What is democracy?

(This post is a preliminary draft - comments are welcome. Last modified on 17th November 2005)

Political scientists and economists love to explore the relationship between democracy and socio-economic outcomes (Barro 1996, 1999; Przeworski et al. 2000; Bates et al. 2003; Glaeser and Shleifer 2005; Persson and Tabellini 2005). By democracy, however, we can mean different things. Unless we define democracy in a certain way, we are unable to develop a theory to understand the effect of democracy on, say, economic growth. It may be the case that some aspects of democracy are good for growth while others are not. Or, all aspects of democracy are complementary - a single aspect of democracy alone is not enough to ensure some outcomes; it works only if other aspects of democracy are also present.

Motivated by Levitsky and Way (2002)'s article on what they call "competitive authoritarianism", here I try to make a checklist for a country to be qualified as "democracy".

There are three stages to look at: (A) pre-election, (B) electoral process, and (C) post-election.

A: Pre-election

In the pre-election stage, the incumbent government lays the ground rule for competition for national political office (executive and legislature). A country is democratic at this stage if the following six criteria are met.

A1: Suffrage is universal - (nearly) all adults can vote.
A2: The executive is directly elected or indirectly elected by legislature.
A3: Elective legislature exists.
A4: Opposition parties are allowed to exist.
A5: No gerrymandering (ie. manipulating electoral rules and districts in favour of the ruling party) is undertaken.
A6: Press freedom is ensured.

Outright autocratic countries are disqualified at this stage - South Africa during apartheid did not satisfy A1 though it did satisfy A2-4. The military government seizing power by a coup is disqualified due to requirement A2, even if it satisfies A3 and A4 as is the case in Brazil during 1964-1985. The communist countries such as China and Cuba do not satisfy A4 while A2 and A3 may be satisfied - elections can take place without opposition such as Saddam Hussein, who claimed 100% voting share in an uncontested presidential election in 2002 (see a BBC article on it).

A1 is more subtle than it appears to be. In the United States South, black voters were de facto denied participation in elections by poll taxes and literacy tests until the 1960s. See Besley et al (2005).

Kenneth A. Bollen provides a dataset of suffrage across countries for 1950-2000. As long as you decide the threshold, you can tell whether or not A1 is satisfied in a country in a certain year with this dataset.

Acemoglu and Robinson (2005)'s theory of democratization mainly looks at A1 though the fact that in their model policies preferred by the median voter are adopted in a democracy implies that A2-A4 are implicitly taken into consideration. (If no opposition candidate is allowed to run for office, the single candidate can choose whatever policy he/she wants and gets elected.)

The minimalist way of defining democracy only looks at A2-4 (sometimes A1 as well). A good example is Przeworski et al. (2000)'s definition of democracy. As they are only concerned with the period after the Second World War, they ignore A1. But when you look back at pre-WWII periods, today's Western democracies initially didn't satisfy A1 by, for example, denying women their voting right. So Carles Boix, when he extends Przeworski (2000)'s dataset to the 19th and early 20th centuries, includes A1 as requirement for democracy.

The minimalist definition of democracy becomes problematic if you see countries such as Singapore, which satisfies A1-4 - it does satisfy A4: look at the list of political parties by Singapore-elections.com - but does not meet requirement A5 and A6. To avoid such complications, Przeworski et al (2000) come up with another requirement, which is to say power has to be alternated under the same electoral rule. This disqualifies Singapore as the same party has ruled the country ever since independence. This qualifying rule is quite successful to exclude "dodgy" democracies - Mexico and Senegal until 2000. But this is an outcome-oriented way of defining democracy. Theoretically, it's not clear whether, say, the Singaporean ruling party keeps power because of its undemocratic feature or because of its popularity among citizens. (Japan and Botswana are cases in point.) If you want to analyze the effect of democracy on political and economic outcomes, this is not appropriate to measure democracy. So I drop this requirement here.

A6 may not be appropriate either. Theoretically, press freedom is important to ensure that the incumbent's bad performance leads to electoral defeat - if voters do not know what the incumbents do, they may not want to vote them out. See Besley and Burgess (2002) and Besley and Prat (2005). So A6 is probably more appropriate to label as an institution complementing democracy, rather than a requirement for democracy. Freedom House's Press Freedom Survey provides yearly cross-country data on the degree of press freedom since 1980.

B: Electoral process

In the electoral process, the incumbents and opposition candidates compete for office given the conditions characterised by the above six aspects of pre-election. A country is called a democracy in this stage if it meets the following four criteria.

B1: Opposition candidates can freely run for office.
B2: Opposition candidates can freely conduct their electoral campaign.
B3: Voters can freely cast their ballots.
B4: Votes are neutrally counted.

Here what Levitsky and Way (2002) call competitive authoritarianism kicks in. It satisfies A1-A6 and B3-B4 more or less but it doesn't meet B1 and B2. B1 can be violated if the government arrests opposition politicians. B2 can be violated if the government denies opposition parties access to media.

B3 is related to the issue of secret balloting. Even in Western countries the secret ballots were introduced quite recently. See Baland and Robinson (2004).

B3 is also violated if voters are physically prevented from voting, as seen in Sri Lanka's 2005 election (in this case it's a rebel group, not the government, who prevented voters from voting, though).

The violation of B4 is what we call electoral fraud. The World Bank's Database of Political Institutions provides cross-country data (since 1975) on this (see variable FRAUD). Levitsky and Way (2002) point out that competitive authoritarianism often breaks down when it resorts to the violation of B4 - Marcos in the Phillipines and Milosevic in Serbia and Montenegro are cases in point. Ukraine's "Orange Revolution" can be included in this case though the incumbent himself didn't run.

C: Post-election

In the post-election stage, even the government elected in elections satisfying all the above conditions can become non-democratic if either of the following conditions is violated.

C1: The executive is checked and balanced by legislature and judiciary.
C2: Non-elective veto players do not intervene politics.

C1 is violated if a democratically elected president disbands legislature - such as Marcos in the Phillipines and Fujimori in Peru - or if the executive fires constitutional court judges after their ruling against the government. POLITY IV's "Executive Constraint" variable (named XCONST or EXCONST) measures this aspect of democracy, and according to North and Weingast (1989) - "Constitutions and Commitment: The Evolution of Institutions Governing Public Choice in 17th Century England," Journal of Economic History, vol.49, pp.803-32 - this is an important factor for secure property rights.

C2 is violated if the military or some religious authority intervenes politics. Examples of the former case are Thailand until the early 1990s and Turkey while Iran is a typical example of the latter.

An Application - Zimbabwe since 2000 -

With this list in hand, this article by The Economist magazine on Zimbabwe's parliamentary elections in 2005 is much easier to understand.

(Note that MDC is the main opposition party in Zimbabwe, and ZANU is the ruling party led by President Robert Mugabe.)

This country satisfies conditions A1-A4. A1 may not be satisfied as

"the 3m-odd Zimbabweans, most of them very likely MDC backers, who have been driven into exile by economic collapse or government repression, are barred from postal voting."

(But it wasn't until 2000 that Japanese citizens living abroad were allowed to vote, either.)

But Zimbabwe violates almost all other criteria.

"Since last time, constituency boundaries have been gerrymandered. A handful of MDC-held seats in populous urban areas have been abolished and new constituencies demarcated in rural areas where land-hungry peasants are friendlier to ZANU. Some urban seats have been merged with neighbouring rural ones, where voters are more pliable and ballot boxes in remote parts more easily stuffed."

This is a violation of A5.

Zimbabwe's most independent newspapers, notably the Daily News, remain closed, and ZANU virtually monopolises radio broadcasts.

This is a violation of A6.

Human Rights Watch, a New York-based group, reported less violence than before but said that intimidation and partisan laws give ZANU a huge advantage. It enumerates dozens of recent cases of MDC people being beaten, kidnapped and harassed by police and ZANU thugs.

B1 seems violated as well.

During the last general election, thugs and veterans of the independence war were paid to kill opposition campaigners... Now, because he wants to avoid shocking observers from South Africa (even though he is letting in only those he thinks most sympathetic), he is adopting subtler rigging techniques.

B2 was violated in the previous election but to a lesser degree in 2005 as

the MDC and Mr Tsvangirai are being given a few minutes of air time on the state television news (followed, of course, by an hour or so of Mr Mugabe and other ZANU leaders).

In the 2002 presidential election, B3 was violated:

Two years later, in 2002, the MDC's leader, Morgan Tsvangirai, a trade unionist, would surely have unseated Mr Mugabe in a presidential election, had the police not beaten up opposition voters, blocked people from reaching polling stations...

In 2005, the situation remains the same as

villagers are being told that ZANU agents will know, by looking through the transparent new boxes, who has voted for the MDC.

In the 2000 general election, B4 was violated:

The MDC lodged complaints about alleged vote-rigging in 37 constituencies which ZANU was adjudged to have won; but the courts, heavily influenced by the president and his friends, have failed in the past five years to deal with a single such case.

In 2005, the situation remains the same:

Of a sample group of 500 voters, barely half were listed correctly. Nearly a fifth of those named were dead; officials ensure that such "ghosts" are loyal ZANU voters.