Stationary bandit theory

Where does government come from, if not from a social contract? Mancur Olson proposed a stark answer in "Dictatorship, Democracy, and Development" (1993). Imagine bands of roving bandits who raid, take what they can and move on. They have no reason to leave anything behind, so production collapses. Now suppose one bandit is strong enough to stay in one place and keep others out. This stationary bandit steals regularly, as taxation, rather than taking everything at once. Because his future income depends on his subjects' output, he has an incentive to leave them enough to invest, to protect them from other bandits and to provide some public goods.

The stationary bandit is thus the first government: a self-interested monopolist of violence whose interest happens to coincide partly with that of his subjects. Olson called this an encompassing interest: the ruler bears a share of the losses his taxes cause and gains a share of the output his public goods create.

The rational autocrat

Olson modelled the autocrat as choosing a tax rate to maximise his own revenue. Too high a rate discourages production and reduces revenue; a revenue-maximising rate below 100 per cent leaves subjects with an incentive to work. The autocrat also spends on public goods such as order and infrastructure as long as each unit increases his tax take by more than it costs.

Time horizons

The theory depends on the ruler's horizon. An autocrat who expects to rule for a long time and to pass power to an heir behaves more like a stationary bandit. One who fears imminent overthrow behaves like a roving bandit, confiscating and plundering. Olson argued that the absence of a reliable succession mechanism makes autocracy inherently insecure and short-sighted.

Democracy as a more encompassing interest

With Martin McGuire, Olson extended the model to majority rule. A ruling majority that also earns market income bears more of the costs of distortionary taxes than an autocrat does, so it chooses lower redistributive taxes and provides more public goods. And because democracy needs secure rights to free speech and to contest elections, the same rights protect property and contracts. Olson concluded in Power and Prosperity (2000) that lasting prosperity requires governments strong enough to protect rights and constrained enough not to violate them.

The quotation

Olson illustrated the theory with the Chinese warlord Feng Yuxiang, who in the 1920s suppressed roving bandits in the area he controlled and was reportedly preferred by the population to the roving bandits he displaced, despite his own heavy taxation.

• Eastern Congo: Raúl Sánchez de la Sierra studied armed groups in the Democratic Republic of Congo and found that a rise in demand for coltan, a bulky mineral that is hard to hide, led armed groups to establish themselves at mines and tax output there, while a rise in the price of gold, which is easy to conceal, led them to establish taxation and administrative functions in villages where gold income was spent. Having a stationary bandit was associated with higher welfare.
• Warlord governance: research on rebel governance in Colombia, Syria, Afghanistan and elsewhere documents armed groups that tax, adjudicate disputes and provide services where they expect to stay.
• Mafias and protection: organised-crime groups that sell protection in stable territories behave more like stationary than roving bandits.
• Early states: archaeologists and historians debate whether early states arose from conquest and tribute in ways similar to Olson's account.

• Benevolence by assumption: the model predicts moderation only if the ruler can commit to his tax rate and expects to stay; real autocrats often face coups and so behave as roving bandits.
• Other goals: rulers may value glory, ideology, religion or personal consumption over revenue, and may wage costly wars.
• Repression costs: Ronald Wintrobe argued that dictators trade off repression and loyalty, so their behaviour depends on how they buy support, not only on tax revenue.
• Coalition politics: selectorate theory and work on authoritarian institutions stress that autocrats depend on coalitions whose demands shape policy, which Olson's single-ruler model ignores.
• History: the model is a stylised thought experiment; it does not explain the variety of early states, including those that arose from religious authority or trade.
• Normative risk: the theory can be misread as praising strong autocrats; Olson's point was that democracy and secure rights do better still.

The stationary bandit is one of the most cited models in political economy because it captures, in a few assumptions, why the monopolisation of violence can improve on anarchy and why time horizons and succession matter for autocratic behaviour. Field evidence from eastern Congo supports its central mechanism. Its limits are its single-ruler simplicity and its reliance on credible commitment. For designers, the theory implies that predictable, long-horizon authority with constraints is better than both anarchy and insecure despotism, and that rights-protecting democracy aligns rulers' interests most broadly with those of the governed.

• Mancur Olson, "Dictatorship, Democracy, and Development" (1993)
• Raúl Sánchez de la Sierra, "On the Origins of the State: Stationary Bandits and Taxation in Eastern Congo" (2020)
• Martin C. McGuire and Mancur Olson, "The Economics of Autocracy and Majority Rule", Journal of Economic Literature 34(1), 1996
• Mancur Olson, Power and Prosperity (Basic Books, 2000)
• Ronald Wintrobe, The Political Economy of Dictatorship (Cambridge University Press, 1998)