Seeds of (climate) change: Private adaptation and subsidized insurance in West Bengal

We use a simple model of decision-making under risk to investigate how subsidized in-
surance impacts the demand for private climate adaptation. Under actuarily fair insurance
pricing, and absent other market failures, farmers would choose the private (and also socially)
optimal level of investment in both insurance and other adaptation technologies. However,
we can then show that subsidized insurance has two counterveiling effects on adaptation
investment, in the spirit of Ehrlich and Becker (1972). On one hand, because climate in-
surance and private climate adaptation both protect agents against the same risks, subsidies
that lower the price of insurance may lead to a “crowd-out” effect. On the other hand,
because climate insurance protects agents from facing the full brunt of the bad state of the
world if a costly adaptation measure fails, these subsidies may instead lead to a “crowd-in”
effect. Which of these effects ultimately dominates is ambiguous.

We set up and implement a sharp empirical test of the impacts of subsidized flood in-
surance on demand for one particular technology – flood-tolerant seeds – in order to shed
important light on the effect of these subsidies more broadly. We take the predictions of the
model to data by implementing a cluster-randomized trial in 300 villages in West Bengal
in conjunction with the International Crops Research Institute for the Semi-Arid Tropics
(ICRISAT), a well-respected international organization known for developing advanced crop
varieties. We randomize villages into three groups (sampling 6 farmers per village): a control
group; a group that receives a fully-subsidized index insurance product that pays out a low
amount in the event of a flood; and a group that receives a fully-subsidized index insurance
product that pays out a high amount in the event of a flood. After offering (treated) farmers
the insurance product, we elicit every sampled farmers’ willingness-to-pay for both a flood-
tolerant seed minikit and a high-yield-variety seed minikit using a Becker, DeGroot, and
Marschak (1964) mechanism. We test for crowd-in vs. crowd-out by estimating the impact
of insurance on demand for flood-tolerant seeds, high-yield-variety seeds, and the difference
between the two. We further measure the extent to which these impacts vary with insurance
payout levels.

In addition, through the BDM process, we induce experimental variation in seed take-
up: in each village, at random, one farmer will be offered the high-yield-variety minikit
for the market price and one farmer will be offered the flood-tolerant seed minikit for the
market price (together making up the “seed control group”); two farmers will receive the
high-yield-variety seed minikit for free; and two farmers will receive the flood-tolerant seed
minikit for free. This design enables us to also study how specialty seeds, insurance, and
their interaction impact ex ante agricultural input decisions and ex post welfare outcomes.
Finally, we test how these impacts change with flood realizations.

2024-06-06

2024-12-30

We will conduct
baseline surveys in early June 2024, before planting has begun. During this survey, we collect
information about households’ demographics, past farming behavior, flood exposure, risk
preferences, consumption, assets and loans, and off-farm work. As part of the baseline survey,
we also provide households with their insurance offer (if applicable), and elicit willingness-to-
pay (WTP) for flood-tolerant and high-yield-variety seeds. Farmers who purchase the seeds
through the WTP elicitation game will receive their seed kits once the survey is completed.
Applicable insurance payouts will occur in November, and our endline survey will take place
after harvest, in November - December 2024. Our endline we measure ex post agricultural outcomes.
In particular, for each plot, we ask how much of each
crop they harvested. We then ask the amount of the harvested crop that is sold, consumed,
spoiled, and slated for future sale.

We use a simple model of decision-making under risk to investigate how subsidized in-
surance impacts the demand for private climate adaptation. Under actuarily fair insurance
pricing, and absent other market failures, farmers would choose the private (and also socially)
optimal level of investment in both insurance and other adaptation technologies. However,
we can then show that subsidized insurance has two counterveiling effects on adaptation
investment, in the spirit of Ehrlich and Becker (1972). On one hand, because climate in-
surance and private climate adaptation both protect agents against the same risks, subsidies
that lower the price of insurance may lead to a “crowd-out” effect. On the other hand,
because climate insurance protects agents from facing the full brunt of the bad state of the
world if a costly adaptation measure fails, these subsidies may instead lead to a “crowd-in”
effect. Which of these effects ultimately dominates is ambiguous.

We set up and implement a sharp empirical test of the impacts of subsidized flood in-
surance on demand for one particular technology – flood-tolerant seeds – in order to shed
important light on the effect of these subsidies more broadly. We take the predictions of the
model to data by implementing a cluster-randomized trial in 300 villages in West Bengal
in conjunction with the International Crops Research Institute for the Semi-Arid Tropics
(ICRISAT), a well-respected international organization known for developing advanced crop
varieties. We randomize villages into three groups (sampling 6 farmers per village): a control
group; a group that receives a fully-subsidized index insurance product that pays out a low
amount in the event of a flood; and a group that receives a fully-subsidized index insurance
product that pays out a high amount in the event of a flood. After offering (treated) farmers
the insurance product, we elicit every sampled farmers’ willingness-to-pay for both a flood-
tolerant seed minikit and a high-yield-variety seed minikit using a Becker, DeGroot, and
Marschak (1964) mechanism. We test for crowd-in vs. crowd-out by estimating the impact
of insurance on demand for flood-tolerant seeds, high-yield-variety seeds, and the difference
between the two. We further measure the extent to which these impacts vary with insurance
payout levels.

In addition, through the BDM process, we induce experimental variation in seed take-
up: in each village, at random, one farmer will be offered the high-yield-variety minikit
for the market price and one farmer will be offered the flood-tolerant seed minikit for the
market price (together making up the “seed control group”); two farmers will receive the
high-yield-variety seed minikit for free; and two farmers will receive the flood-tolerant seed
minikit for free. This design enables us to also study how specialty seeds, insurance, and
their interaction impact ex ante agricultural input decisions and ex post welfare outcomes.
Finally, we test how these impacts change with flood realizations.

Not available

Randomization

Village and Household

Yes

We randomize in two steps. First, we randomly assign each of our 300 villages into an insurance control group (125 villages who receive no insurance product); a high-payout insurance group (125 villages); and a low-payout insurance group (50 villages). We sample six farming households per village. We exclude households that did not cultivate any rice in the last 3 years, since our research design is focused on improved rice seeds. Every sample household in a given village will receive the same insurance treatment. To ensure balance and increase statistical power, we stratify our village-level randomization by block.

Second, we randomize individuals to receive offers to purchase either a flood-tolerant seed minikit or a high-yield-variety seed minikit at either market price or for free. This randomization yields four groups: market-price flood-tolerant seed offer (1/6 of households); free flood-tolerant seed offer (1/3 of households); market-price high-yield-variety seed offer (1/6 of households); and free high-yield-variety seed offer (1/3 of households). Because these seeds are already available outside of the experiment, we treat the market price groups as a ``seed control'' group, which we pool for the purposes of analysis. In this second step, we stratify the randomization by village.

1800

See above

In order to ensure that our design is powered to detect reasonably-sized treatment effects, we conduct a series of power calculations. All power calculations use a standardized outcome variable, with control mean 0 and SD 1; are based on a two-sided hypothesis test with a 5% significance level; and use a sample of 300 villages containing a total of 1,800 households. We present two sets of power calculations: a “two- group” calculation, which compares villages without insurance to villages with insurance (Panel A), and a “three-group” calculation, comparing no insurance to low-payout and high- payout insurance (Panel B). In the two-group calculation, we compare a control group of 125 villages to a treatment group of 175 villages. In the three-group calculation, we compare a control group of 125 villages to a treatment group of 125 villages (representing the control vs. high-payout insurance comparison, our main effect of interest). In both cases, we present ICCs of 0, 0.05, 0.1, and 0.15. In the two-group calculation, we are powered to 80% for effects of approximately 0.15–0.2 SD. In the three-group calculation, we are powered to 80% for effects of approximately 0.2–0.25 SD. Since the insurance treatment is free to farmers, we expect take-up to be close to 100%. In practice, these calculations are likely somewhat conservative, as we will use specifications which control for baseline data, removing residual variation in the outcome. These calculations give us confidence that the experiment is powered to detect treat- ment effects within the literature of impacts across prior agricultural studies in low-income countries (e.g., Mobarak and Rosenzweig (2014); Karlan et al. (2014); Emerick et al. (2016); Carter et al. (2017); Cole and Xiong (2017)). As perhaps a particularly helpful benchmark, Burlig et al. (2024) stimated that an index insurance product which had a maximum pay- out two-thirds the size of that described in our proposed experiment, which was provided 50 villages (with a control group of 100 villages), yielded impacts on agricultural investment of 0.12 SD, statistically significant at the 10% level. Because we are providing insurance to 175 villages (125 at the high payout level and 50 at the low payout level), we expect to be able to estimate precise treatment effects on agricultural outcomes. We ultimately conclude that these power calculations support our study design

Insure_Adapt.pdf