Alexander Coutts

Associate Professor of Economics, Schulich School of Business

Publications
Feedback is vital for growth and learning, yet anecdotal evidence suggests people often hesitate to provide it, and its provision may be shaped by asymmetries and gender-related biases. We study feedback provision across variations in the nature of performance signals, their instrumental value, and the recipient’s gender. We find that a surprising degree of both positive and negative feedback is withheld, with a follow-up experiment suggesting that advisors’ feedback decisions are driven mainly by transparency- and duty-related considerations, or, to a lesser extent, are motivated by self-serving reasons. Additionally, when initial performance signals are vague, advisors are more likely to withhold noninstrumental negative than positive feedback—an effect we conjecture may be stemming from the lower psychological cost of lying (by omission) under uncertainty. Suggestive evidence shows the difference is more pronounced for female recipients, and exploratory analysis traces this to stronger ego-protective concern by advisors for women than for men.
People often receive feedback influenced by external factors, yet little is known about how this affects self-serving biases. Our theoretical model explores how multi-dimensional uncertainty allows additional degrees of freedom for self-serving bias. In our primary experiment, feedback combining an individual’s ability and a teammate’s ability leads to biased belief updating. However, in a follow-up experiment with a random fundamental replacing the teammate, unbiased updating occurs. A validation experiment shows that belief distortion is greater when outcomes originate from human actions. Overall, our experiments highlight how multi-dimensional environments can enable self-serving biases.
Beyond the direct destructive nature of war, conflict may also have complex impacts on social preferences and norms, as well as on risk-taking behavior. This paper examines the enduring impact of conflict on preferences and behavior by studying different versions of public goods games conducted nearly two decades after the Rwandan genocide, leveraging variation in radio reception among villages in 1994, which determined exposure to hate-propaganda messages inciting conflict. Results indicate that communities exposed to greater violence exhibit increased cooperation, active punishment, and risk-taking. Exploiting the heterogeneity in participants' ages of conflict experience, enhanced cooperation is entirely driven by those who were children during the genocide, while norm enforcement through punishment and reward is more prominent among older individuals. Risk-taking increases are observed across all age groups. The results allow for a reassessment of previous research and highlight the psychological imprint of conflict exposure in line with the literature on early childhood development.
Corruption is often harmful for economic development, yet it is difficult to measure due to its illicit nature. We propose a novel corruption game to characterize the interaction between actual political leaders and citizens, and implement it in Northern Mozambique. Contrary to the game-theoretic prediction, both leaders and citizens engage in corruption. Importantly, corruption in the game is correlated with real-world corruption by leaders: citizens send bribes to leaders whom we observe appropriating community money. In corrupt behavior, we identify an important trust dimension captured by a standard trust game.
Use of lab-in-the-field experiments has steadily increased, given benefits of studying relevant populations and their preferences. In the field, researchers must often relinquish the control of a standard laboratory, raising the specter of communication from past to future participants. Little is known about the consequences of such spillovers, and recent literature indicates variation in how authors deal with them. I provide estimates of communication spillovers using existing data from public goods games in Rwanda, leveraging variation in planning the sequence of visiting 147 villages. The resulting order created opportunities for some villages to communicate with past participants. Using ex-post matching of villages with and without these opportunities I find that communication led to substantial increases in cooperation, suggesting that unanticipated spillovers can bias inference. I conclude with advice for creating protocols to deal with communication spillovers.
Natural resources can have a negative impact on the economy through corruption and civil conflict. This paper tests whether information can counteract this political resource curse. We implement a large-scale field experiment following the dissemination of information about a substantial natural gas discovery in Mozambique. We measure outcomes related to the behavior of citizens and local leaders through georeferenced conflict data, behavioral activities, lab-in-the-field experiments, and surveys. We find that information targeting citizens and their involvement in public deliberations increases local mobilization and decreases violence. By contrast, when information reaches only local leaders, it increases elite capture and rent-seeking.
Bayesian updating remains the benchmark for dynamic modeling under uncertainty within economics. Recent theory and evidence suggest individuals may process information asymmetrically when it relates to personal characteristics or future life outcomes, with good news receiving more weight than bad news. I examine information processing across a broad set of contexts: 1) ego relevant, 2) financially relevant, and 3) non value relevant. In the first two cases, information about outcomes is valenced, containing either good or bad news. In the third case, information is value neutral. In contrast to a number of previous studies I do not find differences in belief updating across valenced and value neutral settings. Updating across all contexts is asymmetric and conservative: the former is influenced by sequences of signals received, a new variation of confirmation bias, while the latter is driven by non-updates. Despite this, posteriors are well approximated by those calculated using Bayes' rule. Most importantly these patterns are present across all contexts, cautioning against the interpretation of asymmetric updating or other deviations from Bayes' rule as being motivated by psychological biases.
Optimistic beliefs affect important areas of economic decision making, yet direct knowledge on how belief biases operate remains limited. To better understand these biases I introduce a theoretical framework that trades off anticipatory benefits against two potential costs of forming biased beliefs: (1) material costs which result from poor decisions, of Brunnermeier and Parker (2005), and (2) direct psychological costs of distorting reality, of Bracha and Brown (2012). The experiment exploits the potential of the increasingly popular BDM elicitation procedure adopted to lotteries to distort beliefs in different directions, depending on which costs are most important. Relative to an elicitation procedure without distortionary incentives, beliefs are biased in the optimistic direction. Increasing payments for accuracy further increases belief reports, in many cases away from the truth, consistent with psychological costs of belief distortion. Yet the overall results suggest that theories of optimism based on anticipatory benefits and material or psychological costs fail to explain how beliefs respond to financial incentives.
Working Papers
Collaboration obscures individual contributions, inviting bias. We model how selection into teams can amplify or reverse a gender credit gap, hiding its source. Candidates choose to work alone or in a mixed-gender team before employers decide whom to hire. Our experiments with 2,812 participants randomize whether employers see gender and whether a coin flip overrides choices. Employers who see gender penalize men, ranking the man above the woman in 39 percent of teams. The penalty is unchanged when the coin flip removes selection: a direct bias linked to the mistaken belief that women collaborate more. When employers see gender, high-performing men, not women, withdraw from teams despite expecting favoritism, but consistent with gender norms.
Experimental evidence of dynamic inconsistency is often interpreted as indicating widespread naivety, based on point predictions of future choices. In the canonical effort paradigm, we revisit this interpretation by eliciting full probabilistic beliefs to study sophistication and menu demand. While point forecasts replicate apparent naivety, probabilistic beliefs reveal substantial uncertainty; accounting for it bridges over half of the belief–behavior gap implied by point predictions. At the same time, this approach uncovers that naivety is less widespread but more sharply concentrated: 40% of reversals arise from misspecified respondents, who account for 21% of the sample and assign zero probability to the effort they ultimately choose. This measurement shift also reshapes our understanding of commitment demand. While flexibility choices track anticipated behavior, commitment demand does not: almost all removed options are assigned zero probability, a pattern inconsistent with purely consequentialist models such as the standard present-bias framework, even in stochastic form.
Sub-Saharan Africa (SSA) faces significant challenges in infant and maternal health. The demand for simple treatments and preventive care is surprisingly low. Biases stemming from traditional beliefs are a possible explanation. We run a study in rural Guinea-Bissau, in which we first describe traditional beliefs and health behaviors, namely by using novel behavioral games. We then ran a large-scale RCT in which the main intervention was a health information campaign based on science but referencing traditional beliefs and integrating Traditional Health Practitioners (THPs). We find widespread traditional beliefs in the supernatural and in witchcraft causing health problems. These beliefs are correlated with seeking informal healthcare. The main intervention we study increases knowledge about health, primarily for those individuals holding stronger traditional beliefs. We find a negative treatment effect on observed malnutrition. We conclude that acknowledging traditional beliefs and integrating THPs can improve the impact of information for health in SSA.
Beliefs are central to economic decisions, so accurate belief elicitation is critical. Incentive-compatible procedures are widely used, and conventional advice recommends using larger financial incentives to improve reporting accuracy. We test this claim with a survey of researchers and a laboratory experiment using the widely-adopted binarized scoring rule. While researchers prefer higher stakes, we find the reverse — raising incentives tenfold increases reporting errors by 33%, independent of the level of instruction detail. We find that this effect is driven by individuals with low cognitive reflection performance. Thus, larger financial stakes may undermine the very accuracy they are intended to promote.
In this document I study racial bias across the world, and in doing so examine the extent to which the United States differs from other countries on a well-known psychological measure of racial bias against Black people. Specifically, I use data from nearly 3.75 million individuals across 146 countries who have taken the implicit association test (IAT) over the last ten years, to examine the extent of racial bias. There are five important results. First, there is significant racial bias against Black people and it is present in every country in the sample. Second, the average country has approximately one in five of its residents exhibiting strong pro-White bias and one in four exhibiting moderate pro-White bias. Third, the majority of countries show more bias than the USA, which is ranked, from high to low bias, 95th out of 146. Fourth, all races exhibit a pro-White bias, with the exception of Black respondents, who exhibit no substantial bias. Fifth, supplementary data about racial attitudes in Europe corroborates the IAT as a meaningful correlate of racist attitudes.
Data Collected (Work in Progress)