Search

Prediction markets elicit and aggregate beliefs by paying agents based on how close their predictions are to a verifiable future outcome. However, outcomes of many important questions are difficult to verify or unverifiable, in that the ground truth may be hard or impossible to access. Examples include questions about causal effects where it is infeasible or unethical to run randomized trials; crowdsourcing and content moderation tasks where it is prohibitively expensive to verify ground truth; and questions asked over long time horizons, where the delay until the realization of the outcome skews agents' incentives to report their true beliefs. We present a novel and unintuitive result showing that it is possible to run an $\varepsilon-$incentive compatible prediction market to elicit and efficiently aggregate information from a pool of agents without observing the outcome by paying agents the negative cross-entropy between their prediction and that of a carefully chosen reference agent. Our key insight is that a reference agent with access to more information can serve as a reasonable proxy for the ground truth. We use this insight to propose self-resolving prediction markets that terminate with some probability after every report and pay all but a few agents based on the final prediction. We show that it is an $\varepsilon-$Perfect Bayesian Equilibrium for all agents to report truthfully in our mechanism and to believe that all other agents report truthfully. Although primarily of interest for unverifiable outcomes, this design is also applicable for verifiable outcomes.

Spring-Summer 2020 Forecasting Tournament on COVID-related outcomes (e.g., # of deaths, # of statewide stay-at-home-orders)

We preregister an experimental survey study testing multiple hypotheses about the effect of training and political polarization on forecasting accuracy.

We examine 70,581 felony court cases filed in Chicago, IL during the period 1990-2007. We exploit case randomization to assess the impact of judge assignment and sentencing decisions on the arrival rates of new charges. Relative to prior research, we document an important source of heterogeneity in the impact of incarceration on recidivism. Incarceration creates lasting reductions in recidivism among first offenders but not repeat offenders. We present suggestive evidence that these reductions among first offenders primarily reflect outcomes for offenders who live in lower-crime areas of the city and are not involved in the drug trade. During our sample period, Illinois parole officers were able to issue arrest warrants for former inmates under their supervision. These powers place former inmates at significant risk of returning to prison as punishment for violations of technical conditions of their supervision. However, we find no evidence that these police powers increased the arrival rate of new charges against formerly incarcerated offenders. Incarceration does not reduce the arrival of new criminal charges among repeat offenders, and this outcome is not the result of parole officers over-policing repeat offenders.

Between 1890 and 1921, Andrew Carnegie funded the construction of 1,618 public libraries in cities and towns across the United States. I link these library construction grants to census data and measure the effect of childhood public library access on adult outcomes. Library construction grants increased children's educational attainment by 0.10 years, did not affect wage income, and increased non-wage income by 4%. These income effects are driven by occupational choice. Access to a public library caused children to shift away from occupations like manual labor, factory-work, and mining into safer and more prestigious occupations like farm-ownership, clerical, and technical jobs. I show that compulsory schooling laws had parallel effects on children, increasing educational attainment, non-wage income and occupational prestige without affecting wage income. Economists often rely solely on wage income to measure the returns to education. But public libraries and compulsory schooling laws in the early 1900s increased educational attainment and had positive effects on children's adult labor market outcomes without affecting wage income.

Economists often evaluate public policies by measuring the effects of those policies on participants' wage income. I use two historical settings to argue that this focus on wage income biases estimates of the returns to education, the returns to local investment, and the measurement of inequality. I investigate a rich set of non-wage outcomes that are available in historical decennial Census data, and I show that by augmenting standard economic and econometric analyses with these new outcomes, economists and policymakers can develop a more complete understanding of long-run trends in inequality and the causal effects of public policies on children. In the first chapter of my dissertation, I measure the causal effects of public libraries on children. Between 1890 and 1921, Andrew Carnegie funded the construction of 1,618 public libraries in cities and towns across the United States. I link these library construction grants to census data and measure the effect of childhood public library access on adult outcomes. Library construction grants increased children's educational attainment by 0.10 years, did not affect wage income, and increased non-wage income by 4%. These income effects are driven by occupational choice. Access to a public library caused children to shift away from occupations like manual labor, factory-work, and mining into safer and more prestigious occupations like farm-ownership, clerical, and technical jobs. I show that compulsory schooling laws had parallel effects on children, increasing educational attainment, non-wage income and occupational prestige without affecting wage income. Economists often rely solely on wage income to measure the returns to education. But public libraries and compulsory schooling laws in the early 1900s increased educational attainment and had positive effects on children's adult labor market outcomes without affecting wage income. In the second chapter of my dissertation, I embed income and mortality in a unified measure of black-white inequality. In the 1940 census, white males earned twice as much as black males. But white males born in 1900 had lifetime earnings 3.4 times as large as black males born in 1900. The black-white lifetime earnings gap is so much larger than the cross-sectional earnings gap because of the large black-white gap in life expectancy. 48% of black males born in 1900 died before the age of 30 as compared to just 26% of white males. Economists often use cross-sectional earnings gaps to measure inequality between groups, but a more complete measure of inequality combines income profiles and mortality risk. I develop a model of optimal consumption in a world with mortality and I calibrate the model separately for each cohort of black and white males born between 1900 and 1970. Using this model, I find that the black-white welfare gap shrank by 48% from the 1900 to the 1920 birth cohorts as black and white mortality rates converged, but the black-white welfare gap declined only modestly from the 1920 to 1970 birth cohorts.

We create a new weekly index of retail trade that accurately predicts the U.S. Census Bureau's Monthly Retail Trade Survey (MRTS). The index's weekly frequency provides an early snapshot of the MRTS and allows for a more granular analysis of the aggregate consumer response to fast-moving events such as the Covid-19 pandemic. To construct the index, we extract the co-movement in weekly data series capturing credit and debit card transactions, small business revenue, mobility, and consumer sentiment. To ensure that the index is representative of aggregate retail spending, we implement a novel benchmarking method that uses a mixed-frequency dynamic factor model to constrain the weekly index to match the monthly MRTS. We use the index to document several interesting features of U.S. retail sales during the Covid-19 pandemic, many of which are not visible in the MRTS. In addition, we show that our index would have more accurately predicted the MRTS in real time during the pandemic when compared to either consensus forecasts available at the time, monthly autoregressive models, or other high-frequency data that attempts to track consumer spending. The gains are substantial, with roughly 60 to 80 percent reductions in mean absolute forecast errors.

We identify 16,016 recipients of Covid-19 Economic Impact Payments in anonymized transaction-level debit card data from Facteus. We use an event study framework to show that in the two weeks following a sudden $1,200 payment from the IRS, consumers immediately increased spending by an average of $577, implying a marginal propensity to consume (MPC) of 48%. Consumer spending falls back to normal levels after two weeks. Stimulus recipients who live paycheck-to-paycheck spend 68% of the stimulus payment immediately, while recipients who save much of their monthly income spend 23% of the stimulus payment immediately. Consumer age and location are only marginally correlated with individual MPCs after controlling for each individual's pre-pandemic propensity to save. We use the 2018 American Community Survey to re-weight our data to match the U.S. population. Ignoring equilibrium effects and assuming a constant MPC for each person, we estimate that the CARES Act's $296 billion of payments to individuals will increase consumer spending by $138 billion (47% of total outlays). A stimulus bill of the same size targeted at individuals with the highest MPCs would have instead increased consumer spending by $201 billion (68% of total outlays).

We link the county-level rollout of stay-at-home orders to anonymized cellphone records and consumer spending data. We document three patterns. First, stay-at-home orders caused people to stay at home: county-level measures of mobility declined by between 9% and 13% by the day after the stay-at-home order went into effect. Second, stay-at-home orders caused large reductions in spending in sectors associated with mobility: restaurants and retail stores. However, food delivery sharply increased after orders went into effect. Third, there is substantial county-level heterogeneity in consumer behavior in the days leading up to a stay-at-home order.