Why dictatorships fail

Dani Rodrik points out that, on average, democracies outperform dictatorships in terms of long-term economic growth and economic stability. Personally, I wouldn't make too much of this observation, since it's a small sample size (what if only America and its friends grew in the postwar period), there are a million kinds of selection bias at work (e.g. the "resource curse"), and there's a huge endogeneity problem (what if growth causes democracy instead of vice-versa).

But assume it's true. Why? Matt Yglesias thinks that it's because dictators in the past were just stupid:

[H]istorically [Rodrik's point has been right]. [But] if you’re a modern-day dictator, the lesson of history is clear that the less-corrupt, less-exploitive Singapore model was not only better for the Singaporeans it was better for the dictator. For all the same reasons that over the long term the revenue-maximizing tax rate equals the growth-maximizing tax rate, over the long-term dictatorship is more incentive-compatible than Mobutu seems to have realized.

However, Yglesias thinks, dictatorships have now become much smarter, and will now do significantly better:

Historically, few authoritarian regimes have seen that their own self-interest is best maximized via enlightened policies. But at least one interpretation of what’s happening in China is that the most important authoritarians around have figured this out (Abu Dhabi also seems to have) and this is driving major improvements in human well-being.

I heavily doubt this. Hasn't the human race had thousands of years to experiment with dictatorships? You'd think that if maximizing a country's long-term growth performance was always in a dictator's interests, that dictators would have figured this out long before now. Why should now suddenly be so different?

I think the explanation for the success of democracy lies elsewhere. In the modern era, we've seen many instances of dictatorships that grew rapidly - Imperial and then Nazi Germany, fascist Italy, Taiwan and South Korea and Singapore in the 70s and 80s, Chile under Pinochet, etc. The problem is, they all seem to stall out at a moderate level of income; it's easy for dictatorships to get middle-class quick, but it's very hard for them to get rich.

My guess that it has to do with what political scientist Bruce Bueno de Mesquita calls "the logic of political survival." The theory in a nutshell: no matter whether you're the leader of a democracy or a dictatorship, you have to pay people off to stay in power. In a democracy, you have to pay off voters, which requires public goods (because there are so many voters). In a dictatorship, you have to pay off your oligarch buddies, or the army, or a vast ruling party. Each one of those supporters asks for a lot more than a voter asks for in a democracy. Hence, dictatorships are forced to squander more state resources in direct payouts to regime supporters.

I'd add something to this theory. In a democracy, if you lose power you simply become part of the minority party in Congress or retire and go on posh speaking tours. In a dictatorship, if you lose power there's a fair chance you'll be riddled with bullets and hung on meathooks in the public square. Yglesias seems to think that this is a strength of dictatorships, because it will cause them to be very careful about keeping the country on a firm growth path:

But when growing dictatorships hit economic downturns, what tends to happen is you throw the dictators out of office. I’m not sure whether China’s leaders can keep delivering growth, but if they can’t it’ll be hard for them to stay in charge.

I see it very differently. When losing power means losing everything, dictators have an incentive to increase payouts to regime supporters by huge amounts, as a form of insurance. If a country is ever faced with an opportunity for long-term gain at the expense of short-term pain (e.g. deficit reduction, or curbing wasteful public works projects), a dictator is unlikely to take that opportunity. Prosperity in the long term is not worth the meathook treatment in the short term.

I think we can see this dynamic at work in China. Who are the regime's key supporters? Answer: local government officials. China has five layers of government (compared to our three or Japan's two). All those government officials have their hands in the pie; as Minxin Pei will tell you, much of China's economy consists of "private" factories owned by friends of local government officials. The government officials get massive kickbacks, the factory owners get cheap second-class-citizen migrant labor, license to destroy the local environment, and free land obtained by booting local peasants.

It's easy to see that China will not be able to afford this kind of economic system forever. Right now, the system works because China is so poor that kickbacks are relatively cheap, and because nearly every investment in a poor country tends to pay off. But as the economy grows richer, the kickbacks to the local government officials will become a greater and greater drain on the private sector (this is called Baumol's Cost Disease), even as the need for kickbacks forces China to maintain high levels of investment in projects that will never pay off. Michael Pettis and others report that this inefficiency is already starting to kick in.

Therefore, I do not share Yglesias' faith in the intelligence of dictators. No matter how intelligent they are, the logic of their political survival will force them to choose the short-term over the long-term. This time will not be different.