In a recent conversation with an Indian politician, I asked how they chose to spend their tax money between different groups of voters. He replied that it was prioritized: there would be the largest emphasis on “swing” voters, then on loyal (partisan) voters who vote for the party regardless and finally on voters who were partisan for other parties! In my research we reach a surprisingly similar result based on data on grants from the central government to 14 state governments in India during the period 1975 to 1997. The most important instrument in the control of incumbent parties is the allocation of the budget. When the central government allocates grants to state governments, we find that the alignment of the incumbent party at the state with the incumbent party at the center together with how “swing” a state is predicted to be in a particular state election affects how much is allocated: a state that is aligned and swing receives 16% higher transfers than a state which is unaligned and not swing.
Grants are used by state governments to finance public goods – e.g. road building, electrification of villages, telephone connectivity, water or sanitation facilities in poor neighbourhoods, employment guarantee schemes etc. These projects build goodwill for the party that’s credited with providing the grants. If a state government is aligned with the central incumbent then both levels of government benefit: voters will reward the party at the state level election or the central election. But if the state incumbent party is not aligned, then the credit is shared between two different parties: it is this that causes central governments to limit the grants to unaligned states. Swing states are those where the voters are undecided between parties. A small increase in goodwill can lead to large gains in votes. Of course the benefit only goes to the incumbent party at the center if the state is an aligned one. It turns out that parties allocate expenditures to maximize their chances of getting re-elected and this effect holds both for central as well as state level elections.
In our data, the Congress Party dominated central elections for most of the period under study. The Indian states which turn out to be electorally very important at the state level elections are Bihar, Kerala and Uttar Pradesh. Both Bihar and Uttar Pradesh had multi party contests during the period 1975 to 1997, where a small swing in favour of one party lead to a large change in election results. In Kerala, there were two main alliances: one led by the Congress party and one led by the Communist Party of India. Our measure of alignment is Bihar is high (0.7, where the maximum is 1) relative to other states, hence Bihar ends up being a favoured state.
To illustrate our results even better: compare the two states of West Bengal and Rajasthan: West Bengal has a low alignment with the central government during the period of study (it was dominated by the Communist party during this period) and was not a swing state during this period, Rajasthan on the other hand was aligned and swing. Rajasthan got an average per capita grant of Rs 115 (the highest) in 1980-81 prices while West Bengal got Rs 39 on average. Rajasthan also happens to be marginally higher ranked than West Bengal on per capita income. This shows the problems with letting politics decide the allocation of grants: governments look less at how deserving or needy a particular state is and more at how important the state is for re-election.
So, what can be done? It is generally agreed that grants decided by the Finance Commission are less manipulable for political gains, as they are governed by formulae based on the needs of state governments. A simple policy prescription from our study is to reduce the discretionary elements in grants and base them on economic considerations based on formulae.
India has a parliamentary democracy with a first past the post voting system. Parties are interested in winning the largest number of state elections, as winning in state elections is closely linked to winning in central elections. Our conclusions hold for such countries, but similar studies have been carried out in European countries and Latin American countries (e.g. Sole-Olle and Sorribas for the case of Spanish municipalities, Armesto for Argentina, Leigh for Australia etc) suggesting that while the precise way in which politics affects this decision may vary, the idea that politicians choose budgets not so much to benefit citizens but to increase their own chances of re-election is a robust one.
