Publication Details

Quarterly Journal of Economics, May 2007, v. 122, iss. 2, pp. 601-46. Available From:

Link to Source
Esther Duflo and Rohini Pande
Institutional affiliations
None specified
Grant-holding institution
None specified
South Asia
Agriculture and Rural Development, Energy
Irrigation & Drainage, Hydro
Irrigation & Drainage, Hydro
Equity Focus
None specified
Evaluation design
Instrumental Variables (IV)
Journal Article


This study assesses the impact of dam construction on rural poverty and agricultural production in India. With over 45,000 large dams worldwide, dam construction represents a significant field of investment. However, the overall impact of such constructions still remains widely controversial. On one hand, advocates of large dam investments argue that dam construction can increase irrigation capacity, reduce dependency on rainfall and improve the provision of water and electricity supply. On the other hand, opponents point out that such benefits are unequally distributed across population groups, arguing that downstream populations capture most benefits, whilst upstream groups suffer significant losses of forest and agricultural land as well as reductions in land productivity. Since 1950, Indian states have financed about 4,500 large dams, built mainly for irrigation purposes. The authors evaluate dams’ impact using district panel data on local geography, dam placement, poverty (374 districts) and agricultural production (271 districts) collected between 1971 and 1999.
The study measures through linear regression analysis how the number of dams located in a particular district, or upstream from a district, affects agricultural and welfare outcomes. To account for the fact that annual variation in dam construction across districts may be correlated with unobservable district-specific factors (e.g., political decisions), the authors develop an instrumental variables strategy based on the non-monotonic relationship existing between river gradients and the likelihood of dam construction. In other words, the researchers estimate the number of dams that were constructed in a particular district due to favourable river gradients (rather than unobservable confounders) and use these estimates as an instrument for the quantity of dams constructed in a district for productive purposes. In addition, they control for time-invariant characteristics, state-specific annual shocks and district-specific time-varying factors.

Main findings

Impact estimates show that large dam construction positively affects agricultural production in downstream districts. The authors report a statistically significant increase in gross irrigated area, with a 0.33 per cent increase per additional dam built. Large dams also positively affect downstream districts by increasing agricultural production (0.34 per cent), yield (0.19 per cent) and the production of water-intensive crops (0.47 per cent). In contrast, the results do not show any statistically significant effect on agricultural production in upstream districts. Moreover, the existence of a dam upstream from a district appears to mitigate the adverse effects of a negative rain shock on agricultural production, but on the contrary, the presence of a dam within a district seems to amplify these effects. Results on rural poverty show that the construction of each dam is associated with a 0.77 per cent increase of the poverty headcount ratio in upstream districts and a 0.15 percent decrease in downstream districts.
The authors note that the limited availability of data for poverty outcomes impedes a better understanding of the causal links existing between dam construction and poverty patterns. However, the study does suggest that the costs and benefits associated with large dam construction are unequally distributed between downstream and upstream populations. It appears that institutional quality might represent a determinant factor explaining this uneven distribution. The researchers notably point to the fact that dam construction in India is correlated with higher poverty increase in districts where historical institutional arrangements tend to favour the elite class. The authors therefore conclude by calling for further research on the possible links existing between dam construction and institutional quality.

Scroll to Top