(The Joint Electoral Management Body has a terrific photo gallery.)
The legislative elections in Afghanistan were held on September 18, but the results were not released until a few days ago. With the help of a map offered by the Joint Electoral Management Body, and the slightly easier to read format of the results posted by Adam Carr, the following is a selective analysis of the results. (For background on the electoral system–SNTV–and other aspects of the campaign, please see my Afghanistan subdomain.)
Analyzing the results and their possible national policy-making significance is difficult due to the absence of any real party system. What I want to highlight is the votes breakdown by candidate, given that this is an SNTV system, to see the extent to which my expectation about how this system would perform in the Afghan context proved correct. In a previous post on the elections I noted:
Was that the case? I will take a few provinces, semi-randomly (but including districts of different sizes and from different parts of the country), and take a closer look. My intention is not to perform a scientific analysis, but rather just to get an overview of the results. In this sense, what I will do here is similar to what I did on Liberia in two posts on November 11 and 12.
First, a note about the rules. It is SNTV (first M past the post, where M is the district magnitude), but with a proviso. 25% of the elected legislators were required by law to be women, so one or more male candidates who ranked among the top M vote-getters might be skipped to ensure that the requisite number of female candidates would be elected.
In each of the provinces that I survey, I will give some votes percentages in the following order: Top winner, last elected without gender quota, any women elected only due to gender quota [in brackets], and first two losers.
Kabul (33 elected): 13.8, 0.53, [0.51, 0.47, 0.46], 0.52, 0.51
Herat (17): 3.6, 1.51, [1.0, 0.8], 1.44, 1.37
Kandahar (11): 8.0, 2.8, [2.1, 0.9, 0.9], 2.7, 2.4
Badakhshan (9): 11.4, 3.0, [0.8], 2.8, 2.6
Sar-i-Pul (6): 10.2, 6.8, [3.3], 4.6, 4.1
Paktika (4): 13.3, 7.6, [6.0], 7.1, 5.5
Kapisa (4): 12.2, 7.6, [2.4], 5.1, 4.9
Badghis (4): 7.3, 5.4, [--], 5.3, 5.1
Daikundi (4): 17.2, 5.9, [--], 4.9, 4.3
Kunar (4): 8.2, 6.7, [1.5], 4.9, 4.2
Bamyan (3): 20.1, 6.4, [--], 6.2, 4.3
Nimroz (2): 14.6, –, [4.1], 13.2, 8.7
So, the expectation was largely confirmed. The margins between the last winner and first two losers tend to be small. The overall results are very fragmented, with a fairly typical result being that one candidate is well ahead of the rest of the field. Yet even the leading candidate has a small vote share in most cases. The largest vote share I saw was just over 35% and only a few had over 20% of their respective province. Such candidates, of course, are clearly local notables or even ‘warlords.’ Still, an advantage of SNTV over FPTP (as in Liberia’s lower house) is that other candidates aside from the strongest locally can be elected, and an advantage over PR is that these warlords’ own personal votes elect only themselves rather than a list of candidates. (Of course, if they have sufficient organization, they might be able to coordinate their supporters to divide their votes among several affiliated candidates; from just looking at results by candidate, with no party labels available, it is not possible to tell the extent to which this might have happened.)
It is well worth noting that the smaller-magnitude districts had some tendency to have both a leading candidate with a non-neglible vote share and trailing candidates who were farther behind the last winner than is the case in the larger districts. Take, for example, Kunar, where there is a clear fall-off between the last winner and first loser (ignoring the gender quota). The first loser in Bamyan just missed, but the next one is much farther back. (Yes, to call 2 percentage points “much farther back” is odd, but in this context, it is also accurate.)
The smaller magnitude districts, thus, appear to have had somewhat better coordination than the large ones, just as we would expect under SNTV. Larger magnitudes compound the low-information problem of a first election and a party-less campaign. Kabul shows the lack of information about which trailing candidates were in the running, with the last winner (before gender quota) at only two hundredths ahead of the second loser.
And look at Herat (also a large-magnitude district): the first winner had only 3.6% of the vote!