The devil is in the electoral system design

My colleague David Jandura, a current student in the Georgetown DG program, has a good post at Democracy & Society about electoral system choices in emerging democracies. He argues that broad categorizations like “mixed system” and “proportional representation” belie – perhaps deliberately – our expectations of those systems’ fairness.

In Sudan, low district magnitudes make proportional representation operate more like a party block vote system. In the Philippines, a maximum three-seat-by-party allowance vitiates that country’s reserve PR seats.

Give it a read.

Iraq’s parliament gets bigger

Under the 2009 electoral law,¹ there will be 323 seats in the Council of Representatives. This is an increase from 275 in December 2005. As in December 2005, most seats will be allocated on the governorate level. In that election, however, there were 45 seats allocated nationally to minority groups and parties failing to meet governorate-level thresholds.² This time, there are only 16 compensatory seats.

And, of course, the new electoral system is open-list proportional representation.

More from Iraq and Gulf Analysis, including the distribution of seats by governorate.

1. Score!
2. These thresholds were not formal, but arose as a function of apportionment.

Open lists for Iraq

Reidar Visser reports. More at Fruits and Votes.

I am surprised. Then again, the political leaders who agreed to this are unlikely to lose their seats under the new system. See MSS’ comment on another post, the essence of which supports my prediction.

I do not know yet whether the lists are fully open or just “flexible.” The lists will be open. Candidates will not need quotas of preference votes as they did in the January 2009 governorate council elections. Voters, however, will have the option of voting for the party’s pre-ordered list.

Preference voting for El Sal?

The other day I had the opportunity to attend a talk by Juan Carlos Sanchez from the Foundation for the Study and Application of Law (FESPAD), a civil society organization that seeks to reform El Salvador’s electoral system.

He opened by arguing, quite bluntly, that El Salvador has “one of the worst electoral systems in Latin America.” To demonstrate this, he pointed to a number of specific facets of the system, such as the lack of absentee voting, the politicization of the Supreme Electoral Tribunal (TSE), the laissez-faire approach to parties and campaign regulation, and feckless mechanisms for enforcing the rules of the game.

What struck me most about Mr. Sanchez’s talk was what he did not address: the actual processes by which voter preferences are translated into political representation.

El Salvador utilizes a system of closed-list proportional representation (CLPR), a system known for encouraging strong, often programmatic, yet sometimes deeply ideological and polarized parties, not unlike those found in El Salvador. Briefly, a closed list allows the party leadership to select candidates with little or no input from the electorate, while proportionality provides opportunities for parties to garner substantial representation without necessarily having to reach across the political aisle or even into the center aisle in order to acquire district-wide majorities or pluralities. While it must be recognized that the two main parties – Arena and FMLN – have made considerable strides in moderating themselves since the days of the civil war, they nevertheless remain deeply divided, so much so that many question the degree to which they will respect the legitimacy of a loss in the upcoming presidential election.

Of course, in a country where civil war wounds have not yet fully healed, and where substantial socioeconomic disparities remain a potent political reality, it would be misguided to attribute full blame for the country’s polarized politics to its electoral institutions. Yet, it seems reasonable to begin to question the degree to which this system may be exacerbating, or at least failing to ameliorate, the nature and dynamics of existing political divisions.

To be sure, PR has its virtues and it has been proposed as a means to alleviate the effects of deeply divided societies in a number of contexts. However, such proposals are almost always tied to the caveat of parliamentarianism and the assumption of more than two relevant political parties - two additional factors that would presumably contribute a more conciliatory executive, legislative coalition building and, by extension, a more consociational political atmosphere.

This model, however, does not reflect the political realities of El Salvador, where holdover Cold War manichaeism and deep class divisions have encouraged the emergence of two dominant parties, which are currently involved in a bitter, winner-take-all struggle for the powerful presidency.*

With this background in mind, I asked Mr. Sanchez whether anyone has ever recommended a move away from CLPR, towards a system that provides incentives for existing parties to moderate the selection of their candidates, and for individual candidates to soften their rhetoric, such as the Alternative Vote (AV) or the Single Transferable Vote (STV) (the latter would seem a more likely option for a country already accustomed to proportionality and multi-member districts). The virtue of these systems is that they allow voters to select not only their first choice, but their second, third, or however many candidates decide to run. If their first choice does not receive enough votes to win a seat, their second choice candidate then receives their vote. For this reason, AV and STV systems are both referred to as forms of Instant Runoff Voting (IRV). This can create strong incentives for parties and candidates to attract votes outside their traditional base by moderating their platforms, campaigns, and rhetoric, as they begin to recognize the value of being voters’ “next-best choice.” Given that upwards of 14 percent of the electorate remains undecided going into tomorrow’s presidential election, it seems plausible that there may be a significant underrepresented “center,” whose voice could serve as a force of moderation if amplified through one of these preferential systems.

To Mr. Sanchez’s knowledge, despite the near-universal recognition of the need for a less polarized political dynamic, no one has made such a recommendation. In fact, he confessed that he - ostensibly one of the foremost domestic experts on reforming the Salvadoran electoral system – was unaware of any electoral alternatives for diminishing polarization.

This response surprised me, and I was thus wondering if anyone out there reading this with knowledge of El Salvador or electoral systems has any insight with respect to this issue, especially since our computer time at the hotel is rationed, and opportunities for even basic research are extremely limited. Has anyone proposed a preferential model for El Salvador? Might it help temper the country’s polarized politics? Is it even a plausible option? To what degree are current power holders’ interests tied to existing procedures? Are there potential unintended consequences that one should consider? Might a simple shift from CLPR to open-list PR offer a less drastic means of achieving greater moderation, or might this have the opposite effect? Perhaps what is really needed is a focus on reforming the executive branch vis-a-vis other organs of the state, whether this means a move toward parliamentarianism or simply a curtailment of executive authority.

So many questions. With any hope, the conduct of the parties and their supporters during and after tomorrow’s election will make them all seem a little less relevant.

*The president of El Salvador is selected through a two-round system, which in other contexts has been credited for the success of more moderate candidates (according to the same logic of the aforementioned IRV systems). However, in tomorrow’s election, because none of the smaller parties have put forth candidates, it is understood that there will be no opportunity for a second round of voting.

Preferential voting is not too difficult

Several political scientists have advocated for preferential voting systems as conflict management devices in divided societies.¹ The most common criticism I hear is that they are too complicated. I do not believe this, at least where literacy rates are reasonably high.

Recent American experience with ranked voting systems shows that a little voter education goes a long way. Burlington, VT just held its second ever instant runoff voting election. Of 8,980 total ballots, only four were invalid. That’s a 0.04 percent error rate. Less fatally, only one voter did not use all five of his/her rankings, according to Rob Richie.²

Here is an example of the ballot voters used.

1. Donald Horowitz has advocated instant runoff voting/alternative vote. Ben Reilly and Andrew Reynolds have variously advocated and highlighted the single transferable vote.
2. UPDATE: Actually, the voter did not specify a first choice.

Iraqi governorate elections: thank the open lists

Widely lauded gains for secular forces in Iraq’s provincial elections last month were largely a function of the candidate-centric, open list proportional representation system used.¹

Following the certification of results today, Michael Allen for the NED writes:

People voted on the issues rather than according to identity, and for individual candidates rather than anonymous lists. The poll represents an important step towards consolidating the country’s fragile democracy, but the real test will come with national legislative elections later this year.

And:

Iraqis voted strongly against religious sectarian parties widely perceived to be corrupt and to have failed to deliver security and basic services. “No party in the elections ran with the slogan ‘Islam is the Solution’ since voters were much more interested in who could actually provide services at the local level,” writes the Washington Institute’s J. Scott Carpenter.

Finally:

“Iraq was once defined by sectarian tensions pitting Shiite against Sunni,” writes Shadi Hamid, director of research at the Washington-based Project on Middle East Democracy. “Now, intra-Shiite competition may take greater precedence.”

In sum, personalistic campaigns revolved around everyday governance issues, and there was competition among members of the same sectarian groups. The thread uniting each of the points above is the electoral system. Iraqis did not suddenly take governance to heart. We can assume voters were as concerned with service delivery in 2005 as they were last January. The open list proportional representation system, however, freed candidates to campaign on governance and reduced the costs for Iraqis of voting on those issues.

As this insightful paper points out, moving from closed to open list PR fundamentally changes campaign dynamics.

The 2005 national elections proceeded under closed lists, and sectarian parties dominated. There are institutional reasons for this. First, candidates in closed list systems must curry favor with party leaders for high ranks on party lists. Second, since voters choose among labels and not people, party leaders have incentives to appeal to sect and ethnicity in order to maximize vote (and therefore seat) shares. Third, because the combination of social division and party-centric electoral rules politicizes identity, voters who would have voted on governance issues instead support the party representing their group. After all, what if the other group captures the state? The outcome of the 2005 elections was consistent with these incentives: a system of disciplined parties organized around religious affiliation.

Open lists change the game entirely. A candidate’s prospect for winning depends on his personal level of popularity.² He or she has an incentive to campaign against co-ethnics or co-sectarians. Such a campaign is likely to focus on issues ‘below’ the level of the group. Therefore, it is more likely to focus on “who could actually provide services at the local level,” in Carpenter’s words.

Unless the system changes, national elections later this year again will use closed list proportional representation. It will be interesting to see whether the secular organizations emerging from provincial elections reproduce their gains nationally, or whether Iraqis get another “national identity referendum.”

1. Questions nonetheless remain about the finer details of that open list system.
2. Depending on the way personal votes contribute to the party’s total, or “pool,” a candidate’s success may even depend on being more popular than fellow co-partisans, with more personal votes raising his position on the party list.

Iraqis recognize need for change

According to Reuters, it seems as though many Iraqis have recognized the shortcomings of sectarianism as a basis for political organization due to factors outside of institutionally-determined incentives — namely, the problem of governance and basic administration*:

“Religious parties didn’t keep their promises. They exploited our problems,” said Safaa Kadhim, a teacher in Basra, reflecting anger voiced across Iraq towards the major parties, mostly founded along sectarian lines and seen by many as corrupt and self-serving.

“The voter must be more careful this time, and vote for someone who is deserving,” Kadhim said.

Polling evidence seems to suggest that Kadhim’s sentiment is shared among the broader population:

In an opinion poll by the government’s National Media Centre in November, 68 percent of those questioned rejected the use of religious appeals in the campaign and 42 percent said they favoured secular parties, while 31 percent supported religious parties.

I do not highlight the above to undermine the importance of Jack’s most recent post, which provides a valuable and insightful institutional assessment. I agree with its premise as well as its speculations — institutions matter, and OLPR seems to represent an improvement.

However, I do wish to point out that Jack originally advocated for not just any “candidate-centric electoral system,” but for a specific type of system, SNTV, and I agreed. While SNTV is known to encourage several unsavory consequences over the long term (from highly factional parties to clientilism and political corruption), these appeared palatable in lieu of possible alternatives — whether that meant continuing down the path of CLPR and accepting the long-term institutionalization of sectarianism, or ham-handedly banning religious-based political discourse.

But institutions exist in a world of perceptions, both of which can change over time — the former through decree and the latter through learning, as the Reuters piece illustrates. Perhaps we did not sufficiently consider the speed at which this latter process could take place, particularly amidst Iraq’s dire circumstances (the school of hard knocks, it seems, provides a quick education). If we had, the middle way offered by OLPR — which appears superior to the extent that it does not encourage the same problems of SNTV, avoids some of the pitfalls of CLPR, and can actually be sustained if societal demands for sectarianism are not too overwhelming — might have been more apparent.

I am not sure of the specific ways in which this lesson could be of value in terms of broader application. Recognizing the fact that a dynamic learning process takes place as institutions illustrate their opportunities, advantages and failings over time is one thing; predicting the direction in which this learning process will progress is quite another. At the very least, it serves as a humbling yet necessary reminder that, in our efforts to change the world for the better, we often operate with limited means and in uncertain environments (even those of us as bright as Jack). So, fellow DGers of Georgetown and other future policy shapers, take note, for I imagine it is far more pleasant to gain an understanding of this reality as a student than at any other point hereafter.

*As a side note, while the so-called “surge” and the stability that it has helped bring about seems to have flown under Reuters’ radar, it seems incumbent upon me to point out that this was in many ways an even more basic determinant in leading Iraqi politics away from sectarianism. Falling back on the immediate certainties and familiarities associated with primordial bonds can become an appealing prospect in the midst of chaos, and thus it makes sense that as order is established, this tendency would taper and longer-term priorities (i.e., issues of governance and basic administration) would come to the fore.

Philly Dem would engineer GOP off city council

From Philadelphia comes more news that proportional representation (and its cousins) isn’t just for progressives and minor parties.

While a coalition of the Cincinnati NAACP and local Republicans backs that city’s return to PR-STV, a Philly Democrat wants to boot the only two Republicans from city council by eliminating two at-large seats elected under limited voting.

Of 17 council seats, 10 are elected in single-member districts. Limited voting (here, two-vote MNTV) helps prevent Democratic electoral majorities from sweeping the remaining seven at-large seats.

The Democratic member’s apparent plan is to reduce district magnitude in the non-majoritarian tier, making it more difficult for the city’s few Republicans to win representation.

PR-STV on the ballot in Cincinnati

The Hamilton County, OH Board of Elections yesterday certified a petition by the Cincinnati NAACP to put proportional representation on the November ballot.

If the measure passes, voters will use the single transferable vote to elect a nine-member city council, renewing a 30-year good government experiment that ended in a vitriolic 1957 repeal effort:

The single transferable vote had allowed African Americans to be elected for the first time, with two blacks being elected to the city council in the 1950s. The nation was also seeing the first stirrings of the Civil Rights movement and racial tensions were running high. PR opponents shrewdly decided to make race an explicit factor in their repeal campaign. They warned whites that PR was helping to increase black power in the city and asked them whether they wanted a “Negro mayor.” Their appeal to white anxieties succeeded, with whites supporting repeal by a two to one margin.

I have tried recently to focus on international democracy assistance, but this could be a major development in the history of American democracy and world of electoral systems.

Today only Ireland, Northern Ireland, Australia, Malta, New Zealand, Scotland (h/t to James) and Cambridge, Mass. use STV (of the multimember variety) for governmental elections, so Cincinnati would add a case to that family.

Cincinnati is the next page in a long and underexposed history of election reform in America. From the Progressive Era through the Civil Rights movement, 22 US cities (or 24 depending on definitions) used PR-STV for local elections, many of which were in Ohio. The second to last experiment ended in 2002 with the disbanding of New York City’s school board.

While system performance varied by city and indicator, STV’s overall record was positive:

On the whole, from the available evidence, proportional representation seemed to have a beneficial effect on the cities that adopted it. It clearly produced more representative government and, where voters wanted it, a more diverse party system. Large increases in the number of effective votes were also enjoyed in these cities. It may not have resulted in the substantial increases in voter turnout that proponents predicted, but neither did it produce the increases in voter alienation that critics feared. And finally, even though PR city councils were often more diverse politically, this did not seem to impair their political efficiency or effectiveness.

Good sources for more specifics are Doug Amy’s site linked above and Robert Kolesar in Proportional Representation and Election Reform in Ohio, Kathleen Barber ed., OSU Press 1995.

STV seeks proportional results and maximizes ‘votes that elect’ by transferring votes in excess of a quota to voters’ next-ranked choices. With nine seats in Cincinnati, it will take 10 percent of votes to win each. There are different ways to transfer surplus, and Cincinnati would use the quasi-random “Cincinnati method.”

Who cares? American reformers, for one, but the ends they are pursuing should not be lost on the international democracy assistance community, which has engaged in electoral engineering from Afghanistan to Nepal over the past few years.

As Donald Horowitz, Ben Reilly and others have noted, STV (and its single-winner cousin) can benefit divided societies through the incentive it presents to campaign for second- and third-choice support outside one’s group. Because it’s a proportional system, STV prevents exclusion of significant minority groups, especially as the number of seats to elect increases. As a candidate-centric system, STV emphasizes entrepreneurial campaigns over party labels. Finally, as a system based on multimember districts, it reduces incentives to gerrymander.

Not all contexts would benefit. Innumeracy can be a barrier to a method based on ranking, and places with highly fragmented party systems probably need stronger incentives for cohesion. These caveats notwithstanding, democracy promoters should embrace the wealth of lessons learned – and to be learned – about the growing number of STV cases at home and abroad.

Another candidate-centric Iraq proposal

Via POMED comes a call by Scott Carpenter and Michael Rubin for MMP in Iraq’s governorates. A candidate-centric system, they argue, could dampen sectarian tension by weakening the party system.

Reforming Iraq’s election system on the national level will be difficult… At the local level, however, there is real opportunity… Iraqis should have the right to vote for the best individuals to administer governorates and sit on district councils. The country need not abandon parties or proportional representation, but lawmakers could explore an open-list system that would allow citizens to vote for people they know. Even better would be a mixed system, such as the one practiced in Germany, which combines party lists with the ability to elect individuals.

More on the rationale:

“[Adopting list PR for national elections] was a fateful decision. Rather than vote for individuals, Iraqis voted for political parties, whose leaders compiled lists of candidates. In descending order, one candidate would enter parliament for every 31,000 votes the party received. Under this system, aspiring politicians owed their future not to voters but to the party leaders who compiled the lists. Instead of encouraging Iraqi politicians to debate security, sewage and schooling, the party-slate system encouraged them to engage in the most extreme sectarian or ethno-nationalist rhetoric to prove their mettle to party leaders. Those who preached tolerance or voiced more technocratic concerns found themselves at the bottom of lists.

I have been making the same basic argument since April. The parties are the problem. Institutional choices made in 2005 largely caused them. Present institutional design efforts in the governorates are an opportunity to work on the problem. The system implemented must be highly candidate-centric.

To make that system work, federalism has to be strong enough to put a premium on governorate elections. And to keep federalism from ripping the country apart, there must be inter-governorate revenue sharing.

I applaud Carpenter and Rubin’s careful thinking about an important detail that most democracy promoters ignore. At the same time, open-endorsement SNTV remains preferable to their proposals.

Open-list proportional representation only mildly puts the candidate ahead of the party. Even though one votes for an individual entrepreneur, co-partisans depend on his or her performance for their own chances at winning seats. Open-list PR does not adequately dampen the incentive to run as a team.

Mixed-member proportional representation is problematic for theoretical and implementation reasons alike. One, it requires drawing single-member districts. Those presumably need to be of equal population. Even if the census data existed to allow equal population districts – it does not – districting would raise lots of different questions about gerrymandering (Does the way districts are drawn “naturally” advantage certain groups? Are the districts drawn purposely to do so? Et cetera.)

On the theoretical side, the nominal tier would have to be much larger than the list tier. That is, the proportion of seats elected in districts would have to overwhelm those elected from lists. Otherwise the ‘list logic’ of campaigning that the writers identify would again dominate.

Carpenter and Rubin are thinking in the right terms. Their proposal, however, should be more practical and ambitious. SNTV gets around the districting headaches while even more radically “put[ting] the people ahead of the party bosses.”

H/T to POMED’s Andrew Albertson.