The effect of assassinations on political institutions: Are the Jones–Olken findings robust?

Abstract

Jones and Olken (JO; 2009) report that the outcome of an assassination attempt – whether it is successful or not – matters for political institutions. We question how these institutions are classified. Revisiting their methodology shows that their findings are generally robust to reasonable differences in classification, thereby strengthening their conclusions that leadership and luck matter for political outcomes.

Keywords:

• assassination
• democracy
• leadership

JEL Classification:

• N4
• O1

Notes

¹ See JO (2005, 2009) for more detailed discussion on the relevance of chance, individuals and social forces in history.

² The POLITY2 score goes from −10 (autocracy) to 10 (democracy) in unit intervals.

³ The more commonly used Freedom House measure only begins in 1972, limiting its usefulness for our purposes.

⁴ Since the POLITY score is −9 in 1929 and 2 in 1933, the POLITY2 index replaces the −88 during the 3 transition years with −6, −3 and 0.

⁵ JO themselves are somewhat uncertain about Pakistan 1988. General Zia died in a plane crash, the cause of which is debated. Because they state in their paper that this instance does not drive their findings, we do not further explore this concern here.

⁶ A lingering concern, however, is that the ACLP (2000) only begins in 1946. Presumably, prior differences would have also arisen between the ACLP and Marshall and Jaggers (2004) classifications.

⁷ We also conducted extensions mirroring table 6 in JO (2009), where they not only distinguish between autocracies and democracies but whether leaders had been in power for more than 10 years. They also considered 10- and 20-year horizons after the assassination attempt. Applying these concerns to their extensions also shows their results are robust. These results are available upon request.

⁸ JO do report, however, that that their results are robust to using other indices provided in Marshall and Jaggers (2004).