Recent theorizing on corruption is split between two approaches: corruption is described as a collective action or principal-agent problem. Insights from political science and geography suggest that these theories are not as bifurcated as some of the literature indicates, as their explanatory power is shaped by place-specific factors. This article draws on observations of administrative and community responses to decentralization policy in Papua New Guinea. Despite efforts of policy makers to institutionalize principal-agent theory inspired systems of government, we argue that state-society relations have meant that the potential for corruption is a part of a collective action problem in some places and a principal-agent problem in others. The applicability of these theories is determined by the degree of alignment between cultural and social values and administrative norms, which have been shaped by historic, political and economic factors. We call for a more nuanced understanding – one that better accounts for spatial difference – of the applicability of these theories beyond the scale of the nation-state.