Working papers results

Are the players “commonly meta-certain” of an interactive belief model itself? The paper formalizes what it means by: “a player is (meta-)certain of her own belief-generating map” or “the players are (meta-)certain of the profile of belief-generating maps (i.e., the model).” The paper shows: a player is (meta-)certain of her own belief-generating map if and only if her beliefs are introspective. The players are commonly (meta-)certain of the model if and only if, for any event which some player i believes at some state, it is common belief at the state that player i believes the event. This paper then asks whether the “common meta-certainty” assumption is needed for epistemic characterizations of game-theoretic solution concepts. The paper shows: common belief in rationality leads to actions that survive iterated elimination of strictly dominated actions, as long as each player is logical and (meta-)certain only of her own strategy and belief-generating map

Algorithms are becoming the standard tool for bidding in auctions through which digital advertising is sold. To explore how algorithmic bidding might affect functioning of these auctions, this study undertakes a series of simulated experiments where bidders employ Artificial Intelligence algorithms (Q-learning and Neural Network) to bid in online advertising auctions. We consider both the generalized second-price (GSP) auction and the Vickrey-Clarke-Groves (VCG) auction. We find that the more detailed information is available to the algorithms, the better it is for the efficiency of the allocations and the advertisers profit. Conversely, the auctioneer revenues tend to decline as more complete information is available to the advertiser bidding algorithms. We also compare the outcomes of algorithmic bidding to those of equilibrium behavior in a range of different specifications and find that algorithmic bidding has a tendency to sustain low bids both under the GSP and VCG relative to competitive benchmarks. Moreover, the auctioneer revenues under the VCG setting are either close to or lower than those under the GSP setting. In addition, we consider three extensions commonly observed in the data: introduction of a non-stategic player, bidding through a common intermediary, and asymmetry of the information across bidders. Consistent with the theory, the non-strategic player presence leads to increased efficiency, whereas bidding through a common intermediary leads to lower auctioneer revenue compared to the case of individual bidding. Moreover, in experiments with information asymmetry, more informed players earn higher rewards.

We evaluate how traditional parties may respond to populist parties on issues aligning with populist messages. During the 2020 Italian referendum on the reduction of members of Parliament, we conducted a large-scale field experiment, exposing 200 municipalities to nearly a million impressions of programmatic advertisement. Our treatments comprised two video ads against the reform: one debunking populist rhetoric and another attributing blame to populist politicians. This anti-populist campaign proved effective through demobilization, as it reduced both turnout and the votes in favor of the reform. Notably, the effects were more pronounced in municipalities with lower rates of college graduates, higher unemployment, and a history of populist votes. This exogenous influence introduced a unique populist dynamic, observable in the 2022 national election where treated municipalities showed increased support for Brothers of Italy, a rising populist party, and decreased support for both traditional parties and the populists behind the 2020 reform. A follow-up survey further showed increased political interest and diminished trust in political institutions among the residents of municipalities targeted by the campaign.

In a recent paper, Lin & Palfrey (2024) developed a theory of cognitive hierarchies (CH) in sequential games and observed that this solution concept is not reduced-normal-form invariant. In this note I qualify and explain this observation. I show that the CH model is normal-form invariant, and that the differences arising from the application of the CH model to the reduced normal form depend only on how randomization by level-0 types is modeled. Indeed, while the uniform behavior strategy in the extensive form yields the uniform mixed strategy in the normal form, the latter does not correspond to the uniform randomization in the reduced normal form, because different reduced strategies may correspond to sets of equivalent strategies with different cardinalities. I also comment on (i) the invariance of the CH model to some transformations of the sequential game, and (ii) the independence of conditional beliefs about co-players' level-types.

We study whether a better knowledge of the functioning of pay-as-you-go pension systems and recent demographic trends affects natives’ attitudes towards immigration. In two online experiments conducted in Italy and Spain, we randomly treated participants with a video explaining how, in pay-as-you-go systems, the payment of current pensions depends on the contributions paid by current workers. The video also informs participants about population aging trends in their countries. The treatment increases knowledge of pay-as-you-go systems and future demographic trends for all participants. However, it improves attitudes towards migrants only for treated participants who do not support populist and anti-immigrant parties.

We document the spiral of populism in Europe and the direct and indirect role of economic insecurity shocks. Using survey data on individual voting, we make two contributions to the literature, namely: (1) Economic insecurity shocks have a significant impact on the populist vote share, directly as demand for protection, and
indirectly through the induced changes in trust and attitudes; (2) A key consequence of increased economic insecurity is a drop in turnout. The impact of this largely neglected turnout effect is substantial: conditional on voting, when economic insecurity increases almost 40% of the induced change in the vote for a populist party comes from the turnout channel.

This paper empirically shows that the imbalance between an ethnic group’s political and military power is crucial to understanding the likelihood that such a group engages in a conflict. We develop a novel measure of a group’s military power by combiningmachine learning techniques with rich data on ethnic group characteristics and
outcomes of civil conflicts in Africa and theMiddle East. We couple thismeasure with available indicators of an ethnic group’s political power as well as with a novel proxy based on information about the ethnicity of cabinet members. We find that groups characterized by a highermismatch betweenmilitary and political power are between 30% and 50% more likely to engage in a conflict against their government depending on the specification used. We also find that the effects of power mismatch are nonlinear, which is in agreement with the predictions of a simplemodel that accounts for the cost of conflict. Moreover, our results suggest that high-mismatched groups are typically involved in larger and centrist conflicts. The policy implication is that powersharing recommendations and institutional design policies for peace should consider primarily the reduction of power mismatches between relevant groups, rather than focusing exclusively on equalizing political power in isolation.