Liberia: Would open-list PR, a la Somaliland, be better?

One of my emerging academic (and real-world) interests is in institutions, and the possibility for institutional engineering, for fledgling ethnically divided democracies. I say “emerging” because, while I have been publishing on electoral systems and constitutional design (and occasionally revolution) for many years, I have not turned my academic attention specifically to ethnically divided states thus far. However, I am doing some sessions devoted to this topic in my Policy-Making Processes course in the next couple of weeks, and I will be teaching a new course specifically on this topic in the spring of 2006.

Thus I am most grateful to The Head Heeb (Jonathan Edelstein) for calling my attention to the electoral system used in Somaliland for its recent legislative elections. Edelstein describes this system in a way that makes clear that it is open-list proportional representation, and indeed it might be a better solution for a country like Liberia than the single- and dual- member districts used this year, about which I wrote previously. It is also almost certainly an improvement over closed-list PR, which Edelstein plausibly (though incorrectly) assumed is what I was advocating in my earlier post on Liberia.

In my post yesterday on Liberia’s fragmented congress, in which most legislators were elected with minimal voter support and parties will be represented with almost no connection to their electoral support in society, I noted that this outcome creates the following problem for executive-legislative relations, and thus policy-making, in the coming term of President-elect Ellen Johnson-Sirleaf:

Johnson-Sirleaf’s party caucus will be highly concentrated regionally, much more so than is its actual voter support. This will force her to buy support with patronage in order to form legislative coalitions to a far greater degree than if the system were proportional.

As far as what other institutional options there might have been, I left the matter hanging, other than to suggest proportionality would be beneficial. Even though I suggested that proportionality would be sensible, I concur with Edelstein that in situations of weak parties, PR has its disadvantages. As I always tell my students, any institutional choice has tradeoffs. There is no such thing as an institutional solution to all of a society’s governance problems. There are only solutions to some sets of problems that minimize the creation of other problems, for any given context.

On the idea of proportionality causing problems in contexts like that of Liberia, Edelstein notes, in the comments to my previous post:

Liberia, like many countries in similar situations, has a weak party system – there are lots of parties, but there isn’t much ideological distinction between them and most of them are personality-based. If most of the parties are vehicles for regional bosses, then party-based PR will elect the regional bosses, and will reinforce existing ethnic and feudal power structures. [my emphasis]

And he developed a very similar point at his blog later on (to which I linked above).

He is right, of course. Weak political parties are a problem, and the adoption of “party-based proportional representation” (as I called it) does not guarantee that strong parties will emerge. In fact, what it does is ensure that the various lists of candidates, however constructed, will be represented in proportion to their votes obtained over whatever territorial units the allocation is carried out. Such units must have multiple seats, but may be smaller than the entire country. In Liberia, presumably the districts would be the 15 counties, with varying magnitudes (seats per district) in the lower house, but retaining the current equality of representation for each county in the senate.

The point about which I was not clear is that one does not need to have closed lists, whereby the voter only selects the list, but has no choice over candidates. It is not one or the other (candidates or parties), as is too often assumed. And, as Edelstein notes, Somaliland uses a system that fully qualifies as proportional representation, but in which the candidates on the lists are ranked by votes received as individuals, and not by whoever draws up the list. This is an open list. (As an aside, it is fascinating that Somaliland reverted back to the type of electoral system that was used briefly in the 1960s when Somalia had a previous experience with democracy, and that the open-list concept came to Somalia from Italy, the former colonial power for the larger nation of Somalia, though not for Somaliland, which was a British colony.)

In a closed list, voters essentially delegate to some entity the decision as to which legislators will represent them. The entity might be an organized programmatically defined political party, or it might be a regional (or national) notable who has enough name recognition to draw up his own list and sweep into the legislature several cronies thereby fully dependent on him for their election.

In an open list, the specific candidates to be elected–among those nominated by the list-drafting entity–are determined by a rank based on their individual popularity.

Both forms of PR (and there are intermediate hybrids, too) ensure that the lists as a whole are represented in a way that is proportionate to their votes (in any given multi-seat district, and up to the limits imposed by the magnitude of the district). The difference is in the choices provided to voters (just a list, or a candidate within a list), and thus in how the candidates relate to the voters.

Either type of PR would address the problem of the regionalization of a party’s caucus being exaggerated relative to its distribution of popular support–i.e., one of the problems I was referring to in my original post. Either type of PR would address the severe wasted-vote problem (in the average Liberian county, a majority of the votes did not help elect anyone), by rewarding parties (of whatever form and strength) according to their actual votes, and not according to how they happen to be concentrated within unequally populated districts or how well they coordinated amongst their own candidates (in the 2-seat districts of the Senate) or with other parties (in both houses). Proportional-representation systems are not the only systems that would address these problems. SNTV (as used in Afghanistan, for example), for all its other flaws, would do so as well, and without the potential new problem that open lists would create. (PR, possibly including the open-list variety, was also seriously considered in Afghanistan, but rejected.)

A key problem with open lists is that the “party” may be no more than a collection of greater and lesser notables who share only a weak connection to one another, based only upon the convenience of maximizing one another’s legislative representation by pooling votes on a list. In fact, open lists create incentives for list-drafters to bid for candidates who are popular enough to bring votes to the list. If the list is being drafted by a party concerned with its ideological reputation, then its drafters are not going to put just any candidate with a block of voters on the list. But when there are weak parties to start with, open-list PR most certainly is not going to make them stronger, even though it does generate incentives for notables to work together to a degree that plurality voting (as in Liberia) does not do.

As Edelstein argues at his Head Heeb post on Somaliland, parties there have some degree of ideological distinctiveness–probably greater than is the case in Liberia. If so, then open lists may be less suitable in Liberia than they are for Somaliland.

Yet another possible family of systems–which are true hybrids of candidate and party-based representation–would be some form of mixed-member system. It would have the advantage of retaining single-member districts and plurality voting, but would “top up” a party’s total representation with seats allocated based on a party vote. I would advise MMP (as in Germany and New Zealand or the similar “AMS” systems of Scotland and Wales), not the more disporportional MMM (used in Japan and Thailand, among other countries), because MMM (or ‘parallel‘) systems can be quite erratic in their votes-to-seats conversion. However, once again, if one is concerned about an electoral system that priveleges allocation to lists in a context where the parties that would draft those lists are weak, them MMP might not be the solution. Alas, there is no best system, and the choice needs to be shaped to the context. The system used in Liberia, however, comes pretty close to being the worst, for the way it exaggerates the regionalism problem and wastes so many votes.

So, the bottom line is that I agree with the difficulties that PR (closed, open, or mixed-member) would entail in a situation of weakly organized, regionally concentrated and personality-based parties like in Liberia. I just don’t see how these problems would be worse than the actual result of the plurality system used.

I am most grateful to Jonathan for pushing me on this, and also for ensuring that Somaliland will be a case I will look at more closely now, and probably include as an example in my spring class.

Epilogue: Jeremy (a former student of mine, I might add) reports that the “Elections here have really been some of the most transparant and well organized that I have seen.” And he is an experienced observer of elections.

Liberia election: Africa’s first woman president will face a deeply divided legislature

[UPDATE, Nov. 12: Don’t miss Jonathan Edelstein’s thought-provoking comment to this post; I responded to Jonathan in a subsequent post: Liberia: Would open-list PR, a la Somaliland, be better?]

Liberia only sporadically gets noticed in the United States, despite the deep historical ties between the US and that west African country, founded by freed US slaves (supposedly) in 1847. There is much good news in this past week’s completion of a round of quite free elections. Most of the news and the relatively limited blog focus (e.g. Gateway Pundit, Mensa-Barbie, The Head Heeb, and others) is on the presidential race, but most of my post here today will be about looking ahead to relations between the new president and the congress that was elected last month. The outcome of the congressional elections offers lots of reasons for being cautious about progress towards democracy in that country.

But first, the presidential race. Results are still not complete from Tuesday’s presidential runoff, but that has not stopped former Finance Minister Ellen Johnson-Sirleaf from claiming victory. She will be the first female president in African history. She defeated former international football star George Weah. The divide between the candidates appears to be about 59–41.

Weah had led after the first round on October 11, though with only 28.3% of the vote. Johnson-Sirleaf had 19.8% on the first round, and other candidates trailed with vote percentages of 13.9, 9.2, 7.8, and the rest scattered amongst many others.

Weah seems to have forgotten that often the team that leads the first half does not win the game. He is alleging fraud (though international observers disagree), and today there were clashes between his supporters and UN peacekeeping forces.

As I noted in my previous post on the first round of the Liberian elections, the new president will face a deeply fragmented bicameral congress. Liberia’s system of government is pure presidential, meaning President Johnson-Sirleaf will have the freedom to appoint a cabinet of her choosing, but also meaning that negotiations with such a fracuted legislature will be difficult.

The results of the legislative election–held at the same time as the first round of the presidential contest–show how highly regionalized and personalized electoral politics is. Without any party component to the allocation of seats, “local notables” prevailed in the candidate races for House and Senate seats. Many of these elected notables are closely tied to ousted former dictator Charles Taylor, and some are even his family members. Both candidates in the presidential runoff were courting the pro-Taylor vote. In other words, this election hardly presages Liberia’s turning a corner from the corruption and violence of the Taylor years. Maybe rounding a gentle curve, but not turning a corner.

A closer look at the results reinforces just how fragmented and non-party-based electoral competition in Liberia is.

Both houses of the legislature are elected by plurality. In the case of the Senate, each of the 15 counties elects two senators, with voters having two votes each. In the House, each county is divided into several single-seat districts, elected by first-past-the-post. The population disparities of counties (senate districts) are very large, and while the House districts are roughly apportioned by population, the key word is roughly. Some districts have under 9,000 voters, while others have over 20,000. In short, both houses are significantly malapportioned.

Although the vote by party is somewhat regionally concentrated, most candidates have won their seats with far less than a majority of the vote. The result is no real relationship between party vote strength nationally and the distirbution of seats in either house. For instance, Johnson-Sirleaf’s Unity Party (UP), with 19.8% of the first-round presidential vote, will have about 12.5% of the House seats and only 10% of the senate seats. Runner up Weah’s Congress for Democratic Change (CDC), which won 28.3% of the first-round presidential vote, will have just under a quarter of the House seats and also only 10% of the Senate seats.

In Senate races, in most cases each of the larger parties ran two candidates, but elected only one of them. In fact, in only four counties were both seats won by candidates of the same party. In three of those four, the two candidates won very similar vote shares (UP in Maryland, CDC in Montserrado, APD in Sinoe), indicating that those parties in those counties have party-loyal voters. In the fourth county in which both seats went to the same party (COTOL in River Gee), the first winner won 26.9% and the second only 10.7%. That implies a strong personal vote to the first winner, and little party loyalty in the elecorate beyond this senate candidate. In fact, a candidate for Johnson-Sirleaf’s UP came in a close third in River Gee with 10.3% of the vote and the party had another candidate with 5.1%. Had the UP been able to coordinate on one candidate, it could have won a seat.

There are numerous cases where such coordination could have made a difference in the outcome. There were few cases in which a party ran only one candidate in attempt to avoid splitting its vote. Of course, given how disparate the votes are in most cases for each of a party’s two candidates, it is unlikely that one candidate could have succeeded in claiming all the votes that were cast for the actual two candidates. The personal vote, rather than the party vote, is key here.

(And a cursory look at vote totals in counties across the two houses suggests a fairly significant number of voters did not even use both senate votes–yet another indicator of candidate- rather than party-based voting.)

The result of poor coordination is that the average county-level wasted vote (any vote cast for a candidate who did not win) is over 51%.

The House, with single-seat constituencies, does not show much better coordination, even though each party, of course, nominated only one candidate. Many of the winners have only around a quarter to 30% of the vote, and some had well under 20% of their district’s vote.

The result of malapportionment, regional districting, plurality voting, and poor party coordination is a congress that can hardly be said to be representative of the political divisions of the country, given the absence of any proportional representation. Those who defend the idea of candidate-based plurality representation in divided societies often claim that parties would only reinforce those divisions, and of course they have a point. But at least a party-based proportional system ensures that the various divisions in society obtain a policy-making bargaining power commensurate to their level of support. And the CDC and UP, in particular, do have voter support in most counties (even if it is highly variable by county), but this support will not be reflected in congress. Johnson-Sirleaf’s party caucus will be highly concentrated regionally, much more so than is its actual voter support. This will force her to buy support with patronage in order to form legislative coalitions to a far greater degree than if the system were proportional.

Moreover, the supposed advantage of the plurality system in settings of regional and ethnic divisions is that at least each legislator represents his or her region. But can that be said to be the case when each legislator typically has a personal constituency that is only around a quarter of the votes cast in his or her own district?

I wish Liberians well in building a democracy. But it is hard to be optimistic given not only the legacies of civil war and Taylor’s dictatorship, but also the political fragmentation and the institutions chosen to channel it into the policy-making process.

EPILOGUE: Regarding Liberia’s alleged founding by freed slaves, it seems the story is a bit more complicated than that; however, there is a good reason for why Liberia has a county named Maryland (h/t Jones Blog).

Liberia: A blog about the elections and difficult governance ahead

Liberia held legislative and first-round presidential elections on October 11, and one of the coordinators of an observer mission, working for the International Republican Institute, has a blog about his experiences: Jeremy in Liberia.

In one post, Jeremy says:

I believe that organizations like IRI and NDI are making a real, positive and meaningful impact on the transition. The other day these organizations organized the first ever presidential debate in Liberia’s history, with some of the true ‘heavy-hitters’ making an appearance.

As an academic, I have to say that the jury is still out on just how effective democracy promotion by groups like IRI, NDI, and their counterparts based in other countries, are. As someone who believes in the spread of democracy, often calls himself a political engineer (I have done some consulting at stages prior to several new democracies’ elections), I really want to believe that these efforts can make a difference. And I admire people who will do the hard work that Jeremy is doing now in Liberia. As he relates in several of his posts, and shows with some nice photos, it is not exactly a cushy assignment. But it is an important one.

IRI has just issued a preliminary statement about the elections.

There will be a runoff to determine the presidency some time in early November.

Quite apart from what observers check on—whether the election is procedurally fair—Liberia faces very difficult governance due to the obvious fragmentation of political forces in the country and a set of institutions—presidential system with majoritarian congressional elections—that do little to fairly represent and channel that fragmentation.

Preliminary results show a very fragmented field and a closer race for second place (i.e. between inclusion and exclusion in the top-two runoff) than between first and second. This is quite common for runoff systems, and is one of their Achilles heals: Who places second and who just misses qualifying for the runoff sometimes can be decisive for the ultimate result. George Weah, an international football star, won just over a quarter of the vote, while Ellen Johnson-Sirleaf, who also ran in 1997, won around 14%. Two other candidates are at 11 and 10 percent. I have been able to find nothing about the likely second preferences of the voters who voted for the two who just missed the runoff, but in these kinds of fragmented first-round fields, there is always the risk that one of those who just missed the runoff would prove a stronger challenger to the front-running candidate than the one who does advance to the runoff. So, while runofffs ensure that the ultimate winner has a majority and make it unlikely that an extremist would win, sometimes the winner wins more or less by default (see Fujumori in Peru in 1990 or Chirac in France in 2002; either outcome might have been different if the candidate who narrowly finished a close third instead of second had advanced to the runoff). Add in the low information of a first election after civil war, and the runoff system seems especially risky.

Compounding this risk is the use of majoritarian elections for congress in the context of such fragmentation and uncertainty. No legislative results are out yet (apparently), but the result is sure to show little relationship between votes and seats percentages and to make for difficult governance for whoever is elected president. A fragmented legislature in which no party has a majority yet the many parties are not represented in relationship to their popular support is the worst of both worlds. (And, of course, a majority of seats when the largest party has only a quarter of the votes would be even worse!) Proportional representation or some form of mixed-member system would clearly be a better choice.

The 1997 legislative election was proportional, based on the percentage of votes cast for each party’s presidential candidate—a very odd way to run a PR legislative election, though it has been done in some Latin American countries. In the 1997 election, one candidate, the notorious Charles Taylor, won around 75% of the vote.

According to Adam Carr (as well as other sources I have seen), the lower house (64 members) this time will be elected by plurality in single-seat districts. The senate (26 members, though another source says 30) is elected in two-member districts (though I am unsure of whether voters vote for a slate of candidates, for two candidates separately, or for one).

The term of office for these legislators is extraodinarily long: Six years for the lower and nine for the upper. As far as I know, those are the longest legislative terms in the world. (Nicaragua’s 1987 constitution originally called for six-year terms for its unicameral congress, but once the Sandinistas were defeated in the 1990 election, the term was shortened to five. Four or five years for lower or sole houses is typical; upper-house terms are rarely longer than six, though Chile’s are eight.)

So, here you have a legislature in which many parties will be represented with only weak connection to their popular support and a president who may not really be the majority favorite despite winning a majority runoff, and they have to work with each other for the next six years. That is, assuming democracy survives that long. I hope it does. But Liberia could hardly have picked a worse set of institutions to that end.