Wednesday, December 13, 2017

Insurance products for increasingly risky livelihoods

As climate change increases the frequency and severity of extreme weather in developing countries, a key question is whether agricultural insurance can play a larger role in mitigating the household level losses associated with catastrophic droughts.

Insurance products are particularly relevant in regions where agriculture or pastoralism constitute the primary source of income. In these regions indemnity payouts from insurance products can provide significant consumption smoothing benefits to insurance policyholders in the event of drought and improve material well-being.

A paper published last week by the World Bank's Development Research Group ("Insuring Well-being? Buyer's Remorse and Peace of Mind Effects from Insurance") finds that index-based livestock insurance (IBLI) also improves non-material well being for a sample of pastoralists in southern Ethiopia. Specifically, Tafere et al. (2017) find that a "peace of mind" effect associated with purchasing insurance that is positive and statistically significant. They find this effect outweighs a negative "buyer's remorse" effect on well-being, which arises when households purchase insurance and realize after the uncertainty period ends that they did not need the insurance after all.

They note:

"The implication is that, despite premiums set above actuarially fair rates, IBLI improves buyers' SWB [subjective well-being] even over a period when pastoralists in southern Ethiopia lose money on the policy. The ex ante peace of mind effect dominates any ex post buyer's remorse. In other words, even an insurance policy that does not pay out still improves people's perceptions of their well-being."

Estimation strategy

To estimate the "peace of mind" effect, the authors randomize the provision of 10-80 percent discount coupons and comic book or audio tape information interventions to households in the communities in southern Ethiopia. These incentives act as instruments which increase uptake of IBLI among households in an instrumental variables model. In the reduced form  stage, the measure of well-being is regressed on the probability of IBLI uptake predicted in the first stage.

The authors measure SWB using the question: "On which step do you place your present economic conditions?" with possible responses ranging in the Likert scale from very bad (1) to very good (5). They employ a vignette-based adjustment of the SWB scores by asking respondents to rank their own circumstances with respect to a set of individuals described in short vignettes to improve the comparability of the subjective welfare measurements to one another.

In both of the years in which insurance policies were active there were no indemnity payouts. This allowed the authors to disentangle the ex ante peace of mind effect from the ex post buyer's remorse effect on SWB. It also ensured that the effects measured were not material or payout based.

There are two issues to consider with respect to the empirical strategy:
  1. Depending on how well-being is assessed it is possible that well-being could be positively affected by discounts for the sole reason that it is a discount and is saving people money. If true this would invalidate the instrumental variables assumption that the instrument be exogenous from the reduced form model. Though I considered this possibility, I dismissed it since SWB is assessed in the period after the period in which the discount is applied and insurance purchased, it seems unlikely that said discount would have such a lasting impact on SWB. 
  2. The effect that is measured here (local average treatment effect) is the effect of IBLI on SWB for households whose probability of buying insurance was impacted by the incentives. It does not capture the effect for households who would have purchased insurance even without the incentives (who may better represent the buyers in an actual market). The question then is whether the latter households would face similar improvements in well-being? It is possible those households would have greater improvement in well-being (if they are even more risk-averse than households that would only purchase with discounts) but it is difficult to tell without using a different empirical strategy. 

The implications of these finds are interesting from a market equilibrium perspective: if people experience an improvement in their SWB even in cases where the market rate is not actuarially fair, could it lead people to purchase insurance in cases where the market rate is not actuarially fair? This could potentially increase social welfare if  it induces insurers who could not viably enter the market at lower rates to supply insurance.

However, the key question is whether households would purchase the insurance at non-actuarially fair rates without discounts. Not only would they need to pay a higher rate but they would need to incorporate the impact on their own SWB into their financial decisions. It is not clear that this is the case judging by the number of households that lapsed on their insurance policies between the first and second years of the survey: of the 130 households that purchased insurance in the first sales period only 77/130 renewed it again the next year. Similarly of the 94 households that purchased insurance in the third sales period only 21 renewed it the next year.

One reason for this may be that the same households do not necessarily receive a discount in each sales period ("The randomized assignment of respondents into information treatments and discount coupons with varying discount levels was implemented independently for each sales period.") So the relatively low rate of renewal captures either one or likely both: the price differential between different years and a lack of internalization of the increase in SWB felt from insurance.

Broader context and market viability

There is a real space for agricultural insurance in developing countries. There have traditionally been limited formal insurance products in these spaces due to moral hazard and adverse selection problems. It is difficult on the lender or insurer side to model the risk profiles of unbanked clientele with limited financial records, including limited or no proof of income or credit history, who have had limited experience with formal financing (McKinsey, 2012). There are furthermore high transaction costs for insurers in rural and remote areas (Besley, 1995). Informal insurance systems (agreements that arise between individuals and communities as opposed to in formal markets) are increasingly overburdened as climate variability increases and there are more frequent payouts.

These issues with formal and informal insurance providers are also the case for provision of other financial services for the poor in developing countries. I discuss the disconnect between financial product offerings and the financial needs of the unbanked in a paper that I co-wrote entitled "Tailored Finance and Organic Growth: A Livelihoods Approach to Financial Inclusion in India."

There are a few main areas I think are relevant going forward:
  1. Though informal community-based insurance systems find themselves in an increasingly challenging environment, the survey in this paper indicates that membership of an informal community insurance system (iqub) was negatively correlated with IBLI uptake meaning where it is available, households view it as a viable substitute for formal insurance. 
    1. It would be interesting to see how the rates/coverage offered by the iqub compare to the scheme offered here and whether there are other reasons people prefer informal systems (e.g. they are seen as more trustworthy given limited experience with formal providers). 
  2. It would also be relevant to study the insurance rates with respect to household income patterns to assess pricing and timing of collection. The authors hinted at the fact that seasonal liquidity changes could be a factor in the fewer purchases at the second sales period in a given year than the first. It may make sense to develop insurance schemes that are better tailored towards the income patterns of households in the community being served. 
  3. A final note that this paper specifically looks at index-based insurance products where indemnity payouts are not linked to actual losses but to a weather index. A pre-determined threshold on weather indicators triggers the payout. Interest is growing around these products in developing countries because they reduce insurer costs (e.g. monitoring costs) and control for adverse selection and moral hazard more so than traditional insurance products which may make the market more appealing to insurers. For more details see here. It will be interesting to keep an eye on the supply side for index products and whether it is a more active market than the one for traditional insurance products in this space. 
  1. Tafere, K., Barrett, C.B., Lentz, E., & Ayana, B.T. (2017). Insuring well-being? Buyer's remorse and peace of mind effects from insurance (Policy Research Working Paper No. 8256). The World Bank. 
  2. Baer, T., Goland, T., & Schiff, R. (2012). New credit risk models for the unbanked (McKinsey Working Papers on Risk No. 30). McKinsey & Company.  
  3. Besley, T. (1995). Chapter 36 savings, credit and insurance. In T. N. Srinivasan & J. Behrman (Eds.), Handbook of development economics (Vol. 3, p. 2123-2207). Amsterdam: Elsevier.

1 comment:

  1. Wow nice blog, thanks for sharing this content, i will bookmark this website so that i will read some posts again.

    Vistiting from: