Protection specialists presumably have a comparative advantage in providing protection services. Protectors are needed for deterrence of violence. In turn, specialists in protection require skill in the use of violence to perform their task. Some forms of protection are subject to scale economies. The scale economies to enforcement by force have been discussed extensively in Chapter 3. Protection subject to scale economies is provided efficiently when the same protector or set of cooperating protectors protects many individuals. Protection, therefore, can entail a large-scale assembly of power. The assembled power and its head can be controlled by the protector's clients or by the protector himself. A protector who is able to gain such control may use it to become a dictator.

Creating a collective-action mechanism is a necessary condition for preventing protectors from taking over. The costs of activating the mechanism, however, seem substantial. To reduce the probability of takeover, individuals are expected to undertake measures to restrain the protectors, thereby reducing the burden of activating the collective-action mechanism.

Individuals, besides directly activating their collective-action mechanism, can prevent themselves from becoming easy prey to their protector by the way they form their agreement with the specialized protector and by the restrictions they impose on him. The agreement that individuals make with their protector is expected to be incentive-compatible. Reducing the protector's inclination to take over, however, requires incentives very different from those that promote efficiency in more conventional circumstances.

In the discussion that follows, I argue that as long as the protector does not possess dictatorial power, the clients, rather than the protector, hold the purse strings to the protection budget. Clients determine what taxes to impose and at what rates and how to spend the tax revenue.