Jie Zheng

This article introduces an experimental dataset on social preferences among 3068 Chinese university students, collected in 101 laboratory sessions at two universities between 2019 and 2024. The dataset combines preference-elicitation tasks with individual background information and records three stages of science–humanities sorting: high-school track, university major, and institutional environment. The measures cover social allocation under different private costs and prices of giving. Descriptive and regression analyses show selective rather than uniform preference differences across academic backgrounds. Science-track students display lower levels of giving when it is personally costly, and institutional differences are most visible in dictator decisions. Preference distributions nevertheless overlap substantially across groups. The dataset offers a resource for studying how social preferences relate to successive educational choices.

In this paper, we study a large class of multi‐contestant multi‐battle resource allocation contests, where contestants can be asymmetric in terms of both resource budgets and battle valuations, battles can be heterogeneous in terms of both contest success functions (CSFs) and battle valuations, and the conflict network can be arbitrary. Under a set of axioms satisifed by a broad range of stochastic contests, we prove the existence of pure strategy Nash equilibrium (PSNE) of the contest, which is the first equilibrium existence result regarding resource allocation contests with contestant‐and‐battle‐specific valuations under general conflict networks. Furthermore, we show that proportional strategy profile is the unique PSNE under full participation, contestant‐symmetric valuation, and Tullock CSFs with discriminatory power , while the interleaving strategy profile is the one under ring structure, interleaving valuation, symmetric value and lottery CSFs.

We study whether the focal pull of the equal split or the distribution of bargaining power governs coordination in tacit distributional conflicts. In a coordination game offering both equal and unequal splits of the same total surplus, we manipulate players' relative bargaining power through their disagreement payoffs, shifting the prediction of the Nash Bargaining Solution away from equality and towards the stronger player's preferred outcome. Asymmetric disagreement payoffs reduce coordination on the equal split by 17.9 percentage points and increase coordination on the stronger player's preferred outcome by 11.7 percentage points. When bargaining power reverses, so does the direction of claiming, with the newly stronger player pushing for the unequal division in their own favour. However, the equal split remains the modal outcome. The results show that bargaining power can shape tacit division even when it operates only as an implicit threat embedded in disagreement payoffs, while the equal split remains the dominant convention, constraining how far that power moves the division.

This paper examines the welfare implications and optimal regulation of commodities that generate externalities and induce social comparison. We show that the interaction between these dual market failures is super-additive: the resulting overconsumption and deadweight loss exceed the sum of the distortions generated by each failure independently. When the regulator recognizes and correctly accounts for both mechanisms, Pigouvian taxes and tradable quotas can each implement the first-best allocation. We then consider regulatory unawareness of social comparison: the regulator recognizes the physical externality but omits the positional motive from its perceived model. This unawareness-induced behavioral misperception leaves the first-best aggregate quantity unchanged but causes the regulator to understate the tax required to implement it. Consequently, tradable quotas remain first-best, whereas the misspecified tax induces persistent overconsumption and welfare losses. Our results identify a behavioral rationale for the robustness of quantity-based regulation when a relevant motive for consumption is absent from the regulator's perceived model.

Rising income inequality complicates public goods financing, as institutions make efforts to mobilize contributions across income groups. In fundraising settings such as online platforms, community projects and taxation, contribution decisions are typically structured as either fixed amounts or proportional shares. This paper examines how framing contribution decisions as either fixed amounts or proportional shares shapes giving behavior, and how feedback on the highest contribution interacts with this framing to produce differential effects across endowment levels. Drawing on reference dependence theory, we conduct a laboratory experiment and find that relative framing combined with feedback on the highest contribution significantly enhances contributions among high-endowment individuals. However, in the absence of such feedback, relative framing reduces contributions. An asymmetric design that assigns relative framing to high-endowment individuals and absolute framing to low-endowment individuals mitigates the crowding-out effect on contributions but exacerbates redistributive inequality. Mechanism analyses indicate that asymmetric framing promotes contributions by heightening the salience of the high reference point for different endowment groups, while inducing self-serving bias in fairness beliefs. This asymmetric design, when paired with highest-contribution, corrects individuals' self-serving bias and reinforces the incentive effects of high reference points. This study provides behavioral insights for designing income-targeted framing strategies in public goods financing.

Using a controlled online laboratory experiment, we investigate how leader-selection rules influence public goods provision under fixed matching and random rematching. In our design, a leader is a first mover who chooses a contribution before the other group members, and this contribution is revealed to the later mover(s) before they make their own decisions. In leaderless groups, all group members choose their contributions simultaneously, so no member observes another member’s contribution before deciding. Leadership does not produce a uniform increase in contributions. Among the mechanisms examined, delegated leadership by a randomly selected candidate (DL-RC) exhibited the lowest overall mean contribution, particularly under random rematching; however, the primary analyses did not establish a statistically significant general mechanism effect. Within VL-RC, the accepted one-leader branch exhibited higher observed contributions than the refusal-contingent leaderless branch; within DL-RC, the accepted one-leader branch exhibited higher observed contributions than the refusal-contingent three-first-mover branch. Because these realized branches are generated by candidates’ acceptance decisions, the comparisons are conditional and do not identify causal effects of acceptance, refusal, leader absence, or the number of first movers. Across leader-present states, later-mover contributions were positively associated with the observed mean contribution of the first mover(s). Fixed matching yielded a higher observed mean contribution than random rematching, but the pooled difference was not statistically significant; the clearest differences in matching protocols arose in DL-RC.

On a daily basis, individuals encounter different types of information signals about other individuals' trustworthiness, and decide which signals are most worthy. Using controlled laboratory experiments, we investigate how hard information on an individual's prior track record versus visual information in the form of a personal photograph shape trust in a laboratory trust game. Trustors in three baseline treatments observe one of three possible signals about their randomly assigned trustee: the trustee's firstround action, a non-smiling photo of their trustee, or a smiling photo of their trustee. Two additional treatments use multiple price lists to elicit trustors' willingness to pay for the trustee's first-round track record relative to the two types of photos. Smiling photos are the only information signal that significantly increases trust and total surplus on average. Information about trustees' first-round track record has no average effect, but masks an asymmetry that favorable records raise trust to the same level as smiling photos, while unfavorable records reduce trust below the level induced by non-smiling photos. The multiple price list and baseline treatments together show that trustors are willing to pay to observe their trustee's first-round track record rather than their photo, while smiling photos in fact generate the highest trust levels. This result highlights a disparity between the type of information individuals prefer, and the type of information they respond positively to, with industrial policy consequences for online platforms and other businesses.

We study an incomplete-information environment in which a designer chooses both the number of identical winner slots and a minimum performance standard. Creating each slot entails a committed capacity cost for the designer. The number of winner slots governs how likely an agent is to win, whereas the performance standard, by inducing an ability cutoff, determines which types are excluded from winning. We represent aggregate output as virtual surplus generated by the resulting winning-probability allocation and characterize the marginal value of an additional slot by a Beta-distributionmoment of virtual value. Under committed capacity costs, the optimal induced ability cutoff is the zero-virtual-value type, and the number of slots is then chosen conditional on this cutoff. At this cutoff, the performance standard excludes the negative-virtual-value portion of the marginal winning-probability allocation and raises the return to slot expansion. Hence, the tournament with an optimally chosen performance standard creates weakly more winner slots than the standard tournament. With both committed capacity costs and filled-slot costs, the cutoff becomes a zero-net-virtual-value cutoff, and the complementarity between the performance standard and slot expansion remains robust.