Rainfall Insurance in Gujarat, IndiaPDF version

Shawn Cole
Jeremy Tobacman
Petia Topalova
Partenaires de terrain: 
Centre for Micro Finance (CMF)
Rural Gujarat, India
1900 households in 108 study villages of Gujarat
2006 - 2007
Problématique Politique: 
Accès à l'assurance
Problèmatique politique: 

Despite the gradual introduction of irrigation technology, agriculture in developing countries is highly dependent on natural sources of water, meaning that farmers must rely on erratic seasonal rains. Such dependency affects the income of both farmers and agricultural labors, and when droughts threaten, farmers are often forced to borrow from relatives, friends and neighbors. Informal risk-sharing arrangements are of limited value, however, because everyone engaged in agriculture in a given area is subject to the same productivity and revenue shocks from poor rainfall. To ease this dependency on erratic rainfall and reduce the risk of economic shocks, policymakers are seeking ways to mitigate this important source of risk.

Contexte de l'évaluation: 

While Gujarat is relatively well off compared to many Indian states, it still faces major development hurdles.  Close to fifty percent of the state’s population and most of the rural poor are dependent on agriculture, which is largely non-irrigated and rain fed.  Roughly half of the population in the study area was scheduled caste or scheduled tribe, with an average household of six people with monthly expenditures between Rs 1,100 and Rs 1,200.  

Weather Insurance is a fairly new financial product, which attempts to provide insurance against unpredictable and uncontrollable weather risk. In March 2006, the Centre for Micro Finance (CMF), Harvard University and the Self-Employed Women’s Association (SEWA) began offering this product, in the hope of understanding how well it can reduce risk, and what complementary interventions can make weather insurance more effective.

Description de l'intervention: 

In 2006 a baseline survey was administered to 1500 households in 100 villages of the Anand, Ahmedabad, and Patan districts of Gujarat, to gain data on demographics, income, savings, investment, attitudes towards risk, financial literacy, and experience with insurance.

Following the survey, SEWA and ICICI/Lombard began offering rainfall insurance to all villagers (not just survey respondents) in 32 villages randomly selected from the 100 surveyed.
It offered protection against both deficit as well as excess rainfall. The policy, covering 110 days over 3 phases, had an average premium of Rs. 202 and a total sum insured of Rs. 1485. SEWA marketing teams traveled to each of the 32 villages in order to explain and market the policy through village meetings and household visits. Payout was dependent on rainfall collected at the district level. Later that year the original 1500 households were surveyed again, as well as any other households that had purchased rainfall insurance.  In 2007 with better marketing techniques, weather insurance was offered to an additional 20 villages, though these policies covered only rainfall deficits. Households were resurveyed in March of 2008. Since then, in 2009, 8 entirely new treatment villages were added, bringing the number of villages in the study to 108 and the number of households to 1900.

Rainfall was not low enough to trigger payouts by the terms of the policy in 2006, 2007, or 2008. However, in 2009, households in several blocks of Ahmedabad and Anand districts received compensation for low rainfall according to the policy offered that year, which was underwritten by the Agriculture Insurance Company of India.

Résultats et implications politiques: 

Insurance Adoption: A total of 908 policies were sold to 826 households in the first year. Of the 500 initially surveyed households, approximately 20% eventually purchased insurance, suggesting a high level of trust in SEWA. This includes non-surveyed purchasers who received no treatment marketing. Take-up was positively correlated with household wealth and education, but not correlated with individuals’ expectations of monsoon quality in the coming year. Skill at predicting probabilities and aversion to risk are also highly correlated with the take-up, as is experience with an integrated SEWA insurance product.

Variations in the advertising flyers produced large variation in take-up. Households receiving flyers with “positive language” were 26% more likely to buy insurance than households receiving flyers with “negative language.” “Negative information,” on the other hand reduced take-up probability by 33% when combined with negative language.

Follow-up Effects: In the 7-8 months between purchase and follow-up, people who bought insurance forgot the terms of the ICICI rainfall insurance product policy. It appears that they were not able to figure out how patterns of excess or deficit rainfall would influence payments.

Due to normal rainfall in years the first three years there were no payouts. Analysis of the affect of payouts from year 4 is ongoing.