Elections and Registration in Afghanistan (ERA) Project
Options for an Electoral System for the
Afghanistan Wolesi Jirga
ERA Topical Report #3
February 22, 2003
Article 43 of the 1964 Constitution of Afghanistan provided for the election of members of the Wolesi Jirga (Lower House) of the Shura (Parliament) by plurality in single-member constituencies (i.e., first-past-the-post elections). Two elections were conducted on that basis in the 1960s.
However, the 2001 Bonn Agreement requires that the free and fair elections foreseen "no later than two years from the convening of the Emergency Loya Jirga" (i.e., June 2004) must result in "a gender-sensitive, multi-ethnic and fully representative government" (italics added, JE). The Bonn Agreement also states that Afghanistan's 1964 Constitution is only applicable until the adoption of a new Constitution and insofar as it is not inconsistent with the Bonn Agreement.
"Fully representative government" (italics added, JE) can only be achieved if elections are conducted on the basis of a system of proportional representation (PR), as majoritarian electoral systems can only accidentally provide for a sufficiently representative government.
Even without such guidance from the Bonn Agreement, a reasoned choice based on the eight well established criteria for choosing an electoral system, which are listed below, evidently leads in the direction of a PR electoral system. Consequently, and for both reasons given, the 2004 Wolesi Jirga elections should be conducted as PR elections. Proportional representation electoral systems, however, are not one particular electoral system only, but a whole (even extended) family of electoral systems. This necessitates a choice also among the various PR electoral systems that have developed over the years. The reasoned choice among electoral systems is normally based on a balanced weighting of a considerable number of criteria, among which one usually finds the following eight concerns (here adopted from Reynolds, Reilly, et al., 1997):
- Ensuring a representative parliament
- Making elections accessible and meaningful
- Providing incentives for conciliation
- Facilitating stable and efficient government
- Holding the government and representatives accountable
- Encouraging "cross-cutting" political parties
- Promoting a parliamentary opposition
- Cost and administrative capacity.
Trade-offs between these eight criteria are unavoidable. This is particularly the case in transitional and post-conflict elections, where time constraints, ethnic tensions, representation issues, language and literacy issues, expected outcomes under certain conditions, and other elements are important factors to include when the final decision on the electoral system and its constituent components is eventually taken. Furthermore, various more specific issues in relation to the various PR systems under scrutiny also need to be considered carefully.
One can identify at least seven PR-specific issues to consider and decide upon. Since these issues are interrelated, they must be dealt with simultaneously.
- PR formula (divisor or quota system, and within the two systems: which divisor system, which quota system)?
- One or more (usually only two) tiers?
- Constituency structure (and constituency magnitude) within the tier(s)?
- One or two ballots (only in the case of more than one tier)?
- Electoral threshold?
- Surplus seats (only in the case of more than one tier)?
- List structure (closed, semi-closed, or open)?
As argued above, the 2004 Wolesi Jirga elections should be conducted using a PR electoral system. PR elections require constituencies with more than one seat, so-called multi-member constituencies (or districts). The reason is that only with more than one seat can seats be distributed among contesting parties in such a way that the seat distribution within the constituency reflects–more or less perfectly–the distribution of votes for parties and independent candidates. The more seats in a constituency, the easier it is to reflect reasonably well the vote pattern for parties, so that a party with 17% of the vote will also get approximately 17% of the seats.
Systems where the entire country for all practical purposes is one PR constituency (such as The Netherlands, South Africa, Denmark, Germany and Israel) will therefore have an almost perfect correspondence (at this level) between the distribution of votes and the distribution of seats (for parties above the electoral threshold, if any). If, however, parties are only contesting at the national level and based on national list of candidates, voters often tend to feel that they have no proper local or regional representation and that they are not aware of who their representative(s) is (are).
The remedy is either to allocate all seats in smaller multi-member constituencies (often identical to regional, administrative subdivisions of the country, such as provinces, regions or counties) or to combine the overall national allocation of seats with a regional or local allocation of some of the seats, which are then accounted for in the final computations. If the allocation of seats at the local level takes place as plurality elections in single-member constituencies, one often talks about Mixed Member Proportional (MMP) elections. The MMP electoral system is well known from Germany and has been introduced in a couple of countries over the last 10 years (e.g., New Zealand and Lesotho and the assembly elections in Scotland and Wales) and it has been debated elsewhere also.
The MMP electoral system has a number of advantages but also some disadvantages, not least in traditional countries with low literacy levels and a weak party structure. The reason for these disadvantages–at least in countries with weak administrative structures, weak party structures and low levels of literacy–is that an ordinary MMP electoral system requires:
- that each voter be given two different ballot papers: (1) one national, which will decide the overall composition of parliament, and (2) one local, which decides which candidate will win the local constituency,
- that the entire country be demarcated into constituencies (which is administratively complicated, time-consuming and politically sensitive), and
- that intensive voter education be conducted, especially to allow the less-educated voters (including women) to have their voices heard at par with other voters.
Based on the above reasoning, the options for the 2004 Wolesi Jirga electoral systems are primarily two, even though a third option (introduced below) is also available. This judgment is based on the wording of the Bonn agreement, the various criteria for choosing electoral systems, and the obvious need for a simple system that can be fully understood and appreciated by voters and parties already in these first elections. Taken together, these reasons rule out the use of a straightforward MMP electoral system. The main electoral system options are then these two:
- A system where the entire country is one constituency and where registered
parties run national, so-called closed lists. Independent candidates should
also be allowed to stand, but their chances of election are slim. Examples
of such electoral systems are first and foremost Israel and The Netherlands.
- A similar one-tier system, but with the 31 Afghani provinces and Kabul used as 32 multi-member constituencies. The 32 constituencies would each be allocated a number of seats reflecting the size of their population (or their number of registered voters) as well as their geographical size and accessibility. A good example of a similar system is Finland.
The third option, not to be ruled out beforehand, as it would not require voters to cast two ballots, would be to combine these two systems in the following way:
- The main element of the electoral system under this option would be as under (2), i.e. PR elections conducted in 32 multi-member constituencies. But a certain percentage of all the Wolesi Jirga seats (10-25 per cent, the exact number to be decided later) could be set aside as so-called compensatory seats to ensure a reasonable overall, i.e., national, level of proportionality. Good examples of this kind of electoral system are the three Scandinavian countries (Denmark, Norway and Sweden) and South Africa. This system differs from MMP by using one ballot only (the one cast at the provincial/constituency level) and by taking into consideration all the votes cast for the parties across provinces/constituencies when the national election management body allocates the compensatory seats. Obviously, the computations are published to make the entire operation fully transparent. A system for allocating the compensatory seats to individuals within the parties that are the beneficiaries of such seats must be included in the electoral legislation. It can either be a national closed list or a system for selection of unsuccessful candidates in the provinces, based on how well their lists fared in the individual constituencies where they stood for election.
As these three electoral system options all have considerable merit, it is strongly recommended that they form the main basis for the future discussion about the electoral system to be implemented for the 2004 Wolesi Jirga elections.
- The current lack of reliable information on population size and distribution
across Afghanistan calls for the simplest possible solution to constituency-
and demarcation-related issues. Therefore, the system should not require that
current population figures be available, as it is unclear when such information
will actually be available for use in the election preparation processes.
The system should also relate to a known administrative structure. For these
reasons, the 32 current provinces (or 31 provinces plus Kabul) are the obvious
choice as the basic units.
- The number of parliamentary seats awarded to each province must reflect the number of potential voters who have registered in each of them, in combination with a factor reflecting the accessibility of the terrain, i.e., the physical difficulties for voters to get to registration centers, voter education events, political campaign events, polling stations, etc. This combined number will be sufficiently accurate to provide a solid basis for the seat allocation exercise. It will also include an indirect consideration for the size of the population in each province, as the correlation between population size and voter registration figures will be high (but not necessarily perfect). The main advantages of this proposal are two, namely
- that this will detach the preparation of the election from the completion and publication of results of the pre-census exercise, and
- that the direct link between the number of registered voters in a province and the number of seats allocated for distribution within the province provides a strong incentive to all political movements and groupings to get all potential voters, i.e., also women, to register before the election, as a low registration level in a province would have the immediate consequence of lowering number of seats for the province.
- The proposed use of the 32 actual provinces as multi-member PR constituencies
in options 2 and 3 also addresses concerns over the implementation of the
electoral system. These options can easily and flexibly cater to the diverse
ethnic, political and other ground structures and allegiances across the country
as it allows for such groups and movements (and independent candidates) to
present themselves to the voters of a province as their special representatives.
If a movement (group, "party") has a more regional support base (covering
several provinces), it can present its lists in all the relevant (probably
adjacent) provinces. At the same time, options 2 and 3 also allow for national
political movements, groupings and more or less established "parties" to field
lists of candidates in all (or only some) of the constituencies. Therefore,
these systems also allow national political movements to develop, even though
this objective obviously is furthered more under option 1.
- All three options should allow independent candidates to run, but the chances
for such candidates of winning seats in a nationwide constituency (option
1) are slim.
- Under options 2 and 3, candidates should not be required to reside in the
province where they stand for election, as the voters themselves should be
allowed to decide who will represent them in the Wolesi Jirga.
- Option 3 includes a national layer (tier) of seats to allow for a higher
degree of proportionality than option 2 by allocating a least some compensatory
seats to parties that are under-represented in the provinces. Arguably, such
a system could be as effective as option 1 in developing, at least over time,
national "party" structures. The system could function easily with one ballot
paper only and should, if implemented, be integrated with measures taken in
relation to out-of-country voting and voting for IDPs (as should options 1
and 2 also). On balance, option 3 (with only one ballot paper, relatively
few compensatory seats–maybe 12 to 15%–and allocation of these seats
to "best losers" in the 32 constituencies) has considerable merit. The important
thing to remember is that the voters would only have to cast one ballot and
should not be unnecessarily concerned about the post-election day allocation
of the compensatory seats to the parties that eventually end up underrepresented
- Under options 2 and 3, the PR elections in the 32 multi-member constituencies
should be conducted according to the following basic set of principles and
suggestions, which should be spelled out in the Electoral Law, with computational
and other details to be specified in the Rules and Regulations:
|1.||For the June 2004 elections, the suggested 225 seats in the Wolesi Jirga
under option 2 (and maybe 200 under option 3) are allocated to the 32 provinces
(or whatever the actual number of provinces may be) in strict proportionality
to the sum of two figures: (1) the final number of registered voters in
each province and (2) the geographical area of each province in square kilometres,
maybe multiplied by a factor 10 (the exact factor has to be decided on,
but 10 appears to be a reasonable choice).
|2.|| A tentative allocation of 225 seats to provinces under option 2, based
on average population forecasts for each province plus the area in square
kilometres multiplied by 10, yields the following results:
|2 seats: 1 province (Nuristan)|
|3 seats: 4 provinces (Logar, Kunar, Laghman, Kapisa)|
|4 seats: 7 provinces (Khost, Paktya, Sar-i-Pul,
Badghis, Nimroz, Zabul,
|5 seats: 3 provinces (Wardak, Paktika, Jawzjan)|
|6 seats: 2 provinces (Samangan, Farah)|
|7 seats: 3 provinces (Takhar, Ghor, Parwan)|
|8 seats: 5 provinces (Kunduz, Baghlan, Balkh, Faryab, Uruzgan)|
|9 seats: 1 province (Badakhshan)|
|10 seats: 0 provinces|
|11 seats: 1 province (Hilmand)|
|12 seats: 3 provinces (Ghazni, Nangarhar, Kandahar)|
|13 seats: 0 provinces|
|14 seats: 0 provinces|
|15 seats: 1 province (Herat)|
|24 seats: 1 province (Kabul)|
A similar calculation based on the 200 seats under option 3 has not been conducted, but the distribution will obviously follow the same pattern.
- Within each province the seats available according to (a), cf. (b) are allocated
to parties and independent candidates using the Hare (simple) quota in combination
with largest remainders (details to be spelled out in the Rules and Regulations).
This quota is suggested because it is not biased against either large or small
parties. To provide for the highest possible degree of proportionality, there
should be no electoral threshold in the multi-member constituencies, but the
so-called natural threshold will ensure that small parties will not win seats
in most multi-member constituencies, as most constituencies will have fewer
than ten seats and half of the provinces/constituencies only between two and
six seats. Under option 3, underrepresented parties will be able to win at
least some compensatory seats.
- Parties (small and big) should be allowed to form coalitions for electoral
purposes. This will contribute to coalition building among like-minded parties,
because they will see it as rational to assist each other to cross the (national)
electoral threshold. Within such coalitions (and/or independent candidates),
seats won must be allocated by the same electoral system as used in the multi-member
constituencies. This is a normal procedure and will not complicate matters
- Each voter will get only one ballot paper.
- Under option 3, surplus seats, i.e. seats allocated extraordinarily in case
the ordinary number of compensatory seats is not enough to ensure full proportionality,
should not be allowed, as there is no need to have a floating number of seats
in the Wolesi Jirga.
- Lists presented (both from registered "parties" and independent candidates)
must be supported by a certain number of signatures from registered voters
from the constituency (province), which should not be too small, but which
should also reflect the size of the province. The required number of signatures
should be in the Rules and Regulations.
- Lists presented by the same political movement or "party" in more than one
constituency should be allowed to run under the same name and symbol in all
constituencies. The permission to run under a certain name and symbol should
be granted by the Central Electoral Commission to avoid confusion and overlap
in the use of such names and symbols.
- Lists should be closed, i.e. have a fixed order of candidates who can only
be elected in that order. The order of the candidates cannot be changed after
a certain cut-off date before the election. Candidates are only allowed to
present themselves for election in one constituency and on one list. A list
by an independent candidate will by definition have only one name, while a
list from a political group or "party" can have any number of candidates up
to the number of seats in the constituency, as specified under (a) + 50
per cent (rounded down to the nearest whole number). This last provision is
to allow for filling vacancies from the lists, so that by-elections are not
- If a list runs out of names during the parliamentary term, that particular
seat will not be filled until the next Wolesi Jirga election. If a party (or
an independent candidate) at the election wins more seats than is on
the list, the next party to win a seat will be the beneficiary.
- In the table below, the three options have been allocated scores on eight different criteria of importance (not the same eight criteria as mentioned in paragraph 6). Such a scoring exercise contains a certain element of subjectivity, so readers of this paper are invited to do their own scoring and also to reflect on whether or not it is reasonable to give all eight criteria equal weight (as here). The three options are scored from 1 to 5 on each criterion, 5 being "best", and the order of the criteria is accidental, so no particular importance should be attached to it. The overall conclusion of this exercise is clearly that the three options score equally well (30, 31, and 29, respectively), especially when the crudeness of such a scoring exercise is taken into consideration.
1. Simplicity of system
2. Additional data required to allocated seats to constituencies
3. Number of ballot papers required
4. Expected effects on the formation of national political movements ("parties")
5. Will political, regional and other diversities be well reflected in the Wolesi Jirga?
6. Can voters hold individual politicians directly accountable?
7. How easily can individual candidates be elected?
8. Level of overall proportionality (if no electoral threshold is implemented under options 1 and 3)
Sum of 1. -8.
- The issue of an electoral system for a presidential election (if needed) will have to be debated separately. However, all issues related to voter registration etc. will be as above, especially if a presidential election is conducted simultaneously with the Wolesi Jirga elections. The proposal here will be to conduct a presidential election as a two-round majority system, as is well known from many other presidential elections.
The three electoral system options discussed above must be carefully presented to all stakeholders. The options must be closely scrutinised and debated at the appropriate political levels before a decision is taken on which system to apply in the 2004 Wolesi Jirga elections. But there appear to be no urgent need to finalise this decision-making process as there are certainly more serious problems now in relation to other elements of the election preparation process.
The provisions for the 2004 Wolesi Jirga elections as well as the relevant title in the Constitutional Proposal should at least include the following elements and relevant portions of articles under Title Four in the 1964 Constitution:
- The Wolesi Jirga will have–as an example–225 members, all of whom
will be directly elected in free and fair, secret elections by universal suffrage
and proportional representation. The Electoral Act will determine whether
the elections will be conducted with the entire country as one PR constituency
(option 1), as 32 smaller multi-member constituencies (the provinces) (option
2), or as a combination of the two (option 3). If the provinces are to be
used as multi-member constituencies (under options 2 and 3), the borders of
the constituencies would be changed only if and when the provincial borders
- Before an election under options 2 and 3, each of the 32 multi-member constituencies will be allocated a number of seats, which will represent the entire Afghan nation as well as the provinces in the Wolesi Jirga. The number of seats allocated to each constituency will be in direct proportion to the sum of two figures: (1) the population as established by the most recent population census and (2) the geographical area of the province, multiplied by a factor10 (or something similar), as determined in the Electoral Law. In the first elections (when census figures may not be available) the number of registered voters in each province will replace the population figures when seats are being allocated to provinces. The Electoral Commission will carry out the necessary calculations and take the necessary decisions, which will then be published as decided by the EC. Figures and calculations must also be gazetted, if a Government Gazette is available. If option 3 prevails, 35 seats of the 225 should be set aside as compensatory seats.
An Independent (Central) Electoral Commission should be established. The Electoral Commission will have x members, of which at least one shall have legal qualifications at par with a judge of the High Court. The chairman and the members of the Electoral Commission are appointed by the president, when they have been identified according to procedures laid down in the Electoral Act. The procedures must provide for an open, transparent and unbiased scrutiny of possible candidates, who must not at the time of their appointment or during their term of office be members of any political movement or party. The Electoral Act must also specify (in general terms) how the Secretariat of the Electoral Commission will be structured and for which period the commissioners are appointed. The period should be longer than the expected parliamentary electoral term, which was four years in the 1964 Constitution, and also longer than the presidential term, if an elected presidency is eventually included in the Constitution.
The Constitution should also indicate that there will be some sort of Electoral Tribunal (Court), which will deal with serious electoral queries and complaints as stipulated in more detail in the Electoral Law. Complaints should always first be submitted to the Electoral Commission, which will function as first instance in election-related complaint cases.
Criteria for voter eligibility and membership of both Houses of the Shura (Parliament) must also be included in the Constitution. This is particularly important in relation to internally displaced persons (IDP) and refugees residing outside Afghanistan.
Electoral thresholds are not necessary under option 2; under options 1 and 3 the recommendation is not to implement any electoral threshold as it is more important to have all strands of opinion represented in the Wolesi Jirga than to have them only outside that body.
No special provisions for the representation of women have been included in these recommendations as there is no reason to believe that such provisions will be accepted and will improve the quality of the 2004 Wolesi Jirga elections. However, those preparing lists of candidates should be encouraged to include as many women as possible (and in winnable positions) on their lists.
If a system with an elected president is eventually included in the Constitution, provisions for the election of the president must also be included. A two-round majority system is proposed, as it will ensure that the president has been supported by more than 50% of the voters whether he or she is elected in the first or the second round.