Knowledge that Transforms

Boys are less likely than girls to enter college, a gap that is often attributed to a lack of non-cognitive skills such as motivation and self-discipline.We study how being classified as gifted -determined by having an IQ score of 116 or higher -affects college entry rates of disadvantaged children in a large urban school district.For boys with IQ's around the cutoff, gifted identification raises the college entry rate by 25-30 percentage points -enough to catch up with girls in the same IQ range.In contrast, we find small effects for girls.Looking at course-taking and grade outcomes in middle and high school, we find large effects of gifted status for boys that close most of the gaps with girls, but no detectable effects on standardized tests scores of either gender.Overall, we interpret the evidence as demonstrating that gifted services raise the non-cognitive skills of boys conditional on their cognitive skills, leading to gains in educational attainment.

I examine how disruptions to students' environments propagate to their classmates to understand the mechanisms behind peer interactions in the classroom.I combine administrative and survey data from Chile with detailed measures of housing damages from the 2010 earthquake, one of the most violent ever recorded.Damages to a student's own home reduced achievement and raised self-reported cost of study effort.Average damages among classmates induced schools to reallocate resources towards student support and increased achievement.In contrast, dispersion in classmates' damages had heterogeneous achievement effects across the prior performance distribution, which schools did not appear to mitigate, pointing to peer interactions.Motivated by evidence suggesting students value classroom rank, I show that a game-of-status model of competition for grades rationalizes the findings.The results suggest that, beyond production complementarities and a desire to conform, a desire to compete could shape peer effects on learning.2 Examples of studies using naturally occurring exogenous variation in peer characteristics to account for confounding effects are Hoxby (2000); Angrist and Lang (2004); Hoxby and Weingarth (2005); Lavy, Paserman, and Schlosser (2012); Imberman, Kugler, and Sacerdote (2012).3 I gratefully acknowledge Prof. Sergio Ruiz of the Geology Department at the University of Chile, a leading expert on the seismic vulnerability of Chilean buildings, for feedback on the damage measure.The measure falls within Deterministic Earthquake Loss Estimation, which aims to estimate losses from a specific seismic event.It stands in contrast to Probabilistic Earthquake Loss Estimation, which aims to predict potential losses from many possible seismic events (McGuire (2004)).

We construct a dynamic model of child development where forward-looking parents and children jointly take actions to increase the child’s cognitive and noncognitive skills within a Markov perfect equilibrium framework. In addition to time and money investments in their child, parents also choose whether to use explicit incentives to increase the child’s self-investment, which may reduce the child’s future intrinsic motivation to invest by reducing the child’s discount factor. We use the estimated model parameters to show that the use of extrinsic motivation has large costs in terms of the child’s future incentives to invest in themselves.

A rent guarantee insurance (RGI) policy makes a limited number of rent payments to the landlord on behalf of an insured tenant unable to pay rent due to a negative income or health expenditure shock. We introduce RGI in a rich quantitative equilibrium model of housing insecurity and show it increases welfare by improving risk sharing across idiosyncratic and aggregate states of the world, reducing the need for a large security deposits, and reducing homelessness which imposes large costs on society. While unrestricted access to RGI is not financially viable due to moral hazard and adverse selection, restricting access can restore viability. Private insurers target better off renters to break even, while public insurers focus on households most at-risk of homelessness. Stronger tenant protections increase the effectiveness of RGI.

An informed seller designs a dynamic mechanism to sell an experience good. The seller has partial information about the product match, which affects the buyer's private consumption experience. We characterize equilibrium mechanisms of this dynamic informed principal problem. The belief gap between the informed seller and the uninformed buyer, coupled with the buyer's learning, gives rise to mechanisms that provide the skeptical buyer with limited access to the product and an option to upgrade if the buyer is swayed by a good experience. Depending on the seller's screening technology, this takes the form of free/discounted trials or tiered pricing, which are prevalent in digital markets. In contrast to static environments, having consumer data can reduce sellers' revenue in equilibrium, as they fine-tune the dynamic design with their data forecasting the buyer's learning process.

We study the macroeconomic implications of asymmetric information in capital markets.We build a quantitative capital-accumulation model in which capital is traded in illiquid markets, with sellers having more information about capital quality than buyers.Asymmetric information distorts the terms of trade for sellers of high-quality capital, who list higher prices and are willing to accept lower trading probabilities to signal their type.Led by the model's predictions, we measure the distortions from asymmetric information by studying the relationship between listed prices and trading probabilities in a unique dataset of individual capital units listed for trade.By combining the empirical measurement with the model, we show that information asymmetries can play a quantitatively large role during economic crises when the degree of asymmetric information deteriorates.

We present a model where social media platforms offer plans that intermix entertaining content with digital advertising (“ads”). Users derive utility from entertainment and learn about their valuation for a product from ads. While some users are fully rational, others naïvely perceive digital ads as more informative than they actually are. We characterize the profit-maximizing business model of the platform and show that welfare is lower when the platform monetizes through advertising instead of subscription both for naïfs (because they are targeted by intense digital advertising, which makes them over-optimistic about product quality and over-purchase the product) and for sophisticates (because the inflated demand from naïfs increases the firm’s price). This negative welfare effect is intensified when the platform can offer mixed business models that separate the naïve and sophisticated users into different plans. Our results are robust to firm-level and platform-level competition, because digital ads soften competition between both firms and platforms. We also show how digital ad taxes can improve welfare.<br><br>Institutional subscribers to the NBER working paper series, and residents of developing countries may download this paper without additional charge at <a href="http://www.nber.org/papers/w33017" TARGET="_blank">www.nber.org</a>.<br>

Recent years have seen tremendous advances in the theory and application of sequential experiments. While these experiments are not always designed with hypothesis testing in mind, researchers may still be interested in performing tests after the experiment is completed. The purpose of this paper is to aid in the development of optimal tests for sequential experiments by analyzing their asymptotic properties. Our key finding is that the asymptotic power function of any test can be matched by a test in a limit experiment where a Gaussian process is observed for each treatment, and inference is made for the drifts of these processes. This result has important implications, including a powerful sufficiency result: any candidate test only needs to rely on a fixed set of statistics, regardless of the type of sequential experiment. These statistics are the number of times each treatment has been sampled by the end of the experiment, along with final value of the score (for parametric models) or efficient influence function (for non-parametric models) process for each treatment. We then characterize asymptotically optimal tests under various restrictions such as unbiasedness, α-spending constraints etc. Finally, we apply our our results to three key classes of sequential experiments: costly sampling, group sequential trials, and bandit experiments, and show how optimal inference can be conducted in these scenarios.

Private signals model noisy information about an unknown state. Although these signals are called "private," they may still carry information about each other. Our paper introduces the concept of private private signals, which contain information about the state but not about other signals. To achieve privacy, signal quality may need to be sacrificed. We study the informativeness of private private signals and characterize those that are optimal in the sense that they cannot be made more informative without violating privacy. We discuss implications for privacy in recommendation systems, information design, causal inference, and mechanism design.