How can government promote economic development? This is a perennial question, and it has not gone away with the collapse of the Soviet model 20-plus years ago. The debate often becomes acrimonious and ideological, but through it all, one can be hopeful about the accumulation of experience and understanding. The road from experience to understanding is not straightforward, and requires rigorous analysis of data. I want to illustrate that point with the example of rural roads in India.
India’s government did not always seem to pay enough attention to the importance of connecting rural area with centres of economic activity. This seemed to change in the late 1980s, with programmes to build rural roads. The Pradhan Mantri Gram Sadak Yojana (PMGSY), started in 2000, took this effort to a higher level, with an ambitious plan to connect India’s smaller villages to nearby towns, all through the country. The ministry of rural development has been in charge of the scheme, with the Planning Commission playing its usual role. An earmarked tax proved inadequate as a funding source, and the World Bank stepped in with loans at concessional rates.
By the account of the World Bank and the ministry of rural development, the PMGSY has been a roaring success, improving agricultural practices, raising incomes and land prices and increasing literacy and access to health care. Some of these studies have tried to quantify the benefits in terms of social rates of return, which is rarely done for government spending in India. On closer examination, though, the claims of success are built on case studies that focus on a few selected rural roads or villages in a few selected districts. This makes it difficult to be confident about the overall, nationwide impact. One needs more rigorous quantitative analysis.
Some years ago, I had embarked on a project for RBI’s Development Research Group, in which I proposed to understand economic growth in India using the district as the basic unit of analysis. With my co-authors, I found that, using data for about 200 districts across nine states, there was clear evidence that the extent of a road