In Search of Lost First Bests

“If a little dreaming of first bests is dangerous, the cure for it is not to dream less, but to dream more, to dream all the time.” Maybe M. Proust

Motivation

Some of us are currently faced with a choice: pay for a full year of daycare for our kids as some centers are now open, or not register our kids for daycare. If new lockdowns are implemented, all daycares may shut down. If we pay for daycare, the average outcome may be better, but the worst case outcome (pay for daycare, but daycare shuts down) is much worse; to avoid this risk, we might choose not to sign up our kids for daycare. This choice is particularly hard for low-income families, who may not be able to afford the double hit of daycare costs and substitute expenditures in the event of closures, thus exacerbating inequalities. However, insurance against daycare closures would solve this problem: we’d pay nothing if daycares closed, and get daycare (although at a higher price) if they don’t, recovering the first-best outcome.

Risk in agriculture

In the above example, the presence of uncertainty (absent insurance) caused inefficiently low investment in children’s human capital. Analogously, it is widely believed that the presence of uncertainty causes farmers in developing countries to underinvest in modern agricultural inputs. From an apple farmer in Lesotho to maize farmers in Ghana, farmers may diversify sources of income, including diversifying away from the most profitable sources, to avoid losing everything. If risk is a constraint, insurance should return to first best: insurance enables farmers to effectively sell their uncertainty to insurers, freeing them to focus on profitable investments, increasing incomes and enabling escape from poverty.

Why does traditional insurance fail? 

Unfortunately, it's easy to find reasons why traditional insurance may not work well in many rural contexts. 

First, if farmers were fully insured against losses, they would have no incentive to increase productivity — any increases in production would immediately result in a decrease in insurance payments. This causes moral hazard — farmers would go back to underinvesting, because they no longer benefit fully from the returns to their investments.

Second, insurers do not know which farmers are more or less productive, meaning more productive farmers would receive less insurance payments than less productive farmers. This causes adverse selection, where only less productive farmers would be willing to buy insurance. However, this undoes insurance markets, as these less productive farmers receive more payments, increasing the costs of providing insurance, and pushing out relatively more productive farmers.

Even after solving these challenges, insurance requires measuring farmer production at scale, but the costs of this exercise would likely leave insurance prohibitively expensive for smallholder farmers.

Index insurance as a solution

Addressing the above concerns requires insuring against risk that cannot be manipulated by farmers, does not vary across farmers, and is easy to observe. Since the early 2000’s, rainfall-based index insurance has become a popular solution for this, where insurance payouts are made during periods of drought and/or flood. This solves the above concerns: farmers cannot affect rainfall, and measurement became widely feasible with newly available remote sensing data sources.

Recent work has suggested index insurance allows farmers to increase risky investments: Karlan et al. (2014) find index insurance increased on-farm investment in Ghana, while Hill et al. (2019) find it also caused farmers to shift into higher-value crops in Bangladesh.

What constrains adoption of index insurance?

Despite its clear benefits, index insurance has not yet seen widespread adoption among smallholder farmers; a number of studies highlight constraints that make successful implementation of index insurance challenging.

First, Karlan et al. (2014) note a particularly important factor is basis risk: while traditional insurance might pay out whenever crops fail, index insurance will pay out in drought years even when production is high, and will not pay out in normal years even if crops fail. However, Mobarak & Rosenzweig (2012) note that in many contexts, farmers insure themselves against idiosyncratic sources of basis risk through their social networks. In India, they find that these networks increase demand for index insurance by reducing basis risk, while index insurance offers insurance against aggregate risk that social networks cannot insure against.

Second, Karlan et al. (2014) also show that farmers overreact to recent payouts, increasing demand after payouts and decreasing demand when payouts do not occur (a phenomenon also documented in demand for flood insurance in the United States). However, they also suggest this can be overcome with careful insurance product design.

Third, index insurance typically requires farmers to make payments before planting, when farmers have the greatest need for cash, but payouts do not occur until after harvest, when farmers typically have the least need for cash. Casaburi & Willis (2018) worked with cooperatives in Kenya to allow the deduction of insurance payments from farmers’ revenues at harvest, and found this increased demand for index insurance from 5% to 72%. However, this requires institutional presence (in this case, agricultural cooperatives) to successfully collect premiums in periods without payouts.

In sum, the institutional environment and careful design of index insurance products interact to affect eventual demand; developing these products is rarely easy, but evidence suggests they can reduce the role of risk in constraining farmers’ investments.

When is credit a solution?

However, insurance is not the only financial solution to reducing risk: the idea is really to get money into farmers hands when they need resources the most. As an alternative, carefully designed credit may be easier to implement.

For example, Fink et al. (2018) document that Zambian farmers are forced to sell their labor when they have limited resources. However, they find that access to credit enables these households to work on their own farms instead of working for others, and this in turn increases average productivity. 

Alternatively, Lane (2020) studies the experimental introduction of rainfall indexed guaranteed access to credit in Bangladesh. They find that households receiving access to credit increase risk investments and are able to smooth their consumption, in a manner comparable to the effects of index insurance. However, the loans did not require up front payments from farmers, were implemented at scale, and were profitable for the lender.

When is technology a solution?

In all the above contexts, farmers have access to multiple possible investments, and may choose lower risk and lower return investments to effectively insure themselves. However, technology can adjust this tradeoff: irrigation, to give one example, may reduce the volatility of agricultural production, enabling farmers to invest more in modern inputs and in turn increase productivity.

Alternatively, Emerick et al. (2016) show the introduction of a new variety of rice in India that reduced production uncertainty increased investments and these investments in turn increased productivity.