Presidents Don’t Matter, but Dictators Do

On Monday, Dubner argued that the President of the United States matters far less than people think.

As it turns out, there is some economic research to back him up, at least when it comes to influencing economic growth. Ben Olken (one of my favorite young economists) at the Harvard Society of Fellows and Ben Jones from Northwestern (not the same Ben Jones of Dukes of Hazzard) have two recent papers on the subject.

The first of these papers, published in the Quarterly Journal of Economics in 2005, uses deaths of leaders in office as a quasi-random source of variation in the identity of the leaders. (You don’t want to use switches in leadership that arise when an incumbent gets voted out of office, because poor economic performance might be one of the reasons the incumbent was beaten.) They find that, in democratic countries, leaders don’t matter much. Dictators, on the other hand, appear to exert substantial influence on the economic fortunes of their countries. This result makes intuitive sense, since there are so many fewer checks on the power of an autocrat.

The second paper is slightly farther afield. In this study, Jones and Olken compare a country’s outcome after a leader is assassinated with the outcome after a failed attempt on the leader’s life. Their most notable finding is that successful assassinations of autocratic leaders are more likely to produce transitions to democracy than unsuccessful assassination attempts.


Presidents, or at least presidential party, seems to matter in shaping the distribution of growth in real disposable income since WWII:


That might suggest that totalitarian dictatorships are the best form of government, as long as the dictator is smart and benevolent.


"Their most notable finding is that successful assassinations of autocratic leaders are more likely to produce transitions to democracy than unsuccessful assassination attempts. "

Or as a metaphor: Successful purchases of lettuce and dressing are more likely to produce salads than unsuccessful purchases.
The assassins may not have wanted to create a democracy, but that democracy will not likely come about with the autocrator still in power (exceptions involving King Juan Carlos I of Spain not counted.)

ils vont

Is Democracy something that we really want in this country? Everybody complains about the president and the political climate as if there is a right and a wrong. If things were like x, y and z then our country would be perfect. Why don't we split up the country, liberals on one side and conservatives on the other and give each side their ideal ruler and see what happens.

Rita: Lovely Meter Maid

So....Dictators matter/presidents don't? Wow, that's...quite depressing. What about Kings and Queens? (I guess most of them are mere figure-heads, nowadays, though). Frankly, it all stopped mattering to *me* when the Queen Mum passed on. And I'm not even British, so it can't be that. I just liked her kindly, benevolent face and those rather delightful, fairy-godmother type outfits of soft pastel that she wore. Why must our presidents dressed so hard-edged and business-like? (Okay, I've really nothing to say on this topic, either. I guess I'm a bit inspired by the Lord, God! bird, I must admit).


Did they include Robert Mugabe's opposition on the distribution list?

Rita: Lovely Meter Maid

I meant a bit *more* inspired by the Lord God! bird (than this present topic). Jeez. It's horrible to hit the "submit comment" button and see it doing its thing, when all the while, a glaring error is flashing in front of one's face. But, it's too late!!


ils vont wrote "Why don't we split up the country, liberals on one side and conservatives on the other and give each side their ideal ruler and see what happens."

That's a great idea! As long as us liberals get the western (better) half.

Alex Crosby

I think a glaring problem with these young fellows' research are their methods. How in the world do you measure the effects a president or dictator yields on the economy?

Let's take for instance the Freakonomics example where Roe v. Wade lowered crime rates and perhaps helped give boom to the lucrative 90's. The President appoints a Supreme Court Justice -> they make vital decisions such as Roe v. Wade -> said decision inadvertently or purposefully affects the economy.

Alright, we get it already, the President has no control over the Fed, but virtually every other decision - going to war, not going to war, the patriot act, appointing a Supreme Court Justice - is going to have some ripple effect on the economy. Would it be so outrageous to suggest that these ripple effects over the years have become predictable and are thus utilized by a team of experts under the President?


re:post 2

The Greek philosophers came to the same conclusion, i.e. that a wise benevolent dictator would be the best form of government. Unfortunately, they never could find one, so they settled on democracy.
Or, to paraphrase Winston Churchill:
Democracy is the worst possible from of government, until you consider all the others.


re: post #4

actually ils vont, we sort of tried that once. It was called the American Civil War. But the Northern Liberals would have none of that.

p.s. Remember that war was first about economics and only later about slavery.

Shane Harris

I am unsubscribing from this blog since you are supporting a news source on the internet which requires a subscription. While I realize that it is free to view your blog, it is the association with the NYT online in its present configuration to which I object. Their freedom to make money is equivalent to my freedom to choose to boycott them in lieu of better alternatives. Information wants to be free. Good luck with your experiment.


if that second paper is true, this doesn't bode well for W...


re post 11

Your P.S. is a pretty loaded statement. I don't "remember" the causation as definitively as you apparently do. This isn't the forum to refine your statement exhaustively, but at a minimum, some might say that slavery and economics were intertwined inextricably. Separating one from the other and thereafter establishing singular causation is a bit harder than your off-the-cuff remark would indicate.


I'm skeptical that the differences between dictatorships and democracies are controlling properly for differences in economic systems. At least nowadays, most democracies have more or less laissez-faire economies, while dictatorships practice much more controlled economics (even if they are oligarchical rather than socialist). If that is true, then saying that the controller of a controlled economy has a greater effect than the controller of a free economy is as much of a tautology as saying that if you kill a dictator, his successor is likely to be different.


"That's a great idea! As long as us liberals get the western (better) half."

The proper division seems clear from election results: moderate liberals should get the northeast, conservatives get the south and midwest, and socialists get the west coast; possibly with Alaska saved for us libertarians (who will be the only ones who still want to live there after the federal expenditures are cut off). A lot of people will have to move, but that should at least minimize the dislocations. There doesn't seem to be a nice territorial split between the paleoconservatives and neoconservatives, but I'll let them work that out themselves.



Do you defend pirating software and music? Do you believe it costs the Times nothing to provide their "free" information to you? Or do you believe it's okay to pay with your eyes (i.e., seeing ads) but not okay to pay with your wallet?

Joel Rennie

Well, if Presidents don't matter, but dictators do, and if successful political assassinations (Diem, Allende, etc.) are more likely to foster democracy than failed attempts (Castro, Lamumba, et. al.) does that explain why the ultimate experts on all of these things - our brave civil servants at the Central Intelligence Agency - are reportedly busy spending our tax dollars by using "Company" computers to edit the Wikipedia biographies of Richard Nixon and Ronald Reagan?

What is the economic value of government-fostered disinformation these days?

jackie hayes

I wish you would take more care to spell out the proper interpretation of the Olken-Jones result. It does NOT imply that any president is as good as the next one. That would only be the case if the leaders' replacements had been randomly determined, when surely most of the replacements were ideologically close to the previous leader. It only means that one leader's PERSONAL acumen and charisma are unlikely to matter much, except in a dictatorship.

Given that this comes as a presidential race heats up in the US, some might take this to mean that it doesn't matter if a Republican or Democrat is elected president. In fact, the Olken-Jones result says nothing about this -- it only implies that two democratic leaders who favor the same policies are likely to achieve similar results.

Richard Elliott

I have just read your book twice, once on the way from England to Shanghai, and back......yes the movies were trash.But they never had a chance.
I laughed at the parallels between our two countries. Broken windows being the main one...but also the "low income" names (check Britains peasant underclass...!!Take this site with a bit of humour please.
Please keep up "thinking outside the box"
I will be in touch