Fruits & Votes is the Web-log of Matthew S. Shugart ("MSS"), Professor of Political Science, University of California, Davis.
Perspectives on electoral systems, constitutional design, and policy around the world, based primarily on my research interests.
Also experiences with growing many varieties of fruit (always organic) and other personal interests. Please see the Mission Statement for more. (There is also an explanation of the banner.)
Other "planters" have been invited to contribute. Please check the "Planted by" line to see the author of the post you are reading.
Join the conversation. Comments are always open. Except, that is, when Word Press mysteriously shuts them down, which happens with distressing frequency.
Abstract: Scholars and practitioners express concern that parties in “third wave” democracies are poorly developed, compared to parties in older democracies. We suggest that parties vary in their organizational “capacity”, focusing on parties’ ability to select trustworthy executive agents. Capacity is higher where parties can vet potential executive talent by observing future leaders over time in the legislature – an increasingly available option as democracy matures. The key distinction in parties’ use of this option lies in the delegation structure between a party and the executive. Parliamentary systems offer a clear line of delegation, which parties control. In presidential systems, parties must recruit executive candidates who can win a popular election, requiring characteristics that may not be well correlated with those that make them good party agents. As parliamentary democracy matures, we find a steady increase in prime ministers’ average length of prior legislative service. For presidents, there is significantly weaker growth in prior legislative service. We also theorize about and investigate patterns in semi-presidential democracies. Our findings suggest that the institutional format of the executive is more important for party capacity in new democracies than the era in which a democracy was born. [↩]
Two electoral systems that use “nontransferable preference votes” are commonly used: single nontransferable vote (SNTV) and open-list proportional representation (OLPR). Both systems promote intraparty competition by vote-seeking candidates, but differ on the extent to which the incentives of individual candidates and collective seat-maximizing parties are aligned, or not. We develop “logical models” of expected vote shares of parties’ first and last winners, and test (and confirm) these models using “symmetric regression” on an original data set drawn from over 2000 party-district observations in nine countries. The analysis helps bring us closer to an understanding of the relatively neglected “intraparty dimension” of representation, and allows us to offer some modest suggestions for improving systems of nontransferable preference votes. [↩]
Democratic representation involves tradeoffs between collective actors – political parties seeking to maximize seats – and individual actors – candidates seeking to use their personal vote-earning attributes (PVEAs) to maximize their own chance of election and reelection. We analyze these tradeoffs across three different electoral systems used at different times for the large-magnitude nationwide tier of Japan’s House of Councillors. These electoral systems – closed and open-list proportional systems and the single non-transferable vote – differ in the extent to which they entail candidates seeking individual preference votes and in whether collective vote shares affect overall party performance. We use local resources as a proxy for PVEA and seek to determine the extent to which parties nominate “locals” and how much the presence of such locals affects party performance at the level of Japan’s prefectures. [↩]
Vanuatu, one of the last cases of the Single Non-Transferable Vote, held general elections earlier this month.
Jon Fraenkel of Victoria University in Wellington, New Zealand, sends along the following tidbit about the challenges of vote distribution under SNTV:
Ralph Regenvanu, a reformist MP elected last time out as an independent but this time around seeking to establish a political party, contested the 6-seat Port Vila constituency and tried to get a running mate elected, but apparently failed to divert votes away from himself to that running mate, so a load got ‘wasted’, and his colleague failed.
It looks as though Jordan is going to adopt some form of list PR. David Jandura, writing at Awha Talk and The Monkey Cage, has the details.
If this change happens, it will mean saying goodbye to yet another SNTV system. On the other hand, as best I can tell from David’s description, SNTV was de-facto already abandoned as of the most recent election. In that election, they used a rather odd system of “ghost districts” that I am not sure that I really understand; it seems as if each wider electoral district was subdivided into M sub-districts (where M is the district magnitude), and that each candidate had to beat out only the other candidates in the “ghost” district to win. In other words, it was mechanically FPTP, as the winners would not necessarily be the top M in votes over the wider electoral region. The twist is that no one actually knew which candidates were competing against which other ones for a given seat–that’s the “ghost” aspect. Weird.
Monica Pachon and Matthew S. Shugart, Electoral reform and the mirror image of inter-party and intra-party competition: The adoption of party lists in Colombia. Electoral Studies 29, 4 (December), pp. 648-660.
Abstract:
The Colombian case offers a rare opportunity to observe effects of electoral reform where districting remains constant. Only the formula changed, from extremely ‘personalized’ (seats allocated solely on candidate votes) to ‘listized’: seats are allocated to party lists, which may be either open or closed. Electoral reform has effects on both the inter-party dimension (the number of parties competing) and the intra-party dimension (the extent of competition within parties). Consistent with theoretical expectations, the inter-party dimension features an increased number of parties in the low-magnitude districts and a decrease in the high-magnitude districts. On the intra-party dimension, the impact “mirrors” the inter-party: less competition in smaller districts, yet more in larger districts.
If you have access via an academic library, you can read or download the article from Science Direct.
In the 2008 election, the largest party won 11 of 52 seats. So I imagine holding a coalition together is not so easy. (It’s a pure parliamentary system, with a ceremonial president.)
I have a more-than-passing interest in Vanuatu (pop. 221,552), because it is one of a small (and diminishing) number of countries to use SNTV for its national legislature. I even have Vanuatu data in a forthcoming paper on candidate vote distributions under SNTV and OLPR. So it’s nice to see the country’s politics in the news.
The results of September’s election for the Afghanistan legislature finally have been released. From the CSM:
Early analyses of the final results show that the Hazara community may have snagged a share of the lower house that represents as much as double their actual proportion of the population.
In Ghazni, the last remaining constituency to be counted, preliminary results indicated that all 11 seats went to Hazara candidates, even though the province has a slim majority of Pashtuns with significant Hazara and Tajik minorities. [...]
Wardak province also saw a surge in Hazara representation. Though the region is predominately home to Pashtuns, three of the five seats went to Hazaras.
The story emphasizes the impact of violence on the lack of ethnic proportionality: if turnout by Pashtuns was lower due to their regions being more violent, then other groups would be over-represented.
While violence may well be the main factor, it is worth remembering that the electoral system is single nontransferable vote (SNTV), which is not a proportional system. If Pashtuns had their votes less efficiently distributed across their candidates than did other ethnic groups, for whatever reason, then the result could be disproportional regardless of turnout differentials.
Frequent commenter Bancki noted the following about the Jordanian election of this past week:
When you think you’ve seen it all, Jordan invented a new hybrid electoral system: SNTV with virtual sub-districts: on the one hand, every voter has only one vote for one candidate in his multi-member-district (SNTV), but on the other hand, the district is divided in as many ‘virtual sub-districts’ as there are seats and every candidate stands in the sub-district of his choice. Not the M highest vote-getters are elected (SNTV), but the winner in every sub-district.
I have no idea why this sub-district complexity was added (who enacted the electoral law change anyway?), but it seems to me it’s more difficult (compared with simple SNTV) for a well-organised minority (say worth a Droop quota) to get a candidate elected: if their opponents know in which sub-district the minority concentrates its votes, the majority can overrun the concentrated minority in that sub-district, while winning on low numbers in other (less contested) sub-districts.
Even though someone who was following the Jordanian election closely sent me a detailed description which I was traveling this past summer (and which I subsequently lost), I do not understand this odd twist, either. Maybe someone can enlighten us. The NDI report is not exactly clear on this point. For example:
The new law preserved the single, non-transferrable vote system, which has been controversial in Jordan as some argue that the system favors tribal voting over the development of political parties. It also increased the number of seats in the lower house from 110 to 120, adding four seats for heavily populated areas in Amman, Irbid, and Zarqa, as well as six new quota seats for women. [...]
Jordan’s government tried to address a long-standing complaint about Jordan’s single non-transferable vote system (often described as “one man, one vote”) with the creation of “virtual” sub-districts. In some polling stations, the candidate lists were broken down by sub-districts while in others only the overall candidate list was displayed. Voters had to make their choice without knowing the full list of competitors in each sub-district. This system should be improved or changed for future elections.
If the candidates are actually only in competition with other candidates in a given sub-district, then how could the system be considered SNTV at all?
Finally, a terminological issue. The reference to “one man, one vote” as SNTV is odd, unless one realizes that the alternative (used in some past Jordanian elections) is MNTV. Of course, the term, “one man, one vote” normally refers to an absence of malapportionment, not to the number of votes per voter. If districts had equal voter populations, it would be “one man, one vote” whether it was MNTV, SNTV, or some other system, because all voters would be represented equally.
There is still malapportionment, albeit less than before:
One of the most significant features of the electoral context in Jordan remains the disproportionality among electoral districts. The underrepresentation of urban, largely Palestinian-origin voters, has long been an issue of political contention.
Of the (too) many offices up for election in my area this November, one of the most puzzling is the Ramona Community Planning Group. This is an elected advisory body to the County Board of Supervisors. Ramona is a relatively large community, but is unincorporated.
The Planning Group (can’t they call it a board or a council rather than a group?) consists of 15 members, all elected at-large. In other words, there are no districts. In alternate biennial elections, either eight or seven are up for election. This is a “vote for no more than seven” election. It is a nonpartisan race. The only identifying information on the ballot regarding the candidates is their (self-described) occupation.
So here we have an interesting electoral system. The district magnitude is seven, and the candidates with the seven highest vote totals will be elected. It is a good example of multiple nontransferable vote (MNTV).
MNTV is often called “block vote,” but it really only functions that way if there are, in fact, identifiable “blocks” of candidates in the race and voters who tend to vote “in block.” In other words, if there are de facto parties, which have loyal voters who will go to the trouble of giving all their M (here 7) votes to candidates of the block. Otherwise, it may be more like the limited vote, with many voters using fewer than M votes.
I will certainly be one of those “limited” voters, as I can’t figure out seven candidates I would want to vote for. It is not for lack of choice. There are twenty candidates. But information is somewhat scant. Only six of the candidates submitted statements for the ballot pamphlet. Not that these are ever terribly informative. (One can track down another four on Smart Voter, but information is not extensive there, either.)
There is, however, a “block” within Ramona. It is called Citizens for a Rural Ramona (CFRR). Sounds good, given the character of the region. However, it is a classic NIMBY special-interest organization, comprised mainly of property owners in the vicinity of a proposed road extension. The extension, which would relieve traffic congestion on other streets, has been on the County planning maps for many years, but now that construction is set to start, a neighborhood group is organized to try to take control of the Planning Group.
CFRR has endorsed ten* candidates (overnomination!). Given their organization, they stand a good chance of electing several of their candidates. If their supporters have sufficient “blockness” in their use of votes (using all seven of their votes and voting only for candidates from the endorsed list) they could fill all the seats at stake in this election, even if they are not a majority of the voters. And if they are a majority, they will still be over-represented, because their blockness is sure to exceed that of other groups of voters–many of whom, like me, may cast only two or three votes.
Classic MNTV.
(Three other MNTV races on my ballot are a lot less interesting. Each has M+1 candidates, M of whom are incumbents, with M ranging from 2 to 4.)
_______
* Their website says eleven, but actually lists only ten.
This Saturday Afghanistan holds its second legislative elections since the US invasion. Like the 2005 elections, these will be held with the electoral system that always appears near the bottom of electoral-system experts’ rankings of “best” system: the single nontransferable vote (SNTV).
Under SNTV, the winners are simply the candidates with the top M vote totals, regardless of their party (if any), where M is the magnitude of the district (the number of seats it has in the legislature). Afghanistan has a wide range of M; I have not seen if the magnitudes have been adjusted since 2005, but in that election, districts had anywhere from 4 to 33 seats each, with an average of around 7.*
In 2005, there were no party names or symbols on the ballots. In fact, there were no officially recognized parties at all. Since then, a political parties law has been passed, but a mere five parties have gained the official right to have their symbols on the ballots. So only a tiny minority of candidates will be identified by their party affiliation; the rest will be effectively independent candidates, regardless of whether they in fact have a party affiliation. See Thomas Ruttig at the FP for detail about the parties and the registration process.
Given that SNTV is a party-less electoral system in terms of the process of seat allocation, one could wonder what additional value party labels on the ballot would offer. To vote in SNTV, for any party that has more than one candidate in the district, the voter must know the candidate that he or she favors. Compared to any proportional representation system that uses party lists, or a first-past-the-post system that uses single-seat districts, knowing the partisan identity of candidates is relatively less important.
Key facts about the political consequences of SNTV are:
1. SNTV puts a premium on personal connections (e.g. being a local notable of any kind) rather than party reputation; and
2. SNTV practically guarantees tiny margins of votes between the last few winners in a district and the first few losers, especially in districts electing more than about 5 or 6 seats.
In other words, whether candidates are identified as party nominees or not, it is personal reputations that count above all else. Those personal reputations could be derived from incumbency if the member has stood above others in delivering services or benefits to the region, or from outside electoral politics, such as from being tied to (or being) a local warlord or chieftan. Or it could be a reputation from business or some other pursuit outside politics. What SNTV does not reward, in general, are candidates who try to provide broad public goods or run on ideological appeals.
_______
* There is a gender quota, which does not fundamentally alter the dynamic of SNTV; it mere stipulates that a minimum number of the winners must be women, even if some men had higher vote totals. In a sense, it is two parallel SNTV contests in each district, with one of them reserved for women.
According to the report, these 34 items are “the major points of consensus among Afghan civil society organizations, international observer missions, assistance organizations, and independent election experts.” Notable actors included various UN bodies, ANFEL, the local AREU, various EU groups, IFES, NDI, the OSCE, and so forth. If you want to see all 437 recommendations that those groups made, visit DI’s Afghanistan website.
Recommendation number one:
The use of the SNTV system should be reconsidered: There is broad agreement that the Single Non-Transferable Vote (SNTV) system impedes the development of political parties and prevents fair and accurate representation of Afghanistan’s diverse population. A public consultative process should take place to solicit the opinions of relevant Afghan actors and international election experts to determine the best alternative system for Afghanistan. One alternative which has been consistently presented is a mixed SNTV-proportional system.
Afghanistan clearly does not host a model party system. Yet I wonder whether the ‘strong’ parties that might result from more party-centric electoral rules would be all that great. If, for example, closed-list PR turns divided societies’ elections into “national identity referenda,” would programmatic coherence and party discipline be such great ideas?
It’s nice to see consensus emerging on some form of system that retains a role for the personal vote, whether through an SNTV tier as alluded to above, or maybe through OLPR, as belatedly used in Iraq. This is because I believe that most voters prefer moderates to extremists. Therefore, when a country’s best organized political leaders are extremists, institutions should be used to diminish their control over ballot access and rank.
The verdict on this theory, of course, is still out, but I’m working on it.
A “well informed source” relates the following notes about elections in the Palestinian Territories:
1. They have been having rolling local elections but for some reason they decided to use list PR but didn’t force candidates to link themselves to parties either during nomination, or on the ballot. Thus, when they tried to allocate seats…errr…they couldn’t. So they had to go back to candidates and ask them which party they were, but most refused to tell them, so [and this is what another well informed source told source no. 1] the election commission just did it themselves (literally saying “we know who you really are aligned with) and just made parties themselves – fabulous!
2. The current Hamas central council elections are the block vote [what we call MNTV here at F&V], with 100+ to be elected, 600 candidates and you MUST use all votes.
For Kuwait’s upcoming parliamentary election, the government is instituting, for the first time ever, a campaign against buying or selling one’s vote.
During a recent broadcast on Dubai TV (which I saw via Link TV Mosaic), this poster was shown. I do not read Arabic, but I gather that these voters are happy because they have neither bought nor sold their vote.
Kuwait had parliamentary elections just a year ago, using a form of multiple non-transferable vote (MNTV). Apparently voters were “limited” to voting for fewer than M candidates, where M is the magnitude of the electoral system. That was a change from the previous “unlimited” vote, in which each voter could cast up to M votes, and the candidates with the top M vote totals in the district would win. M-votes MNTV is often misleadingly called “block vote.” However, in the 2003 election when this system was still in use, there was not a lot of “blockness” of voting, which I suggested could be calculated from the ratios of winners’ and losers’ votes.
Non-transferable vote systems–whether voting is “limited” or not–are personalistic, and thus provide the context in which vote-buying can be expected to flourish, absent effective controls of one kind or another. Limits on the number of votes per voter (with SNTV being the limit) would not exactly seem to decrease the personalism that often breeds “vote-buying,” especially in the absence of legally recognized political parties to structure the campaign.
Kuwaitis are going back to the polls so early because the emir evidently did not like the notion of a parliament that would question the prime minister. So he dissolved it.
My latest paper, now under review, is now posted at my Working Papers page (scroll down, past the list of “Research projects in progress,” for a link to the paper in PDF).
Title: ‘Party System Rationalisation and the Mirror Image of Interparty and Intraparty Competition: The Adoption of Party Lists in Colombia” (co-authored with Mónica Pachón).
Abstract: We analyze the adoption of a party-list electoral system in Colombia, replacing a former highly ‘personalised’ system (in which seats were allocated solely on candidate votes). Consistent with theoretical expectations, there were two major changes. First, the party system was ‘rationalised’: the number of parties competing is now more consistent across districts. Formerly one-party dominant districts now feature interparty competition, and districts that formerly had numerous tiny parties now have fewer parties. Second, changes in the intraparty dimension ‘mirror’ those on the interparty dimension: where interparty competition increased, intraparty competition decreased, and vice versa. The results are important to both practical reform efforts and furthering knowledge on the effects of electoral systems and their reform.
Comments on the paper are invited, either here or at one of my e-mail addresses.
This planting inaugurates a new orchard block of work in progress that I hope to use regularly as a way of announcing new postings at my academic website.
And, of course, congratulations to Mónica, who just defended her Ph.D. and is now starting her new academic job at the Universidad de los Andes!
It has been widely reported that in Kuwait’s election, “radicals” gained significantly and one “westernized” candidate narrowly missed becoming the first woman elected to the assembly. Less reported is what the electoral system is.
If it is the same as it was in 2003, the 50 seats are allocated in 25 districts by MNTV. [UPDATE: In a comment, Bancki notes a major change in magnitude; formula apparently remains MNTV, but in its "limited vote" variant.] That is, voters have [or, rather, had, in 2003] two votes, and the top two candidates win. There are no real parties,1 so it is a near-perfect example of personalistic voting and seat-allocation. A quick glance at the 2003 elections (Psephos) shows why this electoral system’s common name–block vote–is misleading.
When voters have M votes, where M is the district magnitude, and the winners will be the candidates with the M highest votes, it is an open question whether there exist “blocks” and whether, if they do exist, voters will vote as if they recognize those blocks as electorally relevant.
If there are blocks–parties, or just teams of local notables who campaign for votes in block–then the top M winners should have identical vote shares. So should the first M losers, the next M losers, etc., if there are multiple blocks contesting the seats.
For Kuwait, or any similar electoral system, we can get a window into the block vs. personal-voting tendencies of voters by simply calculating the ratios of candidates’ votes. Divide the second candidate’s votes by the first, the third by the second, etc.
When M=2, this will produce three ratios of interest: Second winner to first winner (SW:FW), first loser to second winner (FL:SW), and second loser to first loser (SL:FL).
If there is a block-voting tendency within the electorate, SW:FW and SL:FL should be close to 1.0. There is no expectation about FL:SW, because it depends not on intra-bloc cohesiveness but rather on inter-block competitiveness (i.e. the closer it is to 1.0 the more the district has close competition between two blocks–and, of course, the more any slackening of the leading block’s cohesiveness can contribute to the second block’s winning one of of the seats).
In Kuwait, 2003, the mean ratios across the 25 districts were:
SW:FW .858
FL:SW .837
SL:FL .827
Maybe some blockness there, but not much: If there were blocks, we might expect closer to 1.0 for at least SW:FW and maybe also SL:FL (if there are two blocs contesting the election).2 Standard deviations will also tell us something here. If .85 is a typical degree of achievable intrabloc unity, and the winners in each district tend to come close to that, then the standard deviation should be fairly small on SF:FW. On the other hand, if there is variation in inter-block competitiveness across districts, there would be a relatively high standard deviation on FL:SW, as some districts have two candidates well in the lead, while others have three or four bunched closely. The standard deviations are, in fact, .101, .126, and .191. So the top two in a district do indeed tend to be less variable in their ratios, which implies there might be some block tendency, after all, but just that it is hard to get much better than the .85 to .9 range.3
The first two losers’ ratio is more variable than even the FL:SW ratio, implying that it may be harder for the opposition to coordinate its votes–which is a pretty standard finding for this type of electoral system. That is, for electoral systems that are based solely on nominal and nontransferable votes.
Obviously, a leading block with two candidates relatively unchallenged does not have to worry about non-equalization between its two candidates, which might be caused by differential personal votes (or simple voter laziness). There seems little evidence for a notion that the leading block might put more (successful) effort into equalization when a third candidate in the district is more challenging,4 but the absence of party labels makes it impossible to know for sure!
Special note has to be made of the Al-Rumaithiah district, where the top two candidates had vote totals only seven votes apart! Then again, it is impossible to know if that is two candidates from one “block” that had voters generally willing to vote for both, or if it was two candidates from opposing blocks who had just enough personal support to squeak to victory over the third candidate, who missed one of the seats by 98 votes (out of over 11,000 cast). (The fourth candidate missed the second seat by a whopping 518 votes!)
At the opposite end of the spectrum was Al-Salmiah, where the first winner was far above any other candidate (SW:FW ratio of .68), but three other candidates were closely competitive for the second seat (their ratios to the candidate ahead were .86, .88, and .98).
In the absence of cohesive parties and strong party identification in the electorate, these MNTV systems can be rather unpredictable. They function as if they were pure personal-vote systems because, from the mechanical standpoint of how seats are allocated, they are. Blocks are irrelevant to actual seat allocation, and only through careful nomination and campaign strategy–and voter willingness to go along–can they function as block plurality.
What this might actually mean for Kuwaiti politics–or for 2008 (remember, the data I worked with are 2003)–I do not know. Did the “radicals” do relatively well because their ideological cohesiveness allowed them to place in the top two in 12 districts (they are reported to have won 24 seats)? Or did they have one popular candidate in almost every district whose popularity led some pro-government voters to vote for only one of the government candidates and give their second vote to a charismatic radical? Either outcome is plausible in such a system (though the first a good deal more so), and the final results will give us some clue. As for the “westernizers” they probably simply aren’t sufficiently organized to do well under this sort of system.
_______
“Political parties are banned in Kuwait, but various groupings operate as de facto parties,” as the AFP item linked in the first paragraph puts it. [↩]
Ten of the 25 districts have ratios of the top two candidates that are more than .9. [↩]
If the top two were assumed to be from the same block and to have ratios closer to 1.0 when the first loser’s ratio was also close to 1.0, it would point in that direction. But that is very definitely not the case, because some of the highest and lowest ratios for the top two candidates occur in districts with FL:SW>.9. [↩]
As I write, democracy assistance groups are helping lawmakers develop an electoral system for Iraq’s 18 governorate councils. Some creative electoral engineering could take the sectarian sting out of Iraq’s party system. One proposal worth serious thought is using the single non-transferable vote (SNTV) with open endorsements in governorate-wide districts…
A party system that continues to revolve around sects will not help consolidate Iraqi democracy. Luminaries from Lipset to Lijphart have taught that stable democratic politics are about more than race, religion or language. The challenge is to get Iraqi elites talking about more than sectarian interest. What candidates need are incentives to cultivate a personal vote.
SNTV certainly does that, and I thank Jack for the link to a previous F&V planting on those incentives. I would, however, take exception to Jack’s suggestion that political science “luminaries” have said that
Campaigns need to be about what’s-in-it-for-me: jobs, schools, roads and, as a colleague quipped, a shawarma machine in every kitchen.
The case can be made that such a politics oriented around individual legislators’ credit-claiming is a salve to sectarian tensions in countries where party-oriented politics necessarily means sectarian-oriented politics. I am just not aware of that argument having been made by any of the scholars Jack alludes to.
If by my laws you walk, and my commands you keep, and observe them,
then I will give-forth your rains in their set-time,
so that the earth gives-forth its yield
and the trees of the field give-forth their fruit.
--Vayikra 26: 3-4
F&V time: This blog's date function is so set as to start a new day at approximately local sunset.
(Why, if we have "day" and "night," should a new "day" start in the middle of the night?)
FRUITS: Support your local, organic growers; and, plant vines and fig trees and pomegranates for the generations to come...
VOTES: For democratization and full representation, for environmental sustainability, social justice, and peace, always sincerely...
