As I write this, the exit poll results of the third phase have just been declared. These indicate a slight improvement in the electoral fortunes of the Bharatiya Janata Party and its allies compared to their returns from the first two phases. Exit polls have been known to go horribly wrong in India as well as abroad. But, to the extent that they can be believed, it appears that the National Democratic Alliance will perhaps fall just short of the magic figure of 272 seats in the Lok Sabha. Atal Bihari Vajpayee should then have no problem in cobbling together a working majority in parliament. After all, several small parties will gain representation in parliament, and at least some of them can be won over with the right incentives.

Even if he fails to persuade an adequate number of Lok Sabha members to formally join his government, it is almost certain that an NDA government will obtain sufficient tacit support to function for some time as a legitimate government. The anti-NDA opposition parties will fall so far short of a majority that they have no chance of forming a government. And all the politicians who manage to get elected will want the new Lok Sabha to complete its term so that they can reap the fruits of success. So, they will ensure that Vajpayee once again occupies the prime minister's office. Of course, the important question is whether the prime minister can push through any coherent agenda for change if he is subjected to political blackmail from coalition partners.

As many politicians and observers have remarked, coalition governments are here to stay. The first signs of political instability in recent times were witnessed after the general elections in 1967, when coalition governments were formed in several states. Many of these governments were highly unstable. However, although the Indian political system is formally a federal structure, most of the effective power is concentrated at the Centre, and so it is government stability or the lack of it in Delhi which occupies much greater attention in public discourse. Despite the potential for instability in the last decade -and-a-half, Central governments in India have tended to survive. For instance, the last Congress government with P.V. Narasimha Rao at the helm did not have majority support, but managed to last a full term. Similarly, the current NDA government, despite being a patchwork of, at times, 21 parties, has overcome all odds and exhibited remarkable longevity.

A popular misconception about coalition governments is that they are inherently unstable. Indeed, there is overwhelming evidence from around the world which suggests that the average duration in office of single-party majority governments is significantly larger than that of coalition governments. However, there is considerable variation in the duration of coalition governments, with several coalitions showing remarkable stability.

This is particularly true in the case of western Europe, where the evidence indicates a wide variation in the pattern of coalition governments as well as clear country-specific patterns -- some countries seem to have systematically more stable governments than other countries. For instance, Germany, Austria and Luxembourg have been ruled more or less continually by stable coalition administrations in the post-war period, while Italy has witnessed relatively unstable governments. More interestingly, Kerala has had a long history of coalition governments, some of them serving full terms in office.

There is a growing body of research in political science containing several competing hypotheses about cabinet durability. One school actually eschews attempts to identify factors that explain government durability, believing instead that duration is essentially a random event -- it is the occurrence of unpredictable events which trigger governmental collapse. This belief does not square with empirical evidence which suggests that some governments are systematically more successful in warding off shocks than others. At the other end of the spectrum is the "deterministic" school, which conjectures that cabinet duration can be explained by a set of structural factors such as the composition of the legislature or the cabinet, the ideological positions of the parties in the cabinet, as well as the institutional features distinguishing coalitional processes in different countries.

Several political scientists have suggested that the more fragmented the legislature, the less likely it is to produce a stable government. A fragmented legislature is one in which the distribution of seats is dispersed amongst several parties. The more fragmented the legislature, the greater is the number of potentially viable coalition governments. The increase in the number of feasible options makes the overall bargaining environment more complex. This promotes instability because coalitions tend to break up more easily in a complex bargaining environment since even a minor shock may change the preference of parties regarding their most preferred coalition.

Despite the growing tendency towards fragmented legislatures, there is very little discussion in India about steps which can be taken in order to reduce fragmentation. For instance, there is hardly any debate about possible reforms in the electoral system, which is a crucial determinant of the type of legislature.

Conventional wisdom amongst political scientists suggests that plurality rule or the first-past-the-post system, which is the electoral system used in India, is most likely to produce majority parties. Indeed, the so-called Duverger's Law, based on empirical regularities, states that single-member district electoral systems, in which winners are decided by a simple plurality rule, usually produce two-party systems.

Clearly, the electoral experience in India does not fit this general pattern. This suggests a need to examine why Duverger's Law has not been vindicated in India. A possible explanation is that while this "law" implicitly assumes that the distribution of electoral support for the various parties is uniform across all constituencies, the Indian experience indicates that perhaps voters care more about local issues than national ones. This is not to deny that all voters care equally about crucial economic issues such as prices and incomes. But there are equally important issues which are more localized. No national party has been able to formulate a consistent electoral platform which addresses the multiplicity of local issues, and that is why several regional parties have gained electoral success, contributing to a fragmented legislature.

An attractive alternative to the first-past-the-post system is the family of proportional representation systems, which award seats in proportion to the votes polled by any party. This means that PR systems tend to be more favourable to minority groups and small parties since the number of seats won by a party is "close" to its share of the overall votes. This means that a "pure" PR system is more likely to produce fragmented legislatures.

However, several countries in western Europe use some PR system, but with the modification that parties getting less than a stipulated fraction of the valid votes cast in all the constituencies put together are denied any representation in the legislature. This stipulation obviously rules out the possibility of any independent candidate or small parties gaining representation. Although there have not been too many instances of single-party majority governments in these countries, their legislatures usually comprise of 3 or 4 relatively large parties. Moreover, most of these governments have been formed by stable coalitions consisting of two or three parties. An added advantage is that these coalitions tend to be between like-minded parties, thereby promoting coherent decision-making.

The author teaches at the University of Warwick, United Kingdom.