We develop an asset-pricing model with heterogeneous investors and search frictions. Trade is intermediated by risk-neutral dealers subject to capacity constraints. Risk-averse investors can direct their search towards dealers based on price and execution speed. Order flows affect the risk premium, volatility, and equilibrium interest rate. We propose a new solution method to characterize the equilibrium analytically. We assess the quantitative implications of the model in response to a large adverse shock. Consistent with the empirical evidence from the COVID-19 crisis, we find an increase in the risk premium and market illiquidity, and a decline in interest rates.

This paper develops a model of voters’ and politicians’ behavior based on the notion that voters focus disproportionately on and, hence, overweigh certain attributes of policies. We assume that policies have two attributes—resources devoted to two distinct issues (e.g., defense and education)—and that voters focus more on the attribute in which their options differ more. First, we consider exogenous policies and show that focusing polarizes the electorate. Second, we consider the endogenous supply of policies by politicians running for office and show that focusing leads to inefficiencies: voters that are more focused are more influential; distorted attention empowers social groups that are larger and more sensitive to changes on either issue; resources are channelled towards divisive issues. Finally, we show that augmenting classical models of electoral competition with focusing can contribute to explain puzzling stylized facts such as the inverse correlation between income inequality and redistribution.

I model decision-making constrained by morality, rationality, or other virtues. In addition to a primary preference over outcomes, the decision maker (DM) is characterized by a set of preferences that he considers justifiable. In each choice setting, he maximizes his primary preference over the subset of alternatives that maximize at least one of the justifiable preferences. The justification model unites a broad class of empirical work on distributional preferences, charitable donations, prejudice/discrimination, and corruption/bribery. I provide full behavioral characterizations of several variants of the justification model as well as practical tools for identifying primary preferences and justifications from choice behavior. I show that identification is partial in general, but full identification can be achieved by including lotteries in the domain and allowing for heterogeneity in both primary preferences and justifications. Since the heterogeneous model uses between-subject data, it is robust to consistency motives that may arise in within-subject experiments. I extend the heterogeneous model to information choice and show that it accounts for observed patterns of information demand and avoidance on ethical domains.

We present a two-period model examining how the central bank and the elected government jointly shape elections and economic outcomes. An apolitical central bank minimizes a quadratic loss function in inflation and unemployment along an expectational Phillips curve, which is shifted by the government’s quality. Fully rational voters optimally choose between the incumbent, whose quality they infer from unemployment, and a challenger of unknown quality. We find that governments prefer more inflation-averse central banks than the social planner, rationalizing the political success of inflation-targeting in practice. Inflation-targeting, however, has negative economic consequences by allowing lower quality incumbents to be reelected.

Lab and field experiments have proven to be an important source of empirical evidence on discrimination. We show that if average outcomes in a discrimination experiment satisfy simple conditions, then this provides evidence that decision-makers are prejudiced – regardless of what they learned about individuals in each demographic group before making their decisions. We demonstrate our robust test of prejudice using the lab experiment of Reuben, Sapienza, and Zingales (2014) and the field experiment of Bertrand and Mullainathan (2004).

This paper proposes a new theoretical definition of (non-)democracy based on two "political rights" parameters (η , κ) that capture the extensive and intensive margin of the population's ability to replace the incumbent; and an "individual rights" parameter λ that captures the degree to which individual citizens are protected from political retribution. Within the rules of the game specified by (η , κ ), two office-motivated politicians compete for power by making promises to citizens. The policy space features a trade-off between redistribution and public good provision. I study two types of public good: one that delivers egalitarian benefits, the other that delivers non-egalitarian benefits. I find that when political rights are stronger, and/or individual rights are weaker, competition drives politicians to treat citizens more equally, and to provide the egalitarian public good more efficiently. Regimes where political and individual rights are perfectly protected give politicians incentives to treat citizens inequitably for political advantage; these regimes provide the non-egalitarian public goods efficiently.

Uncovering the causes and foundations of increasing political polarization poses a pressing open problem to all social sciences. Here, we report results from an online experiment with a representative sample of the US population, deployed the week before the 2020 US presidential election, to analyze the role of affective polarization and social identity in facilitating ideological polarization at the individual level. Participants were incentivized to predict policy-sensitive statistics a year after the election conditional on its outcome. To update their initial predictions, individuals select or exogenously obtain factually similar articles on the respective topics curated from differently slanted news sources. We employ "political groupiness'' as an empirical measure for individual susceptibility for affective polarization, using behavioral experiments from the economic literature on social identity. We find political groupiness to be systematically associated with ideological polarization, as the partisan biases in both the demand for and processing of information is significantly amplified for groupy subjects. Our results suggest that ideological polarization is founded, at least partly, in the social identity roots of partisanship. Reducing the salience of intergroup distinctions by delabeling information sources decreases the partisan bias in information demand but not in information processing.

We directly optimize the objectives of portfolio management via reinforcement learning—an alternative to conventional supervised-learning-based paradigms that entail first-step estimations of return distributions, pricing kernels, or risk premia. Building upon breakthroughs in AI, we develop multi-sequence neural-network models tailored to the distinguishing features of financial data, while allowing training without labels and potential market interactions. Our AlphaPortfolio yields stellar out-of-sample performances (e.g., Sharpe ratio above two and over 13% risk-adjusted alpha with monthly re-balancing) that are robust under various economic restrictions and market conditions (e.g., exclusion of small stocks and short-selling). Moreover, we project AlphaPortfolio onto simpler modeling spaces (e.g., using polynomial-feature-sensitivity) to uncover key drivers of investment performance, including their rotation and nonlinearity. More generally, we highlight the utility of deep reinforcement learning in finance and “economic distillation” for model interpretation.

To study the effects of neighborhood and place-based interventions, this paper incorporates neighborhood effects into a general equilibrium (GE) heterogeneous-agent overlapping-generations model with endogenous location choice and child skill development. Importantly, housing costs, as well as neighborhood effects, are endogenously determined in equilibrium. Having calibrated the model using U.S. data, we use simulations to show that predictions from the model match reduced form evidence from experimental and quasi-experimental studies of housing mobility and urban development programs. After this validation exercise, we study the long-run and large-scale impacts of vouchers and place-based subsidies. Both policies result in welfare gains by reducing inequality and generating improvements in average skills and productivity, all of which offset higher levels of taxes and other GE effects. We find that a voucher program generates larger long-run welfare gains relative to place-based policies. Our analysis of transition dynamics, however, suggests there may be more political support for place-based policies.

We present a new class of methods for identification and inference in dynamic models with serially correlated unobservables, which typically imply that state variables are econometrically endogenous. In the context of Industrial Organization, these state variables often reflect econometrically endogenous market structure. We propose the use of Generalized Instrument Variables methods to identify those dynamic policy functions that are consistent with instrumental variable (IV) restrictions. Extending popular \two-step" methods, these policy functions then identify a set of structural parameters that are consistent with the dynamic model, the IV restrictions and the data. We provide computed illustrations to both single-agent and oligopoly examples. We also present a simple empirical analysis that, among other things, supports the counterfactual study of an environmental policy entailing an increase in sunk costs.

This paper uses new bond-level data on sovereign borrowing and defaults during 1869-1914 to quantify a channel of contagion via banks’ reputation for monitoring borrowers. Concerns over reputation incentivized Britain’s merchant banks (who underwrote sovereign bonds) to monitor and exert influence over sovereigns. Default signaled to investors that a bank was less willing or able to write and support quality issues, indicating that its other bonds may underperform in the future. Consistent with reputation-based contagion, I find that comovement between defaulting and non-defaulting bonds is six times larger when the bonds share an underwriter. To isolate the causal effect of a shared underwriter, I exploit within-country variation in bonds’ underwriters. Testing predictions from a dynamic game where underwriters build a reputation for monitoring, I find further evidence supporting reputation as the mechanism – as opposed to alternative explanations such as wealth effects. These findings highlight that the reputation of intermediaries that monitor and intervene in crises can be a powerful source of contagion unrelated to a borrower’s fundamentals.

Banks engage in maturity transformation and the term premium compensates them for bearing the associated interest rate risk. Consistent with this view, I show that banks’ net interest margins and term premia have comoved in the United States over the last decades. On monetary policy announcement days, banks’ stock prices fall in response to an increase in expected future short-term interest rates but rise if term premia increase. These effects are muted for nonbank equity, amplified for banks with a larger maturity mismatch, and reflected in bank cash-flows. The results reveal that banks are not immune to interest rate risk.

Individual forecasts of economic variables show widespread overreaction to news, but laboratory experiments on belief updating typically find underinference from signals. We provide new experimental evidence to connect these two seemingly inconsistent phenomena. Building on a classic experimental paradigm, we study how people make inferences and revise forecasts in the same information environment. Subjects underreact to signals when inferring about underlying states, but overreact to signals when revising forecasts about future outcomes. This gap in belief updating is largely driven by the use of different simplifying heuristics for the two tasks. Additional treatments link our results to the difficulty of recognizing the conceptual connection between making inferences and revising forecasts.

Ten trillion dollars are allocated to illiquid vehicles for which investors commit ex-ante to transferring capital on demand – most of which are Private Equity (PE) funds. We design a dynamic portfolio allocation model in which investors commit capital to PE. The effects of commitment risk on investors’ portfolios and welfare are large. Investors are under-allocated to PE, and willing to pay a premium to update their PE allocation when capital is called, which is more than twice the one to eliminate standard liquidity frictions induced by the limited tradability of PE. In contrast, investors are not willing to pay to remove the uncertainty over the timing of capital call.

Banks are becoming “one-stop shops" serving multiple customer and product segments. This expansion in bank scope can generate complementarities across products and customers (e.g., deposits and loans), potentially leading to lower marginal costs and prices. In this paper we quantify cost synergies for banks using new data from a major commercial credit bureau for firm credit in the United States. Moreover, we consider how banks with a wider scope may have incentives to steer some firms into specific products or loan sizes. We develop and estimate a supply and demand equilibrium model of firm credit. The model allows for cost synergies and steering by banks, while in competition with non-banks. We find that banks’ incentives generate product and quantity distortions for firms, while non-banks that are more specialized in their product offerings do not show such distortionary behavior. We use our estimated model to evaluate the counterfactual equilibrium effects of restricting bank scope and regulating firm credit.

We study two dimensions of complexity that may interfere with individual choice. The first one is object complexity, which corresponds to the difficulty in evaluating any given alternative in a choice set. The second dimension is composition complexity, which increases when suboptimal alternatives become more similar to optimal ones. We develop a satisficing-with-evaluation-errors theory that incorporates both dimensions and delivers sharp empirical predictions about their effect on choice behavior. We confirm these predictions in a novel data set with information on hundreds of millions of decisions in chess endgames. First, as the object complexity of an optimal (suboptimal) alternative increases, it becomes less (more) likely to be chosen. Second, even highly experienced decision-makers are more likely to make mistakes when choosing from sets with higher composition complexity. These findings help to shed some of the first light on the effect of complexity on choice behavior outside of the laboratory.

Discrete choice data allow researchers to recover differences in utilities, but these differences may not suffice to identify policy-relevant counterfactuals of interest. This fundamental tension is important for dynamic discrete choice models because agents' behavior depends on value functions, which require utilities in levels. We propose a unified approach to investigate how much one can learn about counterfactual outcomes under mild assumptions, for a large and empirically relevant class of counterfactuals. We derive analytical properties of sharp identified sets under alternative model restrictions and develop a valid inference approach based on subsampling. To aid practitioners, we propose computationally tractable procedures that bypass model estimation and directly obtain the identified sets for the counterfactuals and the corresponding confidence sets. We illustrate in Monte Carlos, as well as an empirical exercise of firms' export decisions, the informativeness of the identified sets, and we assess the impact of (common) model restrictions on results.

We study the relationship between credit expansions, macroeconomic fluctuations, and financial crises using a novel database on the sectoral distribution of private credit for 116 countries starting in 1940. Theory predicts that the sectoral allocation of credit matters for distinguishing between “good” and “bad” credit booms. We test the prediction that lending to households and the non-tradable sector, relative to the tradable sector, contributes to macroeconomic boom-bust cycles by (i) fueling unsustainable demand booms, (ii) increasing financial fragility, and (iii) misallocating resources across sectors. We show that credit to non-tradable sectors, including construction and real estate, is associated with a boom-bust pattern in output, similar to household credit booms. Such lending booms also predict elevated financial crisis risk and productivity slowdowns. In contrast, tradable-sector credit expansions are followed by stable output and productivity growth without a higher risk of a financial crisis. Our findings highlight that what credit is used for is important for understanding macro-financial linkages.

This paper provides evidence that customer discrimination in the market for doctors can be largely accounted for by statistical discrimination. I evaluate customer preferences in the field with an online platform where cash-paying consumers can shop and book a provider for medical procedures based on an experimental paradigm called validated incentivized conjoint analysis (VIC). Customers evaluate doctor options they know to be hypothetical to be matched with a customized menu of real doctors, preserving incentives. Racial discrimination reduces patient willingness-to-pay for black and Asian providers by 12.7% and 8.7% of the average colonoscopy price. Further, providing signals of provider quality reduces this willingness-to-pay racial gap by about 90% suggesting statistical discrimination as an important cause of the gap. Actual booking behavior allows cross validation of incentive compatibility of stated preference elicitation via VIC.

We propose a methodology for effectively modeling individual heterogeneity using deep learning while still retaining the interpretability and economic discipline of classical models. We pair a transparent, interpretable modeling structure with rich data environments and machine learning methods to estimate heterogeneous parameters based on potentially high dimensional or complex observable characteristics. Our framework is widely-applicable, covering numerous settings of economic interest. We recover, as special cases, well-known examples such as average treatment effects and parametric components of partially linear models. However, we also seamlessly deliver new results for diverse examples such as price elasticities, willingness-to-pay, and surplus measures in choice models, average marginal and partial effects of continuous treatment variables, fractional outcome models, count data, heterogeneous production function components, and more. Deep neural networks are particularly well-suited to structured modeling of heterogeneity in economics: we show how the network architecture can be easily designed to match the global structure of the economic model, giving novel methodology for deep learning as well as, more formally, improved rates of convergence. Our results on deep learning have consequences for other structured modeling environments and applications, such as for additive models or other varying coefficient models. Our inference results are based on an influence function we derive, which we show to be flexible enough to to encompass all settings with a single, unified calculation, removing any requirement for case-by-case derivations. The usefulness of the methodology in economics is shown in two empirical applications: we study the response of 410(k) participation rates to firm matching and the impact of prices on subscription choices for an online service. Extensions of the main ideas to instrumental variables and multinomial choices are shown.