The Potential of Direct Democracy: A Global Measure (1900–2014)

V-DEM distinguishes among 7 high-level principles of democracy (e.g. liberal, participatory, deliberative, etc). It also disaggregates into dozens of lower-level components of democracy such as regular elections, judicial independence, direct democracy, and gender equality, and provides disaggregated indicators for each concept and each component. V-DEM also provides an estimate of measurement reliability for each rating whenever possible. Overall, the database is composed of more than 15 million data points. The data, including those used in this paper, are public, free of charge, and fully available to researchers at V-DEM’s homepage (https://​v-dem.​net/​en/​).

This measurement serves as the basis of the direct democracy index in the V-DEM database (Coppedge et al. 2016a, b), which constitutes one of the components of the participatory variety of democracy as described in greater detail by Coppedge et al. (2011).

While there is neither “universal referendum terminology” (Suksi 1993: 10), nor a unique typology (see Hug and Tsebelis (2002), Vatter (2009)), here I employ the terminology used by the National Conference of State Legislatures (http://​www.​ncsl.​org/​), the Initiative & Referendum Institute of the University of Southern California (http://​www.​iandrinstitute.​org/​), and the Centre for Research on Direct Democracy of the University of Zurich (http://​www.​c2d.​ch/​).

They claim “Plebiscites have nothing to do with initiatives and referendums; on the contrary, they are often used by governments who want to get a special legitimacy on their policies by bypassing existing laws and constitutional rules” (p. xix).

This is not a particularity of relatively healthy democracies such as Uruguay or Switzerland (Papadopoulos 2001) or the United States (Gerber and Lupia 1995) where this phenomenon has been studied in great detail. This can also happen in non-democratic regimes such as Iran, as evidenced by Erdbrink (2015), where the threat of a referendum may be sufficient to change politics at the highest levels. Of course, this is a hard thing to measure, but it is a good example of where “de jure” features of a constitution may matter more than we think.

This type of approach was taken by Tolbert et al. (2001).

These studies have either countries as units of analysis, such as those in Continental Western Europe (Gross and Kaufmann 2002; Vatter 2009), Latin America (Breuer 2011; Madroñal 2005), and South East Asia (Hwang 2005), or subnational units, most notably Swiss cantons (Freitag and Stadelmann-Steffen 2010; Trechsel and Serdült 1999; Frey et al. 2001; Stutzer 1999), American states (Gerber 1999; Bowler and Donovan 2004), or German Länders (Scarrow 1997; Schiller 2011; Vatter and Stadelmann-Steffen 2013)).

Popular votes do not occur in an institutional vacuum and the extent to which they are free and fair is crucial in the same way it is for regular elections. Perhaps, as with any electoral procedure, a popular vote held in an autocratic setting is notably different from the same type of vote in a democratic context.

I am aware that there were some cases in which MDDs occurred in non-independent states. The major problem I face in coding them is that in several cases, the votes took place in territories that are circumscribed to different current countries (e.g. French and British Togoland, British Southern and Northern Cameroons, Southern Rhodesia, Danish West Indies).

Let me illustrate with two prodigious users of direct democracy: Switzerland and Uruguay. In Switzerland it is extremely easy to qualify a popular initiative as the requirement for a minimum number of signatures is one of the lowest cross-nationally (100,000 signatures, which represent less than 2 % of the electorate); in Uruguay, however, an initiative must have at least 10 % of the signatures of the electorate to qualify. At the same time, the chances of approving a qualified CI-MDD in Switzerland are lower than in Uruguay because of the requirement of double majorities (citizens and cantons). In Uruguay, a majority of the electorate determines the fate of the initiative (as long it represents 35 % of the electorate), which is far lower than a double majority.

Of course, I could think that in the long run, knowing popular votes’ previous results might influence the prospect of some groups attempting to qualify a CI-MDD, but even so it might not necessarily undermine the chances of triggering a CI-MDD per se. The question, therefore, is which characteristics are crucial for triggering and which are crucial for approval.

If the country under consideration is federal, then this index should ideally correspond to the average of the sum of each subnational unit (the same logic applies for the rest of the terms). In this research, however, only the national dimension is covered. Finally, as it happens with most democracy indices, this index does not include subnational uses of/rights to use MDDs regardless of how intensive their use may be (American states, Swiss cantons, German länders, etc).

I could have opted for a logarithmic transformation, but the shape of the obtained line does not fit my theoretical expectations, particularly when only a few days are available to gather signatures.

As Maniquet and Morelli comment: “quorums are a simple way of protecting the status quo” (Maniquet and Morelli 2010: 2), “a low turnout in referendums is seen as a threat to their legitimacy” (Qvortrup 2002: 164). It has been shown elsewhere that participation quorums may have a pernicious effect on the process of direct democracy as they produce incentives for strategically derailing certain proposals through vote abstention, thus helping in not reaching the quorum. Not only are they expected to decrease electoral participation, but they may also violate the secrecy of the vote (see Altman 2011: 18–24; Aguiar-Conraria and Magalhães 2010a, b). Since the incentives to abstain disappear under approval quorums, they are considered superior—better institutional tools—than participation quorums.

Regardless of whether the decision is binding, any decision taken directly has a great dose of legitimacy that is hard to undermine, particularly under a democratic regime. Thus, a consultative vote is more than “half” but less than a binding vote.

A similar rule also applies for plebiscites and obligatory referendums just in the case a country has the legal architecture to use a particular TD-MDD but never had one. The threat score of that TD-MDD equals 0.1.

As the assumption that the cycle endures for 20 years is rather controversial, I have also explored this decay function with a shorter cycle of about of 10 years. As in the long cycle, during the first five years the threat equals 1, but then it loses 0.15 yearly in case of success. In case of failure, the first five years equals 0.9 and then it loses 0.2 yearly. No substantial differences were found between the 10- and 20-year cycles.

By definition there is no signature gathering for obligatory referendums or authorities’ plebiscites.

Sometimes, leaders call for plebiscites without the legal framework needed to do so. These “ad hoc” plebiscites are usually justified by governments as a means to bypass national emergencies or crises, and are recurrently based on “façades of legality” through sometimes-obscure administrative acts. The question is what status to assign to a regime that has no permanent constitutional authorization for plebiscites but uses them nonetheless, perhaps even regularly. There are two alternative ways to deal with this problem: treat them either as single events or as lasting characteristics of the regime where they occur. Whether we treat them as single events (“flashes” of direct democracy) or lasting characteristics of the regime depends on the research question at hand. Sometimes we need to measure discrete events of direct democracy, other times we are more interested in the ongoing character of a regime. Here I treat them as flashes of DD rather than a lasting characteristic of a regime for two reasons: first, I do not have reliable information about the length of each regime in each country on earth; and second, treating them as flashes provides valuable information which enriches the analysis particularly if combined with polyarchy indices. In any case, given the index for assessing the potential for plebiscites was composed by two terms, the first being (∃) * (D) and the second [(1 − SQS) * (AQ)], for “ad hoc” plebiscites the first term will be zero for the whole period that a particular government was in power.