Olga Gorelkina

We study methods to enhance statistical privacy in blockchain transactions. We analyze economic mechanisms for privacy-aware transaction owners whose utility depends not only on the outcome of the mechanism but also negatively on the exposure of their economic preferences. First, we consider an order flow auction, where a user auctions off to specialized agents, called searchers, the right to execute her transaction while maintaining a degree of privacy, and characterize the optimal auction, which is a sealed-bid auction. We then analyze a variant of a Dutch auction in which the user gradually decreases the price and the degree of privacy until the transaction is sold. Finally, we introduce a two-sided privacy marketplace and propose a posted-price mechanism that guarantees a constant approximation of the optimal social welfare while maintaining incentive compatibility on both sides of the market and budget balance.

This paper presents a novel solution concept, called BAR Nash Equilibrium (BARNE), and applies it to the Verifier's Dilemma, a fundamental problem in the design of blockchain consensus. BARNE extends the Nash Equilibrium to accommodate interactions among Byzantine, altruistic, and rational agents, known as the BAR setting in the distributed computing literature. We prove the existence of BARNE in a broad class of games and introduce two refinements: global stability and local stability. We demonstrate that in the classical quorum-based blockchain protocol, honestly following the prescribed strategy is not a locally stable BARNE, whereas free-riding is a globally stable BARNE. To address this, we designed a more robust protocol with the same classical guarantees, by incorporating fines and forced errors. Under this new protocol, honesty becomes the unique globally stable BARNE, and free-riding is never a BARNE.

This paper explores the effects of the straight-ticket voting option (STVO) on the positions of politicians. STVO, present in some U.S. states, allows voters to select one party for all partisan elections listed on the ballot, as opposed to filling out each office individually. We analyze the effects of STVO on policy-making by building a model of pre-election competition. STVO results in greater party loyalty of candidates, while increasing the weight of non-partisan voters' positions in candidate selection. This induces an asymmetric effect on winning probabilities and implemented policies in the two-party system.

We study the division of trade surplus in a competitive market environment by conducting a natural field experiment on German eBay. Acting as a seller, we offer Amazon gift cards with face values of up to 500 Euro. Randomly arriving buyers, the subjects of our experiment, make price offers according to eBay rules. Using a novel decomposition method, we infer offered shares of trade surplus and find that the average share proposed to the seller amounts to 29%. Additionally, we document: (i) insignificant effects of stake size; (ii) poor use of strategically relevant public information; and (iii) behavioural differences between East and West German subjects.

This paper studies a firm's optimal capital structure in an environment where the firm's stock price serves as a public signal for its credit worthiness. In equilibrium, equity investors choose how much information to acquire privately, which induces a positive relation between the amount of equity issued and the stock price signal's precision. Thus, through its capital structure, the firm can internalize the informational externality that stock prices exert on bond yields. Firms with a strong fundamental therefore issue more equity and less debt than they would if the informational spill-over did not exist.

Mechanism design theory relies on the concept of Nash equilibrium. However, studies of experimental games suggest that Nash equilibria are rarely played and provide evidence that subjects may be thinking only a finite number of iterations. Our purpose is to find out whether the expected externality mechanism (d'Aspremont and Gérard-Varet, 1979) retains its properties under finite-iterations thinking. While efficient implementation fails under certain conditions, our results provide a vindication of the mechanism in the convex quasi-linear environment with finitely rational agents.

In this paper, we investigate one factor that can directly contribute to, as well as indirectly shed light on the other causes of, the gender gap in academic publishing: time and length of peer review. Using detailed administrative data from an economics field journal, we find that referees spend longer reviewing female-authored papers, are slower to recommend accepting them, manuscripts by women go through more rounds of review and their authors spend longer revising them. Less disaggregated data from 32 economics and finance journals corroborate these results. We conclude by showing that all gender gaps decline, and eventually disappear, as the same referee reviews more papers. This pattern suggests novice referees initially statistically discriminate against female authors, but are less likely to do so as their information about and confidence in the peer review process improves.

We investigate the impact a straight-ticket voting option, a.k.a. the Master Lever, has on U.S. senators' roll-call voting records in Congress. Using a difference-in-differences framework, we find the Master Lever leads to a 3 to 6 percent rightward shift in senators' policy positions. The effect is largely driven by the Republican party. To interpret our results, we analyse the Master Lever's impact on electoral incentives and outcomes. Our findings suggest that ballot design has a non-negligible impact on policy-making. They also imply that electoral outcomes in moderate to right-leaning Master Lever states may be especially vulnerable to right-wing, non-partisan voters.

Individuals often have to decide to which degree of risk they want to expose others, or how much risk to accept if their choice has an externality on third parties. One typical application is a household. We exploit the opportunity to run an experiment in the German Socio-Economic Panel with two members from 494 households. Participants have a good estimate of each other's risk preferences, even if not explicitly informed. They do not simply match this preference when deciding on behalf of the other household member, but shy away from exposing others to risk. We model the situation, and we find four distinct types of individuals, and two distinct types of households.

This paper studies collusion via information sharing in the context of auctions. The model of collusion via information sharing builds on Aumann's (1976) description of knowledge. Robustness of auction mechanisms to collusion via information sharing is defined as the impossibility of an agreement to collude. A cartel can agree to collude on a contract if it is common knowledge within that cartel that the contract is incentive compatible and individually rational. Robust mechanisms are characterized in a number of settings where some, all, or no bidders are bound by limited liability. Finally, the characterization is used in a simple IPV setting to design a mechanism that is both optimal and robust to collusion.

This paper introduces a virtually efficient mechanism in a setting with sequentially arriving agents who hold informative signals about future type distributions. To reveal the information the principal organizes betting on future type reports. An agent's betting reward depends on how accurately the prior updated on his report predicts the type reports observed in the following period. The mechanism satisfies participation constraints and generates no deficit after any reported history.

I study collusion in one-shot auctions, where a buyer can bribe his competitors into lowering their bids. I modify the single-unit Vickrey auction to incite deviations from the designated-winner scenario and thus undermine collusion. The construction of the mechanism does not require the knowledge of the colluding bidders' identities or distributions of valuations, in which sense it is entirely detail-free.

Collusive bidding in auctions jeopardizes the revenue to the seller. This paper describes a way to preclude strong and tacit subgroup collusion in a non-repeated auction environment, when cartels can commit to transfer exchange ex post, but not to reallocation. The robustness is attained by an optional assignment rule in a sealed-bid Vickrey auction, which is not applied in equilibrium, but serves as a credible threat to any collusive agreement aimed at generating an extra surplus.