This paper examines the impact of refugee flows, following the 1947 Partition of British India, on long-term agricultural development in receiving districts in India.
The authors employ a difference-in-differences approach for a subset of districts for which consistent agricultural data are available for the pre- and post-Partition periods. Data on refugee presence are drawn from the 1951 census, while data on agricultural yields (from 1957 to 2009) are drawn from the Indian Agriculture and Climate Dataset (IACD) and the Village Dynamics in South Asia Dataset (VDSA). Pre-Partition agricultural data from the Agricultural Statistics Reports of British India (for four years between 1910 and 1940) are used to evaluate whether refugees might have moved to places pre-disposed to agricultural growth. The authors control for agricultural characteristics of the district (soil types), altitude, latitude and longitude, state fixed effects, calendar year fixed effects, and state-time trends (pre- and post-Partition).
Several interesting findings emerge from the analysis:
Districts with a larger refugee presence in 1951 had significantly higher agricultural yields in the decades following India’s independence. Compared to districts with a lower refugee presence in 1951, districts with a larger refugee presence saw average annual wheat yields increase by 9.4 percent between 1957 and 2009. There is a particularly large and statistically significant effect during the decade of 1977–1987—the height of the Green Revolution period in India—in the high refugee districts. The figure below, which plots the year and refugee interaction coefficients on the y-axis, illustrates this finding: refugee presence in 1951 appears to be uncorrelated with trends in wheat yields before partition and for several years after the partition. However, after the start of the green revolution there is a clear ‘take off’ in high refugee areas;
The relationship between refugee presence and agricultural yields are only present for crops that were affected by the Green Revolution;
Districts with a larger refugee presence in 1951 were more likely to take up new agricultural technologies in the 60 years following Partition. Refugee presence at the district level is correlated with the adoption of high yielding varieties (HYV) of seeds, as well as the use of fertilizers and tractors;
The increase in agricultural yields and adoption of agricultural technology coincide with the beginning of the Green Revolution in India in the 1960s.
There does not appear to be a systematic correlation between factors that predict refugee flows and factors that may predict the take up of the Green Revolution. There is no correlation between pre-Partition canal irrigation and aquifer depth (endowments associated with the Green Revolution, which transformed Indian agriculture in the 1960s) and refugee presence. Additionally, there is no correlation between refugee migration and the presence of other infrastructure variables such as banks, post offices, length of roads, and hospitals by 1961 (before the Green Revolution). This mitigates the concern that, even if migrants did not choose districts based on agricultural yields, migrants may have chosen districts based on some other characteristic that was important for the spread of the Green Revolution (like roads, banks, or schooling);
Controlling for the British taxation system on agricultural lands, which has been shown to affect agricultural yields and the take up of the Green Revolution, does not affect the main estimates.
Refugee presence in 1951 appears to be uncorrelated with trends in yields for wheat prior to Partition, and even for many years after the Partition. However, there is a clear “take off” in districts with a larger refugee presence immediately after the start of the Green Revolution in India. The authors argue that refugee presence enabled the adoption of new agricultural technologies once the Green Revolution made it possible to do so.
The authors explore the possible mechanisms driving these results and present some preliminary evidence that refugee literacy played an important role in the take up of high yielding varieties of seeds; Indian districts that received refugees at Partition experienced a net increase in their literacy rates. There is also some suggestive qualitative evidence that the higher literacy of the refugees had already led them to engage in superior farming practices even before Partition. This mechanism is consistent with the broader literature demonstrating a positive relationship between education and agricultural development, based on the underlying hypothesis that better education enables farmers to understand, evaluate and respond to new developments, making them more likely to adopt new ideas and technologies.
Other potential mechanisms for the authors’ findings include: the involvement of Hindu refugees in money lending, which may have supported the take-up of new technologies; and land consolidation occurred in areas that had more refugees, which may have been more likely to see investments in technology during the green revolution.