Affirmative action (AA) is highly controversial. We ask whether opposition to AA reflects a genuine commitment to merit, or  reflects deeper identity-based considerations. Specifically, does the resistance to AA vary with the social identity of those who benefit from the policy? Using an incentivized online experiment with university students in India, we compare perceptions and rewards toward participants selected through caste-based versus income-based affirmative action. Evaluators assess test-takers’ competence and allocate monetary rewards under different selection rules. We find that beneficiaries from India’s historically stigmatized caste groups (Scheduled Castes, or SC) are perceived as less competent than non-marginalized beneficiaries, even under income-based affirmative action. This suggests that beliefs about competence of affirmative action beneficiaries  vary systematically with their social identity, rather than reflecting a generalized aversion to preferential treatment alone. Income-based beneficiaries from non-marginalized groups are seen as more competent and rewarded accordingly, whereas low-income SC income-based beneficiaries gain neither. These findings show that shifting from caste to income criteria does not make identity irrelevant, underscoring the persistent salience of social identity in redistributive policies.

Governments frequently pursue multiple objectives through a single policy instrument. Language policy provides a particularly revealing example because decisions regarding the medium of instruction often simultaneously serve educational, economic, political, and symbolic goals. Drawing on the concept of instrumental mismatch, this paper examines Rwanda's transition from French to English as the principal language of instruction. Rather than asking whether the reform ``worked," we examine how instrumental mismatch manifested itself during implementation. We examine four observable manifestations. First, the reform was introduced despite substantial implementation capacity constraints, including limited teacher English proficiency and shortages of instructional materials. Second, repeated policy reversals reflected tensions between competing educational and developmental objectives. Third, descriptive evidence reveals elevated grade repetition and weaker basic learning outcomes among cohorts initially exposed to English-medium instruction. Finally, we document a redistribution of linguistic capital, as English evolved from a language largely nested within the francophone population to the dominant form of foreign-language capital, shifting the burden of linguistic adaptation towards francophone rather than anglophone Rwandans. Taken together, these findings illustrate how attempts to pursue multiple objectives through a single policy instrument can generate institutional tensions during implementation. More broadly, the paper argues that instrumental mismatch provides a useful framework for understanding reforms that seek to achieve multiple objectives operating over different time horizons, and can result in incongruous institutions.

We examine how education relates to women’s employment and occupational attainment across India’s major caste and religious groups using the 2019–21 National Family Health Survey. We find substantial differences in how education maps into economic outcomes across groups. Higher employment does not necessarily correspond to greater educational or occupational attainment, and similar levels of education are associated with markedly different probabilities of employment and professional work. These patterns suggest that women’s economic trajectories in India cannot be understood from educational attainment or employment alone. Understanding these differences requires attention to both whether women work and the kinds of employment they attain.

This paper studies whether identity reformulation can mitigate inherited social stigma by examining religious conversion among the Scheduled Castes (SCs) in India. Exploiting within-jāti variation in religious affiliation across three nationally representative datasets spanning nearly three decades, we document systematic differences in education, occupation, assets, and exposure to stigma across religious groups within the SCs. SC-Christians and SC-Buddhists/Sikhs exhibit substantially better outcomes than SC-Hindus, while SC-Muslims fare worst. To address concerns about selection into religious affiliation, we exploit a historical episode of mass conversion among the Mahar caste and instrument Buddhist conversion using distance from Nagpur, the site of Dr. B.R. Ambedkar’s 1956 conversion. Instrumental variables estimates indicate large gains in schooling among Buddhist Mahars. The findings suggest that religious identity is an important dimension shaping caste disparities and that identity reformulation, under certain institutional conditions, can weaken the persistence of caste-based disadvantage.

Can empathy by analogy reduce discrimination in societies with entrenched inequalities? We study this in the context of caste-based discrimination by teachers in India. In an RCT with 1,708 teachers, we use structured analogy to foster empathy: teachers observe racial discrimination against a high-caste individual in the West, then analogous discrimination against a low-caste individual in India. In a correspondence study on grading, control group teachers assign lower grades to low-caste students for low-quality answers, but slightly higher grades for high-quality answers. The intervention eliminates negative discrimination while leaving positive discrimination unchanged, especially among high-caste teachers and those with higher baseline empathy.

A distinct feature of the Protestant Reformation was Martin Luther's intentional use of German, rather than Latin, in his writings. This paper examines the reformers’ reliance on the vernacular and its impact on the market for ideas. We document a significant increase in the printing of vernacular works after the start of the Reformation. The increased use of the vernacular expanded the ranks of authors, increased the supply of works from authors with a non-elite background, and broadened thematic coverage. Consequently, a much broader share of the populace could participate in the consumption, creation, and dissemination of printed information.

This paper documents a language change in printing from Latin to the vernaculars, the spoken tongues, in the immediate aftermath of the Protestant Reformation of 1517. As a result, the share of vernacular titles in Europe rose from around 30% in 1500 to almost 60% in 1600. With the increased use of the vernaculars in printing, the availability of knowledge and ideas increased at the city level and became more diverse in terms of authors and themes. Finally, we study long-run consequences. Using linguistic differences across cities as a source of exogenous variation in the number of vernacular titles printed in cities, we document a positive effect of vernacular printing output on upper-tail human capital and city growth. This suggests that the turn to the vernaculars in printing was an important driver of European dynamism in the early modern period.