We study prisoner’s dilemmas played in continuous time with flow payoffs over 60 seconds. In most cases, the median rate of mutual cooperation rises to 90% or more. Control sessions with 8-time repeated matchings achieve less than half as much cooperation, and cooperation rates approach zero in one-shot control sessions. In follow-up sessions with a variable number of subperiods, cooperation rates increase nearly linearly as the grid size decreases and, with one-second subperiods, they approach the level seen in continuous sessions. Our data support a strand of theory that explains how the capacity to respond rapidly stabilizes cooperation and destabilizes defection in the prisoner’s dilemma.

My dissertation consists of three chapters that use a behavioral economic framework to enrich our understanding of the political economy of environmental policies.

Chapter 1 explores the role of cognitive frictions in the perceived distributional impacts of externality pricing. Despite their desirable properties, carbon prices have received very low public support. Leading explanations attribute the rejection of carbon pricing to preferences and ideology. I show that, instead, much of this rejection is due to mistakes voters make when reasoning about the consequences of such policies. Using laboratory experiments, I show that subjects have specific blind spots in reasoning about a key pricing policy structure, even in abstract tasks that remove scope for preferences and ideologies. Based on these findings, I introduce a new policy that is isomorphic to a leading carbon pricing proposal, but engineered to be robust to these cognitive frictions. The new policy (which I call ``carbon penalty and reward'') garners significantly more support than the original proposal (known as ``carbon fee and dividend''). Using independent cognitive measures, I show that the increase in support is concentrated in people exhibiting systematic difficulty reasoning about a key policy structure.

Chapter 2, which is joint work with Guglielmo Zappalà, details how exposure to environmental policy in the past shapes preferences for them. Low public support has been an obstacle to the enactment of stronger environmental policies. Yet if policies are enacted, support for them may change. Using surveys covering 38 countries around the world, we study the dynamics of environmental policies and individual preferences over time. Exploiting within-country, across birth-cohort variation in exposure to environmental policy stringency, we document that cohorts exposed to more stringent policies in the past are more supportive of environmental policies at the time of the survey, with the effect largely driven by exposure during a period of early adulthood known as the formative age window. This result even holds when evaluating exposure to a specific policy instrument, environmental taxes. Past exposure to these taxes improves support for them, but not for other environmental policies. This relationship suggests that a society's environmental policy attitudes evolve endogenously, with implications for normative frameworks used in welfare economics.

Chapter 3 evaluates how quick, inattentive decision-making systematically distorts perceptions of the effectiveness of pricing in reducing pollution externalities. When people are asked to make policy choices that affect a good's price -- for example, a tax to reduce a negative externality -- we traditionally assume that they accurately perceive quantities demanded conditional on price. But failures to consider substitutes and other adaptations to a permanent price shock can result, ex ante, in a perception of inelasticity and policy ineffectiveness. In past surveys, perceptions of ineffectiveness are among the main reasons voters do not support a carbon tax, believing it would fail to cut emissions unless the revenue were spent on offsetting projects. I implement an incentivized survey experiment to test whether increasing time and promoting hypothetical thinking increase subjective perception of responsiveness of demand to price, and I find meaningful, directional effects.

This dissertation consists of three chapters that explore why individuals make seemingly suboptimal decisions in risk management, why they fail to use information in a Bayesian manner when updating their beliefs, and how novel methodological tools can be developed to advance modelling and inference about subjective beliefs or perceptions, considering cognitive limitations.

Chapter 1 studies the underlying mechanisms behind a classical behavioral puzzle in risk management, called Probability Matching. Probability matching refers to people's tendency to randomize between different risky options, or even match their choice frequency to the outcome probability, when choosing over binary lotteries that differ only in their probabilities. Why? I present an experiment designed to distinguish between three broad classes of explanations: models of Correlation-Invariant Stochastic Choice (mixing due to factors orthogonal to how outcomes are jointly determined, such as non-standard preferences or errors), models of Correlation-Sensitive Stochastic Choice (e.g., deliberately mixing due to misperceived hedging opportunity), and Framing Effects (indecisiveness due to frame-sensitive heuristics e.g., similarity heuristic: attending to dissimilar but irrelevant attributes (outcomes), while ignoring relevant attributes (probabilities)). My experimental design uses a diagnostic approach, differentiating between their testable predictions over a series of treatments. The results suggest that a substantial proportion of mixing behavior aligns with models of Correlation-Sensitive Stochastic Choice, while the other classes have limited explanatory power.

In Chapter 2, a joint work with Menglong Guan, ChienHsun Lin, and Ravi Vora, we experimentally investigate how people value and utilize different statistical characteristics of a set of realized binary signals, referred to as sample features, to understand why individuals deviation from the Bayesian benchmark when updating beliefs. We find that, subjects systematically under-infer the information contained in each sample feature. Furthermore, the magnitude of under-inference significantly varies across sample features. Specifically, under-inference is least severe with Sample Proportion (the relative frequency of different outcomes in the realized signals), compared to more informative features such as Sample Count (the absolute number of different outcomes in the realized signals). We also find that the standard measure of informativeness used in information theory does not fully explain subjects' preferences for sample features. Subjects demonstrate a strict preference for the information contained in the Sample Proportion over those without it and undervalue the usefulness of sample size. Combining preference and belief updating behaviors, we find that subjects deviate less from the Bayesian benchmark when provided with a more-preferred feature than a less-preferred one. These results suggest that some biases in signal usage is more likely an intentional deviation rather than a result of inattentive heuristics.

In Chapter 3, a joint work with Xin Jiang, we introduce a novel elicitation method, called the Dynamic Binary Method (DBM), designed to address the common challenge individuals face in pinpointing the best point estimate of their beliefs, particularly when their beliefs are imprecise. Unlike Classical Methods (CM), which require respondents to make absolute judgments and form a point estimate of their true beliefs, DBM guides them through a series of binary relative judgments, enabling them to express interval beliefs by exiting the process at any step. To assess the empirical validity of DBM, we conduct both within-subject and between-subject experiments using a diverse range of perception tasks drawn from previous literature and CM as a benchmark of performances in each task. We find that DBM does not perform significantly differently from CM at the aggregate level, regardless of whether the perception questions use artificial/laboratory settings or real-life settings, and irrespective of the measurement used. Notably, DBM outperforms CM when the objective truth is extreme. Furthermore, we find a negative correlation between the length of stated beliefs in tasks using DBM and their accuracy. Additionally, we find that the length stated in DBM can predict respondents' performance in CM tasks at the aggregate level, albeit not strictly in a monotonic manner. Finally, we explore methods to use DBM-collected data for predicting stated point beliefs in DBM, offering insights into potential applications of the method beyond its immediate implementation.

This dissertation consists of three experimental studies on information economics, exploring the topics of the demand for information, the choice and use of information, and information design within strategic contexts.

Chapter 1 studies how people choose sets of information sources (referred to as information bundles). The findings reveal that subjects frequently fail to choose the more instrumentally valuable bundle in binary choices, largely due to the challenge of integrating the information sources within a bundle to identify their joint information content. The mistakes in choices can not be attributed to an inability to use information bundles. Instead, these mistakes are strongly explained by subjects' tendency to follow a simple but imperfect heuristic when valuing them, which we call "common source cancellation (CSC)''. The heuristic causes subjects to mistakenly disregard the common information source in two bundles and focus solely on the comparison of the sources that the two bundles do not share. As a result, choices between information bundles are made without adequately considering the joint information content of each bundle. Notably, CSC emerges as a robust explanation for the information bundle choices for all subjects, including those who make perfect use of information bundles to make inferences.

Chapter 2, based on a joint work with Ryan Oprea and Sevgi Yuksel, studies how people’s demand for information structures is shaped by their informativeness—the reduction in uncertainty they produce. To do this, we introduce new methods that remove confounds for information demand like failures of Bayesian reasoning. We show that people (i) strongly demand informativeness when it has instrumental value but also (ii) display a sharp aversion to informativeness when it cannot be used to improve choice, sometimes leading to costly errors in information choice. Several strands of evidence suggest that this aversion is driven by subjective information processing costs that rise with informativeness.

Chapter 3, based on a joint work with Sen Geng, explores theoretically and experimentally whether information design can be used by trustees as a signaling device to boost trusting acts. In our main setting, a trustee partially or fully decides a binary payoff allocation and designs an information structure; then a trustor decides whether to invest. In the control setting, information design is not available. In line with the standard equilibrium analysis, we find that introducing information design increases trustworthiness and trusting acts, and some trustees choose full trustworthiness with the most informative structure. We also find systematic behavioral deviations, including some trustees' choosing zero trustworthiness with the least informative structure and trustors' overtrusting in low informative structures. We finally provide a model of heterogeneity in prosociality and strategic sophistication, which rationalizes the experimental findings.

This dissertation contains three chapters on experimental economics and microeconomics. In the first chapter, Dynamic Investment and Preferences over the Resolution of Risk, I report the results of a laboratory experiment which attempts to explain the finding that individuals invest less in risky assets when risk is gradually resolved over time, rather than all at once. Though the literature has traditionally attributed this behavior to a cognitive error, Koszegi and Rabin (2009) recently characterized this finding as the result of non-standard preferences over the resolution of risk. My results reject the traditional "cognitive errors" explanation in favor of Koszegi and Rabin's "non-standard preferences" explanation. In the second chapter, Kidney Co-operative: A Mechanism to Improve on Human Kidney Markets, myself and coauthors propose a mechanism called the kidney co-operative which is designed to provide sufficient incentives to alleviate the human kidney shortage, while at the same time addressing the concerns regarding the potential losers to such a reform. We show that it is reasonable to expect that the number of transplants will be larger under the kidney co-operative mechanism than under either the status quo or the conventional market mechanism. In the third chapter, Charity in the Laboratory:

Matching, Competition, and Group Identity, myself and a coauthor study the effects of donation matching, competition, and group membership on charitable donations using a laboratory experiment. We find that providing matching donations to all subjects or

having individuals compete for the privilege to have their donations matched (we match the top half of donations in each session), raises donation levels modestly. However, arbitrarily assigning subjects to teams which competed for matching funds substantially raised donation levels. We appeal to the notions of group identity and team dynamics to explain our results.

The first chapter studies gender differences in persistence in response to performancefeedback. The decision to persist in stratified career trajectories is often dynamic in nature: people receive feedback and decide whether to persist or to drop out. I show experimentally that men are on average 10 percentage points more likely to persist in an environment that rewards high performance than equally performing women who received the same feedback. About one-third of this gap can be explained by gender differences in beliefs about the future. In the laboratory as well as a field study, men are more optimistic about their future performance even when compared to women who are similarly confident about their past performance. Another 30% of the gender gap in persistence is attributable to men seeking, and women avoiding exposure to additional feedback.

The second chapter is based on joint work with Mireille Jacobson and Heather Royer.We observe that fewer births occur on major US holidays than would otherwise be ex- pected. We use California data to study the nature and health implications of this birth date manipulation, and document 18 percent fewer births on the day of and just after a holiday. C-sections account for roughly half of the decline. “Missing” holiday births are moved to the roughly two-week window both before and after the holiday. High-risk births are more likely to be re-scheduled than low-risk births. Despite the documented change in timing, we find little evidence of any adverse health consequences for babies born around a holiday.

The third chapter is based on joint work with Hazem Alshaikhmubarak, David Hales,Molly Schwarz and Kent Strauss. We experimentally study how mutual payoff informa- tion affects play in strategic settings. Subjects play the Prisoner’s Dilemma or Stag Hunt game against randomly re-matched opponents under two information treatments. In our partial-information treatment subjects are shown only their own payoffs, while in our full-information treatment they are shown both their own and their opponent’s payoffs. In both treatments, they receive feedback on their opponent’s action after each round. We find that mutual payoff information initially facilitates reaching the Pareto-efficient outcome in both games. While play in the Prisoner’s Dilemma converges toward the unique Nash equilibrium of the game under both information treatments, mutual pay- off information has a substantial impact on the equilibrium selection in the Stag Hunt throughout all rounds of the game. Using a belief-learning model and simulations of play, we provide evidence that these effects are driven not only by initial play but also by the way subjects learn. We propose that strategic uncertainty is a probable channel through which payoff information affects play.

This dissertation consists of two essays in experimental economics and one in applied microeconomics. The first chapter, Preferences over Exploration: Simple Bandits in the Lab, presents an experimental study about how people view the trade-off between exploration (trying a new option) and exploitation (taking a familiar option). This study documents a large and persistent tendency of people to under-explore, even in the simplest possible bandit setting. By simplifying the environment, the study can directly control for risk preferences, abstract from other issues such as ambiguity, Bayesian updating, or failures to reduce compound lotteries. The second chapter, Complexity and Procedural Choice: Evidence from Experimental Bandits, presents an experiment about how subjects select procedural rules to make decisions in a simple bandit task. The optimal strategy involves a relatively sophisticated 4-state decision rule. When complexity costs are low, subjects conform to the optimal behavior; when complexity costs are higher, holding other aspects of the problem constant, subjects systematically employ lower-complexity rules. This suggests that aversion to complexity causes subjects to play simpler strategies and earn lower payoffs. The final chapter, The Economics of the Montana Liquor License System, is an applied microeconomics paper about the the liquor license system in the state of Montana. The state has a complicated system of quotas for liquor licenses, involving multiple license types, which are tradable within quota area (and occasionally between quota areas). The study offers a simple model of the system, providing testable hypotheses. Empirical results support these hypotheses, indicating that gambling revenue, non-permanent population, and income all positively affect prices, while the emergence of small-scale producers (e.g., craft breweries) negatively affect license prices.

Human players in our laboratory experiment converge closely to the symmetric mixed Nash equilibrium when matched in a single population version of the standard Hawk-Dove game. When matched across two populations, the same players show clear movement towards an asymmetric (and very inequitable) pure Nash equilibrium of the same game. These findings support a distinctive prediction of evolutionary game theory.

This dissertation consists of three essays that analyze the impact of behavioral biases on labor market and health outcomes. The essays use tools from both experimental economics and applied econometrics. The common thread that runs through this research agenda is the goal of understanding how biases and suboptimal behaviors impact long-term outcomes crucial to well-being: labor market and health outcomes.

The first essay asks whether dismissal threats are more motivating than other types of incentives. In a laboratory experiment, workers earn a fixed wage per period and complete real effort tasks to reduce their chance of being fired at the end of each period. Behavioral motivators are purposefully activated in addition to monetary incentives. The design innovates on previous literature by implementing dismissal threats in a quantifiable way and by collapsing preference elicitation over incentive along with random assignment to incentives into the same round. The experiment produces two main results. First, workers produce significantly more output under dismissal threats than they do under piece rates, even though the marginal benefit of output is lower. Second, the productivity gains from strengthening dismissal threats on the margin have a large self-selection component but significant heterogeneity in pure incentive effects. Workers who prefer higher pay with steeper dismissal threats appear to respond positively to this environment, but these high-pressure incentives backfire among workers who want to avoid them.

The second essay implements a lab experiment to investigate the effects of self-image concerns on search behavior. Subjects play a simple sequential search game in which they decide how many times to search for a wage offer before giving up. Feedback from search contains both instrumental information about search prospects and signals about subjects' relative performance on an intelligence test taken earlier in the experiment. Treatments isolate and shut down two mechanisms: biased belief updating and information avoidance. Despite replicating results from the literature on overconfidence in incentivized reporting of initial beliefs, subject search behavior does not differ between treatments with or without self-image concerns during search. These results seem to suggest that people are more likely to state overconfident beliefs when these beliefs are directly elicited, but that people act much closer to the rational Bayesian benchmark when actions only indirectly reveal self-relevant beliefs.

The third essay is joint work with Michael F. Pesko. We estimate the effect of county-level e-cigarette indoor vaping restrictions on adult prenatal smoking and birth outcomes using United States birth record data for 7 million pregnant women living in places already comprehensively banning the indoor use of traditional cigarettes. We use both cross-sectional and panel data to estimate our difference-in-difference models. Our panel model results suggest that adoption of a comprehensive indoor vaping restriction increased prenatal smoking by 2.0 percentage points, which is double the estimate obtained from a cross-sectional model. We also document heterogeneity in effect sizes along lines of age, education, and type of insurance.

An irreversible investment opportunity has value V governed by Brownian motion with upward drift and random expiration. Human subjects choose in continuous time when to invest. If she invests before expiration, the subject receives V−C : the final value V less a given avoidable cost C . The optimal policy is to invest when V first crosses a threshold V ∗ = (1 + w∗ )C , where the option premium w∗ is a specific function of the Brownian parameters representing drift, volatility and discount (or expiration hazard) rate.

We ran 80 periods each for 69 subjects. Subjects in the Low w∗ treatment on average invested at values quite close to optimum. Sub jects in the two Medium treatments and the High w∗ invested at values below optimum, but with the predicted ordering, and values approached the optimum by the last block of 20 periods. Behavior was most heterogeneous in the High treatment. Subjects underrespond to differences in both the volatility and expiration hazard parameters. A directional learning model suggests that sub jects react reliably to ex-post losses due to early investment, and react much more heterogeneously (and on average more strongly) to missed investment opportunities. Simulations show that this learning process converges on a nearly optimal steady state.