The resource curse literature often posits that public employment in resource-rich economies tends to be large, driven by motivations beyond public goods provision. One set of theories attributes this phenomenon to (ethnic) clientelism, but lacks robust empirical support. Another set of theories challenges the notion of (ethnic) clientelism, suggesting that the expansion of public sector employment serves as a form of social insurance in the absence administrative tools to redistribute income. This study diverges from the conventional story of an invariably positive association between resource abundance and public sector employment. Instead, I show that the positive correlation between resource abundance and public sector employment only holds when ethnic fractionalization is low. In countries with high levels of ethnic fractionalization, the relationship weakens and can even reverse. This is because an increase in public sector employment tends to dilute public sector wages, diminishing the attractiveness of public sector jobs as clientelist rewards. The empirical analysis supports this claim: when ethnic fractionalization is low, more resource rents correspond to higher public sector employment and lower (or even negative) public sector wage premiums; when ethnic fractionalization is high, the opposite holds: more resource rents correspond to fewer public sector jobs but larger public sector wage premiums.