Does the type of regime — democracy or autocracy — determine economic performance? The question is old, politically charged, and easy to answer badly. This module gives you the tools to answer it well: a model that explains why regimes provide different mixes of public and private goods, and the careful empirical evidence that cuts through the anecdotes.
Why the raw correlation misleads
Point to fast-growing autocracies (China, Singapore, Botswana for decades) and you 'prove' autocracy works. Point to stagnant ones (Zaire, North Korea) and you 'prove' the opposite. The variance among autocracies is enormous — far larger than among democracies — which is the first clue that 'regime type' is too blunt a variable. Jones and Olken (2005) showed that for autocracies, the individual leader matters enormously for growth (the death of an autocrat shifts the growth path); for democracies, leaders matter far less because institutions constrain them. The lesson: look inside the regime at the structure of power, not the label.
Selectorate theory
Bueno de Mesquita and co-authors (The Logic of Political Survival, 2003) offer the most useful framework. Every leader, democratic or not, survives by satisfying a winning coalition (W) — the group whose support is essential to staying in power — drawn from a larger selectorate (S) — those with a say in selecting leaders. The ratio W/S drives everything.
The logic of W and S
Large W (broad coalition, as in a democracy where you need millions of votes): the cheapest way to retain support is public goods — roads, schools, security, growth — that benefit the whole coalition, because private bribes to millions are unaffordable. Small W (narrow coalition, as in a personalist autocracy where you need a few generals and cronies): the cheapest way to retain support is private goods — cash, jobs, licences, rents — targeted at the few, because public goods would 'waste' resources on people whose support you don't need. The size of the coalition, not the regime label, predicts whether a leader builds schools or buys loyalty.
The stationary bandit
Mancur Olson (1993) explained why some autocrats nonetheless invest in their economies. A 'roving bandit' who will move on tomorrow steals everything today. A 'stationary bandit' who expects to rule the same territory for decades has an encompassing interest in its prosperity — because he taxes it. A secure autocrat with a long horizon will therefore provide enough order, infrastructure, and property security to grow the pie he is taxing. This explains the secure, long-horizon developmental autocracies. It also explains their fragility: the moment the autocrat's horizon shortens (age, instability, a contested succession), the encompassing interest evaporates and predation returns.
The evidence: democracy does cause growth
Acemoglu, Naidu, Restrepo, and Robinson (2019), using a large panel and careful handling of the fact that democratisations cluster in regional waves, estimated that a country democratising raises its long-run GDP per capita by around 20–25%. The mechanism is not the act of voting but what democracy tends to bring: investment in public goods, schooling, and health, broader taxation, and constraints on expropriation. The effect is real but slow — it accrues over decades through capacity and human capital, not in the first electoral cycle, which is why citizens often feel democracy 'hasn't delivered' in the short run.
Why African democratisation looked disappointing
The 1990s democratisation wave across Africa produced mixed growth results, which fed a narrative that democracy doesn't work in Africa. Selectorate theory and the evidence above explain the disappointment without that conclusion: many transitions enlarged the selectorate (more voters) without genuinely enlarging the winning coalition (power stayed with a narrow elite who now bought votes through clientelism rather than serving a broad coalition). Elections without the underlying shift to public-goods provision deliver the form of democracy, not its growth dividend. The constraint is the structure of the coalition, not the ballot.
Exercise
Compare two resource-rich states. State X (think Botswana) used diamond revenue to build schools, roads, and reserves, growing for three decades. State Y (think Mobutu's Zaire) used mineral revenue to enrich the ruler and a small circle while infrastructure collapsed. Both were, for long periods, non-democracies. (1) Using selectorate theory, explain the divergence in terms of W and S rather than the democracy label. (2) Using Olson's stationary-bandit idea, explain how time horizons reinforce the difference. (3) The 2019 evidence says democracy causes growth, yet State X grew under a dominant-party non-democracy. Reconcile this — is the evidence wrong? (4) What does the comparison imply for a reformer who can influence the structure of power but cannot instantly create a full democracy?