A leader of some country somewhere has been assassinated in 2 out of every 3 years since 1950. That’s one of the attention grabbing facts in a recent paper by Benjamin Jones and Benjamin Olken, which exploits the supposed randomness in the success or failure of an assassination attempt to find out whether assassinations actually matter in terms of the social and political impact of assassinations. The idea is that while the probability of an assassination attempt is endogenous to social conditions, i.e. an assassination may be more likely when change is afoot, the probability that an assassination succeeds conditional on the attempt being made is random.

The authors have collected data on all 298 assassination attempts from 1875 to 2004, which I think is an interesting enough dataset to merit a paper of just descriptive statistics. For instance, 75% of all assassination attempts fail. At 28%, gun attempts have the second highest success rate, behind only  the”unknown” category, which has a 40% success rate. Knife attempts have a success rate of 13%, and on average .3 bystanders killed, which seems high to me. Attempts are made by solo rather than group attackers 59% of the time.

Overall number of assassinations worldwide experienced a boom in the early 1900s, and trough in 1940, another boom in the 60s, and has been generally trending downward over the past 50 or so years while still remaining well above the low point reached in the 40s. If you were a leader in 1910 the probability of being assassinated in a given year was around 1%. That has fallen to .3% today.  In contrast,  the conditional probability of success given an attempt has remained approximately constant at 25%. I could go on and on with facts, but you’ll have the read the paper yourself if you want more.

The main result of the paper is derived from the apparent randomness of success or failure conditional on the attempt. The authors cite the fact that had Hitler not left a Munich beer hall 13 minutes early due the weather he probably would have been killed by an assassins bomb, and Idi Amin had a grenade bounce off of his chest and he survived, while Kennedy was killed in a moving car 265 feet from his assassin. Clearly, chance plays a large part in the outcomes here.

They find that successful attempts of autocrats produces institutional change, and raises the probability of becoming a democracy. They also find that successful attempts increase the intensity of moderate wars, and may end large-scale conflicts faster. Surprisingly, they find no evidence that success leads to new wars.

Overall I have a hard time not reading this paper as a call for the assassination of autocrats. When they say “Our results point to the individual autocrat as a cornerstone of institutions, which suggests mechanisms (through leader selection and leader change) that can lead to institutional change”, I think they are ignoring the elephant in the room, which is the mechanism of successful assassinations.

The paper has many other interesting implications as well, for instance it challenges the deterministic view of history and lends support for the “Great Man” theory of history.