Can landed elites prevent mass education? If so, under what conditions? This article examines the long-term human capital impacts of the Spanish hacienda, the predominant system of agrarian estates with coerced labor granted to local elites during the colonial era in Latin America. Using original district-level data on the placement of haciendas in Peru since 1876 and over a century of population census records, we exploit variation in hacienda crop-suitable areas, age cohorts, and predetermined elite family connections to show that hacienda prevalence persistently reduced literacy rates throughout the 20th century. Consistent with a supply-side mechanism, exposed districts received lower investments in schools and teachers, without comparable reductions in other public goods, and exhibited negligible impacts on demand-side proxies such as local economic activity, agricultural specialization, or broader development outcomes. Moving beyond explanations centered on early market integration or public-goods provision, we show that a larger share of workers in haciendas explains long-term human capital losses even among regions with comparable structural conditions. This effect was strongest where labor control was concentrated in fewer haciendas, allowing landed elites to block schooling investments. Because voting required Spanish literacy, limited schooling in exposed areas translated into persistently lower Indigenous political participation. Although industrialization, urbanization, and top-down land redistribution later emerged, they failed to reverse the human capital deficits entrenched by the hacienda system.

Can a brief period of economic prosperity leave a legacy of long-term adversity? This study examines the lasting impact of the Amazon rubber boom around 1900 on contemporary income, inequality, Indigenous presence, and forest conservation. Identification combines variation in historical rubber distribution with an instrumental variable strategy using FAO-based rubber suitability and a Regression Discontinuity design around concession boundaries. Municipalities with greater rubber presence experienced short-term gains in 1920 but long-run reversals by 2010, showing lower income, population density, and higher inequality and Indigenous extinction. Grid-level analyses across the Amazon further show that historical rubber suitability is associated with weaker economic activity and population today, alongside greater deforestation. The findings, consistent across Brazil, Colombia, Peru, and Bolivia, indicate that the rubber boom's short-lived wealth reinforced extractive institutions and violence against Indigenous peoples, leaving long-lasting economic, social, and environmental scars across the Amazon.