We study bureaucratic corruption in a model in which a constituency sets required levels for a given set of activities. Each activity is carried out by an external provider, and its realization is supervised by a bureaucrat. While bureaucrats are supposed to act on behalf of the constituency, they can decide to be corrupt and allow providers to deliver lower activity levels than contracted in exchange for a bribe. Given this, the constituency sets the optimal activity levels weighing off the value of activity levels, their costs, as well as the possibility for the bureaucrats to be corrupt. We use this setup to study the impact on equilibrium corruption of the degree of decentralization of corruption. To do this we compute equilibrium corruption in two different settings: 1) Each bureaucrat acts in such a way as to maximize his own individual utility (competitive corruption); 2) An illegal syndicate oversee the corruption decisions of the population of bureaucrats in such a way as to maximize total proceeds from corruption (organized corruption). We show that, since average corruption payoff is increasing in the activity levels set by the constituency, and since the latter responds to high levels of corruption by reducing required activity levels, in equilibrium the illegal syndicate acts in such a way as to restrain the total number of corrupt transactions, so that corruption is lower when it is organized than when it is competitive.