Principal investigator: Bruno Benevit
Original title: Subsidy Policies and Insurance Demand
Authors: Jing Cai, Alain de Janvry and Elisabeth Sadoulet
Location of the Intervention: China
Sample Size: 3.442 Families
Sector: Political Economy
Primary Variable of Interest: Agricultural insurance coverage
Type of Intervention: subsidy
Methodology: OLS, IV, RDD
Summary
Household income and consumption fluctuate in response to various forms of risk, especially in rural areas. Formal insurance policies to protect against such risks, however, are not widely adopted in either developed or developing countries. This scenario presents challenges for many countries in expanding voluntary insurance coverage for individuals. To assess how subsidy policies impact agricultural insurance adoption, this study evaluated a subsidy and instruction policy for climate insurance in China for rice-producing families. The results showed that families who benefited from financial education permanently improved their adoption rate. Conversely, farmers with little knowledge about insurance continuously updated their adoption decisions based on recent experiences with disasters and payouts.
- Policy Problem
Individuals deal with risks in different ways. The decision to opt for agricultural insurance is influenced by several factors, including the level of financial education, risk perception, prior experience with insurance, and trust in the institutions that offer these products. In many cases, the low valuation of insurance is related to a lack of knowledge about long-term benefits and the rarity of events that trigger insurance payouts. Furthermore, economic barriers, such as the cost of policies and the availability of government subsidies, also play a crucial role (CAI; DE JANVRY; SADOULET, 2020).
Government initiatives aimed at increasing agricultural insurance uptake prioritize access to subsidies, dissemination of information, and improving confidence in insurance. However, evidence in the literature (BANERJEE et al., 2019; COLE et al., 2013) indicates low effectiveness of these policies in addressing how knowledge and experience regarding insurance influence individuals' decisions to take out this protection.
According to the authors (CAI; DE JANVRY; SADOULET, 2020), there are three channels through which initial exposure to insurance can affect long-term insurance demand: (i) the direct effect of experiencing a payment, with an expected positive effect on adherence if an insured shock occurred and a payment was received, and a negative deterrent effect if a premium was paid and no shock occurred or a shock occurred without a corresponding payment; (ii) the social effect of observing payment experiences in the network, which follows the same process of positive and negative effects in relation to stochastic payments; and (iii) a habit-formation effect, with past use of the product influencing current demand.
- Policy Implementation Context
Rice is the most important agricultural product in China, representing almost 50% of the country's agricultural workforce. In this context, in 2009 the Chinese government requested the People's Insurance Corporation of China (PICC) to create a rice production insurance policy for rural families in order to maintain food security and protect farmers from climate shocks. Initially implemented in 31 pilot counties, the program was expanded to 62 counties in 2010 and 99 in 2011.
Rice production insurance is based on the yield index of the area and covers natural disasters such as heavy rains, floods, windstorms, extreme temperatures, and droughts. Unlike climate index-based insurance, payments depend on the average loss in the area. If the yield loss is 30% or more, farmers receive payments proportional to the loss rate, ranging from 60 RMB per mu for a 30% loss to 200 RMB per mu for a total loss.
Rice production costs are approximately 400 RMB per mu, while the average gross income is around 800 RMB per mu. Insurance covered 25% of gross income or 50% of production costs, with a fair actuarial price of 12 RMB per mu per season. The policy was marketed in February, before the growing season in March. Premiums were deducted from a rice production subsidy deposited into the farmers' bank account in January, eliminating liquidity issues. Insurance payments were deposited directly into the same bank account.
- Evaluation Details
To assess the impact of subsidy policies on insurance uptake, the authors conducted a two-year field experiment between 2010 and 2011, in which subsidies were randomly granted to 134 villages with 3.442 households in rural China. In the first year, subsidy policies were implemented at the village level, where rice producers in the selected villages were offered agricultural insurance with a partial subsidy of 70% of the fair actuarial price (monthly cost of 3,6 RMB).
After the families decided to participate in the program, the experiment separated the villages into two groups: the "non-free sample" group, comprising farmers from 72 villages who participated in the program and received the partial subsidy initially offered, and the "free sample" group, comprising 62 villages where all farmers received the full subsidy (regardless of participation in the program). The samples showed similar participation rates, ranging from 40% to 43%, such that the insurance beneficiaries from the "non-free sample" and the "free sample" covered 43% and 100% of the families, respectively.
In the second year of the study, the subsidy level was randomized between 90% and 40% of the fair price at the household level, where farmers were randomly assigned a price from three available options in their village. The program also included a financial education program on insurance products for farmers randomly selected from 86 of the 134 sampled villages. Additionally, approximately 40% of the sampled villages were randomly assigned to the option defaults, where participation in the program was predetermined, it was up to the farmers to indicate whether they would not participate if they did not want the insurance.
The validity of randomizing the groups and prices was corroborated by the authors through regressions on the observed characteristics of the farmers, such as gender, age and education of the head of the family, family size, rice production area, risk aversion, and future perception of natural disasters.
- Method
The study considered a Bayesian theoretical model to formalize the learning process. In this way, this process was impacted by the price paid in the first year through three distinct effects: (i) a reach effect, where subsidies increase adherence and, consequently, the opportunity to experience or observe payments; (ii) an attention effect, where a lower insurance cost for the individual leads to less attention to the information generated by payment experiences; and (iii) a price anchoring effect, where low prices in the past reduce the current willingness to pay.
To estimate the impact of the agricultural insurance subsidy policy in China, the authors conducted several ordinary least squares (OLS) regressions, examining channels through which initial exposure to insurance can affect long-term insurance demand. First, the aggregate effect of the intervention on insurance uptake was verified, estimating the demand curve at insurance prices and the impact of the total subsidy policy on rice producers.
Subsequently, the direct effect of experiencing the payment of the agricultural insurance premium was analyzed. For this analysis, the effect of directly receiving the premium payment and the effect of witnessing the premium payment by more than 50% of the farmer's contact network were estimated separately. In addition to the OLS method, a regression discontinuity (RDD) was also conducted around the critical point of 30% crop losses. For this analysis, only farmers who initially joined the program were considered.
To verify the existence of a long-term habit formation effect, the study observed how past receipt of premiums in the first three years and access to a financial education course affected farmers' adherence to agricultural insurance in the fourth year. To this end, a model was estimated considering the isolated effect of having received the insurance premium for each year, identified from binary variables, and the effects of receiving the premium in more than one year (1st and 2nd year, 1st and 3rd year, 2nd and 3rd year, and 1st, 2nd and 3rd years). Finally, the authors conducted tests to verify loss memorization using a two-stage least squares (2SLS) model, considering as instruments the interaction between farmers' agricultural losses and binary variables identifying farmers in the "free sample" group and the [unclear] group. default and according to the price drawn in the 2nd year.
- Main results
The results regarding the aggregate effects of the program revealed a 6 percentage point (pp) higher participation rate for farmers who received full subsidy in the first year compared to farmers who received partial subsidy. The price elasticity of demand for insurance was -0,44 for the subsidized price level of the program (3,6 RMB). The addition of covariates to the models did not significantly alter the results.
Regarding the effect of receiving a premium in the first year, farmers who received a partial subsidy in the first year showed an increase in the enrollment rate in the second year of approximately 10 to 36 percentage points, reducing the price effect by about 80%. For farmers who received a full subsidy in the first year, this increase ranged from approximately 5 to 17 percentage points. The results were significant for both empirical strategies. The results also demonstrated a 20 percentage point increase in insurance enrollment for farmers who did not join the program in the first year and witnessed other producers receiving the insurance premium, and no significant effect for those who had joined in the first year.
The analysis regarding the identification of long-term habit formation showed results indicating that exposure to financial education courses influenced the impact on farmers' adherence to insurance. The adherence of farmers who did not receive financial education was only significantly positively influenced by receiving premiums in the 3rd year, the period prior to the year analyzed (4th year), while the adherence of families who received financial education was also significantly affected by receiving premiums in the 1st and 2nd years, indicating a greater understanding of the benefits of insurance on the part of these farmers.
- Lessons in Public Policy
This article investigated the impact of subsidy policies on rice crop insurance uptake against climate shocks among farmers in China. To this end, a two-year randomized field experiment was conducted in 134 villages across the country, where partial and full subsidies were offered, in addition to a financial literacy program.
The evidence from this study highlighted that farmers are moderately sensitive to the level of subsidized insurance prices. The experience of receiving the insurance premium significantly increased adherence over time, even when subsidies were reduced, highlighting the relevance of a successful initial experience with insurance. Furthermore, the study emphasized the importance of farmers' perceived gains for adherence to insurance subsidy policies. The design of this policy reveals how a better understanding of the program by policyholders can induce continued long-term adherence, contributing to financial security and resilience in the face of climate shocks to agricultural productivity.
References
BANERJEE, A. et al. The Challenges of Universal Health Insurance in Developing Countries: Evidence from a Large-scale Randomized Experiment in Indonesia. Cambridge, MA: National Bureau of Economic Research, Aug. 2019.
CAI, J.; DE JANVRY, A.; SADOULET, E. Subsidy Policies and Insurance Demand. American Economic Review, vol. 110, no. 8, p. 2422–2453, 1 Aug. 2020.
COLE, S. et al. Barriers to Household Risk Management: Evidence from India. American Economic Journal: Applied Economics, vol. 5, no. 1, p. 104–135, 1 Jan. 2013.
