A study of triggering events: When do political regimes change?

Abstract

Political regimes are stable most years, but sometimes they jump. The stable years are periods of political status-quo equilibrium. To break a status quo requires a triggering event. The paper is an attempt to identify and classify what close observers at the time thought were the triggering events in a sample of 262 larger regime changes between 1960 and 2015 in 170 countries. The sample consists of all changes in the Polity index with a numerical rating above 3 (i.e. of 4 or more). The source for the triggering events is country-relevant articles in The Economist. Triggering events are classified in a (2 × 2) table with four cells: domestic political (DP), domestic economic (DE), external political (XP), and external economic (XE), which remains empty. By far the most common is DP, but the domestic political events prove to be very different. Thus, most jumps are exogenous in the perspective of development.

Appendix: Is the sample of larger regime changes a good sample?

The sample of major regime changes includes all differences over the threshold of three points (i.e., for 4 or more points) in the Polity index from 1960 to 2015, as explained. The threshold is chosen according to the model in Fig. 3. Within those frames, the sample is transparent and complete.

Other political regime indices exist. Some are binary, but others are based on scales allowing a selection of the larger changes. The alternative indices yield alternative data samples. Most large changes surely would be included in all samples, but some Polity jumps would have been excluded, and other extra jumps would have been identified. As long as those differences are random relative to Table 6, they would not have influenced the results.

One alternative index is the Freedom House’s FH index. To see if the sample is robust, I have repeated the exercise with the FH index. It starts in 1972 and has two parts: Political Rights and Civil Liberties; I use the average of the two. Both parts use an integer scale from 7 for the least democratic to 1 for the most democratic countries. Thus, the scale is falling when democracy increases. The FH index does not include the categories of anarchy and foreign intervention. The FH index was started for a period that overlapped 2 years, but then the overlap changed until it became a calendar year and, in the process, 1 year was lost. That change is very visible when the larger events are matched, but it is easy to control for. As general characterization, the FH Index has more small events and more sequences than Polity does. The closest matching to a Polity change of four points is a FH change of 1½ point.

The two indices are matched for 169 countries from 1973 to 2015. For 127 countries, both indices point to the same large events. In the data for the last 42 countries, some differences are evident. Polity typically has larger changes, i.e., more changes pass the threshold. An example is the events in Myanmar since 2014, which Polity counts as a large move towards democracy (+ 11 points), while Freedom House only reports a small adjustment (− ½). There is about 20 such cases. The difference in the sizes of the changes are especially large since 2010 and explaining why the FH index shows falling democracy since 2010, while the P-index reveals no such decline. By my assessment, the FH index misses about 20% of the larger changes, while Polity misses only 3% of them. A few countries are treated quite differently, notably Bangladesh, Cambodia, Congo Ki, Guyana, Haiti and Pakistan. It is clear that such measurement contains error. It appears that the missing cases are distributed randomly across Table 6.