How to think about democracy and autocracy in 2023

An unfinished horse race

This is the kind of post that should be an academic paper, but it has a lot of loose threads and uncertainties, and you can’t admit those in a paper. The point is, it isn’t a journalistic comment on how autocracies are doing versus democracies right now, instead, it’s me trying to think through the right “model” to have of democracy versus autocracy. How are the wheels connected, which handle turns which crank? Of course, your choice of model will have implications for how you interpret current events.

When I think about this, I start from Mançur Olson’s paper Dictatorship, Democracy and Development.

Olson’s idea

Olson was one of social science’s great hedgehogs. Everything he wrote was a development of his basic insight that political action is a public good. This doesn’t mean that politics produces public goods like bridges and clean air and national defence: it’s broader than that. For example, take a tax break for a particular group — doctors, say — which is a pure piece of redistribution, with no other consequences. That’s not a public good in the conventional sense. But if, as a doctor, you lobby for that tax break, then you will be doing all your fellow doctors a favour. Your lobbying is a public good among the set of people that benefit. From this single idea many things follow: lobbying will be undersupplied relative to the benefit it produces; small groups will undersupply lobbying less (if a specific monopoly lobbies for a tax break, it is doing nobody a favour but itself); and we end up with a whole vision of politics as a game with a very uneven playing field.

When Olson came to think about democracy and dictatorship, he applied the same logic, this time not to political lobbying, but to a more basic question: why does the state exist at all? The existence of a state, he said, benefits every citizen, because the state provides order and stops people robbing each other. But equally, we shouldn’t expect all citizens to bear the huge cost of setting up a state that does those things. Keeping strictly to the assumption of self-interest, why would any one citizen do his fellows a favour like that? The state itself is a public good.

So instead of a contractarian theory, where states are set up by mutual agreement among citizens, Olson introduces the famous stationary bandit. Bandits live by taking other people’s product. A roving bandit is like a stereotypical Viking, plundering wherever he goes and never returning to the ruins he leaves in his wake. (Maybe this is unfair to the real historical Vikings: don’t @ me.) But a stationary bandit stays in one place, so he wants to be able to steal from his victims more than once. In fact, he’d even like them to grow rich, not because he loves them, but so he can steal more from them. Thus he becomes a shepherd of a human flock, which he keeps fat and regularly shears. He stops other bandits stealing from them — i.e. he provides “peaceful order”. And he might even build bridges and roads and provide other public goods, all to increase his tax take. Boom, now you’ve got the state.

Kubla Khan settles down and begins the Yuan dynasty in China

This is Olson’ theory of dictatorship and also of the state in general, and unlike Brad Delong, he realises that most of human progress took place under these dictatorships, so they must have been at least a bit beneficial. But he also wants to compare this old-established state form with the new state form of democracy. (Bear in mind this is written in 1993, at a high point of liberal democratic triumphalism after the collapse of the USSR.)

The stationary bandit provides important goods, but he isn’t completely efficient. First, he taxes as much as he can. That distorts the economy and makes everyone else poorer at his expense. Second, he only provides public goods to the extent that he can tax the resulting gains:

If an autocrat received one-third of any increase in the income of his domain in increased tax collections, he would then get one-third of the benefits of the public goods he provided. He would then have an incentive to provide public goods up to the point where the national income rose by the reciprocal of one-third, or three, from his last unit of public good expenditure.

An efficient government would provide more public goods, up until the last unit of expenditure has a benefit of just one, not three. So dictators don’t do enough. Can democracies do better?

Olson thinks of democracy as a system where the majority rules. To discipline his analysis, he makes the majority just as selfish and greedy as the dictator. They own half the economy. They’re happy to extract money from the economy, but that is limited by their ownership of it. And they provide more public goods, because they benefit not just by taxing the resulting product, but as private owners of half of the untaxed share. Though still not perfectly efficient, they do better than dictators.

This beautiful, deep and simple theory explains:

• how the state arises in the first place;

• why the world’s richest countries are democracies (or oil producers, which don’t count);

• what makes some dictators better than others: a stable dictator who is sure of staying in power will invest for the long run, while an unstable dictator will grab what he can like a roving bandit.

But it also raises some questions.

First, can’t the majority focus their resource extraction on the minority, rather than taxing the whole economy at the same rate? Potentially, they could even expropriate all their assets. (A dictator can’t do this, because he can’t farm a million acres without bodies to do the labour. But a million and one farmers might be happy to cultivate two acres each instead of one.)

Second, is there really a stable, defined majority? If not, won’t it matter that there are shifting majorities and won’t this make democracies less foresightful? Olson acknowledges this point.

Third, is it really true that democracies reliably implement the interests of the majority? Here the question of rational ignorance bites.

Rational ignorance

Rational ignorance comes from the fact that no individual’s vote is decisive in large elections. Almost surely, when next you vote, your vote will not change who gets elected. It doesn’t matter. This is true even if the outcome matters to you massively — say you’re a millionaire and one candidate will raise your tax by a lot. Because nothing you can do will change the outcome, you actually have no incentive to vote. You also have no incentive to learn about the candidates, or to think about politics.

“But, but,” you say, “people do vote and they do learn about politics.” Sure, but they don’t have an incentive to. Something else must be driving them, therefore: fashion, or the desire to be a good person, or a bee in their bonnet. Unfortunately this something else probably won’t guarantee they have the information to vote correctly for their long-run interests. (If someone is very well-informed about global warming, does that mean he has a very accurate estimate of the true social cost of carbon? Experience teaches that instead he is usually a zealot, who has ingested one-sided information so as to bore people at parties.) Indeed, the fact that people don’t know much about politics is one of the best-supported generalizations in political science.

This matters because choosing the best policy is hard and requires research effort. If the right policies were obvious, voter ignorance wouldn’t matter. Most politicians talk as if this is true, the right policy is obvious, and the other side aren’t doing it because they are eeevil. No important policy choices are like this.

Rational ignorance should be central to the analysis of democracy; political economy without rational ignorance is Hamlet without the prince. Despite that, most political economy models of elections assume rational ignorance away and make voters vote for their true interests. As a result, much of the field is an exercise in building castles in the air.

Rational ignorance means that elections don’t reliably advance the interests of the majority. Instead, parties can win power based on cheap rhetoric, fads, or voter emotions. This changes the trade-off between more and less democratization. Olson thinks more democracy is always better, because when the rulers comprise a larger share of society, they will have a more encompassing interest in society. (An interesting implication of this argument, which Olson did not mention, is that it is better for the richest 50% to be in power than the poorest 50%.) Now we can see that this gain has to be balanced against a loss from democratization: as more people are involved in decision-making, policies are less likely to accurately reflect anyone’s interest, and more likely to be short-sighted, foolish or random.

Normally, when there is a trade-off, you expect the best solution to be somewhere in the middle. This raises the interesting question of why we don’t see more oligarchies, i.e. political systems where a large proportion of the citizens make decisions, but not all of them. Most of the world’s countries are either democracies where everyone can vote, or autocracies where nobody can vote or votes are meaningless.

I can think of two possible reasons. The first is to do with the nature of rational ignorance. Rational ignorance kicks in as soon as electorates are big enough that no individual vote is likely to matter. That is probably already true in electorates of a few hundred — here’s a quick simulation where people are equally likely to vote for either party, showing what proportion of elections result in a tie:

So, there’s not much difference between an electorate of a thousand and a million, in terms of motivating people to understand politics and vote for sensible policies. If so, then the trade off between democracy and dictatorship might look like this:

Democracy versus dictatorship. Here I’ve drawn welfare as higher under dictatorship than democracy, but it could equally be higher under democracy, or it might vary depending on other factors. The point is the U-shaped relationship: welfare is worse in the middle than at the extremes.

You always want to choose either full dictatorship, with one decision-maker who has strong incentives to maximize his own welfare, or full democracy. Adding a little democracy to a dictatorship doesn’t help: even a few more decision-makers swiftly start making dumber decisions, and at the same time, they are still a small elite who don’t care about the welfare of the people.

The second reason is just historical chance. There have been limited democracies and oligarchies before, like the system in Victorian Britain which had a property qualification to vote. We live in a democratic age, the US has pushed democracy on its allies, and the main alternative is that a dictator seizes power or a communist party imposes a non-democracy in the name of the people. You can take this argument in lots of ways, the point is that the political systems we see don’t exist because they maximize citizen welfare, but due to some other process of political change. That brings me to the fourth question about Olson’s idea.

What is the state?

What drives Olson’s model of state formation is that the existence of the state is a public good. It’s costly to set up a system of “peaceful order”, and the system benefits everyone who lives under, it even if they don’t help to set it up. At the same time, Olson doesn’t really specify how the state works. It is just treated as a kind of machine which a bandit possesses, which lets him tax people and pay for public goods. Again, most political economy models do the same, and treat the ruler’s ability to get his way as a black box.

That’s fine as shorthand, but in fact, the state is made up of people who obey the ruler. Let’s digress a bit into the definition of influence, leadership, authority and the state. Doing this might help us towards a theory of political change, which in turn might help us understand how democracy emerges.

Suppose a group of people have a choice to make. They can’t each make it independently: it’s the sort of choice that they must all make together. For example, as winter approaches, should a nomadic band of herders stay in the hills or move to the lowlands? They all want to make the best decision, but nobody wants to go it alone. Or: should one city-state make war on another? If its citizens make war, they had better all do it together.

This is a coordination game and a common way to solve it is that one person decides and the others follow them. Even among egalitarian hunter-gatherers, some people have more influence, meaning that their decisions often get followed, perhaps because they have a reputation for choosing right. If it is common knowledge that one person’s decision will always be followed (perhaps in a specific arena), then that person is a leader. Leadership doesn’t require any formal authority structure: gangs of boys have leaders who emerge naturally. Leadership also doesn’t require any coercive apparatus, state or law. The structure of the coordination game provides coercion enough: if everyone else will follow the leader, then you have to follow them too.

But in some situations, actors do have coercive power over each other. People can threaten each other with violence. In the state of nature everyone is roughly equal; in some parts of history coercive power is unevenly distributed, like when armoured knights can boss poorly-armed peasants. Coercive power can be used to extract resources from people, by threatening them. In turn, that can be used to fine bad guys, or just to steal from the vulnerable.

Other things being equal, coercive power is best exercised by many against one. That makes coercion and resource extraction a coordination game. Often, the majority in society, and sometimes a smaller group, can bend an individual to their will, so long as they all pick the same target.

A ruler is a leader in the coordination game of coercion.

He’s the person who gets obeyed when he yells “Guards!” and points at someone. The guards could change their mind and obey someone else. But this is very risky, because if just one guard does that, the ruler will yell “Guards!” while pointing at him. This makes rulership very stable. It is what accounts for the mysterious fact that over great swathes of history, the tail wags the dog, and a single man can exert power over a great many. Nobody is a ruler by virtue of their strength in single combat, or because they drive a tank. They are rulers because they tell the tanks where to drive.

A ruler can be a single individual. But individuals are mortal, and many groups have procedures by which rulers is chosen. One procedure is to pick the male heir of the old ruler. Another is for the nobles or big men to choose a new ruler by acclamation. Another is for the ruler to be the child who picks out certain objects that were owned by the old ruler, when the Lamas present them to him. Another is to have a vote every five years. There are also cases where ruling requires a procedure. You may only be able to shout “guards” if someone has broken a written law. (If Donald Trump had shouted “guards” after the 2020 election, he would not have been obeyed by his commanders.) All of these are possible equilibria. There are many forms of rule. Some of them may be less stable. If the ruler is voted on every five years, he had better not be an absolute ruler, otherwise he may cancel the next election and shout “guards” at anyone who disagrees.

When the procedure for picking a ruler is formalized enough, the ruler rules by virtue of occupying a certain formal position. Then we have what anthropologists call the state, as opposed to merely chiefship. (But political theorists, including Mançur Olson, count any ruler as a state, by contrast with anarchy.)

State stability and change

There’s nothing very deep or new about this analytic framework, but it already tells us some useful things.

1. Many different forms of state power are possible, and most of them are stable in the game theory sense that everyone is making optimal choices given what everyone else is doing. (If everyone else accepts a certain ruler or regime, then it is risky to be the one Tiananmen Tank Guy who doesn’t.) That fits the great variety of forms of state we see across time and space, from early chiefships to theocracies to monarchies to modern democracies and dictatorships.

2. A lot of modern theorists say we shouldn’t “reify” the state, i.e. treat it as a “thing” with an objective existence. This has a natural interpretation here. Someone is only ruler because others accept him as ruling. A particular state only exists because people accept a particular way of picking out the ruler. Political authority is like paper money: a consensual hallucination that exists insofar as people think it does. (That doesn’t mean the state is imaginary, any more than the value of money.)

3. Maybe you didn’t notice, but there is nothing here about the provision of law and order! Olson makes this a central good that the state provides, and the traditional view is that a “minimal” state enforces contracts and prevents crime. But in fact, many or most states throughout history have not provided law and order, and relatedly have not sought the “monopoly of violence”, in Max Weber’s famous phrase. Instead, they have been empires, which, as Charles Tilly puts it “sought tribute much more than they sought the stable control of the population”. Under feudalism, justice was administered by local lords, and only gradually did kings seek to develop a law-and-order monopoly. Any ruler will have to maintain his status against rival rulers at the same level in order to survive; that doesn’t mean he needs to put down every local thug. More usually, they will take the thug as given, so long as he swears allegiance and provides produce or levies.
   But now we might start to question the idea that the state itself is a public good. Things that the state does, including law and order, are indeed public goods, and Olson has a good explanation for why only a stationary bandit can provide them. But the state itself exists just as soon as people are prepared to obey someone’s commands. And being a ruler’s henchman, one of the “guards”, is not necessarily costly! If you get a share of the spoils, then it can pay for itself.
   So now we can come back and think about contractarianism. Contractarians believe that states are founded when everyone agrees to obey a given ruler. The kernel of truth here is that any ruler survives only because the guards — the possessors of coercive power — go along with him. This agreement need not be unforced, and usually it isn’t. All the guards might prefer a different ruler, but unless they can all agree to change simultaneously, they are stuck. Nevertheless, as Thomas Hobbes said, ultimately “the power of the mighty hath no foundation except in the opinion and belief of the people”.

4. Lastly, as well as understanding why many different state forms can be stable, we are also learning something about how they might change. A state is an equilibrium. A change in the state, via revolution or conquest, is a change from one equilibrium to another. In game theorists’ terms, struggles over state power are struggles over equilibrium selection.

   This gives an interpretation to the mysterious concept of “legitimacy”, which political scientists and theorists see as so important to the exercise of power. Legitimacy is not best thought of as a matter of deep loyalty to a given state, though deep loyalties may support legitimacy. Instead it’s a matter of mutual expectations. If I expect that you see Lord Farquhard as the ruler, and you expect that I expect that,… and so on, then Lord Farquhard has legitimacy. Legitimacy collapses not when people have doubts about Lord Farquhard, but when their private doubts become public.

The idea that struggles over power are about equilibrium selection is not as helpful as one might like, because equilibrium selection is a very difficult problem for economic theory. Economists’ main tool — the pursuit of self-interest — doesn’t have much purchase in explaining which equilibrium gets chosen, because all equilibria are by definition compatible with every actor’s self-interest. (This is true even though one equilibrium may be better for everybody!) We don’t yet know much about predicting equilibrium selection in simple lab experiments, let alone in the huge messy field of political history.

We can be sure that fighting over real world equilibria often involves big expenditures of effort and resources, just the kinds of expenditure that are Olsonian public goods. The Russian and Chinese revolutionaries, despite their materialist theory of history, had a hard-nosed sense of legitimacy and focused great efforts on gaining the allegiance of the people when necessary. Lenin at least had an equally hard-nosed sense of how, after consolidating power, the people’s will could be discarded. Wars of conquest, too, are about equilibrium selection. Wars are not usually aimed at total physical destruction of the enemy, but at breaking his will. Physical violence is a way to rebalance the incentives between loyalty and surrender.

(As a sidenote, the fact that intergroup conflicts involve multiple equilibria is one reason why even rational actors may fight wars. It’s possible for both sides to know the balance of material resources perfectly, and yet to still tell themselves different stories, each compatible with common knowledge of rationality, about which side will crumble and which will win.)

One thing we certainly don’t know is that the “best” equilibrium always wins. There are lots of examples of very evil states, like the ancient Hittites or the modern USSR, persisting for very long. And possibly economic development was held up for thousands of years by inefficient states — that’s the Acemoglu and Robinson view in their famous book. So, even if democracy (or dictatorship) were better for sure, this doesn’t mean we could just expect it to win out.

Democracy versus autocracy: the showdown

Turning back to that comparison, which is better? Here are the takeaways from my “model”.

There’s no deep reason to think that democracy should be more efficient than autocracy. Instead, their relative performance depends on the tradeoff between an autocrat pursuing his self-interest intelligently, and the majority pursuing their self-interest erratically.

Lee Kuan Yew as a child

In particular, autocrats who face external constraints might be very efficient. Lee Kuan Yew in Singapore is a famous example. Qatar and UAE may be two more. (Loud complaints about the misery of migrants to these places never ask why they keep arriving in such numbers. Could it be that the pay is vastly better than the alternatives in their home countries?)

Private organizations are rarely democratic. Put another way, democratic organizations have not succeeded in the marketplace. A company is basically a dictatorship, perhaps moderated by an oligarchic board. Presumably competition in labour markets helps to prevent within-organization rent-seeking by the CEO — not completely, but enough that “autocracy” becomes more effective than its rivals.

From this point of view, interstate competition might be more important than within-state democracy as a way to ensure efficient government and economic growth. (Perhaps also to ensure human rights and just development, since democracies are not always especially good at providing these either.) Certainly in history, times and places of interstate competition, within the framework of a common culture, make some of the famous examples of innovation and ferment: ancient Greece, Renaissance Italy, and post-1500 Europe, for example.

This all sounds quite pessimistic about democracy. But there are some doubts on the other side. Economic theory assumes rulers are rational, but in real life an individual autocrat might be an idiot, with disastrous consequences, like Putin plunging his fists ever deeper into the Ukraine tar baby.

In deglobalizing eras, when interstate exit gets harder for capital and people — so maybe, eras like today — democracy may be important as a prevention of autocracy’s worst excesses. Von Hayek compared democracy to a “protection against plague”, rarely needed, but vital on those rare occasions. There’s more evidence that it protects against the other horsemen: war and famine. Voters may not know much, but they can work out who to blame when things go really bad. Democracies don’t produce many Lee Kuan Yews, but they may avoid Stalins and Pol Pots.

So, the tradeoffs are still quite unclear in theory. Can empirical analysis settle the matter? I doubt it, but that’s a topic for another time.

---

If you enjoyed this, you might like my book Wyclif’s Dust: Western Cultures from the Printing Press to the Present. It’s available from Amazon, and you can read more about it here.

You can also subscribe to this newsletter (it’s free):