What do workers at the margin of informality value at work? We answer this question using novel survey data on the employment preferences of residents of Brazil's largest favela complex. Recent work shows that workers value job amenities beyond wages, including meaning and purpose (Ashraf et al. 2025) and scheduling flexibility (Mas and Pallais 2017), but this evidence is predominantly from developed countries. Models of informality emphasize compensating differentials between formal and informal employment (Meghir, Narita, and Robin 2015; Haanwinckel and Soares 2021), yet direct evidence on the preferences of workers at the formal-informal boundary remains scarce. We combine open-ended questions with discrete choice experiments to estimate willingness to pay for amenities specific to formal employment (unemployment insurance, parental leave, and termination notice) and for highly ranked attributes not specific to it (learning opportunities and commute time). We report four findings. First, open-ended responses suggest that workers value formal employment benefits, flexibility, and learning opportunities. Second, workers' primary complaints about current or past jobs center on poor management, difficult customers, and inflexible schedules. Third, discrete choice experiments reveal high willingness to pay in forgone wages for all formal sector amenities on average: 34 percent for unemployment insurance, 24 percent for parental leave, and 18 percent for termination notice. Learning opportunities are valued at 28 percent and shorter commutes at 12 percent. Fourth, valuations vary systematically by employment sector in a way that is consistent with sorting. Formal sector workers value formal sector amenities most, while the self-employed value them least (parental leave) or not at all (termination notice). The informally employed exhibit mixed valuations-unemployment insurance (UI) as highly as formal workers, but not termination notice, similar to the self-employed - while the unemployed value UI and learning opportunities most. These patterns are also consistent with learning on-the-job (about the value of amenities) and endowment effects. We find suggestive evidence for both. These results indicate that pay and other tangible benefits, rather than meaning or purpose, drive labor supply decisions for workers on the cusp of informality, and that workers sort, especially into self-employment, based on their preferences for sector-specific amenities. The role of learning and endowment effects in shaping the supply side of informality merits further investigation.
