HOW likely is it that the PML-Q and its allies will gain equal to or more seats than they held in the last parliament? This is a fundamental question that is being debated in each and every voting household in Pakistan.

The answers to this question range from partisan political proclamations on both sides of the party political divide; casual armchair empiricism; subjective judgments about the PML-Q’s performance in government over the past five years; and a reading of the recently released international pre-election opinion polls.

The judgments being made by individuals and the expectations that result from them are going to fundamentally affect the legitimacy of the outcome of the 2008 general election in Pakistan.

If the result flies in the face of people’s expectations then at best we are going to inherit an extremely demoralised citizenry and at worst there is a serious possibility of agitation breaking out in the country.

An independent and objective answer to the question can be provided by analysing the historical election data put together by the Dawn Election Cell. It is important to bring the insights from hard data regarding vote swings and changes in seat shares across different election years and across different provinces to bear on this question, if only to provide a degree of objectivity to the debate.

One way to answer this question is to try and ascertain, using the National Assembly data of all elections between 1988 and 2002, whether we find a consistent historical pattern that shows the incumbent (or ruling) party and its allies gaining/losing vote shares relative to the vote share of the main opposition parties across consecutive elections.

In short, we ask whether voting in Pakistan exhibits a systemic pro- or anti- incumbency (ruling party) bias.

A systemic anti-incumbent bias would tell us that there is a tendency in Pakistan for a political party to lose voter shares relative to its main opposition in any election that it contests as a ruling party.

That is, all else being equal, the PML-Q is likely to lose its vote and seat share and is highly unlikely to make considerable gains. On the other hand a systemic pro-incumbent bias would suggest that we ought to be open-minded about the possibility of PML-Q making gains in the forthcoming elections rather than judge the electoral outcome with certainty today.

The anti-incumbency bias is not uncommon in developing countries and a number of studies suggest that it is particularly strong in India at the constituency level.

Therefore, it is an important question to ask whether the data suggests some inherent patterns in voting trends that would help us form expectations about the outcomes of the forthcoming general elections.

It would be sensible to calculate the incumbency bias separately at the provincial level on account of the ifferences in the structure of politics, the relationship of the provinces to the state establishment and the configuration of political parties across the provinces and the nature of party competition within them.

We calculate the incumbency bias in vote shares for Punjab and Sindh provinces because these were the two provinces in which the Chief Minister’s office was held by the PML-Q.

Figure 1(a) plots the difference in the vote share of the PPP with the PML-N in the Punjab and with the PML-N and MQM in Sindh across different elections years. Recall PPP was an incumbent in 1990 and 1997 and was a non-incumbent in all other elections. Figure 1(b) plots the difference in the vote share of the PML-N with the PPP in the Punjab and the difference in the vote share of the PML-N and the MQM with the PPP in Sindh across the different elections years.

Recall that the PML-N was an incumbent in 1993 and we add the MQM’s vote and seat shares to the PML-N as it was in opposition to the PPP in these years.

Both figures 1(a) and 1(b) show a strong anti-incumbent bias in voting in the provinces of Punjab and Sindh across the general elections of 1988-2002.

The crucial question is whether the anti-incumbency bias in voting behaviour results in an equivalent anti-incumbency bias in terms of seat share gained or lost by the incumbent party. Figures 2(a) and 2(b) conduct the same analysis as figures 1(a) and (b) for the number of seats won or lost by the PPP and PML(N) across the different elections.

It is now important to calculate whether the gain or loss in the difference in the vote share between the incumbent and the opposition parties led to a decline or an increase in the incumbents’ seat shares across the same two consecutive elections.

That is, whether the change in the vote share actually costs the incumbent (ruling) party in terms of percentage seats lost. Figures 2(a) and 2(b) show a strong anti-incumbency bias in terms of seat shares in the provinces of Sindh and Punjab.

In fact, the strength of the anti-incumbency bias is much stronger in the Punjab. Even if we ignore the 2002 elections because of the splintering of the PMLs we find that PML-N as an incumbent in 1990 lost almost 30 per cent of its seat share between the 1990 and 1993 elections in the province. Similarly, the PPP as the incumbent in 1988 lost 20 per cent of its seat share and in 1993 lost 30 per cent of its seat share in the province.

The data, therefore, suggests that if voting behaviour is in line with historical tendencies and elections are relatively fair and free, the PML-Q should lose its seat share in both Punjab and Sindh. It is certainly not likely to sustain or increase it. Furthermore, the loss in its seat share should be non-trivial and the suggested ranges of the loss are expected to be fairly high. It is important to realise that the anti-incumbency bias is likely to get exacerbated if there is a sympathy wave for the PPP following Ms Bhutto’s death, if the acute wheat and utilities crisis is seen as evidence of failure of performance, and if there is a grave distaste as a result of the aftermath of Nov 3 (imposition of emergency).

It could, of course, be argued that voting behaviour can deviate from the past behaviour. However, we find no reason to support that assumption given that the political personalities, political parties and voters remain the same today as does the fact that the country continues to grapple with crisis after crisis.

What is different about the current elections is that the establishment is backing the PML-Q, whereas the establishment tended to dislodge each and every incumbent government during the 1988 to 1990 period. However, if the only explanation for a PML-Q sweep is that the establishment backed it, then those who raise concerns about the legitimacy of the electoral process and the partisanship of the state will have a legitimate complaint to make.