Knowledge that Transforms

This paper documents that production networks play an essential role in the job search and matching process. We document five facts about worker mobility in production networks using employer–employee data matched with the universe of firm-to-firm transactions for the Dominican Republic: (1) workers move between buyers and suppliers almost twice as much as predicted by standard labour market characteristics, (2) movers between buyers and suppliers experience larger earnings increases than other movers, (3) incumbent workers earnings increase when their firm hires from its buyers or suppliers, (4) firm-to-firm trade increases following supply chain hires, and (5) hiring from buyers or suppliers is associated with stronger firm growth. Survey evidence points to supply chain-specific human capital and better information about job applicants as the main reasons for hiring within the supply chain. These results reveal a new channel through which factors affecting the supply chain, such as international outsourcing or contracting frictions, impact labour markets.

Wealthy households disproportionately invest in equity, causing equity returns to generate large and persistent fluctuations in top wealth inequality. Motivated by this observation, I study the joint dynamics of asset prices and wealth inequality in a model where a subset of agents (entrepreneurs) hold levered positions on the economy. In the model, as in the data, the wealth distribution is stochastic and it exhibits a Pareto tail, with a tail index that depends on the logarithmic average return of top households. The model features a feedback loop between asset prices and wealth inequality, which amplifies the effect of aggregate shocks on the economy. The model, calibrated to the U.S. data, can account for a substantial portion of the fluctuations in asset prices and top wealth shares over the 20th century.

I introduce a generic method for inference about a scalar parameter in research designs with a finite number of heterogeneous clusters where only a single cluster received treatment. This situation is commonplace in difference-in-differences estimation, but the test developed here applies more broadly. I show that the test controls size and has power under asymptotics where the number of observations within each cluster is large but the number of clusters is fixed. The test combines weighted, approximately Gaussian parameter estimates with a rearrangement procedure to obtain its critical values. The weights needed for most empirically relevant situations are tabulated in the paper. Calculation of the critical values is computationally simple and does not require simulation or resampling. The rearrangement test is highly robust to situations where some clusters are much more variable than others. Examples and an empirical application are provided.

This paper proposes that strong institutions provide comparative advantage in clean industries, and thereby improve a country’s environmental quality. I study financial, judicial, and labour market institutions. Five complementary tests evaluate and assess implications of this hypothesis. First, industries that depend on institutions are clean. Second, strong institutions increase relative exports in clean industries. Third, an industry’s complexity helps explain the link between institutions and clean goods. Fourth, cross-country differences in the composition of output between clean and dirty industries explain an important share of the global distribution of emissions. Fifth, a quantitative general equilibrium model indicates that strengthening a country’s institutions decreases its pollution through relocating dirty industries abroad, though increases pollution in other countries. The comparative advantage that strong institutions provide in clean industries gives one under-explored reason why developing countries have relatively high pollution levels.

We estimate impacts of male job loss, female job loss, and male unemployment benefits on domestic violence (DV) in Brazil. We merge individual-level employment and welfare registers with different measures of DV: judicial cases brought to criminal courts, the use of public shelters by victims, and mandatory DV notifications by health providers. Leveraging mass layoffs for identification, we first show that both male and female job loss, independently, lead to large, and pervasive increases in DV. Using a regression discontinuity design, we then show that access to unemployment benefits does not reduce DV while benefits are being paid, and it leads to higher DV risk once benefits expire. Our findings can be explained by the negative income shock brought by job loss and by increased exposure of victims to perpetrators, as partners tend to spend more time together after displacement. Although unemployment benefits partially offset the income drop following job loss, they reinforce the exposure shock as they increase unemployment duration. Since our results cannot be explained by prominent DV theories, we propose a simple model formalizing these mechanisms.

We develop a new class of random graph models for the statistical estimation of network formation—subgraph generated models (SUGMs). Various subgraphs—e.g. links, triangles, cliques, stars—are generated and their union results in a network. We show that SUGMs are identified and establish the consistency and asymptotic distribution of parameter estimators in empirically relevant cases. We show that a simple four-parameter SUGM matches basic patterns in empirical networks more closely than four standard models (with many more dimensions): (1) stochastic block models; (2) models with node-level unobserved heterogeneity; (3) latent space models; and (4) exponential random graphs. We illustrate the framework’s value via several applications using networks from rural India. We study whether network structure helps enforce risk-sharing and whether cross-caste interactions are more likely to be private. We also develop a new central limit theorem for correlated random variables, which is required to prove our results and is of independent interest.

Diversity can pose fundamental challenges to state building and development. The Tanzanian Ujamaa policy—one of post-colonial Africa’s largest state-building experiments—addressed these challenges by resettling a diverse population in planned villages, where children received political education. We combine differences in exposure to Ujamaa across space and age to identify long-term impacts of the policy. Analysis of contemporary surveys shows persistent, positive effects on national identity and perceived state legitimacy. Our preferred interpretation, supported by evidence that considers alternative hypotheses, is that changes to educational content drive our results. Our findings also point to trade-offs associated with state building: while the policy contributed to establishing the new state as a legitimate central authority, exposure to Ujamaa lowered demands for democratic accountability and did not increase generalized inter-ethnic trust.

Industrial policies are widely used across the world. In practice, designing and implementing these policies is a complicated task. In this paper, we assess the long-term performance of different industrial policy instruments, which include production subsidies, investment subsidies, entry subsidies, and consolidation policies. To do so, we examine a recent industrial policy in China aiming to propel the country’s shipbuilding industry to the largest globally. Using firm-level data from 1998 to 2014 and a dynamic model of firm entry, exit, investment, and production, we find that (i) the policy boosted China’s domestic investment, entry, and international market share dramatically, but delivered low returns and led to fragmentation, idle capacity, as well as depressed world ship prices; (ii) the effectiveness of different policy instruments is mixed: production and investment subsidies can be justified by market share considerations, while entry subsidies are wasteful; (iii) counter-cyclical policies, firm-targeting, and shortening the intervention horizon can substantially reduce distortions. Our results highlight the critical role of firm heterogeneity, business cycles, and firms’ cost structure in policy design. Finally, when exploring potential rationales, we find support for nonclassical considerations, such as reducing freight rates to boost Chinese trade.

Calibration, the practice of choosing the parameters of a structural model to match certain empirical moments, can be viewed as minimum distance estimation. Existing standard error formulas for such estimators require a consistent estimate of the correlation structure of the empirical moments, which is often unavailable in practice. Instead, the variances of the individual empirical moments are usually readily estimable. Using only these variances, we derive conservative standard errors and confidence intervals for the structural parameters that are valid even under the worst-case correlation structure. In the over-identified case, we show that the moment weighting scheme that minimizes the worst-case estimator variance amounts to a moment selection problem with a simple solution. Finally, we develop tests of over-identifying or parameter restrictions. We apply our methods empirically to a model of menu cost pricing for multi-product firms and to a heterogeneous agent New Keynesian model.

This article studies the problem of optimally allocating treatments in the presence of spillover effects, using information from a (quasi-)experiment. I introduce a method that maximizes the sample analogue of average social welfare when spillovers occur. I construct semi-parametric welfare estimators with known and unknown propensity scores and cast the optimization problem into a mixed-integer linear program, which can be solved using off-the-shelf algorithms. I derive a strong set of guarantees on regret, i.e. the difference between the maximum attainable welfare and the welfare evaluated at the estimated policy. The proposed method presents attractive features for applications: (i) it does not require network information of the target population; (ii) it exploits heterogeneity in treatment effects for targeting individuals; (iii) it does not rely on the correct specification of a particular structural model; and (iv) it accommodates constraints on the policy function. An application for targeting information on social networks illustrates the advantages of the method.