Data on dams is from the World Registry of Large Dams, maintained by the International
Commission on Large Dams (ICOLD). The registry lists all large dams in India, completed
or under construction, together with the nearest city to the dam and date of completion.
We use city information to assign dams to districts in the year of completion.
Data on district area, river kilometers, district elevation and gradient and river gradient are
collated from two GIS ﬁles: GT OP O30 (elevation data, available at
http://edcdaac.usgs.gov/gtopo30/gtopo30.html), and ’dnnet’ (river drainage network data,
available at http://ortelius.maproom.psu.edu/dcw/). The ﬁles were processed by CIESIN,
Earth Institute Columbia University using ARCGIS software. Polygon-wise GIS data exists
for every district. District gradient and elevation was computed as % district land area
in diﬀerent elevation/gradient categories (summed across polygons in district). For river
gradient we used the same process but restricted attention to polygons through which the
river ﬂowed. We identiﬁed neighboring districts, and within them upstream and downstream
districts, from District Census Maps.
These data are from the Evenson and McKinsey India Agriculture and Climate data-set
(available at http://chd.ucla.edu/dev-data ), with an update. The data-set covers 271 In-
dian districts within 13 Indian states, deﬁned by 1961 boundaries. Kerala and Assam are
the major excluded agricultural states. Also absent, but less important agriculturally, are
the minor states and Union Territories in Northeastern India, and the Northern states of
Himachal Pradesh and Jammu-Kashmir. Data on volume produced, fertilizer used and area
cropped are from the original data-set (1971-1987). We use the average 1960-65 crop prices
to obtain monetary production and yield values. Data on irrigated and total cultivated
area and male agricultural wages span 1971-1994. All monetary variables are deﬂated by the
state-speciﬁc Consumer Price Index for Agricultural laborers in Ozler and Ravallion. (1996),
base year 1973-74.
Rural Welfare data
We use household expenditure survey data collected by Indian National Sample Survey
(NSS). These are All India surveys with a sample size of about 75,000 rural and 45,000
urban households. Households are sampled randomly within districts.27 Only NSS for 1973
regional averages were obtained from Jain, K.Sundaram, and S.D.Tendulkar (1988). For
the 1983-84, 1987-88, 1993-94 and 1999-2000 (“thick”) rounds, Topalova (2004) computed
district-wise statistics using the poverty lines proposed by Deaton (rather than those of the
Indian Planning Commission, which are based on defective price indices over time, across