The Question
This question is posted on behalf of Julie Broadbent, member of the ACE Practitioners' Network

I would like to enquire about people’s experiences with a Proportional Representation Open List voting system.

I am working in a post-conflict environment and there is pressure from all stakeholders to have an open list system whereby the voters can mark one preference vote on the list. There will be 18 political entities on the ballot paper with an average of about 18 to elect in each of the constituencies.

The international community has reservations because of the potential number of candidates and the technical challenges this will bring, such as multi-leaf ballot paper, increased financial costs, length of time to vote, voter education, etc.

I am very keen to hear what other experiences practitioners have had with an open list system.

Summary of responses
It is noted that not all open lists are entirely open, because in so-called ‘flexible list’ systems political parties impose a threshold of preference votes that a candidate needs to receive before the order of the list is altered. This used to be the case in Indonesia for example and in Bulgaria in 2007. 

Open list systems are said to take power away from political party leadership.  Local candidates in Indonesia are seen to have benefited most from the adoption of an open list system, with voters choosing them over national candidates who previously would have been placed in top positions by party leaders under a closed list. One expert describes this as the “disciplining effect “of open list PR on political party leadership, citing Bosnia and Herzegovina as another example.

Another characteristic of open list PR is that candidates have more incentives to distinguish themselves from their parties (as seen in Iraq in 2009 and 2010).

With regards to the impact of an open list system on ballot paper design, one expert refers to studies that show a negative correlation between ballot complexity and voter turnout.

In an effort to minimise the length and complexity of the ballot paper in Iraq in 2009 and 2010, a referential ballot design was used, dividing the ballot into two sections: one with a list of parties and the other with numbered boxes representing candidates. A list of all the candidates and their corresponding numbers was kept in each polling station for reference by voters if needed. This meant the ballot was kept to a single sheet and the resulting invalid ballot rate “compared favourably with similar post-conflict operations and PR elections with the region.” A contrasting example is that of the Democratic Republic of Congo, where in 2011 the largest ballot had 56 pages.

Extra financial costs for adopting open list PR in Iraq were associated mainly with the counting and tallying process, which requires particularly clear and detailed procedures. In Indonesia, on the other hand, there was no noticeable difference in cost. Several replies mention that different criteria and more thought are needed with an open list system during the counting process, especially over determining what constitutes an invalid ballot and the intent of the voter.  One expert notes that open list PR need not, however, slow down the announcement of preliminary results and results by party can be released as quickly as under a closed list system.

In order to save on time and money, great care should be put into minimising the need to amend ballot papers once the candidate nomination process is over.

It is pointed out that female candidates do not generally benefit from an open list PR system and integrating gender quotas can be problematic. In Indonesia, however, it is noted that “women actually did better under the fully open system than they would have under the de facto closed system.”

Examples of related ACE Articles and Resources
Encyclopaedia:
• List PR
• Open, closed and free lists

External Resources
• Election Resources on the Internet: Elections to the Estonian Riigikogu (Parliament), Manuel Álvarez-Rivera 
• OSCE/ODIHR Election Mission Assessment Report, Estonian Parliamentary Elections, 2011  
• Voter education poster from 2010 Iraqi parliamentary elections 
• Photo showing campaign posters and referential ballot paper design from 2010 Iraqi  parliamentary elections  

Names of contributors
1. Horacio Boneo
2. Jack Santucci
3. David Jandura
4. Mykhaylo Nakhod
5. Sean Dunne
6. Kevin Evans
7. Mirimo Mulongo
8. Melle Bakker
9. Julie Broadbent
10. Velko Miloev
11. Hugh Fulton
12. Ramlan Surbakti

There is relevant information which is not provided in your question.  Could you please add if:

1) Voters have as many votes as candidates?  That is, if a constituency have those 18 candidates to elect (as it seems to be in your example), each voter would have up to 18 votes?

2) If that is the case, are ballots which have less than 18 marks valid?

3) What is the average size of connstituencies? Is the 18 seats constituency just one case (as it was Kinshasa in DRC), or it is the norm?

4) Is panachage, or cross-voting allowed?

Julie, here are a few general comments on OLPR:

1. Many electoral systems masquerade as open-list PR because they permit voters to express preferences over candidates within party lists. In reality, though, party leaders have substantial control over the orders of those lists. This can arise under so-called "flexible list" systems in which individual candidates must achieve some threshold of preference votes before those preference votes affect the order of the list. It may also arise under multi-stage seat allocations (e.g. Austria, though this does not appear relevant in your seemingly straightforward case of OLPR).
2. Conventional scholarly wisdom posits an affinity between individual forms of clientelism and open-list systems. The logic is that well-resourced candidates may distribute patronage to raise their positions in voters' preference orderings.
3. Conventional wisdom also suggests that open-list PR is associated with weak party discipline. In other words, legislative caucuses may be less cohesive because pure OLPR removes control of list order from party leaders' hands. That control is a resource that party whips can use to control their members' votes. And when party cohesion is low, it may be difficult for that party to negotiate with other parties, chambers, executives, and so forth.
4. On a related note, open-list PR is said to provide individual candidates with incentives to differentiate their campaign appeals from those of their parties. In other words, OLPR reduces the utility to a candidate of a party "brand." In a sense, much of the above can be said to flow from that.

While not able to speak to some issues you mention, such as administrative costs, there are some other considerations for OLPR. 

While I’ve always I wondered if we overestimate the impact of ballot complexity, there is some evidence that it can have an impact on turnout.   In two papers presented at APSA this year, Aina both Gallego of Stanford University and Saul Cunow of UCSD found some evidence that ballot complexity has a negative impact on turnout.  There were some methodological issues with both paper, which both use field experiments, but they provide strong evidence that this is the case. 

I would also say that such a high district magnitude for open list PR would increase the likelihood of both corruption and political violence, as the incentive to cultivate a personal vote will increase substantially. 

I would like you to look at the Estonian experience. It looks like you looking for. http://electionresources.org/ee/

http://www.electionguide.org/country.php?ID=69

http://www.osce.org/odihr/77557

Iraq transitioned from a closed list PR system in 2005, to adopt an open list PR (OLPR) system for their 2009 Governorate Council elections and subsequently also adopted the system for their parliamentary elections in 2010.

There was initial concern over the ballot design implications, similar to those that you have outlined. However, Iraq was able to keep the ballot to a single sheet by employing a referential system. That is, the ballot was divided into two sections, one area allowed the voter to select the party (one tick), and the other section was numbered boxes representing each candidate position within a list. For example, if you wanted to select candidate 11 in party list A, you would tick party A and then tick candidate box 11 on the left side. A picture of a voter education poster is available at this link (http://graphics8.nytimes.com/images/2008/12/29/world/07ballot01.jpg).

.)

The average invalid ballot rate was about 4.3% (as low as 2% and as high as 6% in different Governorates, which was attributable to different VE approaches and efforts), which compared favorably with similar post-conflict operations, and PR elections within the region; although, it was a kick up from the 2005 closed list PR system with an invalid ballot rate of about 1.1%. Voter education was heavily emphasized through a range of campaigns including broadcast media, billboards, national and international NGOs, and information seminars with parties etc.

We conducted time-work flow studies with groups to determine the throughput estimates and were able to allocate about 400 voters to each polling station, over the course of a standard 10hr polling period. Iraq uses a voters' list assignment to the polling station, which slows throughput at the polling center - due to the inability to balance the workload - so the 400 was a maximum capacity estimate per polling station, which might be increased if you use a polling center level approach to manage the voters' list.

The real time-cost factor and complications for implementing OLPR arose with the counting and tallying processes, due to the extra steps necessary to count the votes. In Iraq, counting was done at the polling station, while tallying was conducted at a nationally centralized location, with a custom IT solution in place. (Data center investment highly recommended, but may also be a strategic investment for any census or other data heavy processes to be conducted in the future.) The OLPR necessitates clear and detailed procedures, in particular, for the counting process and the criteria for determining an invalid ballot (extra time and effort on election official training is a consideration). Strong development of related administrative forms is also essential - especially the counting forms. 

Overall, the OLPR was an extra investment at several levels (compared to the CLPR), but was driven by the national political demands. OLPR was seen as attractive for its benefits in allowing for party structure, but also strengthening the direct connection between the electorate and their representative. Candidates were often less loyal to the party platform, as they had to distinguish themselves from other candidates within their own list (against whom they also had to compete). Of note, this intra-party competitiveness may be an extra source of electoral violence, beyond possible inter-party violence.

The OLPR also weakened the ability to form localized coalitions (compared to the closed list system, where list position is a major negotiating position), but broad national coalitions formed as a result, which allowed for  parties within coalitions to perform well in their specific areas. Post election cohesiveness of electoral coalitions is weak however. A major consideration is that women do not appear to perform well in OLPR (a noted aspect of the electoral system in Iraq as well as Latin America). In Iraq a quota was applied, where a coefficient was determined to select winning lists in a Governorate, where the lowest ranked winning male candidates had to be replaced by the woman within the list with the highest votes but without a seat, to meet a Governorate level quota.

Hope this assists.

Indonesia adopted and applied a closed list system since its first national elections in 1955. This approach was used in free and fair (1955 and 1999) and non-free and fair elections (all those between). After 1999 the public begun to agitate for a more open system. The parliament replied with a de jure open but de facto closed list sytem (which included a virtually unsurmountable threshold for a candidate to "jump the queue" on their party list). Agitation rose towards the 2009 elections and the parliament reduced the threshold to jump the queue from a full quota to 30% of a quota). All parties then preselected their candidates on the basis that the system would remain de jure slightly open but de facto closed. 

Some parties as a sop to public angst at frustrations that the list remained effectively closed signed legally questionable internal "deals" among their candidates that they would respect the preferred wishes of the voters anyway, that is if the party won say 1 seat and candidate listed at 5 gained the most votes, candidates 1, 2, 3, and 4 would all step aside and not fill the seat - allowing candidate number 5 to fill the seat. Most commentators either questioned the legality of such a move or worse suggested it was a cheap ploy to seek support from frustrated voters on the eve of an election. Many doubted that such "deals" would ever be followed through especially if a favoured son or daughter of the party establishments was to be defeated by a candidate further down their list. 

This is where it began to get interesting!

Two candidates placed in low (unwinnable) positions petitioned the Constitutional Court arguing the system was unconstitutional as it was a fix in that they could theoretically, but not practically, win. In a stunning decision for those arguing to empower the voters with the collective authority to determine who (not just whose party) wins, the Court agreed and declared that there was no longer a candidate preference order in the list. In short all candidates were equally able to win against their fellow list members. Thus we had two election systems for the same elections! A very rare political experiement in the real world. To the point of candidate selection, we had a pseudo open list. By the time of the campaign and polling day we had a fully open system.

The form of the ballot paper was not changed. All candidates would continue to be listed under their party symbols on the basis of the order initially submitted to the electoral authorities. With 34 competing parties and district magnitude of between 3 and 10, the ballot paper was large.

Candidates did certainly begin to campaign individually, essentially accentuating a trend begun in 2004 when the first efforts at prying open the candidate's lists was undertaken. Of note is that patronage based politics formed around primordial (ethnic) alligiences, most evident in the easternmost regions of Indonesia, was reflected in voting patterns. In more cosmopolitan or ethnically homogenous regions, such voting patterns were less evident, with minority candidates also being elected.

The decision to force open the candidate's list was aggressively opposed in particular by women's actvists who argued that only a closed list with women placed at the upper levels could guarantee women got elected.

The results indicated that women actually did better under the fully open system than they would have under the de facto closed system. Overall therefore it may be accurate to suggest that the party elites tend to be more resistent to women in elected office than the wider eletorate as such. Looking under the national results women did worse in regions with poor records of gender equality but better in regions where records of gender equality are also better. In effect the new system has simply allowed the underlying gender dynamics in each region of the country to be reflected in its electoral outcomes. 

The clearest beneficiaries of the open list were local candidates (both men and women) who did jump the queue to defeat the politically connected "Jakarta drop-ins" placed at the top of the list in the centralised candidate selection processes.

In terms of ballot cost, the cost was the same. The printing of just the party symbols as was the case before 2004 would have been cheaper. We do however need to be careful not to chase false economies. Presumably no elections would be even cheaper again (certainly in the short term), but the costs of poor political accountability end up being much more expensive in the longer term when political and socio-economic meltdowns become the only way to remove leaders or representatives whose used by date have passed. There are other ways to steadily reduce election management costs especially where the electorate is quite computer literate.

Indonesia manages to maintain rigorous internal parliamentary faction discipline through "recall" means that permit elected MPs to be sacked and replaced by their party organisations if they are deemed to be disloyal. This made some sense in the days of a closed list system. It will be a matter of note how the logic of this system can be applied when candidates are winning in their own right, that is where the voters are determining who (not just whose party) wins a seat. There is currently not much focus by civil or political society on the implications for changing the recall regulations in the light of the new electoral system.

At a technical level of voting, voters can vote for the party symbol or a candidate of that party. This needs to be made clear to voters as part of voter information campaigns. For EMBs there also needs to be flexibility in understandig teh intent of a voter. For example if the voter selects Party A, then candidate C from Party A, what they are doing is expressing a preference for Candidate C. The vote should be deemed valid, despite the double marking as the intent was clear. They wanted Party A anyway as they expressed a preference for Candidate C. The vote should be counted a a vote for Candidate C primarily. This is especially so as it is not valid to vote for Party A and a candidate from another party.

The final verdict. Most candidates elected were indeed elected in the order they were originally placed on the list. So opening the list did not destroy the parties! However sufficient leap frogging did take place to suggest that unliked candidates can be defeated by voters regardless of how much their party leaders wanted them. This does suggest that the voters were able to hold their MPs to political account. 

There can be little doubt that opening the candidate list has helped boost the quality of Indonesian democracy and empower its voting citizens, while not destroying its party system.

Bonjour Julie,

J'aimerai partager mon experience entant qu'ancien membre de la Commission Electorale Indépendante de la RDC. En 2006, nous avons organisé les premières élections libres, démocratiques et transparentes mais aussi post conflit. Pour 500 sièges, nous avons enregistré plus de 9700 candidats pour 169 circonscriptions et le plus grand bulletins avait 8 pages avec plus de 870 candidats. Il s'est posé plusieurs problèmes notamment dans l'éducation civique et électorale. Le marquage de bulletin était un défis majeur avec une population à plus de 50% analphabète. Défis logistique mais aussi financier. En 2011, il y avait 18300 candidats pour 500 sièges, le plus grand bulletins avait 56 pages.

Voyez vous là où peut amener le choix du système électoral par le législateur. Evidemment chaque système électoral a ses avantages et ses inconvénients.

The Netherlands have proportional representation in combination with open lists since 1918. The number of seats in our Second Chamber is 150. The max. number of candidates on one list is now 50 and 80 for parties with more than 15 seats in the Second Chamber. To gain a seat a partylist needs a number of votes equal total number of votes (in the country as a whole) divided by 150. In our country at average that is about 62.000. The seats are allocated to the candidates on the list in the order of the list. If a candidate recieves a number of votes equal to 25% of the number required by the party for on seat (so about 15.450 votes) he or she is elected. In 2010 32 candidates recieved this number of votes; 30 of them would have been elected anyway (because of there place on the list).

At average about 20 parties participate in elections. New (not represented) parties need to pay a certain amount of money (€11.250,- for Second Chamber, €225,- for local alections), which is returned to them if they recieve at least 75% of the votes neede for one seat. Furthermore each lists of a new party needs to be supported by 30 people in each of the 20 electoral districts (so 600 in total). Purpose of the required number of signatures and money is to make sure that only serious parties particpate.

At average about 20 parties participate in national elections. Voter turnout is high (between 70 and 80% of the total electorate (people don't have to register; all nationals (living in the Netherlands or elsewhere) over 18 years of age are registered automatically and recieve their voting cards by post).

Melle Bakker,

secretary-director Dutch Electoral Council

I’ve been part of some initial discussions re introduction of open lists in Bosnia and Herzegovina, Bulgaria and Iraq and my impression is that common citizens were quite enthusiastic about this novelty. It appears, that in a transitional democracy, after the first several rounds of free elections, there was a widespread frustration with was considered “political party dictate” in the process and lack of accountability of those elected.

When first applied in Bosnia and Herzegovina (for the 2000 municipal elections, i.e. district magnitude between 11 and 20+; no threshold) the open lists worked well and roughly 25% of the new municipal councilors in the areas I observed would not have been elected under the closed list PR. In subsequent elections this ratio eventually went down, but most importantly, in my view, was the “disciplining effect” of the OLPR on parties’ HQs in the candidate nomination process. Parties could not afford anymore placing high on the list highly unpopular candidates, who, eventually, would have been considered “important” for the parties’ hidden interests.
In my country Bulgaria, the OL system introduced for the first time for the elections to the EU parliament in 2007 did not work due to the 10% threshold required for a candidate to be “moved up” the list. The parties sitting in the legislature, while formally meeting a loud and clear public demand, essentially killed the open lists through this threshold. It still remains unreasonably high.

As a member of the EU Election Assessment team to the Iraqi 2010 Council of Representative elections, I can fully support the detailed contribution of Sean Dunne above. Attached is a sample ballot from the Basra governorate, as an example, described already by Sean, how OLPR can be accommodated within a simple and user’s friendly ballot design. Interestingly and possibly useful to know, none of the candidates or party HQ representatives interviewed by our team expressed officially any criticism against OLPR, despite its political implications (campaigning “against” party peers, coalition bargaining, etc.) Apparently, once the decision was taken, they had to live with it.

A real problem is that OLPR does not fit well with gender quotas. In Iraq, 2010, the contradiction between the constitutionally guaranteed share for women in the parliament and the open lists was resolved through a highly sophisticated seat allocation formula that is likely to have remained quite non-transparent for many stakeholders. In Bosnia and Herzegovina, where parties were obliged to nominate on every third, sixth, etc.  place of the list a candidate from the “opposite gender”, female candidates did poorly under the OL. Eventually, women did not vote for women and the “every third” rule was subsequently abandoned, while preserving the OL on various levels of elections..

While OLPR does indeed put an additional burden on EMBs (voter education, ballot design and printing, counting, training – with the related financial implications) it does not have to slow down the announcement of preliminary results. Under a carefully designed counting/tabulation process, results by parties can be released as quickly as under a closed list system, while individual seat allocation is usually not such an urgent matter.
Possibly more important is to lay down clear rules and deadlines regarding candidates’ nomination, resignations and replacements, so as to not jeopardize the costly and time-constrained processes of ballot design, printing and delivery.

In one region where I have worked I note the following additional commentary:

1.        The impact upon parties which lack cohesion or have internal factions. Issues are (1) an open list reduces party leader’s internal control and may exacerbate further party functionality, and (2) may encourage fraud as factional leaders / individual candidates have an incentive to influence polling and counting teams – to add a tick for the appropriate candidate wherever it is blank, or to annul the ballot by adding additional marks when a vote for a candidate has been added.  The debate is between supporting “internal party democracy” and reducing party efficacy / encouraging fraud – this is relevant depending upon the specifics of the location in question

2.        Depending upon the number of parties and candidates contesting the election, there may be a requirement to produce a candidate booklet with obvious implications for voter education, time it takes to vote, assisted voting, procurement and financing.

3.        With system redesign comes the requirement for measured and appropriate voter information programme.

4.        On another level one must consider what is appropriate for the current context – a very relevant reform driver in this example is the need for the EMB to manage the process independently from international support – simplicity is a key requirement in this regard.  It is noted that the political drivers do not often allow for this.  

In summary, in the region where I saw this in operation – voting required a ballot paper with parties registered on the left (voter to tick one party) and a numbered box on the right of the ballot (voter to check a number which corresponded to a candidate). Voters could reference the candidate number by access to a candidate booklet contained therein were party name and list of their candidates. Parties were listed in the same order as shown on the ballot. Voters could also vote for the party only; any more ticks than was permitted would invalidate the ballot; one district was in operation.

The particular issues associated with this design were (1) to aggravate internal party factional issues, (2) to encourage fraud as polling / counting teams were in some cases susceptible to influence, (3) results compilation was also not without issue as reconciliations were hard to achieve also due to factional controlled fraud.

However, voter information issues were helped by candidates as in their platforms they would promote party and their candidate number, and as it was in their interest to gain maximum votes they assisted in explaining to voters how to vote. There appeared to be no major issue related to an increase in voting time – helped in part by good voter information and a low turnout. Partially sighted and blind voters did have problems especially if they were not sure which candidate(s) to vote for.

Clearly, once a system change has been made, stability of system must occur thereafter which requires stakeholder buy-in – the danger is continual adjustments to continually fine tune the system a number of times thereafter as is the case in the region I have noted causing voter confusion and fatigue.

This reply is posted on behalf of Ramlan Surbakti, a member of the Malaysian Centre for Commonwealth Studies.

Please see attached an article about the experience of Indonesia in practicing Open List PR Electoral System, which was published in the Indonesian press and then translated by IFES-Indonesia into English.

Julie, it's not necessarily strange that voters do not have a party-only option on their ballots. All this means is that votes cast for candidates within a list will count toward the respective party's seat total as well as affect the position of the respective candidate within that list (provided this is not a "flexible list" system in which some quota of preference votes is required to change the party's pre-ordered list).