Rady [KR11] studied a strategic experimentation problem in which the characteristics of the risky options to be explored are negatively correlated across players; see Figure (a). Rady et al. [HRS15] analyzed how privacy of payoff information affects the equilibrium level of experimentation. Rady et al. [PRT17] investigated optimal experimentation by a platform provider who is uncertain about the externalities that the participants exert on each other.                          

Information aggregation. Centralized markets were among the first institutions studied and advocated for their ability to elicit and aggregate information. In recent work, Lauermann et al. [LW16, LW17, LMV17] addressed information aggregation in decentralized search settings. Moldovanu et al. [JM01, JMMZ06] advanced the general normative analysis of institutions in the presence of interdependent preferences, identifying the theoretical limits imposed by incentive constraints in any market or non-market context. Tran established connections between the theory of mechanism design and tropical geometry. Analogously to markets, elections – and institutions for collective decision-making more generally – solve informational problems when participants have different priorities and may be uncertain about what candidate or policy best serves their interests. Applications here include voting in committees and legislatures, general elections and shareholder voting. Moldovanu and coauthors’ recent contributions [GMS16, KM17] to optimal voting rules with private information – see Figure (b) – provide benchmarks and suggestions for the design of good institutions.

Information design. Traditionally, mechanism design takes individuals’ information as given and optimizes over different forms of interaction among individuals. Information design reverses this perspective by taking the interaction among individuals as given but allowing the designer to control the information to which individuals get access. Interest in this problem was boosted by Kamenica and Gentzkow’s analysis of a sender-receiver game in which the sender can successfully manipulate the receiver’s actions by only controlling the latter’s access to information. Identifying an order on information structures in games of incomplete information, Bergemann and Morris showed how to approach such Bayesian persuasion in multi-player settings. An important application of information design is disclosure in markets. Sellers frequently provide information to consumers prior to the purchasing decision, for instance in the form of product samples. Hoffmann–Inderst [HI11] characterized the optimal discriminatory disclosure policy when increasing the quality of pre-sale information is costly. Krähmer–Strausz [KS15] identified the information rents a monopolist has to concede by providing additional information and showed that they may cause inefficient partial disclosure. Dilmé–Li [DL16] highlighted the inefficiency that may arise when private information is revealed through actions rather than disclosure. As shown by Pollrich [Pol17], sophisticated communication is also crucial in principal-agent relationships with auditing.

 

 

Research program

Information acquisition. While search theory typically postulates that all traders know the distribution of the terms of trade from which they are sampling, we will study search under uncertainty about this distribution, allowing participants to learn from experience. The Poisson-process modeling approach adopted in previous work by Rady promises to be particularly useful for deriving closed-form solutions. It may also help us examine how agents will optimally learn when they can choose among various options (or markets), in addition to choosing a stopping time as in conventional search models. Learning dynamics will also be at the center of our investigation of information acquisition via experimentation. Building on [KR11], we will consider more general correlation structures between the characteristics of the different players’ risky projects. We will also address experimentation with restless bandits whose characteristics change stochastically over time, so that there is an incentive to learn even in the very long run. Further research into the dynamics of market platforms with cross-group externalities will build on [PRT17]. In all these settings, a pre- cise understanding of the stochastic process of posterior beliefs is crucial. The analysis of the relevant stochastic differential equations will benefit from cooperation with RA B1.                        

Information aggregation. Challenges in the theory of information aggregation arise both in settings where monetary transfers are feasible and in settings where they are not. In decentralized markets with large frictions, it is important to understand how uncertainty about aggregate market conditions affects market outcomes through its impact on bargaining and search, and whether the release of public information can enhance efficiency. We will address these questions in equilibrium search models that build on [LMV17] and incorporate learning, and in auction models that build on [LW17] and feature an auctioneer searching for bidders. While these approaches take a particular trading institution as given, we will also contribute to the theoretical foundations of trading institutions by advancing the theory of mechanism design with transfers. This will have important repercussions on our understanding of auctions and two-sided matching and, at the same time, raise interesting technical issues. The generalization of revenue-maximizing mechanisms to environments with convex preferences, for instance, will require tools from convex optimization and majorization, together with a characterization of the extreme points of particular sets of monotone functions in L1via doubly stochastic operators. The work on matching will use tools from transport theory, establishing a close link with RA B3. When transfers are infeasible, collective choices are often made by means of voting procedures. We will study the effectiveness of particular voting institutions in the presence of interdependent values, emphasizing ‘informal elections’ such as petitions, social media, and protests. This will be embedded in investigations of the general design of mechanisms without transfers. An important open problem here is the identification of settings in which there is equivalence between Bayesian and dominant-strategy implementation; for mechanisms with transfers, this equivalence was established by Moldovanu et al. in [GGK+13] using tools from the mathematics of computed tomography. In the specific context of voting problems, Moldovanu and coauthors [GMS16, KM17] showed that sequential binary voting procedures possess strong optimality properties under single-peaked preferences. The challenge is to extend these results to environments with interdependent valuations, for which it is well known from analogous auction settings that dynamic procedures can improve upon static ones. Like in our research on dynamic information acquisition, stochastic dynamic programming and Bayesian learning will play a central role here.                            

Information design. Existing work on information design largely abstracts from constraints imposed by the (market) environment in which information transmission takes place. Thus, a major challenge in information design is to understand the extent to which plausible economic constraints affect the designer’s ability to induce the desired outcome. Such constraints can often be operationalized with techniques from convex analysis. As a leading application, we will study information disclosure in auctions, where a designer is typically unable to inform a bidder about the valuations of competing bidders. We will also extend existing frameworks by allowing a seller to randomize over various information disclosure rules, creating potentially useful correlation between the seller’s and the buyers’ information. The standard information-design paradigm assumes that information is directly passed on to a decision-maker. By contrast, we will study information design when the information is provided to a sender in a game of strategic information transmission. Since the sender acts as a strategic filter, the communication channel puts natural constraints on the outcomes that can be induced through information design. We plan to extend this analysis to market settings, where existing theory assumes that players receive information directly from non-strategic sources, whereas in practice many information providers (such as central bankers or market makers) are clearly motivated by strategic considerations. We will examine the optimal timing of disclosure by strategic agents, the inefficiencies that delayed information disclosure involves, and the value of committing to a particular disclosure strategy. One major aim is to determine policy interventions that minimize the welfare cost of inefficient disclosure strategies or induce agents to commit to a socially desirable disclosure strategy.  

 

 

Summary

Combining a broad range of economic approaches and mathematical tools, RA C2 addresses open questions in information acquisition, information aggregation, and information design. Its aim is to produce fundamental insights into the workings of important market and non-market institutions for resource allocation and collective decision making. The application of transport theory and the analysis of stochastic dynamics create connections with RAs B1 and B3.

 

 

Bibliography

[DL16] F. Dilmé and F. Li. Dynamic signaling with dropout risk. Am. Econ. J.: Microecon., 8(1):57–82, 2016.

[GGK+13] A. Gershkov, J. K. Goeree, A. Kushnir, B. Moldovanu, and X. Shi. On the equivalence of Bayesian and dominant strategy implementation. Econometrica, 81(1):197–220, 2013.

[GMS16] A. Gershkov, B. Moldovanu, and X. Shi. Optimal voting rules. Rev. Econ. Stud., 84(2):688– 717, 2016.

[HI11] F. Hoffmann and R. Inderst. Pre-sale information. J. Econ. Theory, 146(6):2333–2355, 2011.

[HRS15] P. Heidhues, S. Rady, and P. Strack. Strategic experimentation with private payoffs. J. Econ. Theory, 159(A):531–551, 2015.

[JM01] P. Jehiel and B. Moldovanu. Efficient design with interdependent valuations. Econometrica, 69(5):1237–1259, 2001.

[JMMZ06] P. Jehiel, M. Meyer-ter-Vehn, B. Moldovanu, and W. R. Zame. The limits of ex-post implementation. Econometrica, 74(3):585–610, 2006.

[KM17] A. Kleiner and B. Moldovanu. Content-based agendas and qualified majorities in sequential voting. Am. Econ. Rev., 107(6):1477–1506, 2017.

[KR11] N. Klein and S. Rady. Negatively correlated bandits. Rev. Econ. Stud., 78(2):693–732, 2011.

[KS11] D. Krähmer and R. Strausz. Public procurement with pre-project planning. Rev. Econ. Stud., 78(3):1015–1041, 2011.

[KS15] D. Krähmer and R. Strausz. Ex post information rents in sequential screening. Games Econ. Behav., 90:257–273, 2015.

[LMV17] S. Lauermann, W. Merzyn, and G. Virág. Learning and price discovery in a search model. Rev. Econ. Stud., 2017. DOI:10.1093/restud/rdx029.

[LW16] S. Lauermann and A. Wolinsky. Search with adverse selection. Econometrica, 84(1):243– 315, 2016.

[LW17] S. Lauermann and A. Wolinsky. Bidder solicitation, adverse selection, and the failure of competition. Am. Econ. Rev., 107(6):1399–1429, 2017.

[Pol17] M. Pollrich. Mediated audits. RAND J. Econ., 48(1):44–68, 2017.

[PRT17] M. Peitz, S. Rady, and P. Trepper. Experimentation in two-sided markets. J. Eur. Econ. Assoc., 15(1):128–172, 2017.

[Sza05] D. Szalay. The economics of extreme options and clear advice. Rev. Econ. Stud., 72(4):1173– 1198, 2005.

[Sza09] D. Szalay. Contracts with endogenous information. Games Econ. Behav., 65:586–625, 2009.