Economists are often interested in the mechanisms by which a treatment affects an outcome. We develop tests for the “sharp null of full mediation” that a treatment D affects an outcome Y only through a particular mechanism (or set of mechanisms) M. Our approach exploits connections between mediation analysis and the econometric literature on testing instrument validity. We also provide tools for quantifying the magnitude of alternative mechanisms when the sharp null is rejected: we derive sharp lower bounds on the fraction of individuals whose outcome is affected by the treatment despite having the same value of M under both treatments (“always-takers”), as well as sharp bounds on the average effect of the treatment for such always-takers. An advantage of our approach relative to existing tools for mediation analysis is that it does not require stringent assumptions about how M is assigned. We illustrate our methodology in two empirical applications.

This paper combines new data and a narrative approach to identify variation in political pressure on the Federal Reserve. From archival records, I build a data set of personal interactions between U.S. Presidents and Fed officials between 1933 and 2016. Since personal interactions do not necessarily reflect political pressure, I develop a narrative identification strategy based on President Nixon's pressure on Fed Chair Burns. I exploit this narrative through restrictions on a structural vector autoregression that includes the President-Fed interaction data. I find that political pressure to ease monetary policy (i) increases the price level strongly and persistently, (ii) does not lead to positive effects on real economic activity, (iii) contributed to inflationary episodes outside of the Nixon era, and (iv) transmits differently from a typical monetary policy easing, by having a stronger effect on inflation expectations. Quantitatively, increasing political pressure by half as much as Nixon, for six months, raises the price level by about 7% over the following decade.

I discuss recent works that evaluate the effects of technology and knowledge transfers in the twentieth century. Less developed economies cyclically rely on such transfers to promote industrialization and close the gap with the most advanced countries. I focus on embodied technology transfers through capital goods; disembodied transfers via patents, licensing, and worker mobility; the diffusion of managerial know-how; and industrial policy interventions. The analysis highlights the central role of absorptive capacity—human capital, institutional quality, and organizational capability—in determining whether transfers yield lasting productivity gains.

We investigate how common ownership between lenders affects the terms of syndicated loans. We provide a novel view on the role of common ownership in mitigating information asymmetries on the quality of borrowers and the contractual distortions of lending conditions. Empirically, we show that higher levels of common ownership lower loan spreads, decrease the share of loans retained by the lead bank, and relax liquidity constraints at issuance. We use a novel exclusion restriction based on deposit multimarket contact to identify the effect of common ownership on loan pricing after accounting for its impact on lenders’ participation in the syndicate.

How do emotions affect policy views? How do they influence the way people process factual information? We address these questions using a survey experiment in Italy, which randomly exposes 12,000 participants to (i) sensational news about immigrant crimes, (ii) statistical information about immigration, or (iii) the combination of both. First, we find that highly emotional news stories significantly increase the demand for anti-immigration policies, while less emotional news have smaller or no impact. Consistent with a causal role of emotions, this differential effect persists when we provide statistical information that holds posterior beliefs constant across the different news treatments. Second, we find that providing information generally helps correct factual beliefs, even though learning is slightly disrupted when participants are emotionally triggered. Third, we show that emotions strongly influence whether belief updating leads to changes in policy views. When presented alone, factual information reduces anti-immigration attitudes. When paired with emotional news, the emotional reaction overrides the effect of information, leading participants to adopt anti-immigration views as strongly as when exposed to emotional news alone. Once negative emotions are triggered, having more accurate factual knowledge no longer matters for forming policy views.

We provide detailed estimates of how the marginal propensity to consume out of wealth (MPC) varies along the distribution of household wealth and by asset composition, and analyse the sources of MPC heterogeneity across euro area countries. To do this, we i) build a household-level panel dataset combining wealth and consumption surveys for five European countries, and ii) use instrumented household-level panel regressions. First, we find heterogeneity across the wealth distribution with lower MPCs for high-wealth households. Second, we account for asset composition and show the significant role of housing wealth in all countries. We show that our results are indicative of a collateral channel. Third, cross-country differences in MPCs are mostly explained by country-specific institutional and socio-economic characteristics in Germany (compared to Spain) and by differences in consumption behaviours for Belgium, Cyprus and Italy. We show that MPC heterogeneity is related to homeownership rates, mortgage markets, demographics, and wealth inequality. Finally, we investigate to what extent heterogeneous MPC and wealth inequality affect consumption inequality.

Conflicts between management and workers are common in newly industrializing countries. Combining ethnographic, survey and administrative records from a Bangladeshi sweater factory, we study how workers responded when management laid off a quarter of the workers following a period of labor unrest. After the unrest, the factory experienced a substantial drop in productivity. Among surviving workers, those who likely had strong social connections - friends - among fired co-workers suffered relatively larger declines in productivity. Additional evidence on potential mechanism indicates a deliberate shading of effort to punish the factory’s management.

Stark and Byra (2025) identify a mistake in how capital income is calculated in the supplementary files of Battisti et al. (2018) and point out that our wage bargaining equation does not correspond to the Nash bargaining solution in the presence of wage taxation. We acknowledge both points and thank the authors for pointing out these issues. The bargaining solution we use in Battisti et al. (2018) is the Kalai (1977) proportional bargaining solution with fixed surplus shares, not Nash bargaining. In our reply, we show that if we calibrate the model to wage tax rates, correcting the coding mistake has only a minor effect on our original results if maintaining the original Kalai bargaining. If we also switch to the Nash bargaining solution, the estimated gains from immigration become smaller. Welfare effects depend on the precise nature of bargaining and tend to be larger with Kalai proportional bargaining. Average total welfare gains from immigration are then 0.37% for both low-skilled and high-skilled natives, instead of 1.25% for high-skilled and 1.00% for low-skilled natives in Battisti et al. (2018).

Decarbonising the power sector requires investments in renewable energy and storage. Although often viewed as complements, these technologies can also act as strategic substitutes. When renewable output coincides with high demand, storage may reduce renewable profits, and vice-versa. When renewable technologies produce at different times, storage can benefit one technology while disadvantaging the other. These findings suggest that, in solar-dominated systems, an initial push for solar may be needed before storage and renewables become reinforcing. Simulations of the Spanish electricity market confirm that, at high solar penetration, storage increases solar profitability but lowers wind revenues.

We report results from a large-scale randomized controlled trial aimed at empowering young women during the transition from secondary school. The trial includes four school-based arms: economic empowerment, reproductive health empowerment, their combination, and a control group. The analysis draws on a rich dataset combining survey and laboratory measures collected over more than five years. The economic program increased self-employment and income, while the reproductive health program strengthened self-control and increased committed relationships. Both interventions also increased early pregnancy, which we attribute to positive income and relationship effects in a context characterized by high fertility and self-employment.

Concerns over the adequacy of retirement incomes have led governments to incentivise low- and middle-income earners to contribute more to private pensions. In this study we exploit several reforms using a simulated instruments approach and administrative data to estimate the impacts of a targeted national contribution-matching scheme in Australia across 50%, 100% and 150% match rates. Overall, we find that responses increase with the match rate, are modest in size and are mostly unwound when eligibility is lost. Sub-group analysis highlights the possibility that responses are limited by liquidity constraints. We find little evidence that the scheme crowds-out other savings.

As society becomes increasingly diverse, will changes in an individual’s exposure to diversity influence their interactions with others? I study prosocial behaviour in an experiment on a large-scale U.S. sample, in which participants are exogenously exposed to social contexts with varying levels of nationality diversity. I find that diverse contexts amplify participants’ ingroup bias—the tendency to favour one’s own group—driven by higher allocations towards fellow nationals and lower allocations to foreigners, relative to giving in homogeneous contexts where such bias is not present. I provide suggestive evidence that changes in perceptions of social proximity correspond to a driver of the effect of diversity on allocations. The findings are consistent across subgroups of the population, which suggests that the study identifies a general process through which individuals identify with groups, where social context—and not only individual characteristics—is key for the emergence of ingroup bias.

I investigate the unintended effects of publicly informing individuals that their selection for a beneficial opportunity, an international training programme, is based on their group identity. In a natural field experiment with a Colombian university, I target 4, 831 students and disclose to some that selection is based on demographics. Public identity disclosure reduced take-up by 27% and completion by 20% relative to a condition in which eligibility criteria were concealed. Three complementary online experiments with 1, 113 students suggest that identity-based selection raises image concerns, which discourage programme participation. The results highlight that the way institutions communicate beneficial opportunities can itself become a barrier to take-up.