Who responds to incentives for participating in transactions that demand one informs oneself before one can make a decision?

In our model of costly information acquisition we find that subjects for whom acquiring and processing information is more difficult will respond more strongly to changes in the incentive, and decide to participate based on worse information. Consequently, with higher incentives, the pool of participants consists of a larger fraction of individuals with a worse understanding of the consequences of their decision. Our behavioral experiment confirms these predictions, both for experimental variation in the costs of information acquisition, and for various laboratory and naturally occurring measures of information costs, including cognitive ability.

These findings are relevant for any transaction in which an unambiguous incentive contrasts with complex but learnable downsides, in diverse fields across economics. Our results also inform the debate about incentives for transactions like living organ donation, human egg donation, or surrogate motherhood. They clarify the relation between incentives and the ethical principle of informed consent, and thus help address ethical concerns with incentives.

In this paper we study the extent to which social connections are important factors determining legislative efficacy of members of the U.S. Congress.

We propose a new model of legislative effectiveness in congress that formalizes the role of social connections and generates simple testable predictions. The model predicts that the equilibrium effectiveness of legislators is proportional to their Katz-Bonacich centrality in the network of social connections. We then propose a new empirical strategy to test these predictions. The strategy consists in implementing a two-step Heckman correction to deal with network endogeneity which is based on an original instrument: the legislators' alumni connections.

The empirical findings support the model's predictions and provide new insights of how social connectedness interact with other factors such as seniority, partisanship and legislative leadership in determining legislators' effectiveness.

Behavioral responses to taxation affect the trade-off society faces between implementing equality and efficiency. Several influential theoretical papers have used heterogeneity in beliefs about behavioral responses to taxation to explain variation in people's support for redistribution of income in society.

In this study, we use a purposefully simple task to elicit incentivized beliefs from a representative sample of 4200 Americans about how taxes affect people's effort choices. The design allows us to assess the empirical validity of theoretical models suggesting a key role for beliefs about behavioral responses by investigating whether these beliefs are a good predictor of people's support for redistributive policies.

We find that while equality-efficiency preferences strongly predict individual support for redistributive policies in society, beliefs about behavioral responses are not a significant predictor of support for redistributive policies. The findings suggest that preference heterogeneity is more important than differences in beliefs to explain people's attitudes toward redistribution of income in society.

Non-verbal communication (e.g., facial expressions or gestures) characterises many economic and social interactions; yet, their effects have received little attention in the literature.

We design an experiment which tests for the impact of non-verbal communication on pro-social behaviour as measured by a one-shot social dilemma game with and without punishment opportunities. Specifically, we are concerned with “abstract social cues”, such as smiling labels, to which we refer as a type of non-verbal communication in which elements of the social environment (e.g., gaze or touch) have been ruled out. This allows us to focus on the pure effects of non-verbal communication absent of confounding elements which may affect its effectiveness.Our design varies i) whether our social cues are costly or costless, and ii) whether one or both members in a group are given the opportunity to use such social cues.

We find that when smiling labels are costly only a small minority of subjects is willing to use them as opposed to when they are costless. As a result, subjects contribute significantly less, and subsequently, sanctioning is harsher. We also find that welfare (as measured by average net earnings) is significantly lower when smiling labels are costly. Overall, our findings provide new evidence that abstract social cues embody information that influences pro-social behaviour in social interactions.

This paper provides evidence of persistent overconfidence among managers in a firm, who face a high-powered tournament incentive scheme, and have very detailed feedback as well as substantial experience.

We estimate a structural model of rational belief formation, but reject the model, based on the extent of overconfidence in observed manager beliefs about their future performance in the tournament. We turn next to investigating a potential mechanism, posited by models of motivated beliefs, in which individuals make use of an ability to forget or distort negative signals, because they want to maintain positive beliefs (Benabou and Tirole, 2003). We elicit manager recall of recent past performance, and find that managers who performed poorly tend to be inaccurate and recall better than actual performance. We then show empirically that these biased memories of the past help explain overconfident predictions for the future.

The overconfidence effect is a well-established and pervasive bias. A myriad of evidence suggests that people make substantial errors when they evaluate their skills, knowledge, abilities, and attributes; in particular, they overrate themselves. While overconfidence has mostly been studied in individual decision-making settings, there is a small and growing literature on the effects of overconfidence on markets.

We turn our attention to committee decision-making and voting (whether in a market or non-market setting). Feddersen and Pesendorfer (1996, 1999) showed that even if voting has zero cost, less-informed agents rationally abstain and delegate the voting decision to better-informed agents. Experimental studies by Battaglini et al. (2010) and Morton and Tyran (2011) find that significant numbers of subjects with low-quality information abstain. But are agents willing to delegate when they have subjective beliefs about their knowledge and skills?

The aim of our project is to analyze the effect of overconfidence bias on voting and information aggregation.

Voluntary giving in the real world involves giving to a few out of many, or little to each of many. In contrast, giving in lab experiments typically involves giving to just one or a few select partners.

We present a new experimental paradigm that studies how giving and preferences for redistribution vary by group size, and find that voluntary giving decreases as group size increases. This suggests that group size may be a key factor for why generous voluntary giving is so common in the lab. Preferences for redistribution, measured by the propensity to vote for mandatory transfers, are however invariant to group size, suggesting that redistributive government policy may be vital for implementing societal preferences – including the preferences of those taxed. Perceptions of relevant groups are not immutable: assigning a potential donor to a partner increases voluntary giving, more so in large groups.

Our results are consistent with a simple model of scale-free distributional preferences.

Most experimental setups involve choices made among options simultaneously available to the decision maker. However, in real-life situations, most choices - even consumer choices - involve options considered sequentially, and require maintaining a memory trace of the value associated with the options already met until all options have been considered. Thus, memory and choice are essentially linked in choice.

We study experimentally the role of memory in choice with a new design where participants are presented sequentially with two options separated by a random delay. Subjects choose after the two options are presented again simultaneously. We found that the option presented first is, everything else being equal, significantly and substantially less likely to be chosen.

This behavioral effect has a natural explanation in the information processing that brain imaging analysis reveals: the value of the first option is stored in a leaky working memory system in posterior parietal regions. When the second option is presented, its value is encoded in ventro-medial Pre-Frontal Cortex (vmPFC), while the value of the first option is recalled and comparison of the two produces the selection of one option. These results uncover a mechanistic path of the role of memory in choice, and highlights how theory of choice should include an explicit role for memory.