The share of workers in wage employment is low in developing countries. Recent research argues high exit rates out of wage employment, rather than low entry rates, drive low levels of wage work overall. To shed light on the causes of elevated exit rates in low-income labor markets, I conduct a panel survey of job-seekers and a complementary survey of firms in Accra, Ghana. I document three new facts. First, exits are dominated by quits in Ghana while layoffs play a negligible role, in strong contrast to the USA where layoffs dominate and quits are infrequent. Second, workers in Ghana see income gains after quits, also in strong contrast with the USA, where workers see substantial income losses following quits. Third, I show quits are most common among job-seekers experiencing a temporary loss in non-wage income at baseline, but by endline have re-gained access to a source of income flows. I find no evidence information frictions drive exit. Instead, I argue volatility faced by workers outside the wage sector drives high exit rates and build a general equilibrium model of job search to quantify the strength of this mechanism. When calibrated to match the experience of jobseekers in both the USA and Ghana, my model attributes 25% of the difference in exit rates to differences in the volatility of non-wage income.

How does the spatial distribution of income in cities vary with income per capita? We address this question using new granular evidence from 127 cities in 26 developed and less-developed countries. We document that household income levels decline significantly more rapidly with distance to city centers in less-developed countries. Urban neighborhoods with natural amenities—particularly hills and rivers—are poorer than average in less-developed nations but richer than average in developed ones. We evaluate potential explanations for these patterns using a quantitative spatial model disciplined by commuting gravity equations that we estimate for each city. We find that the spatial income patterns are largely explained by nonhomothetic preferences over amenities, coupled with higher commuting costs and spatially more concentrated jobs in less developed cities. Our model predicts unequal welfare gains across households as citywide income increases.

Standard in-person business training programs are costly and difficult to scale to the millions of microenterprises in the developing world. We conduct an experiment to test the feasibility, cost-savings, and impact of delivering live training sessions over Zoom to microentrepreneurs in Mexico and Guatemala. We demonstrate that it is now feasible to both recruit and train self-employed women online, covering a wide geographic area, with few technology issues. However, the cost-savings over in-person classes are less than expected. Training improved business practices and performance over 2 months, but impacts had dissipated within 6 months.