A Haitian NGO gives a goat and training to households that own no livestock and expects a herd to follow. Thompson and Magnan surveyed 169 such recipients in the hills inland from Leogane, all of whom received an animal between mid-2010 and early 2012, and measured what actually happened: a quarter of the transferred does never produced a live kid, about thirty-seven percent of kids born died before weaning, average herds stood below two animals, and only twenty-seven households had ever made a sale. The finding sits in the variation. Regressing outcomes on household characteristics, they show that wealthier recipients build smaller and less valuable herds, which fits a model in which households with better alternatives spend their effort elsewhere. Access to land cuts kid mortality, women lose roughly ten percent fewer kids than men, and village effects are large, with remote zones outperforming peri-urban ones. Screening applicants on wealth, land and location would sharpen targeting.