DEMOCRACY DESIGN FORUM

AN ELECTORAL SYSTEM FOR IRAQ

David Chapman

July 6th 2004

 

Summary

In a divided society like Iraq, the conventional types of electoral system tend to produce ethnically based parties, which can exacerbate the ethnic conflict. A new electoral system is therefore put forward, designed to encourage the formation of pan-ethnic parties, moderate parties which are responsive to, and which get their votes from, all ethnic groups. The system uses small constituencies each electing one candidate. These constituencies are ethnically balanced, each consisting of an area inhabited by Shia, one inhabited by Sunni, and one by Kurds. In each constituency, in order to get elected, a       candidate is required to get an adequate percentage of votes from each of these ethnic areas.

 

A modified form of the system can be used to guarantee single-party government, by electing one party to be the sole government party.

 

If it is decided to have a national president elected by popular vote, then the proposed electoral system could be used for this purpose. This would provide the president with a strong pan-ethnic incentive, promoting more equal responsiveness to each of the different communities into which Iraq is divided.  

 

Lastly, in order to strengthen Iraq’s “civil society”, and to make the democratic system more representative, a proposal is made for a scheme of elected and publicly funded lobbies.

 

1. The need for pan-ethnic parties

In early 2005, Iraq is to hold a popular election, in order to establish some form of democracy . However, Iraq is an ethnically divided country, and in such countries, democracy rarely functions well. Under the usual types of electoral system, the parties tend to divide on the same ethnic lines as the electorate, and each party tends to become mono-ethnic, that is, it draws its politicians from only one ethnic group, and it appeals to and draws votes from only that ethnic group. Thus in Iraq there is likely to be a set of Shia parties competing with each other for Shia votes, a set of Sunni parties competing for Sunni votes, and a set of Kurdish parties competing for Kurdish votes. The Shia parties, for example, will have little incentive to respond to the needs of Sunni or Kurdish electors, since even if they try, they will almost certainly be unable to obtain votes from them. Thus ethnic extremism will be encouraged, and a party will have little or no electoral incentive towards compromise. Indeed, such cooperation with parties of the other ethnic groups can actually lose the party seats to a more extreme party of the same ethnic group. This ethnic division between parties is especially problematic where the parties of one ethnic group, the Shia, are likely to get a majority of seats, enabling them to form an exclusively Shia government, and perhaps retain it indefinitely, for election after election.   What therefore seems to be required is an electoral system which discourages ethnic nationalism, and gives each party the incentive to become pan-ethnic, that is, to draw its candidates from each ethnic group, to respond to each group in its policy, and to seek votes from each ethnic group. With such pan-ethnic parties, the ethnic power-sharing which democracy requires would take place not so much between the parties, but instead within each party. Instead of requiring mono-ethnic parties to negotiate after the elections in the attempt to reach a compromise, the system would give each party the incentive to draw up before the elections its own compromise policy which responded to the concerns of each ethnic group.

If such a pan-ethnic system could be achieved, the ethnic minorities would be well protected. If one party got a majority of seats and formed the government, it would always be a pan-ethnic party, and the government would be responsive to the needs of the ethnic minorities, as well as to those of the majority. If no party obtained a majority of seats, a normal majority coalition could be formed. Since the parties would be pan-ethnic, there would be no need, in order to protect the minority, to introduce special constitutional measures (which might be unworkable and prone to deadlock), such as  forming a “grand coalition” of all major parties, or giving the parties of each ethnic group the power of veto.

 

2. The Distributed Alternative Vote

It is often asserted that the introduction of democracy into a divided society is likely to exacerbate the ethnic conflict. This assertion may well be justified, if the new democracy uses one of the conventional electoral systems, which allow or even encourage the formation of mono-ethnic parties. However, it is possible to design new electoral systems which provide a strong incentive to form pan-ethnic parties, moderate parties which are responsive to, and which get their votes from, all ethnic groups. This paper proposes such a system for Iraq, which is referred to as the Distributed Alternative Vote, or DAV for short. DAV is a modification of the Alternative Vote (the system currently used in Australia to elect the lower chamber of the legislature), this modification being designed to provide the incentive for each party, and also each candidate, to get votes from each of the main ethnic groups, that is, from the Shia, from the Sunni, and from the Kurds.

To explain DAV, we must first explain the Alternative Vote on which it is based. The Alternative Vote uses “single-member” constituencies, that is, ones which elect one representative. Electors vote by putting the candidates standing in the constituency in order of preference, writing in “1” against the first-preferred candidate, “2” against the second-preferred, and so on. In the first stage of the counting, each vote (that is, each ballot paper) is allocated to the candidate for whom the voter has indicated first preference. The candidate who has fewest votes is excluded, and his votes are transferred, each vote going to that continuing candidate for whom the voter has indicated next preference. The candidate who then has fewest votes is excluded. And so on, for as many stages as are necessary, transferring votes and excluding a candidate, until one candidate has at least half the total of votes held by the continuing candidates, and is declared elected.

DAV is similar to the Alternative Vote in that it uses single-member constituencies, and in that the electors vote by putting the candidates in order of preference. However, under DAV, these single-member constituencies are ethnically balanced. Each constituency consists of four separate areas, referred to as “tracts”. Three of them are “ethnic” tracts, of which one is inhabited by Shia, one by Sunni, and one by Kurds, and the other is the “mixed” tract, where the electors are from several different communities. The system is set up so that a candidate has the incentive to get an about-equal percentage of votes from each of the ethnic tracts in his constituency, that is, from the Shia, from the Sunni, and from the Kurds.

In more detail, DAV takes advantage of the fact that each of the three main ethnic groups have areas of concentration in which they form 100 per cent, or fairly close to 100 per cent, of the population. The Kurds have an area or areas of concentration in the north, the Sunni in central Iraq, and the Shia in the south. Let us suppose that the assembly to be elected has 300 members, so that 300 constituencies are required. The Kurdish area of concentration is divided into 300 tracts, each containing about the same number of electors. Similarly, 300 Sunni tracts and 300 Shia tracts are created, in their respective areas of concentration. The remaining areas of the country, where the electors are from several different communities, are divided up into 300 “mixed” tracts, each containing about the same number of electors. Then 300 ethnically balanced single-member constituencies are created, each constituency consisting of a Kurdish (or at least nearly all Kurdish) tract, a Sunni tract, a Shia tract and a mixed tract. (Note that a constituency will consist, not of one single compact area as it does under other systems, but of four widely separated areas.)    

The method of election under DAV is as follows, to elect the single representative for any one constituency. Each vote is allocated to the candidate for whom the voter expresses first preference. Each candidate’s number of “points” is then calculated, points being a measure of the evenness of distribution of the candidate’s votes between the different ethnic groups. A candidate’s number of points is whichever is less, either two-thirds of his percentage of the votes in the whole constituency, or his lowest percentage of the votes in any one of the three ethnic tracts. (For example, suppose a candidate gets 30 per cent of the votes in the whole constituency, 10 per cent in the Kurdish tract, 40 per cent in the Sunni tract and 32 per cent in the Shia tract. Thus his number of points will be the least of the following: two-thirds of 30 = 20, 10, 40, 32. That is, his number of points will be 10.)

The candidate with fewest points is excluded, and each of his votes is transferred to the continuing candidate for whom the voter expresses next preference (if there is a next preference expressed). The points of each continuing candidate are recalculated, to take account of the votes which have been transferred to them, and the candidate who then has the lowest points is excluded. And so on, if necessary, at each stage transferring votes, recalculating points, and excluding another candidate, until only one candidate is left, who is declared elected.

 

3. The political effect of DAV

What then is the political effect of this new system of DAV, as used to elect the assembly? Essentially, DAV operates by excluding, at each stage of the counting, that candidate who has the lowest number of “points”. A candidate’s number of points is whichever is less, either two-thirds of his percentage of the votes in the whole constituency, or his lowest percentage in any of the three ethnic tracts, Shia, Sunni or Kurdish. Thus any candidate who gets a percentage of the votes from any one ethnic tract which is less than two-thirds of his percentage in the whole constituency, will lose points, and will have a greater risk of being excluded.

But what is the effect of this upon the parties? In order to avoid having many of its candidates excluded, a party has the incentive to get an adequate percentage of votes from each ethnic group, that is, to get from each ethnic group at least two-thirds of the percentage it gets over the whole nation. In other words, it has the incentive to become pan-ethnic, appealing to each ethnic group and getting its votes about equally from each ethnic group.

If DAV is introduced, mono-ethnic parties of different ethnic groups will have the incentive to merge to form pan-ethnic parties, in the expectation that this will enable them to get more seats. Thus a Kurdish party, a Sunni party and a Shia party might merge into one party, and adopt a policy responsive to the needs of each of these ethnic groups, a compromise between their different concerns. The party will place Kurdish candidates in some constituencies, Sunni candidates in other constituencies, and Shia candidates in others. This mix of candidates will help each candidate to get votes from electors in his constituency who belong to ethnic groups other than his own. A Kurdish candidate, for example, will be enabled to get more votes from the electors of the Sunni and Shia tracts of his constituency, because they will be reassured by the fact that in other constituencies, perhaps neighbouring ones, his party has run Sunni and Shia candidates.

To investigate the workings of DAV in more detail, let us consider the situation in one constituency, in which the candidates of five parties are competing, these being K, U and S, three extreme mono-ethnic parties, Kurdish, Sunni and Shia respectively, and M and N, two pan-ethnic parties, each formed by the merger of a moderate Kurdish party with a moderate Sunni party and a moderate Shia party. A likely election result produced by this situation, is as follows.

In this constituency, a large majority of voters in the Kurdish tract vote solely according to Kurdish interests, either voting KMN (that is, K first, M second, N third) or KNM, while a minority of them vote MNK or NMK, doing so either out of loyalty to the original Kurdish parties which merged into M and N, or out of an active preference for compromise and moderation. In the same way, a large majority of voters in the Sunni tract vote UMN or UNM, and a minority of them vote MNU or NMU. Similarly, a large majority in the Shia tract vote SMN or SNM and a minority MNS or NMS. Very few voters in the Kurdish tract vote first preference for U or S, certainly many fewer than the minority of them which votes first preference for M, or for N. Similarly very few voters in the Sunni tract vote first preference for K or S, or in the Shia tract for K or U.

In the first stage of the vote-counting, each vote (that is, each ballot paper) is allocated to the candidate it shows as first-preferred. K gets close to zero points, because of his very low percentage of votes in the Sunni tract. Similarly, U gets very low points because of a low percentage in the Shia tract, and S gets very low points because of a low percentage in the Sunni tract. M and N get a greater number of points, because each gets a more substantial percentage of votes in each of the three ethnic tracts, Kurdish, Sunni and Shia. Of the five candidates, K has fewest points, and so is excluded.

In the second stage, K’s votes, almost all in the Kurdish tract and the mixed tract, are transferred, some going to M and some to N. Of the four continuing candidates, U has fewest points, and so is excluded.

In the third stage, U’s votes, almost all in the Sunni tract and the mixed tract, are transferred, some to M and some to N. Of the three continuing candidates, S has fewest points, and so is excluded.

In the fourth stage, S’s votes, almost all in the Shia tract and the mixed tract, are transferred, some to M and some to N. M gets 48 per cent of the votes in the whole constituency, 55 per cent in the Kurdish tract, 51 in the Sunni tract and 47 in the Shia tract. Thus M will have points equal to the least of the following: two-thirds of 48 = 32, 55, 51, 47, that is, points equal to 32. Similarly, N gets 52 per cent of the votes in the whole constituency, 45 in the Kurdish tract, 49 in the Sunni tract and 53 in the Shia tract. Thus M will have points equal to the least of the following: two-thirds of 52 = 34.67, 45, 49, 53, that is, points equal to 34.67. Thus M has fewer points, and so is excluded, and N is declared elected.

This example illustrates how, under DAV, when pan-ethnic parties and mono-ethnic parties are competing for a seat, the pan-ethnic parties have the advantage, and the seat goes to one of the pan-ethnic parties. Note that this moderate result is achieved, even when a large majority of voters are extremists, voting first preference for one or other type of extreme mono-ethnic party. The explanation for this apparent paradox is that while the candidate of a pan-ethnic party gets a moderate percentage of votes from each ethnic group, the candidate of an ethnically extreme party gets a very low percentage from voters of other ethnic groups, and so gets excluded. Another factor explaining the success of pan-ethnic parties is that when the first-preferred candidate of these extremist electors is excluded, their votes are passed on to their second preference, a moderate pan-ethnic candidate.

But it may well be asked, will these putatively extremist electors in fact be prepared to vote second preference for a moderate candidate, or indeed to express a second preference at all? At least when they become familiar with the system, most of them will surely realise that their second preference will be used only when their first-preferred candidate is excluded and has no chance of election, and that by using their second preferences they will help to elect a candidate they prefer more, in place of a candidate they prefer less. In particular, they are likely to be concerned to prevent the election of an extreme candidate of another ethnic group, if there seems any possibility of this. Thus it can be expected that most electors will be prepared to express their second and lower preferences, and that as a consequence, DAV will operate in favour of the pan-ethnic parties, in the manner illustrated by this example.       

Clearly, DAV gives any mono-ethnic party a powerful incentive to broaden its appeal, and become pan-ethnic. If the party does not do this, it will simply lose its seats to more pan-ethnic parties. Thus in a short time, perhaps even at the first election, the use of DAV can be expected to produce a legislature consisting of pan-ethnic parties, ensuring a moderate government which is responsive to each ethnic group. 

We have seen that DAV promotes moderation in respect of ethnic concerns. But what is the effect of DAV in respect of any other dimension on which the electors might differ, such as the “left-to-right” dimension? For example, consider two parties, both equally pan-ethnic, but one a centre party and one to the right. Their candidates are competing in the final two-candidate stage of the vote-counting under DAV, all other candidates having previously been excluded. The candidate of the centre party will tend to beat the candidate of the right-wing party, getting a majority of votes comprising not only centre but also left-wing electors. Similarly, a centre candidate will tend to beat one to the left. DAV thus provides an incentive to move towards a centre or moderate position, on any dimension on which the electors differ.

The conclusion of this investigation is that DAV can be expected to exert over time, on each party and each candidate, a steady incentive towards compromise and moderation, towards a policy responsive to the needs of each group, and towards a centre or moderate position, not only on the ethnic dimension, but also on any other dimension which is of importance to the electors. DAV can thus be expected to promote convergence in policy between parties, towards this moderate centre position. This convergence between parties has implications for the stability of government. If two parties get all or nearly all the seats in the assembly, and alternate in office as single-party governments, then because there is little difference in policy between government and opposition, the electorate will tend to give more willing consent to be governed, and the changeover between parties in office will tend to be less disruptive. If on the other hand no party gets a majority of seats, and a coalition has to be formed, these moderate and convergent parties will more easily be able to form a government, and that government can be expected to be stable and effective, and responsive to each ethnic group.

 

4. DAV with guaranteed single-party government

Under DAV, as under any other single-member-constituency system, it is difficult for smaller parties to get seats, even if they are pan-ethnic, since their candidates tend to get a lower percentage of votes and so are more likely to get excluded. Thus smaller parties will tend to lose seats to larger ones, and it is likely, though not certain, that after several elections, one party will get a majority of seats and will be able to provide single-party government. However, if it is wished to have a guarantee of single-party government even from the first election, this can be achieved by the following modified form of DAV, which elects one party to be the sole government party.

            Ethnically balanced constituencies are created as before, each consisting of three ethnic tracts and one mixed tract. Electors vote by putting the candidates in order of preference, as before. However, in the counting of the votes, one party is elected as the sole government party by the same DAV procedure as was used to elect an individual candidate, but here applied to the ballot papers of the whole nation (a vote for a candidate being counted also as a vote for the candidate’s party). Thus the whole nation is treated as one national constituency, consisting of four tracts, a national Shia tract consisting of all 300 Shia tracts, and similarly a national Sunni tract, a national Kurdish tract, and a national “mixed” tract.

Only parties which have placed a candidate in each constituency are considered for election as government party. Votes are allocated each to its first-preferred candidate. Each party’s number of “points” is calculated. A party’s number of points is whichever is less, either two-thirds of its percentage of the votes in the whole national constituency, or its lowest percentage of the votes in any one of the three ethnic national tracts, Shia, Sunni or Kurdish. The party with fewest points is excluded, and each of its votes is transferred to the continuing party for which the voter expresses next preference (if there is a next preference expressed). The points of each continuing party are recalculated, to take account of the votes which have been transferred to it, and the party which then has the lowest points is excluded. And so on, for as many stages as is necessary, transferring votes, recalculating points, and excluding another party, until only one party is left, which is declared elected as the sole government party. The party elected as government party is assigned a fixed minimum majority of seats, let us say 54 per cent (which will be 162 seats in an assembly of 300).

            It is then determined which of the individual candidates are to be elected to the assembly. For this, the points of each candidate in each constituency of the nation are calculated, by the same formula as used before for calculating the points of the candidates. The candidate who has fewest points in the whole nation is excluded, and each of his votes is transferred to that continuing candidate in the same constituency for whom the voter expresses next preference. The points of each continuing candidate are recalculated to take account of the votes transferred, and the candidate who then has fewest points in the whole nation is excluded. And so on, transferring votes, recalculating points, and excluding the candidate with fewest points in the whole nation, until either there is only one candidate left in each constituency and each of them is declared elected, or the number of continuing candidates of the elected government party is reduced down to its fixed minimum majority. In the latter case, the candidates of the government party are declared elected, and in the other constituencies, exclusions continue of the candidates with fewest points, until there is only one candidate left in each constituency, and each of them is declared elected. 

            This form of DAV thus elects, as the sole government party, the party with most “points” on a national basis, and guarantees this party a minimum majority of seats, which are given to those of its candidates who have obtained most points. Thus single-party government is guaranteed, and this government party, and also each of its candidates, has the incentive to be pan-ethnic, fully responsive to each community, Shia, Sunni, and Kurdish.

 

5. Election of the president by DAV

If DAV is adopted as the electoral system for the assembly, then this could be expected to provide a more coherent parliamentary system, such as to obviate the need for a popularly elected president. However, if it is nevertheless decided to institute a popularly elected president, then whatever the electoral system used for the assembly, this paper would recommend that DAV (of the same form as that proposed for election of the sole government party) should be used to elect the president. This would provide the president with a strong pan-ethnic incentive, motivating him to respond more equally to each of the different communities into which Iraq is divided.  

 

6. Advisory Representatives

An assembly elected by DAV will consist of generalist parties, ones which seek a compromise policy acceptable to all parts of the electorate, rather than seek to articulate and represent the needs of particular sections. There will be an absence of niche parties, parties which cater for some special interest or section, and there might be some dissatisfaction with this, especially if the comparison is made with some forms of proportional representation, which would allow diverse small parties to get elected, to match the divisions in the electorate.   

One answer to this problem might be to have, along with the DAV-elected legislature, a second chamber elected by some form of proportional representation, with powers in the legislative process in parallel with those of the first chamber. However, this arrangement would tend to create a situation of frequent conflict between the two chambers, especially if, as would be likely, this second chamber consisted largely of mono-ethnic parties from different ethnic groups.

Therefore, this paper proposes, as an adjunct to the DAV-elected legislature, a system of elected Advisory Representatives, who have only consultative status, and are without any actual powers in the legislative or governmental process. In brief, electors vote for these Advisory Representatives (ARs for short), and the ARs are paid income in proportion to these votes. Thus with this income, the ARs have both the incentive, and the resources, to lobby on behalf of the electors who have voted for them. 

To explain the scheme in more detail, the elector has one vote, which he may give to one AR, or may split between any number of ARs, in any way he likes. Each AR’s total of votes is then found, and he is given a certain amount of income for each vote obtained (let us say, for each vote, 1/3000 of the average per-person income in Iraq). However, there are two exceptions: (1) the AR gets no income if he gets fewer votes than some minimum, let us say 500; (2) he gets no more income than some maximum, let us say six times the average income.

The voting is carried out not by use of the normal ballot paper, but by the "coupon" method, which is as follows. A coupon is printed for each AR, at the AR’s expense, and it bears the AR's name and message, and most important, his personal bar-code.  The ARs distribute their coupons to electors they think might vote for them, and also make their coupons available at any polling stations where they think they might have support. The elector votes in the privacy of the polling booth by putting the appropriate coupons in a special envelope provided to him by the polling officer, putting in more than one coupon for a given AR if he wishes. To count the votes, each envelope is opened, and the coupons in it are read by a bar-code reader attached to a computer, which calculates the total votes of each AR. For example, if an elector has put in his envelope one coupon from A and two coupons from B, this will be counted as one-third of a vote for A and two-thirds of a vote for B.

The coupon method should be used for the first election of the Advisory ARs. But in later elections, a ballot paper could be used, bearing the names of those ARs who got most votes in that area in the previous election (let us say the top 50 ARs). Any other ARs who sought support in the area could also be voted for, but only by means of coupons.

How then can these Advisory Representatives be expected to operate in practice, and what functions can they be expected to perform? Since they will have no powers in the legislative process, and since they will be extremely diverse in their interests and concerns, it seems unlikely that they will want to meet as a body, in the same way as a conventional second chamber would do. Instead, the ARs will act separately, each lobbying on behalf of the electors who have voted for them with government or parliament or perhaps with business, operating in a similar way to the interest groups which exist at present.

These ARs will be better able to articulate the various interests that exist in society, than would any large generalist party, which would be primarily concerned to find a compromise which had wide support. The ARs will be continually seeking new interests and causes to represent, in order to attract votes from electors who support them. Thus latent interests which at present are not organised at all, will be articulated and will then be able to exert influence. This scheme can be expected to reduce the likelihood of demonstrations and riots, since it provides an alternative way, one which is more peaceful and probably more effective, of making known to the authorities the needs and grievances of the people.

In conclusion, the Advisory Representatives scheme can be regarded as a means of developing and strengthening Iraq's "civil society". It is a neutral procedure for introducing public money into the informal aspects of the democratic process, which could make the democratic system vastly more representative, while avoiding the potential bias of arbitrary grants.

Return to top of page

Return to home page

Democracy Design Forum, Coles House, Buxhall, Stowmarket, Suffolk IP14 3EB, UK
Tel [int. 44] (0)1449 736 223
Fax [int. 44] (0)1449 612 274
e-mail: chapman@democdesignforum.co.uk