Note: This case study is from before the 2005 elections  

Organizing elections in Afghanistan that will be viewed by all major stakeholders as free and fair presents a major technical and logistical challenge to the United Nations (UN) [1] – a challenge that is even further complicated by the tight timeframe and the ongoing security issues. Two of the many problems facing the UN in planning for the parliamentary and local elections in Afghanistan are the delimitation of provinces and districts and the allocation of parliamentary seats to provinces on the basis of population.

Background

Afghanistan has been at war for most of the last 25 years. The Soviet Union invaded in 1979 (following a communist coup within the country), but was finally forced to withdraw 10 years later by anti-Communist mujahidin forces supplied and trained by the US, Pakistan, and others. Fighting subsequently continued among the various mujahidin factions, giving rise to a state of “warlordism” that eventually spawned the Taliban.

Backed by foreign sponsors, the Taliban developed as a political force and eventually seized power over most of the country, aside from Northern Alliance strongholds primarily in the northeast. Following the September 11th 2001 terrorist attacks, however, a U.S.-lead invasion forced the Taliban from power.

Shortly after the fall of the Taliban, representatives of various Taliban opposition groups met under the auspices of the United Nations in Bonn, Germany, and agreed on a plan for the formulation of a new government. Hamid Karzai was inaugurated Chairman of the Afghan Interim Authority (AIA) on December 22, 2001 and in June 2002 he was elected President by a national Loya Jirga (General Assembly).

The Transitional Authority was given an 18-month mandate within which to convene a Loya Jirga and adopt a constitution, and a 24-month mandate to hold nationwide elections. A constitution was ultimately adopted on January 4, 2004 (several months after the deadline); elections were originally scheduled for June 2004, but these were postponed until October 2004. Although both the Bonn Agreement and the Constitution specify that, if at all possible, the first elections should include both presidential and parliamentary elections, only presidential election will be held in 2004. Parliamentary elections are currently scheduled for April 2005.

Delimiting Boundaries for the Parliamentary Elections

The Constitution dictates that the National Assembly in Afghanistan be composed of two chambers: the Wolesi Jirga and the Meshrano Jirga. Only when both houses are in session can the National Assembly fulfill its legislative functions (Chapter 5, Article 87 & 94).

Both parliamentary and local elections must be conducted in order to convene the National Assembly: Parliamentary elections are required to select Wolesi Jirga representatives and local elections (provincial and district) must be held in order to indirectly elect members to the Meshrano Jirga.

Before either set of elections can occur, however, consensus must be reached on the configuration of provinces and districts within Afghanistan because both parliamentary and local elections depend on these administrative units to serve as electoral constituencies.

Constituencies for Parliamentary and Local Elections

The boundaries of electoral constituencies must be established before parliamentary and local elections can proceed.

Using Provinces as Electoral Constituencies for the Wolesi Jirga Because a regional rather than a single-constituency List Proportional Representation (PR) electoral system is being introduced for election to the Wolesi Jirga, some delimitation of constituencies will be necessary. This process is likely to be limited to the adoption of provinces as electoral constituencies and the apportionment of parliamentary seats to these provincial constituencies on the basis of population.

In Afghanistan, this process will be complicated by several factors, one of which is a possible change in the number, and therefore the boundaries, of the provinces[2]. In early March 2004 (and for many years prior to this), there was general agreement that 32 provinces exist; however, pending before the Ministry of the Interior were plans for at least three additional provinces. At the end of March, one of these proposed provinces was declared a new province by President Karzai. More decrees establishing new provinces may follow even closer to the election.

Electing Representatives to the Meshrano Jirga According to the Constitution, elected members of the Meshrano Jirga are to originate from two sources[3]:

• Each provincial council is to elect one member of the council to serve as a member of the Meshrano Jirga
• From among the district councils of each province, the respective councils are to elect one person to serve as a member of the Meshrano Jirga

One important consequence of this constitutional provision is that provincial and district (local) elections must occur before the Meshrano Jirga can be convened. Local elections will require clearly defined provinces and districts. However, as mentioned above, provincial lines may be in flux. District boundaries are even more problematic.

Delimiting District Boundaries

There is no consensus on the number of districts, or on what the boundaries of these districts might be in Afghanistan. The Appendix provides a comparison of some of the more relevant lists of districts in Afghanistan: the United Nations Assistance Mission in Afghanistan (UNAMA) list of districts by province used by the Electoral Component of UNAMA for voter registration purposes; an updated list of provinces and districts provided by the Afghan Central Statistics Office (CSO) in March 2004; and the list of districts used for the Emergency Loya Jirga (ELJ) elections in 2002 and the Constitutional Loya Jirga (CLJ) elections in 2004.

A comparison of district lists from these sources provides only an indication of the problem – even if all of the lists concur on the number of districts to be found in a given province (which they do not), this does not mean that there is agreement on what the boundaries of these districts are, and hence what villages are assigned to each district. On the other hand, if the total number of districts per province does not correspond across lists, then clearly there is disagreement as to village-to-district-to-province assignments.

According to the village-to-district-to-province assignment list provided by the CSO to UNAMA in November 2003, there are 32 provinces and 387 districts. Although UNAMA regarded this information as static, the CSO is in fact updating this information as the pre-census enumeration process continues and more up-to-date information is obtained.

The CSO began the pre-census process in January 2003 with what was known to be an outdated list of provinces, districts and villages – the list was compiled in 1979 for the only attempted census in Afghanistan, which was never completed. The CSO census team has been updating this list on a rolling basis[4]; as of March 2004, district lists for 22 of the currently existing 32 provinces have been updated.

UNAMA designed the voter registration process on the basis of the village-to-district-to-province list provided by the CSO in November 2003; the UNAMA list has not updated since this date, despite periodic updates in the village-to-district-to-province list issued by CSO.

The UNAMA staff has recognized the problematic nature of their village-to-district assignment list[5]. Internal checks on the voter registration data collected to date, for example, have determined that:

• 12% of the villages identified by registered voters are in “unlisted” villages (that is, do not appear on the UNAMA list)
• 3% of those registered thus far have registered to vote in an “unlisted” village and this percentage is likely to increase substantially as the registration process moves from the urban to more rural areas[6].

Even more problematic, however, than the discrepancies in district lists among various agencies is the fact that even the most recent list of districts (as represented by the updated CSO list) may not reflect reality. Even though there are ostensibly rules and regulations for the establishment of new provinces and districts, these rules are not strictly enforced, especially with regard to the formation of new districts[7]. As became evident during preparations for the Emergency Loya Jirga (ELJ) elections in 2002, quite a number of districts have “emerged” that the CSO seems unaware of.

In many cases, the new districts are the result of provincial governors rewarding supporters with administrative positions, or local commanders declaring the existence of new districts to award themselves with administrative positions. These districts, although not necessarily “recognized” by the CSO or the Ministry of the Interior, are usually acknowledged by the local elders and accepted by local voters. In order to conduct ELJ and CLJ elections, “negotiations” often had to occur between the ELJ/CLJ Commission and the local communities, usually resulting in the de facto recognition of these “new” districts for the ELJ and CLJ elections[8]. Ultimately, 465 districts were included in the ELJ/CLJ elections.

Data for Apportioning Wolesi Jirga Seats to Provincial Constituencies

Wolesi Jirga seats are to be apportioned to electoral constituencies on the basis of “population” according to the Constitution. The Constitution does not specify the type of population data to be used, but the apportionment process is almost always based on either census enumeration data or voter registration data[9].

Population data may also be required to determine the number of members to sit on the provincial and district councils. Although it is not strictly necessary, presumably the same data that is used to apportion parliamentary seats will be used to gauge the size of local councils.

At this point in time (March 2004), there is a complete absence of reliable population data on which to apportion parliamentary seats. No census, with the exception of an aborted pre-census in the late 1970s[10], has ever been conducted in Afghanistan. Nor is there a pre-existing civil registry or voters’ list available. By Election Day, however, there will hopefully be several sources of population data available for apportionment purposes. These include:

• Pre-census enumeration data
• Voter registration data
• Voter turnout data

 

Census Data A pre-census household enumeration is currently being conducted by the Central Statistics Office (CSO), with the assistance of the United Nations Population Fund (UNFPA). This phase of the census process, referred to as the household-listing phase, entails visiting every household in Afghanistan and counting the total number of persons in each household. The enumeration includes not only the total number of persons, but number of males and females and the number of persons under the age of 18 and over the age of 18 per household.

The enumeration process commenced in the province of Kunduz in January of 2003 and 22 provinces have since been completed. The anticipated completion date for this project, assuming no major obstacles are forthcoming (i.e., security issues), is June 2004. Depending on whether this schedule is maintained, and when the parliamentary elections are scheduled, this enumeration data may be available for apportioning parliamentary seats to provincial constituencies.

Voter Registration Data The Electoral Component of the United Nations Assistance Mission in Afghanistan (UNAMA) is currently compiling a voters list which will, of course, have to be completed by Election Day – how many days, weeks or months before Election Day, however, remains a question. This voter registration data could be used to apportion seats to parliamentary constituencies.

The advantage of using voter registration data, as compared to census data, is the fact that this data is likely to be less outdated than census data for seat apportionment purposes, at least in the long run. (For this first election, both the enumeration data and the voters’ list data should be equally up-to-date.) The disadvantage of utilizing this data is that the voters’ list may not be reflective of the total population, and may be more biased in some regions than others. This is especially true in areas where potential voters have decided to boycott the registration process for social or political reasons.

Voter Turnout Data Apportioning seats on the basis of posteriori votes is another, albeit far less common, possibility[11]. Theoretically, this approach offers an increased incentive for voters to turn out and cast a ballot. However, this supposition has never been tested and proven. What we do know is the following disadvantages of using posteriori votes: 

• The ambitious election timetable currently in place in Afghanistan may not be able to accommodate the added pressure on the civic education program to ensure that community leaders and voters understand all of the ramifications of NOT voting, and the voter registration program to make certain that ALL eligible voters can cast a ballot.
• Any inclination to boycott the voter registration process or vote (or any bias in the registration process) will have not only immediate partisan ramifications (number of parliamentary seats won), but ramifications for geographic representation that will remain in place until the next set of elections. 
• Given the currently scheduled release date for the CSO pre-census enumeration data, election officials (and any other stakeholders interested in carrying out the exercise) will be able to calculate the "effectiveness" of the registration exercise and to ascertain any bias in seat apportionment due to "ineffective" registration even before Election Day. If such a bias exists, this cannot be corrected (and may, in fact, be exacerbated) by apportioning seats on the basis of voter turnout.

Problematic Draft Electoral Law

Although the draft electoral law (dated 29 February 2004) on “Electoral Boundaries” provides for a streamlined process and a dispute resolution mechanism, there are some problems with the law as drafted. The major problem is the lack of a clearly delineated (or insufficient) timeframe for the identification of electoral boundaries and the resolution of boundary disputes. The lack of specification in portions of the “Electoral Boundaries” draft law may also cause difficulties. What follows are some comments on each of the provisions encompassed in Chapter III.

Article 11. Use of Existing Administrative Boundaries For the purposes of electing members of the Wolesi Jirga, provincial councils, and district councils, the provinces and districts entitled to elect representatives will be those designated by the President in a decree no later than 90 days prior to the election.

Article 11 does not make clear whether the President’s decree merely lists the names of the provinces and districts entitled to elect representatives or specifies the boundaries of these provinces and districts (and records the villages assigned to each district, and the districts assigned to each province). If the boundary delimitation process is to proceed in a timely manner, it is important that the decree include all information necessary to assess the electoral constituencies (provincial and district boundaries) and assign eligible voters to the correct district and province.

The time period established for the release of the Presidential decree – 90 days prior to the election – does not provide sufficient time to resolve possible boundary disputes and still organize elections effectively. (If a boundary dispute arises, this must be resolved before election administrators can assign voters to the correct districts and provinces.) In fact, the entire process, including the resolution of boundary disputes, should almost certainly be completed approximately 90 days prior to the election. Since the boundary dispute process is liable to take a minimum of 60 days to resolve (from the initiation of the evaluation to the resolution of any disputes), this suggests that the 90 day deadline for issuing the decree must be moved back at least an additional 60 days to 150 days prior to Election Day.

Article 12. Electoral Boundary Disputes (1) Disputes relating to the precise boundary of any province or district shall be heard and resolved for electoral purposes only by the JEMB (Joint Election Management Body).

Article 12 does not indicate who is allowed to initiate a dispute. If no limitations are placed on potential complainants, then the dispute process may be subject to manipulation. On the other hand, the process should be as open and democratic as possible, so limiting possible complainants excessively would not be wise. At a minimum, only citizens that are residents of the district or province being disputed should be permitted to lodge a complaint.

Article 12 also does not specify how a dispute is to be recorded. For example, is a written complaint submitted to the JEMB? Or should the JEMB (or perhaps some local entity like the Regional Electoral Coordinator) be required to hold a hearing, or series of hearings, allowing citizens to object to provincial/district boundaries? A public hearing process would be time-consuming, but would be more inclusive.

There is no time period stipulated for submitting a dispute to the JEMB. But if disputes are to be resolved in a sufficiently timely manner (approximately 90 days prior to the election), then they must be initiated well before 90 days prior to the election. (Allowing disputes closer to the election could result in the disruption of an election.)

(2) The Regional Electoral Coordinator shall gather information regarding the dispute from the disputing parties and other local sources and forward that information to the JEMB.

A time limit should be placed on the Regional Electoral Coordinator for gathering and forwarding information so that the dispute resolution process is not delayed. Furthermore, if the JEMB is to decide boundary disputes on the basis of “technical criteria” (Article 12.3), then a requirement that the Regional Electoral Coordinator gather information related to these “technical criteria” should be included.

(3) The JEMB, following consultation with the Minister of Interior, shall decide the dispute on the basis of technical criteria, taking into account the information made available to it by the Regional Electoral Coordinator, and shall notify the Electoral Secretariat, the disputing parties, and local government offices of its decision.

No indication of what “technical criteria” might be is provided in the text. Perhaps the purpose of including this stipulation is to avoid a decision motivated by political considerations, in which case “objective criteria” might be a better choice of words. On the other hand, perhaps “technical criteria” is meant to refer to such “traditional”, or “internationally accepted”, districting criteria as the size of the population, geographic contiguity and compactness, and remoteness of territory. Some consideration should be given to rephrasing the “technical criteria” requirement.

If the dispute resolution leads to a change in the boundaries as provided by the Presidential decree, some thought should be given to implications of this. Despite the admonishment that the boundary resolution is “for electoral purposes only”, the election of a district council to a district not included in the original Presidential decree, for example, has ramifications far beyond Election Day. For instance, district and provincial councils presumably have administrative functions associated with them. Are the President and Ministry of the Interior obliged therefore to accept the JEMB resolution? If so, perhaps the resolution of boundary disputes should not reside solely with the JEMB (in consultation with the Minister of the Interior). It may be that a separate Delimitation Commission, with representatives from both the JEMB and the Ministry of the Interior, should convene to resolve boundary disputes.

Finally, no deadline for resolving the dispute and notifying the Electoral Secretariat, the disputing parties and the local government offices, is included in the provision. But any boundary dispute must be resolved in sufficient time to allow election administrators and candidates to prepare for the election. Therefore the inclusion of a timetable for the dispute resolution process, and an adjustment to the deadline for the Presidential decree, is quite important.

Conclusion

Before parliamentary and local elections can take place, a decision will have to be reached as to the number of districts and provinces and the boundaries of these districts and provinces (or, if not the precise boundaries, then at least the village-to-district-to-province assignments). If new provinces (and districts) are to be created, acknowledgment of this must be forthcoming immediately. Furthermore, a moratorium on the formation of additional provinces and districts must be declared well before elections are to be held.

Although it is premature at this point to decide what data to use to apportion Wolesi Jirga seats, an analysis of potential data sources, and the reliability of these sources, should be conducted closer to Election Day. Assuming that the pre-census process has been completed and the enumeration data has been deemed reliable, this data may be the best option for apportioning parliamentary seats and determining the size of provincial and district councils. This is because the pre-census program is likely to reach more people than the voter registration process[12].

The electoral law, as currently written, should be revised. The major problem with the current draft is the inadequate timeframe for identifying electoral boundaries and resolving boundary disputes. A lack of specification in portions of the “Electoral Boundaries” draft law may also cause difficulties.

SEE TABLE in APPENDIX

Notes:

[1] The United Nations was responsible for organizing and conducting the upcoming elections in Afghanistan.

[2] According to the Cartography Department in Afghanistan, three potential provinces have been mapped, not just one – but these maps are not to be released until/unless the Interior Minister approves the establishment of the new provinces.

[3] The President is to appoint one-third of the Meshrano Jirga members; the other two-thirds are elected from the provinces, one-third from among the members of the provincial councils and the other one-third from among the district councils within each province.

[4] The list for each province was updated in consultation with the provincial governor, district leaders and village elders. Any disagreement in the list was to be resolved by the provincial governor.

[5] The information reported here was supplied by Matthew Blakely, Data Manager, Electoral Component of UNAMA.

[6] In one relatively rural province in which voter registration is currently underway (Bamyan), 9% of those that have registered have identified “unlisted” villages as their place of residence.

[7] The rules and regulations for creating new provinces and districts are the responsibility of the Local Areas Management Body, located within the Ministry of the Interior.

[8] The negotiation process at times was quite “delicate” and even led to violent clashes on occasion, according to informed sources.

[9] The choice of whether to use census data or voter registration data may be guided by either practical or theoretical concerns. For instance, census data may not be the best option if a general enumeration of the population is unavailable, outdated or inaccurate. On the other hand, registration data may not adequate for districting purposes if it fails to include information (i.e., demographic data) that is essential given the specific country context. From a theoretical perspective, delimitation based on registration data is likely to produce constituencies that are more equal with respect to the number of voters contained within them, but a counter-argument could be made that representatives serve all persons, not simply voters, and therefore districts should contain equal numbers of persons, not voters.

[10] The census was aborted in 1979 when the Soviets invaded Afghanistan.

[11] According to research in ACE's Comparative Data, only one country (Belarus) surveyed uses voter turnout data for apportioning seats to the legislature.

[12] The first phase of the census is relying on door-to-door contact rather than centrally located registration sites; the enumeration program is also less likely to be boycotted for social or political reasons.