Knowledge that Transforms

We study how endogenous innovation and technology diffusion interact to determine the shape of the productivity distribution and generate aggregate growth. We model firms that choose to innovate, adopt technology, or produce with their existing technology. Costly adoption creates a spread between the best and worst technologies concurrently used to produce similar goods. The balance of adoption and innovation determines the shape of the distribution; innovation stretches the distribution, while adoption compresses it. On the balanced growth path, the aggregate growth rate equals the maximum growth rate of innovators. While innovation drives long‐run growth, changes in the adoption environment can influence growth by affecting innovation incentives, either directly, through licensing of excludable technologies, or indirectly, via the option value of adoption.

At an initial time, an individual forms a belief about a future random outcome. As time passes, the individual may obtain, privately or subjectively, further information, until the outcome is eventually revealed. How can a protocol be devised that induces the individual, as a strict best response, to reveal at the outset his prior assessment of both the final outcome and the information flows he anticipates and, subsequently, what information he privately receives? The protocol can provide the individual with payoffs that depend only on the outcome realization and his reports. We develop a framework to design such protocols, and apply it to construct simple elicitation mechanisms for common dynamic environments. The framework is general: we show that strategyproof protocols exist for any number of periods and large outcome sets. For these more general settings, we build a family of strategyproof protocols based on a hierarchy of choice menus, and show that any strategyproof protocol can be approximated by a protocol of this family.

How much capital should financial intermediaries hold? We propose a general equilibrium model with a financial sector that makes risky long‐term loans to firms, funded by deposits from savers. Government guarantees create a role for bank capital regulation. The model captures the sharp and persistent drop in macro‐economic aggregates and credit provision as well as the sharp change in credit spreads observed during financial crises. Policies requiring intermediaries to hold more capital reduce financial fragility, reduce the size of the financial and non‐financial sectors, and lower intermediary profits. They redistribute wealth from savers to the owners of banks and non‐financial firms. Pre‐crisis capital requirements are close to optimal. Counter‐cyclical capital requirements increase welfare.

This paper proposes a tractable model of Bayesian learning on large random networks where agents choose whether to adopt an innovation. We study the impact of the network structure on learning dynamics and product diffusion. In directed networks, all direct and indirect links contribute to agents' learning. In comparison, learning and welfare are lower in undirected networks and networks with cliques. In a rich class of networks, behavior is described by a small number of differential equations, making the model useful for empirical work.

We investigate learning at the workplace. To do so, we use German administrative data that contain information on the entire workforce of a sample of establishments. We document that having more‐highly‐paid coworkers is strongly associated with future wage growth, particularly if those workers earn more. Motivated by this fact, we propose a dynamic theory of a competitive labor market where firms produce using teams of heterogeneous workers that learn from each other. We develop a methodology to structurally estimate knowledge flows using the full‐richness of the German employer‐employee matched data. The methodology builds on the observation that a competitive labor market prices coworker learning. Our quantitative approach imposes minimal restrictions on firms' production functions, can be implemented on a very short panel, and allows for potentially rich and flexible coworker learning functions. In line with our reduced‐form results, learning from coworkers is significant, particularly from more knowledgeable coworkers. We show that between 4 and 9% of total worker compensation is in the form of learning and that inequality in total compensation is significantly lower than inequality in wages.

We develop a model of banking industry dynamics to study the quantitative impact of regulatory policies on bank risk‐taking and market structure. Since our model is matched to U.S. data, we propose a market structure where big banks with market power interact with small, competitive fringe banks as well as non‐bank lenders. Banks face idiosyncratic funding shocks in addition to aggregate shocks which affect the fraction of performing loans in their portfolio. A nontrivial bank size distribution arises out of endogenous entry and exit, as well as banks' buffer stock of capital. We show that the model predictions are consistent with untargeted business cycle properties, the bank lending channel, and empirical studies of the role of concentration on financial stability. We find that regulatory policies can have an important impact on banking market structure, which, along with selection effects, can generate changes in allocative efficiency and stability.

We examine intergenerational mobility (IM) in educational attainment in Africa since independence using census data. First, we map IM across 27 countries and more than 2,800 regions, documenting wide cross-country and especially within-country heterogeneity. Inertia looms large as differences in the literacy of the old generation explain about half of the observed spatial disparities in IM. The rural-urban divide is substantial. Though conspicuous in some countries, there is no evidence of systematic gender gaps in IM. Second, we characterize the geography of IM, finding that colonial investments in railroads and Christian missions, as well as proximity to capitals and the coastline are the strongest correlates. Third, we ask whether the regional differences in mobility reflect spatial sorting or their independent role. To isolate the two, we focus on children whose families moved when they were young. Comparing siblings, looking at moves triggered by displacement shocks, and using historical migrations to predict moving-families' destinations, we establish that, while selection is considerable, regional exposure effects are at play. An extra year spent in a high-mobility region before the age of 12 (and after 5) significantly raises the likelihood for children of uneducated parents to complete primary school. Overall, the evidence suggests that geographic and historical factors laid the seeds for spatial disparities in IM that are cemented by sorting and the independent impact of regions.

Standard experimental designs are geared toward point estimation and hypothesis testing, while bandit algorithms are geared toward in‐sample outcomes. Here, we instead consider treatment assignment in an experiment with several waves for choosing the best among a set of possible policies (treatments) at the end of the experiment. We propose a computationally tractable assignment algorithm that we call “exploration sampling,” where assignment probabilities in each wave are an increasing concave function of the posterior probabilities that each treatment is optimal. We prove an asymptotic optimality result for this algorithm and demonstrate improvements in welfare in calibrated simulations over both non‐adaptive designs and bandit algorithms. An application to selecting between six different recruitment strategies for an agricultural extension service in India demonstrates practical feasibility.

We propose a bootstrap procedure for data that may exhibit cluster‐dependence in two or more dimensions. The asymptotic distribution of the sample mean or other statistics may be non‐Gaussian if observations are dependent but uncorrelated within clusters. We show that there exists no procedure for estimating the limiting distribution of the sample mean under two‐way clustering that achieves uniform consistency. However, we propose bootstrap procedures that achieve adaptivity with respect to different uniformity criteria. Important cases and extensions discussed in the paper include regression inference, U‐ and V‐statistics, subgraph counts for network data, and non‐exhaustive samples of matched data.

A ruler who does not identify with a social group, whether on religious, ethnic, cultural, or socioeconomic grounds, is confronted with a trade‐off between taking advantage of the out‐group population's eagerness to maintain its identity and inducing it to “comply” (conversion, quitting, exodus, or any other way to accommodate the ruler's own identity). This paper first nests economists' extraction model, in which rulers are revenue‐maximizers, within a more general identity‐based model, in which rulers care also about inducing people to lose their identity, both in a static and an evolving environment. This paper then constructs novel data sources to test the implications of both models in the context of Egypt's conversion to Islam between 641 and 1170. The evidence supports the identity‐based model.