Political economy of infrastructure spending in India

Stuti Khemani of World Bank has this very interesting paper on the topic.

The author says general public usually puts infrastructure projects like roads, power etc on top priority list. Things like social development- education, employment etc are lower on the list. However, government usually ends up spending more on social development and not on infrastructure. If you say government does what voters demand, there is an apparent puzzle here. 

This paper poses and examines a significant puzzle in the political economy of infrastructure in India—the co-existence of relatively low shares of capital spending in public budgets alongside evidence of large demand from poor voters. Several separate household surveys in different states provide systematic empirical support for the conventional political wisdom that bijli, sadak, and pani (electricity, road, and water) rank at the top of voter demands, and dominate electoral rhetoric. Yet, across the major states of India, public spending on capital works has barely averaged 13 percent of total spending (excluding debt servicing and interest payments), in contrast to over 21 percent on recurrent education expenses. Education, however, is cited far below infrastructure on citizen rankings of priorities for public action. 

The author points to several surveys which show people putting higher weight on infrastructure vs social spending:

In a large survey of over 5000 rural households in the southern Indian states of Andhra Pradesh, Karnataka, and Tamil Nadu undertaken by Besley et al (2004) between September-November 2002, the data show that 89 percent of respondents list water or infrastructure as among their top 3 priorities, while less than 6 percent list education.5 These three states are at the sample average for the shares of spending devoted to capital versus social programs and education. Ban and Rao (2009) report that for a sub-sample of these households, for which they gathered transcripts of village assembly meetings with public officials, 38 percent rank water as their top priority, another 38 percent rank roads, 7 percent each rank electricity and housing, 5 percent rank health, 2 percent rank employment, and 1 percent rank education. Infrastructure outstrips education by several orders of magnitude for every sub-group of household respondents—landed versus landless, low versus high caste, Hindu versus Muslim, women versus men (Ban and Rao, 2009, Table 5).

Data on similar questions in household surveys in other states show the same pattern—an overwhelming number of respondents rank village infrastructure as priorities, versus very few responses ranking education (Banerjee et al, 2007, for Uttar Pradesh; Chattopadhyay and Duflo, 2004, for West Bengal).

Why does this happen? Most Indians know the answer. It is because political parties are not sure whether infrastructure will lead to votes. As infrastructure projects have a huge gestation period and parties are not sure whether they will be voted till the project is complete. Plus, it it is a costly option as well.

The argument offered here to explain this puzzle is that public infrastructure projects are particularly suited to political rent extraction, the benefits from which are traded-off against the electoral gains from delivering infrastructure as a broader public good.3 At the same time, targeted provision of employment and welfare transfers to the poor can be more effective in ensuring their favorable vote, while the vote returns to delivering infrastructure as a public good can be more uncertain (relative to the provision of targeted benefits). Infrastructure provision,even when swing voters at large demand it, may therefore be too costly a means to win votes— both because if it is provided as a broad public good to win broad electoral support, it implies giving-up large rents; and because there is uncertainty in whether swing voters at large will indeed vote for the incumbent after receiving the broad public good.

Here the author comes with a very interesting analysis to prove the rent extraction theory:

The first testable implication of this argument is that when institutions that constrain rentseeking are weaker, the share of infrastructure in public spending should be higher. Second, when re-election requires winning votes from more voters (such as when more voters turn out at the polls, or when political competition is more vigorous), the share of infrastructure should fall, in favor of those types of public spending that deliver jobs and transfers.

 Some new evidence, albeit largely suggestive rather than conclusive, is offered on the variation of public spending composition across states, and within states over time, that are consistent with this argument. The share of capital spending is estimated to be greatest in states that are reputed to have poorer quality governance compared to others, and declines within these states when they experience greater voter participation. In contrast, the share of spending on social programs, including education as its largest component, designed to provide employment and welfare transfers to the poor, is lower in weaker-governance states, and increases in all states with increases in voter turnout.


This is quite different from the usual/intuitive thinking. Intuitively you would say that states where there is poor governance infra spending will be low and vice-versa.  But because infra projects are best suited for rent seeking, it is in weak governance states that you see higher infra spending. And as voter turnout increases, you see higher spending on social welfare programs.

Interestingly you get similar evidence in other countries as well:

This evidence across the Indian states, and its interpretation offered here, is consistent with recent cross-country research which finds greater public investment spending in countries with lower ratings for governance, political checks and balances and electoral competition (Keefer and Knack, 2007). They argue that the more plausible explanation for this pattern of variation is that pubic investment projects are vehicles for rent-seeking in weak governance environments. Other recent work on public expenditure composition in the Indian states has found correlations consistent with greater political competition reducing spending on capital projects and increasing social and education spending. Saez and Sinha (2009) find that education spending in the Indian states is greater when there are greater numbers of political parties competing for office. Teitelbaum and Thachil (2010) find the opposite effect of the number of competing political parties on economic services, consisting largely of capital spending.

 Read the paper for more details

One Response to “Political economy of infrastructure spending in India”

  1. Political economy of infrastructure spending in India « Mostly … « Builder Says:

    […] The author says general public usually puts infrastructure projects like roads, power etc on top priority list. Things like social development- education, employment etc are lower on the list. However, government usually ends up … infrastructure – Google Blog Search […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: