Some months ago, I was looking forward to two cases of (semi-)presidential democracy in East Asia that would be having nonconcurrent presidential and legislative elections with a short gap between each country’s two elections. South Korea (a pure presidential system, notwithstanding the existence of an office of ‘Prime Minister’) had its presidential election on 19 December. It has its National Assembly election on 9 April. Taiwan (a semi-presidential system) had its legislative election on 12 January, and its presidential election just this past Saturday, 22 March.
One of my research interests dating back to graduate school days is on the effect of electoral cycles in systems with elected presidencies. In fact, this was the topic of one of my three papers that formed the core of my dissertation ((The dissertation had the snappy title, Duverger’s Rule, District Magnitude, and Presidentialism. One of its points was to explore the extent to which the presence of presidential elections, and their temporal relationship to legislative elections, impacts the Duvergerian effects expected on the legislative party system from variations in district magnitude.)) Electoral cycles and their impact on party systems and representation are also the theme of a 1995 article I published in the American Political Science Review.
Most presidential systems seem to have concurrent elections–presidential and legislative elections on the same day (or legislative on the same day as the first round of presidential elections in the case of runoff systems). I have not done a systematic count of systems in a while, but that seems to remain the case. Almost all semi-presidential systems, on the other hand, have nonconcurrent elections. In most cases, that is because the term lengths of the two institutions are different, not because of any systematic design principle in having legislative elections come ‘x’ number of weeks/months before/after a presidential election. These democracies with different term lengths thus may have the legislative election at almost any point in the presidential term, varying by presidency. ((There is some variation in these systems in whether they are on the same date when they come up in the same year. For instance, in El Salvador (5 years presidency, 3 legislative), the legislative and first-round presidential election are concurrent every 15 years (next in 2009). But when France had both elections in 2007, they were on separate dates, as was the case under the (pre-coup) Chilean constitution of 1925.))
The result of most electoral cycles being either concurrent or for completely separate term lengths is that there are relatively few cases of legislative and presidential elections that take place in close temporal proximity to each other very closely (but not on the same day). In my grad-school work and subsequent publications I have referred to legislative elections that come shortly after presidential elections as honeymoon elections. The reverse situation, where legislative elections are followed closely by presidential, are counterhoneymoon elections.
The expectation–as suggested by the term–is that a honeymoon legislative election would result in a boost to the legislative representation of the newly elected president’s party (or bloc). And, almost without exception, that is empirically the case. One of the rare exceptions is from one of this year’s East Asian cases: In South Korea, 1987, there was almost complete stasis between the vote share obtained by the president and that obtained by his party at the legislative election shortly after. It was, of course, a three-way election, with the outgoing dictatorship’s candidate winning on 36% of the vote. So, perhaps the honeymoon effect does not work with “departing” dictatorial parties, or perhaps it does not work with winners who have relatively small pluralities instead of majorities or near-majorities. It is hard to say because we don’t have a lot of cases and because we really do not have a good theory of the causal mechanism at work in the honeymoon effect (or its rare absence). Does it result from the excitement and mobilization of change, in which case it may require alternation (thus “excusing” the Korean case because the former dictator’s party was being reelected)? Calling it a “honeymoon” election assumes, of course, that there is a rally around the new president, even by those who might have voted for other candidates–something that clearly did not happen in Korea in 1987.
The counterhoneymoon concept suggests something quite the opposite of the honeymoon effect. Whereas the honeymoon election allows the new (or newly reelected) president to take advantage of his or her maximum public support and translate it into a politically supportive legislature, the counterhoneymoon election may be a signal of the relative support of various parties and factions leading into the presidential campaign.
Both effects, if present, would be expected to result in relative political harmony between the two branches, but through contrary mechanisms. In the counterhoneymoon phenomenon, the outcome of the legislative elections serves as a sort of “primary,” sorting out the alliances, parties, and factions that are best positioned to win the voters’ favor in the looming presidential election. The identity of the next president is not yet known (assuming a competitive campaign), but the various contenders may be the heads of parties or intraparty groups that are jockeying to be president. Obviously, in the honeymoon, the president is both known and just getting started, and the legislative election takes place before the opposition has time to regroup.
Normatively, then, a honeymoon cycle is “best” if one wants to maximize the chance that the president has political domination over the legislature. Is a counterhoneymoon cycle normatively preferred if one wants to make the presidency maximally dependent on the electoral sorting out of competing parties and factions? ((If, on the other hand, one wants to maximize the potential for divergence, one would presumably want legislative elections to occur only at the presidential midterm. The only such case I know of is the Dominican Republic since 1996 (although in the most recent example there, the president’s party actually scored dramatic gains).)) Perhaps. However, the cycle is so rare. Only Colombia has used it regularly, and it appears that the legislative elections have indeed served such a function, though I do not think the pattern has been systematically analyzed from this perspective. ((Before the introduction of presidential primaries by the Liberal party in 1990 and the leftist alliance in 2006, and especially when the quasi-SNTV system was used before 2006, the intraparty competition for legislative seats (and other offices before subnational elections were moved to a separate cycle) may have affected the presidential nominating process. And in years when the primary has been competitive, the various pre-candidates not only were competing to win the nomination, but also to develop their own “coattails” within the party. With the new electoral system being open list (parties may present closed lists, but most lists are open), this competition may be only somewhat attenuated, if at all. A research agenda going forward!))
What has all this to do with South Korea and Taiwan? Alas, not as much as I had hoped when looking ahead to these two countries’ nonconcurrent elections in the Decemberâ€“April period. The initial election of the current cycle in each country–legislative in Taiwan and presidential in Korea–was not competitive enough to be of much interest from this “electoral-cyclist” perspective.
In Taiwan in January, the incumbent DPP was defeated quite badly, and the opposition KMT and its allies wound up with three quarters of the seats. Not a lot of counterhoneymoon sorting out going on there. Nonetheless, there may have been some, on the intra-alliance dimension, as the KMT’s almost total dominance over its own allies may have altered the internal balance. A hypothesis to that effect would be somewhat more convincing if the allied People First Party had announced the opening of discussions of a possible merger with the KMT between the legislative and presidential elections, rather than immediately after the latter. That is, suppose the two parties had been jockeying for leadership of their alliance and had decided on their presidential candidate only after the legislative result was clear. As it was, the KMT presidential candidate, Ma Ying-jeou, backed by the PFP, won Saturday’s election, 58.4%-41.6%, an outcome that was not really ever in doubt. ((Despite the efforts of the ruling DPP, which takes a harder line on relations with (Mainland) China, to exploit pro-sovereignty sentiment by holding a concurrent referendum on the status of Taiwan/Republic of China and Taiwan’s possible admission to the UN. There was also a KMT-sponsored referendum on the ballot, proposing a more “flexible” strategy. Both referenda failed.)) And certainly whether the candidacy would go to the KMT leader or that of the PPP was not in any way sorted out by the counterhoneymoon legislative election. In fact, given the change in the legislative electoral system for this most recent election (from mostly SNTV to MMM with a small proportional-list tier), most of the intra-alliance balance had to have been negotiated before the legislative election.
In the South Korean case, Lee Myung-bak of the Grand National Party won the presidential election decisively, 48.7% to 26.2% for his nearest challenger, Chung Dong-young of the United New Democratic Party. There may not seem to be much of a honeymoon increase possible from that large a victory. However, considering that the current president’s party won only 35.8% of the vote in the 2004 legislative election (when it was in opposition, having lost a relatively close 2002 presidential election), any substantial increase from that share by the party so soon after winning the presidency would indeed look like a honeymoon effect. We will have an idea of the extent of such an effect in just over two weeks. As for a good case of counterhoneymoon elections in action, Taiwan failed to give us one, and we may not have another example soon. That is unfortunate, because I have long thought that the counterhoneymoon cycle just might be the normatively preferred one.