Targeting weather insurance markets

• Published date: 01 September 2021
• Author: Anita Mukherjee,Shawn Cole,Jeremy Tobacman
• Date: 01 September 2021
• DOI: http://doi.org/10.1111/jori.12334

Received: 27 October 2018

|

Revised: 26 October 2020

|

Accepted: 12 November 2020

DOI: 10.1111/jori.12334

ORIGINAL ARTICLE

Targeting weather insurance markets

Anita Mukherjee

1

|Shawn Cole

2

|Jeremy Tobacman

3

1

Department of Risk and

Insurance, Wisconsin School of Business,

University of Wisconsin‐Madison,

Madison, Wisconsin, USA

2

Finance Unit, Harvard Business School,

Harvard University, Boston,

Massachusetts, USA

3

Department of Economics, Alfred

Lerner College of Business and

Economics, University of Delaware,

Newark, Delaware, USA

Correspondence

Anita Mukherjee, Department of Risk

and Insurance, Wisconsin School of

Business, University of Wisconsin‐

Madison, Madison, WI, 53706, USA.

Email: anita.mukherjee@wisc.edu

Funding information

National Institute of Aging,

Grant/Award Number: P30AG012836;

Wharton Risk Management and Decision

Processes Center of the University of

Pennsylvania, Grant/Award Number:

Russell Ackoff Fellowship

Abstract

The suitability of insurance products often depends

greatly on individual circumstances. This paper ex-

amines the challenges of heterogeneity in a relatively

new product, weather‐indexed insurance. This index

insurance product has been launched in over a dozen

countries, with the goal of enabling households

engaged in agricultural activity a means to manage

risk. Using data from a large‐scale field experiment, we

build and calibrate a model which accounts for

household investment decisions, including the scope

for self‐insurance via labor markets to (risky) wage

work. Our results show that insurance is most valuable

to households with reduced access to wage labor, or to

those who face wages that are sensitive to rainfall risk.

These findings have important implications for areas

where index insurance is most effective.

KEYWORDS

index insurance, labor markets, self‐insurance, self‐protection

JEL CLASSIFICATION

D14, G22, O13, O16, Q14

1|INTRODUCTION

A majority of rural households in developing economies rely on agricultural income and face

significant weather risks. The agricultural income for these households is not only a source of

livelihood, but also a key source of food security. Weather insurance markets have emerged

over the past two decades in an attempt to protect farmers from severe income shocks, but the

J Risk Insur. 2021;88:757–784. wileyonlinelibrary.com/journal/JORI

|

757

© 2020 American Risk and Insurance Association

penetration and success of such markets remains limited. According to The World Bank, only

6% of the world's population working in agriculture was covered by some form of insurance in

2014—the number is much lower in emerging economies, as wealthier countries like the

United States have high levels of coverage with nearly 90% of farmers insured in 2016 (USDA

Risk Management Agency, 2017).

1

The setting of our study is in India, the country with the

largest weather index insurance market covering more than nine million farmers (Clarke

et al., 2012).

In principle, weather‐indexed insurance can be provided on any externally measurable data.

For example, recent products use satellite data to observe normalized difference vegetation indices.

Yet, most of the experience with weather insurance has been with rainfall, and we use that index as

a context for this study. Rainfall shocks are particularly damaging in rural areas because they can

cause entire villages to suffer: a drought (or excess rain) affects not only farm production, but also

local prices and the wages offered in nonfarm labor. Because rainfall shocks can be highly localized,

they may not garner the political support needed to trigger governmental safety nets. Thus, agri-

cultural households must consider rainfall and other idiosyncratic risks in selecting their invest-

ments in farm production, wage labor, and other aspects of their portfolios.

We make three key contributions in this paper to extend the existing literature. First, we

develop a model of portfolio choices that explicitly incorporates the decision of labor allocation

to own‐farm versus wage work. This feature is motivated by the empirical findings in Kochar

(1999) and Johnson (2009), which provide evidence that households use labor markets to insure

individual shocks in farm production. In our analysis, we allow the labor markets to also insure

the rainfall shock in farm production and examine the decision of the farmer to use this form of

labor market “insurance”before the realization of shocks. Throughout, we consider labor

allocation as a household decision based on recent work showing that such decisions depend

on household composition (LaFave & Thomas, 2016). With a few simplifying assumptions, we

obtain the closed form solutions of the key decision variables in our model and interpret their

comparative statics. For example, the model shows that agricultural investment declines as the

potential earning from wage work increases.

Note that our treatment of labor market options as a form of self‐insurance that competes

with index‐based insurance is rooted in prior work. Mishra and Goodwin (1997) provides a

model of such decisions, and Kochar (1995) shows that agricultural households are generally

well‐insulated from idiosyncratic shocks because of the availability of wage labor. Additionally,

De Janvry and Sadoulet (2001) shows that farm households in Mexico regularly engage in a

variety of “off‐farm”labor, though their participation is limited by regional demand for such

labor. Labor is so valuable that recent experiments have documented large effects simply by

incentivizing migration during the agricultural lean seasons (Bryan et al., 2014) or by making

such migration easier through improving road infrastructure (Brooks & Donovan, 2017). The

unique aspect of our application is the interactions of decisions on labor allocation to own farm

versus wage work, purchase of rainfall‐indexed insurance, and agricultural investment.

Our second contribution is modeling the labor allocation decision under wage uncertainty.

We allow for wages to be affected by rainfall shocks, more closely mirroring the village

1

The World Bank statistic is from its Global Findex Report, Retrieved from https://globalfindex.worldbank.org/

archives/2014-global-findex. Statistics specific to rainfall‐indexed insurance are more difficult to obtain. In the United

States, this insurance type is concentrated in a federal product developed to “provide livestock producers with a tool to

mitigate drought risk”; the product is called the Rainfall Index‐Pasture, Rangeland and Forage. Of the 400 million acres

that are in this category (USDA, 2019), about 160 million acres (FCIC, 2020) were insured by rainfall indexed in-

surance, implying a coverage rate of about 40%.

758

|

MUKHERJEE ET AL.