4  Referendum and regime type

4.1 Referendum type worldwide

In Figure 4.1 and Figure 4.2, we analyze the institutional trigger type of referendums worldwide since 1900. This variable denotes the way in which a referendum was initiated. As we can see, the differences between the world regions are quite pronounced:

  • In Africa, most referendums were top-down and some automatic. There were no bottom-up referendums.
  • In the Americas, most referendums were top-down, with some automatic and few bottom-up referendums.
  • In Asia, most referendums were either automatic or top-down. The number of top-down referendums remained low.
  • In Europe, the frequency of top-down and automatic referendums was similar. The number of bottom-up referendums peaked at the turn of the millennium.
  • In Oceania, there were many automatic referendums and few bottom-up and even fewer top-down referendums.

In summary, top-down referendums dominated in Africa and the Americas. In Asia and Oceania, there was an increase in automatic referendums. In Europe, there were the most bottom-up referendums, and the number of top-down and automatic referendums was similar.

Figure 4.1: Referendums by institutional trigger type per decade since 1900

Figure 4.2: Referendums by institutional trigger type and region per decade since 1900

(a) Africa

(b) Americas

(c) Asia

(d) Europe

(e) Oceania

4.2 Referendum type by regime type

In figure Figure 4.3, we differentiate the type of referendum by regime type, using the Regimes of the World (RoW) measure by V-Dem (Coppedge et al. 2023). We find the following:

  • In closed autocracies, mostly top-down and a few automatic referendums were held. There were only three bottom-up referendums in closed autocracies.

  • In electoral autocracies, the number of top-down and automatic referendums was similar and higher than in closed autocracies. Again, there were almost no bottom-up referendums.

  • In electoral democracies, again mostly top-down referendums were held. The number of automatic and bottom-up referendums was about equal.

  • In liberal democracies, the number of bottom-up referendums was highest, followed by automatic referendums and then top-down referendums.

  • The number of referendums in countries without a Regimes of the World value (NAs) was very high. This is due to the fact that the RoW measure is not available for many of the small island states making up the bulk of automatic referendums.

In summary, we see that referendums took place across democracies and autocracies. However, the trigger type varies. Top-down referendums dominate in closed autocracies and electoral democracies. The number of top-down and automatic referendums is similar in electoral autocracies and liberal democracies. Bottom-up referendums were almost always only observed in liberal democracies.

Figure 4.3: Referendums by regime type and region per decade since 1900

(a) closed autocracy

(b) electoral autocracy

(c) electoral democracy

(d) liberal democracy

4.3 Rank number of ballot dates by regime type (Regimes of the World)

Figure 4.4 shows the number of ballot dates differentiated by Regimes of the World (RoW) (Coppedge et al. 2023):

  • Around 260 ballot dates took place in liberal democracies. Excluding Switzerland with close to 300 ballot dates, the frontrunners were New Zealand (39), Ireland (31) and Italy (23). Note that Lithuania is among the top-ten for both liberal (7) and electoral democracies (5).
  • A total of around 150 ballot dates took place in electoral democracies. Here, the frontrunners are Ecuador and Uruguay.
  • Around 260 ballot dates took place in electoral autocracies. The most frequent countries were Egypt (18), Syria (11) and Zimbabwe and the Philippines with 9 each.
  • Around 200 ballot dates took place in closed autocracies. Here, the most frequent countries were the Maldives (13), Morocco (11) and again Egypt (9).
  • For around 270 ballot dates, there was no information concerning regime type available. This is where the country names could not be matched with the RoW dataset. This can be due to sovereign states not coded by RoW (e.g. Liechtenstein, Palau or Micronesia) or it can be due to RDB coding territorial units that are not sovereign states (e.g. Norfolk Island: Australia, Northern Mariana Islands: USA, Virgin Islands: USA, etc.).

Figure 4.4: Top-ten countries since 1900 by number of ballot dates and regime type (RoW), without Switzerland and Liechtenstein

(a) liberal democracies

(b) electoral democracies

(c) electoral autocracies

(d) closed autocracies

(e) NA

4.4 Rank number of ballot dates by regime type (Freedom House)

In Figure 4.5, we re-run the analysis according to type of referendums by regime type, this time using the Freedom House measure of regime type (Freedom House 2023). Because Freedom House only gathered data since 1973, the overall numbers are lower and the ranking order changes. What is new with the Freedom House data is that smaller countries such as San Marino (free), the Comoros (partly free) and the Maldives (not free) are added to the list, thus making for a more complete analysis.

However, even using Freedom House data, there were more than 230 cases (NA) where there was no measure of regime type available for a given year.

Figure 4.5: Top-ten countries since 1973 by number of ballot dates and regimetype (Freedom House), without Switzerland and Liechtenstein

(a) free

(b) partly free

(c) not free

(d) NA