How much will the vote swing in 2014?
Correspondent | Friday April 4, 2014 11:42
As the parties finalise their candidate lists and begin launching their manifestos and candidates, voters and observers will begin to assess their prospects. Since 2009, there has been considerable change in the party system and in political affiliations of individual politicians.
How will voters respond? Should we expect equally dramatic changes in voter behavior or more incremental movements in the 2014 elections? An analysis of vote swings in past elections provides some perspective.
The following chart tracks changes in vote shares (%) by party over time:
1965 to 1969: BDP -12.1%, BPP -2.1%, BNF +13.5%, other + 0.6%
1969 to 1974: BDP +8.3%, BPP -5.5%, BNF – 2.0%, other – 0.7%
1974 to 1979: BDP -1.4%, BPP +0.8%, BNF +1.5%, other -0.9%
1979 to 1984: BDP -7.2%, BPP -0.8%, BNF +7.4%, other + 0.6%
1984 to 1989: BDP – 3.2%, BPP -2.3%, BNF +6.6%, other -1.1%
1989 to 1994: BDP -10.4%, BPP -4.3%, BNF +10.1%, other 4.7%
1994 to 1999: BDP +2.8%, BPP 0, BNF -11.1%, BCP +11.9%, other, -3.9%
1999 to 2004: BDP -5.5%, BPP +1.9%, BNF +0.1%, BCP +4.7%, other -1.1%
2004 to 2009: BDP +1.6%, BPP -0.5%, BNF -4.2%, BCP +2.6%, other 0.8%
Average change between elections*
BDP: -3.01%
BPP: -1.42%
BNF: 2.43%
BCP: 6.40%
other: -0.11%
The BMD is not included in this chart (or the next one) because it has not yet contested elections. The average has been calculated for nine inter-election comparisons for the BDP, BPP, and BNF. Since the BCP first contested elections in 1999, however, its average has been calculated for three inter-election comparisons only. The average rate of change for the BCP to date should not be taken as a predictor for its performance in the 2014 elections because (1) there are so few data points, (2) its initial performance weighs heavily in the results, and (3) the rate of growth for the party has been slowing from one election to the next. A more realistic expectation would be that, if the trend continues, support for the BCP would continue to grow but at a lower rate, perhaps around 1.5% - 2%.
It is not evident, however, that trends will continue. The figures for the other parties reveal a considerable degree of movement back and forth over time. While the vote share of the BDP, for example, has generally declined over time, it has done so in fits and starts. Likewise, even if support for the BNF expanded at an accelerating rate from 1974 to 1994, there is no clear trend since the BCP broke away.
The biggest shifts in vote shares occurred in 1969 (-12.1% for the BDP and + 13.5% for the BNF), 1994 (- 10.4% for the BDP and +10.1% for the BNF), and 1999 (-11.1% for the BNF and + 11.9% for the BCP). All three examples involved swings in vote shares of more than 10% from one party to another. Two of these swings followed an important realignment of the party system: the formation of the BNF and Bathoen’s decision to join it prior to the (early) 1969 elections and the split of the BNF and formation of the BCP prior to the 1999 elections. Although no new party formed in 1994, those elections were preceded by several other important developments, including a public sector strike, mobilization of the women’s movement, high profile corruption scandals, slowing of economic growth, the consolidation of the opposition behind the BNF, and the intensification of factional conflict within the BDP.
The situation leading up to the 2014 elections shares several features with previous election years that saw relatively large vote swings, the most notable of which are the formation of the BMD following a split in the BDP, ongoing conflict within the BDP, the recent public sector strike, high profile corruption scandals, and economic problems, including those related to unreliable supply of electricity and water.
The BNF and BCP received 13.5% and 11.9% of the vote in their respective inaugural elections. Should the BMD expect a similar performance? There are some important differences between the formation of the BMD and the formation of the BNF and BCP that can be expected to influence the magnitude of its electoral impact. First, both the BNF and the BCP had to contest elections for the first time within a year or so of their formation. In fact, the formation of the BNF prompted early elections in 1969 (instead of 1970). According to the conventional wisdom, these new parties were at a disadvantage because they had not had time to fully regroup before confronting the electorate. Since the BMD formed four years ago, it has had more time to get organized and might hope to perform even better than previous new parties. On the other hand, the excitement surrounding the formation of a new party inevitably subsides over time. If it were contesting elections on its own, the BMD might expect its inaugural performance to be more comparable to that of second elections contested by the BNF (11.5%, down from 13.5%) or the BCP (16.6%, up from 11.9%). Second, however, and unlike the BNF and the BCP, the BMD will contest its inaugural election as part of an alliance in the form of the UDC. Since the BMD as such will not contest in all constituencies, the aggregate vote of the BMD candidates is inevitably limited.
The UDC, of course, hopes that this partnership will galvanise opposition voters, so that BNF voters support BMD and BPP candidates and vice versa. According to cross-national research, however, opposition coalitions only improve competitiveness when major parties are involved (Resnick 2013). It remains to be seen to what extent the UDC will have the desired mobilising effect without the BCP.
Regardless of the effect of the BMD, the public sector strike, corruption scandals, conflicts within the BDP, and economic problems create a climate that seems comparable to that of 1994, when there was a roughly 10% swing in vote shares even without the formation of a new party. Again, the situation is not identical. Corruption scandals have become so common place that they may be expected to have less influence in 2014 than in 1994. On the other hand, tensions between government and the civil service have only gotten worse. The economic situation appears to be better than in 1994 in that GDP is growing rather than contracting. (According to Statistics Botswana (2014) the rate of GDP growth picked up in 2013. This reflects positive developments the mining sector, which outweighed problems in the energy sector). Disruptions in electricity and water supplies create an uncertain investment climate, however, and the recent electricity supply crisis can only act as a break on economic performance in 2014.
There is currently no civil society mobilisation around elections comparable to that of Emang Basadi prior to the 1994 elections and, rather than consolidating behind a single party, the opposition remains divided. Lack of mobilisation around the elections (as seen in low rates of voter registration) means that any disgruntlement in the electorate may not be fully reflected in voting patterns.
Considering that the BDP suffered a split and faces a tough socio-economic context, previous vote swings under similar conditions suggest that the BDP can expect to see its vote share decline by at least 10%; a 15% decline in its vote share cannot be ruled out. If the BDP’s campaign goes incredibly well, something happens to discredit the opposition parties, or disgruntled citizens abstain rather than vote for the opposition, the BDP might be able to limit the decline in its vote share to 5 - 6%.
These estimates suggest that the BDP will receive between 38.3% and 48.3% of the vote in 2014. It is more difficult to evaluate to extent to which votes are likely to swing to the UDC or the BCP. Past party splits have seen a near equivalence between the inaugural vote share of the new party and the decline in vote share of the party that suffered the split.
While this “new party effect” should benefit the UDC, or at least the BMD candidates in its areas of strength, any shift in voter behavior related to the general social malaise is likely to boost support for the BCP as well as the UDC. The responses of activists and the broader public to the UDC – and to the absence of the BCP from the UDC – have been sufficiently varied to date that they provide little indication as to who is likely to benefit the most from any swing away from the BDP. Indeed, some may opt to case protest votes for independent candidates rather than support either of the main opposition formations.
There is a tension between the expectation of a substantial shift in electoral support based on a comparison of current conditions with past elections and the lack of any statistically significant shift in party support in the by-elections that have been held since 2009. Again, the lack of evidence of any significant shift in party support over the course of the by-elections was surprising precisely because there has been so much change in the party system and in party affiliations. By-elections, of course, are not general elections. As some in the opposition will point out, the one-at-a-time nature of by-elections allows parties to concentrate their resources and, although all parties benefit from the ability to focus their energies, the BDP is expected to benefit disproportionately because it has access to more resources than any of the opposition parties.
Even if the advantage the BDP is thought to gain from the sequential nature of by-elections has been over-stated, the constituencies that held by-elections are not representative of all constituencies. There were only four parliamentary by-elections, of which only three were included in my earlier analysis (because of the unusual circumstances surrounding the Francistown West by-election). These three constituencies were all in either Central district or Kweneng, traditional strongholds of the BDP. While the council by-elections were scattered across 16 constituencies in six districts plus Gaborone and Francistown, nearly half (7) of them took place in Central district. It is entirely possible for the BDP to retain support in historical heartland areas like Central district and rural parts of Kweneng district even if its vote share drops sharply in urban and peri-urban areas. In fact, one could imagine that a perceived threat to the BDP’s hold on urban constituencies might prompt defensive mobilization of the party’s loyalists in heartland constituencies. While the lack of any statistically significant shift in voting patterns in the by-elections since 2009 thus does not preclude the possibility of substantial shifts in electoral support in the general elections of 2014, it does suggest that some caution is in order. In particular, although the opposition might hope that the combination of a party split with challenging socio-economic conditions will produce a stronger shift in voter support than has been observed in any previous election, the record of the by-elections and the low rates of voter registration (among other things) suggest that shifts in electoral support may be more modest (e.g., 5 – 10%).
The next question is how any shift in voter support will affect the distribution of seats in parliament. In countries that, like Botswana, elect representatives from single member districts on a first past the post basis, shifts in vote shares do not translate directly into shifts in seat shares. The next chart shows changes in parliamentary seat shares (%) by party over time.
Change in seat share …
1965 to 1969: BDP -12.9%, BPP 0%, BNF +9.7%, OTHER +3.2%
1969 to 1974: BDP +7.0%, BPP -3.4%, BNF -3.4%, other -0.1%
1974 to 1979: BDP +6.3%, BPP -3.1%, BNF 0%, other -3.%
1979 to 1984: BDP -8.3%, BPP -0.2%, BNF +8.5%, other 0%
1984 to 1989: BDP +8.8%, BPP -2.9%, BNF -5.9%, other 0%
1989 to 1994: BDP -23.7%, BPP 0%, BNF 23.7%, other 0%
1994 to 1999: BDP 15.0%, BPP 0%, BNF -17.5%, BCP +2.5%, other 0%
1999 to 2004: BDP -5.3%, BPP 0%,, BNF 6.1%, BCP -0.7%, other 0%
2004 to 2009: BDP +1.8%, BPP 0%, BNF -10.5%, BCP +7.0%, other 1.8
Average change in seat shares between elections
BDP: -1.3%
BPP: -1.1%
BNF: +1.2%
BCP: +2.9%
other: 0.2%
These data reveal even more volatility than the data on vote shares. In fact, the correspondence between changes in vote shares and seat shares is weak. In some years, the electoral system dampened swings in vote shares. In other years, changes in parliamentary representation amplified swings in voter support. Sometimes, shifts in vote shares and seat shares even move in opposing directions.
In 1979 and 1989, for example, the BDP’s seat share increased despite declines in its vote share. In a similar manner, the BCP’s seat share decreased in 2004 despite growth in its electoral support. Nonetheless, the years that saw swings in voter support of 10% or more produced swings in seat shares of equal or greater magnitudes. Shifts in seat shares have exceeded 10% for at least one party in four elections: 1969 (a 12.9% decline for the BDP), 1994 (a swing of 23.7% from the BDP to the BNF), 1999 (a 17.5% decline for the BNF and a 15% recovery for the BDP), and 2009 (a 10.5% decline for the BNF).
The aggregate data suggest that, if the BDP does in fact experience a drop in electoral support of more or less 10% in 2014, it should expect to see its seat share decline by at least 10% and perhaps even as much as 20%. Even if the BDP’s seat share does fall by 10 to 20%, however, it would remain in power with 33 to 39 seats.
Much hinges on the spatial distribution of changes in voter behavior and partisan competition. If there is a new party effect that favors the BMD, it is most likely to be felt in urban and peri-urban constituencies plus, perhaps, the North West. Urban and peri-urban constituencies have also been the most severely affected by utilities disruptions. Thus, the BDP is likely to register its lowest levels of support in these constituencies. Because the BCP has substantial support in these areas and is not part of the UDC, competition between opposition parties will be greatest in these areas as well. In constituencies with tight three ways races, a candidate can win with as little as 35% and the outcome is anybody’s guess. Elsewhere, as in many constituencies in Central district and western Kweneng, the BDP can retain many seats even if its vote share plummets by 15%. Nonetheless, it is likely that, at a minimum, these elections will see a sharp increase in the proportion of constituencies that are competitive.
While a projection based on careful analysis of constituency level analysis will have to wait for another day, an initial, very rough, and relatively conservative analysis suggested that, even if the BDP’s aggregate vote share declines by less than 5%, to 49% - an outcome that would represent a real accomplishment given challenging circumstances - it could see its representation in parliament drop from 43 to 34 seats.
Again, a lot hinges on the spatial distribution of shifts in electoral support and the prevalence of tight three-way races. Much can change between now and the elections. For the moment, the main observation is simply that there are good reasons to expect a substantial swing in voter support in 2014 and that, if such a swing does in fact materialize, it may produce dramatic changes in the parliament.
*Amy Poteete is a Canadian researcher