This paper proposes a new perspective for understanding the existence of labor informality, which is commonly observed in developing economies, through a political economy framework. We develop an overlapping-generations model with heterogeneous agents, in which policies are determined as an outcome based on a probabilistic voting scheme à la Lindbeck and Weibull (1987). The government decides on the degree of enforcement of taxation and social security contributions (with full enforcement eliminating informal employment). We show that a political economy equilibrium with imperfect enforcement exists because allowing labor informality (especially for low-productivity workers) serves as a means of redistribution. We also calibrate the model to Chile, which has continuously had a significant share of informal employment (around 30%) despite tripling its GDP per capita (from about $5,000 in 1995 to over $15,000 in 2018). Given the inequality and institutional features in Chile, the model is able to generate labor informality rates quantitatively consistent with the data. In addition to the theoretical analysis, we provide empirical evidence to support the model's predictions. We use Chile's largest cross-sectional survey data, Socio-Economic Characterization Survey (CASEN), to show that 1) even among public-sector employees, where the monitoring cost is close to zero, the enforcement of social security contribution is far below 100%; 2) using probit and logit regression models, the probability of enforcement increases with permanent income levels, which is consistent with our model.