2. The electable dictators give the people the policies they want. Rational choice theorists might assume this just re-states #1, but they're wrong. War-mongering and scapegoating are often popular; so are wealth-destroying economic policies. Mubarak has given Egypt peace, but most Egyptians I've talked to tell me that free elections would put bloody-minded Islamists in power. (Aside: Here's my forthcoming paper about the mechanisms that lead dictators to heed their subjects' wishes).
3. The electable dictators are the best brain-washers. Would Kim Jong Il suffer a crushing defeat, or an overwhelming victory? It's a tough question. But if you think that a man who rattled nuclear sabers while his people starved en massemight win a free election, it says volumes about political man.
4. Alternative theories? Please share.
It might surprise you, but even I think there is a lot of truth in #1. Democracy yields a lot of disappointing results, but it rarely rewards leaders for sharp declines in living standards or drawn-out wars.
At the same time, #2 is also a big part of the story: At least since the dawn of modern communications, successful dictators are usually demagogues.
As for #3, I see a lot of variation. In the 20s and 30s, Communist propaganda moved the souls of millions; but by the 70s it was a giant waste of rubles. If I had to bet, I'd say that despite all the juche propaganda, North Koreans would throw Kim Jong Il out on his ear. The real challenge would be to convince the voters that after sixty years of terror, it was finally safe to say no to famine.
Gordon Tullock has often wisely remarked that social scientists spend little time studying dictatorship even though it has almost always been the predominant form of government. If we can answer my questions here, I think we will have learned some deep lessons about not only dictatorship, but democracy, and human nature itself.