Poverty Traps and Index-Based Risk Transfer Products

World Development Vol. 36, No. 10, pp.
17661785, 2008
2008 Elsevier Ltd. All rights reserved
0305-750X/$ - see front matter
www.elsevier.com/locate/worlddev
doi:10.1016/j.worlddev.2007.10.016
Poverty Traps and Index-Based Risk

Transfer Products
BARRY J. BARNETT
Mississippi State University, MS, USA
CHRISTOPHER B. BARRETT
Cornell University, Ithaca, NY, USA
and
JERRY R. SKEES *
University of Kentucky, Lexington, KY, USA
Summary. A growing literature suggests that in low-income countries, households with few as-
sets can be trapped in chronic poverty. This article reviews relevant threads of the poverty traps
literature to motivate a description of the opportunities presented by innovative index-based risk
transfer products. These products can be used to address some insurance and credit market failures
that contribute to the persistence of poverty among households in low-income countries. Applica-
tions are considered at the micro, meso, and macro levels.
2008 Elsevier Ltd. All rights reserved.
Key words index-based risk transfer products, insurance, poverty traps, risk management
1. INTRODUCTION This article reviews relevant threads of the

poverty traps literature to motivate a descrip-
A growing literature suggests that in low-in- tion of the opportunities presented by index-
come countries, households with few assets based risk transfer products (IBRTPs). These
can be trapped in chronic poverty. Due to high products can be used to address some insurance
uninsured risk exposure, households may adopt and credit market failures that contribute to the
low-risk, low-return strategies for using pro- persistence of poverty among households in
ductive assets, reducing the likelihood that they low-income countries. Applications are consid-
can accumulate the assets needed to escape ered at the micro, meso, and macro levels. We
poverty through autarchic savings and invest- discuss as well some of the key limitations of
ment. Thus, those with few assets may accu- IBRTPs in low-income rural areas, where agri-
rately perceive that time is not an ally in their cultural production is a dominant, though not
daily struggle to climb out of poverty. The necessarily exclusive, economic activity and
resulting hopelessness can be both probabilisti- where formal insurance and credit markets
cally accurate and self-reinforcing. Others suer are likely to be most limited. The use of
uninsured asset losses that suddenly cast them IBRTPs to pre-nance safety net and disaster
into poverty and possibly onto a downward assistance programs is also considered.
spiral from which they have a dicult time Separate literatures exist on both poverty
re-emerging. These themes from the emerging traps and on IBRTPs. Our innovation is to
literature on poverty traps underscore the rela- highlight that the poverty traps hypothesisin
tion between risk and persistent poverty, as well
as the opportunities aorded by innovations in
risk management. Final revision accepted: October 4, 2007.
1766
POVERTY TRAPS AND INDEX-BASED RISK TRANSFER PRODUCTS 1767
particular, the possibility of poverty traps based how the combination of exogenous shocks,
on multiple dynamic equilibriasubstantially especially covariate shocks, and incomplete
increases the stakes with respect to risk manage- insurance and credit markets generates condi-
ment. One of our primary objectives is to dem- tions that can trap households in poverty.
onstrate the largely overlooked connections These factors also aect meso and macro level
between the poverty traps and IBRTP litera- institutions in ways that further constrain eco-
tures. While a poverty trap is not necessary for nomic opportunities at the household level.
IBRTPs to be relevant in low-income coun- Section 3 describes why insurance and credit
triesa simple market failure argument suces markets often fail in rural areas of low-income
to justify new risk transfer interventions, as we countries. Many of the same factors that limit
explain in Sections 3 and 4the synergistic the availability of formal insurance and credit
interaction between these two phenomena of also limit the availability of insurance and cred-
growing interest merits more explicit attention it through informal channels. Section 4 de-
than it has received to date. scribes IBRTPs and recent eorts to extend
While many factors contribute to the exis- availability of these instruments into low-in-
tence of poverty traps, limited access to insur- come countries. IBRTPs can be used as a mar-
ance and credit instruments is commonly ket mechanism to transfer covariate risks
identied as a primary causal factor. In many outside of the country or region. Alternatively,
low-income countries, formal insurance and they can be used by governments, donors, or
credit markets are limited due to poor contract even non-governmental organizations (NGOs)
enforcement, asymmetric information, high to pre-nance safety net or disaster assistance
transactions costs, and high exposure to covar- eorts. However, experience to date suggests
iate risk. With limited access to credit or insur- that IBRTPs targeted to the needs of low-in-
ance, households often have a dicult time come countries will not materialize without
managing the myriad risks they face. In recent the coordinated eorts of national governments
years a number of innovative IBRTPs have and donors. Section 5 therefore discusses
been developed for transferring covariate risks implementation issues that aect the long-run
outside the low-income rural economy. Rela- sustainability of IBRTPs. We conclude, in Sec-
tive to traditional insurance products, IBRTPs tion 6, with some summary thoughts and con-
are characterized by fewer asymmetric informa- cerns about the prospects and limitations for
tion problems, lower transactions costs, and IBRTPs as a tool for addressing persistent pov-
simpler contract designs, as we describe below. erty in rural areas of low-income countries.
While to date most commercial applications of
these instruments have been in OECD coun-
tries and India, substantial eorts are underway 2. POVERTY TRAPS
to extend these instruments to low-income
countries. This is a potentially important inno- As research on and measurement of poverty
vation because social and institutional mecha- has evolved over the past decade or two, in-
nisms for coping with covariate risk exposure creased attention has been paid to how well-
are typically quite limited in low-income coun- being evolves over time, with much interest in
tries, especially among the rural poor. resolving the important puzzle of why some
Behavioral responses to risk can limit eco- individuals, households, communities and na-
nomic growth. Thus, IBRTPs hold promise as tions remain mired in extreme poverty for ex-
an important development tool. However, since tended periods and how others are able to
both national governments and donor organi- avail themselves of new market and technolog-
zations face budget constraints, there is an ical opportunities to lift themselves out of pov-
opportunity cost to using scarce resources to erty. 1 Increasingly, the former experience has
develop risk management programs based on become summarized as a poverty trap. This
IBRTPs. Thus, it is critically important that concept has proved extremely inuential in
decisions regarding IBRTPs be based on a clear development policy circles, perhaps most
understanding of both the advantages and lim- clearly manifest by the United Nations Millen-
itations of IBRTPs and how these instruments nium Project (2005).
may t into broader development strategies. In the economics literature, there exist multi-
The plan for the rest of the paper is as fol- ple sorts of poverty traps associated with dier-
lows: Section 2 reviews the relevant poverty ent mechanisms by which these might emerge
traps literature, with particular emphasis on (for details, see Azariadis & Stachurski, 2005;
1768 WORLD DEVELOPMENT
Barrett & Swallow, 2006; Bowles, Durlauf, & growth potential and investment incentives
Ho, 2006; Carter & Barrett, 2006). One special (Bardhan, Bowles, & Gintis, 2000; Carter &
class of poverty traps requires the existence of Barrett, 2006; Dercon, 2005; Eswaran & Kot-
multiple dynamic equilibria, at least one of wal, 1989, 1990; Rosenzweig & Binswanger,
which lies below a standard poverty line. This 1993). Among the poorest of the poor, a subsis-
particular sort of poverty trap is uniquely rele- tence threshold likely exists. If household asset
vant to the insurance literature because it is levels fall below this threshold, the path dynam-
characterized by at least one critical threshold ics suggest that the household would not be ex-
(an unstable dynamic equilibrium, in somewhat pected to generate sucient income to meet the
more precise, formal terms) above which the most basic nutritional requirements (Zimmer-
expected dynamics of the system are character- man & Carter, 2003), thereby collapsing into
ized by asset accumulation (i.e., growth and a nutritional poverty trap (Dasgupta, 1993,
improvements in standards of living) and below 1997).
which decumulation prevails. Unlike poverty Risk can thus have two distinct, crucial ef-
traps based on a unique, low-level stable equi- fects in a system characterized by multiple equi-
librium, threshold-based poverty traps raise libria. First, ex ante eorts to reduce risk
the stakes for risk management because only exposure can dampen accumulation, thereby
in the presence of multiple equilibria can shocks creating a low-level equilibrium. Second, the
exogenously shift the accumulation dynamics, ex post consequences of a shock can knock peo-
which bifurcate at the unstable equilibrium. ple back into a poverty trap.
In a world without multiple dynamic equilib- Of course, if markets exist to permit people to
ria, everyone follows a growth path towards a insure against shocks ex ante, or to borrow ex
unique, long-run standard of living. This can post so as to achieve quasi-insurance through
occur at dierent rates and there may be tempo- ex post loan repayment (rather than ex ante
rary disruptions along the way due to various insurance premium payment), these adverse ef-
shocks, even dierent equilibria for dierent co- fects of risk should be attenuated. The existence
horts, as enshrined in the concepts of club con- of risk need not then contribute to the existence
vergence and conditional convergence in the of poverty traps. Unfortunately, credit and
macroeconomic growth literature. But, in such insurance instruments are routinely undersup-
a world, shocks should have no permanent ef- plied in most low-income areas, and especially
fect, although they can take some years to fully to the poorest peoples (Besley, 1995). Financial
play themselves out. Risk merely adds noise to market failures thereby contribute both directly
the inexorable process of convergence. and indirectly to the persistence of chronic pov-
In the presence of a critical threshold, by con- erty (Carter & Barrett, 2006).
trast, shocks can have permanent conse- Many of the factors that contribute to pov-
quences, ipping people from one growth erty traps at the household level (e.g., barriers
path onto another. Rare, favorable shocks that create scale economies and limited access
(e.g., winning a lottery or receiving a signicant to insurance or credit) can also exist at more
asset transfer) can suddenly make new invest- aggregate levels of analysis (Barrett, Carter
ments worthwhile and lead a poor beneciary et al., 2006; Barrett, Marenya et al., 2006; Bar-
to grow towards a higher-level equilibrium. rett & Swallow, 2006). Poverty traps at higher
By contrast, shocks that push people below levels of aggregation necessarily constrain eco-
the threshold can set them onto a downward nomic opportunities at lower levels of aggrega-
spiral into destitution (a low-level equilibrium) tion and thus, accentuate poverty traps at the
from which they do not recover, or keep them household level (Carter & Barrett, 2006; Meh-
from growing their way out of persistent pov- lum, Moene, & Torvik, 2005). For example,
erty by regularly knocking them backwards as the next section describes how, at a local level,
they struggle to climb out of the trap, a real- covariate risk exposure may limit the availabil-
world Sisyphean tragedy (Carter & Barrett, ity of informal credit or insurance. But various
2006; Carter, Little, Mogues, & Negatu, 2005; meso and macro level institutions may also be
Dercon, 1998, 2005; Krishna, 2006; McPeak exposed to high levels of covariate risk. Absent
& Barrett, 2001; Santos & Barrett, 2006a). some mechanism for transferring this risk,
Knowing this, people may go to extraordinary these institutions will be reluctant to invest in
lengths to manage risk exposure, for example illiquid but highly productive assets (e.g., trans-
by selecting low-risk, low-return portfolios that portation infrastructure, processing facilities,
reduce the risk of greater suering but limit etc.). These choices then further constrain the
opportunities available to households at the mi- cyholders are misclassied. Those who are mis-
cro level (Dercon, 2004). classied to their benet (detriment) are more
(less) inclined to purchase. As a result the insur-
ance program is likely to experience losses that
3. LIMITED ACCESS TO INSURANCE exceed the projections used to establish pre-
AND CREDIT mium rates. In response, the insurer may in-
crease premium rates for all classes. But this
(a) Insurance market failure only compounds the problem and leads to an
even more adversely selected group of insur-
In rural areas of low-income countries, for- ance purchasers (Barnett, 1995). Unless the
mal insurance markets are typically incomplete underlying information asymmetry can be ad-
and often nonexistent. This is particularly true dressed, adverse selection will cause insurance
for insurance that protects against crop produc- markets to fail.
tion shortfalls or livestock mortality. A com- Moral hazard, the second common asymmet-
mon reason for insurance market failure is the ric information problem, occurs when, as a re-
lack of eective legal systems to enforce insur- sult of purchasing insurance, policymakers
ance contracts. But even when eective contract engage in hidden activities that increase their
enforcement mechanisms are in place, insur- exposure to risk. This behavioral response
ance markets often fail due to strong covariate leaves the insurer exposed to higher levels of
risk exposure, asymmetric information prob- risk than had been anticipated when premium
lems, and high transaction costs. rates were established (Barnett, 1995). Unless
the insurer can eectively monitor policyholder
(i) Covariate risk behavior so as to enforce policy provisions,
Insurance is based on the statistical law of moral hazard will also cause insurance markets
large numbers which implies that, for a pool to fail.
of uncorrelated observations, the variance of Adverse selection and moral hazard prob-
the pool decreases with the number of observa- lems can be addressed, in part, by making cer-
tions (Priest, 1996). However, if insured units tain that the insured continues to hold some
face highly covariate risks, the variance reduc- risk. This is why insurance policies typically
tion that can be obtained by pooling is greatly contain deductible and/or co-insurance provi-
reduced (Skees & Barnett, 1999). Spatially cor- sions. However, even with these provisions,
related catastrophic losses can then exceed the serious adverse selection and moral hazard
reserves of the insurer leaving unsuspecting problems still plague agricultural insurance
policyholders unprotected. Such experiences programs in the United States and other OECD
explain why crop insurance policies are gener- countries (Chambers, 1989; Coble, Knight,
ally available only in countries where govern- Pope, & Williams, 1997; Goodwin, 2001; Just,
ments take on much of the catastrophic risk Calvin, & Quiggin, 1999; Quiggin, Karagiannis,
exposure faced by insurers (Binswanger & & Stanton, 1994; Skees & Reed, 1986; Smith &
Rosenzweig, 1986; Miranda & Glauber, 1997). Goodwin, 1996). Information asymmetries are
The presence of highly covariate risk is a major likely even more pronounced in rural areas of
cause of insurance market failure in many low- low-income countries. In addition, since the
income countries. scale of agricultural production tends to be
small in low-income countries, the cost of
(ii) Asymmetric information underwriting and monitoring activities to ad-
The principal-agent literature identies two dress those information asymmetries is a much
primary types of asymmetric information prob- higher percentage of the insured value.
lems: adverse selection (or hidden information)
and moral hazard (or hidden action). In insur- (iii) Transaction costs
ance markets, adverse selection occurs when The transaction costs of oering insurance in
potential policyholders have proprietary rural areas are much higher than in urban areas
knowledge about their risk exposure that is due to the distances that must be covered by
not available to the insurer (Rothschild & Sti- sales agents and loss adjusters and the relatively
glitz, 1976). Insurance underwriters assign po- small number of policy-holders in each locale.
tential policyholders into risk rating classes. These costs are amplied by limited transporta-
Because underwriters do not have access to all tion and communication infrastructure (Binsw-
the relevant information, many potential poli- anger & Rosenzweig, 1986).
Crop insurance, in particular, is character- rates of ination can signicantly reduce incen-
ized by extremely high transaction costs. It is tives for monetary savings (Besley, 1995; Der-
not easy to determine the policyholders ex- con, 1998; McPeak & Barrett, 2001).
pected yield since expected yields vary tremen- Due to macroeconomic uncertainty and cul-
dously across regions, among farms in the tural preferences, household savings in many
same region, and even across parcels for the areas are often held in semi-liquid productive
same large land holder. Assessing crop losses assets such as livestock rather than in currency
is both dicult and time consuming. Further- (Dercon, 1996). If necessary, these assets can be
more, loss assessment is required more fre- liquidated to temporally smooth consumption
quently for crop insurance than for other lines (Rosenzweig & Wolpin, 1993). Market condi-
of insurance. The magnitude of these transactions, however, can limit the eectiveness of
tion costs tends to be largely independent of this self insurance strategy. In the aftermath
the size of the policy. Thus, as a percentage of of a highly covariate adverse shock (e.g.,
the insured value, the transaction costs of sell- drought that aects an entire nation or multi-
ing and servicing insurance are much higher national region), market supply of the asset
for small policies than for large policies. 2 As can increase dramatically, driving down the va-
indicated above, costs associated with address- lue of household savings just when it is most
ing information asymmetry problems are also needed (Dercon, 1996). This can also happen
much higher in rural areas than in urban areas. with localized adverse shocks if markets for
Thus, high transaction costs are another impor- the asset are not spatially integrated (Rosen-
tant cause of insurance market failure in rural zweig & Binswanger, 1993; Zimmerman & Car-
areas of low-income countries. ter, 2003). Liquidating productive assets may
also not be a viable self-insurance option for
(b) Informal risk management mechanisms the poorest of the poor. Evidence suggests that
extremely poor households recognize the dan-
While formal insurance and credit markets ger of subsistence traps (or other undesirably
are limited in rural areas of most low-income low-level equilibria) and thus beyond some
countries, various informal risk-coping mecha- point choose to forego consumption rather
nisms are widely utilized. In general, these than further liquidating assets (Kazianga &
mechanisms can be classied as risk mitigation, Udry, 2006; Zimmerman & Carter, 2003). In
self insurance, and risk transfer. Rural house- other words, they smooth assets rather than
holds can mitigate risk by choosing to produce consumption. Such a decision may require re-
lower risk outputs (e.g., cassava instead of duced expenditures on school fees (i.e., remov-
maize), employing risk reducing inputs (e.g., ing children from school), health care, and food
irrigation), share tenancy, and diversifying in- consumption (Barrett, Carter et al., 2006; Bar-
come sources. However, the extent to which rett, Marenya et al., 2006; Carter, Little, Mo-
households can utilize any of these strategies gues, & Negatu, 2006; Foster, 1995; Morduch,
is highly conditioned on local climatic, techno- 1995). Resulting health and educational de-
logical, and market conditions as well as on ciencies can reduce the value of human assets,
household asset levels (Barrett, Bezuneh, & further trapping the household in poverty (Der-
Aboud, 2001; Barrett, Bezuneh, Clay, & Rear- con & Hoddinott, 2005; Hoddinott, 2006;
don, 2005; Little, Smith, Cellarius, Coppock, Hoddinott & Kinsey, 2001; Jacoby & Skouas,
& Barrett, 2001; McPeak & Barrett, 2001; 1997; Thomas et al., 2004).
Reardon, 1997; Reardon, Delgado, & Matlon, Other common self-insurance strategies in-
1992; Reardon & Taylor, 1996). Further, the clude household migration, movement of
implied risk premia on risk mitigation strategies range-fed livestock to better pasture, or more
can be very high (Morduch, 1995; Rosenzweig intensive use of common natural resources. As
& Binswanger, 1993; Zimmerman & Carter, with risk mitigation, there is an implied risk
2003). premium for all self-insurance strategies. The
Rural households also employ various meth- implied risk premium for self-insurance strate-
ods to self-insure against adverse shocks. Cur- gies is either the explicit or the opportunity cost
rency-denominated savings can be used to of undertaking the strategy. An example of the
smooth consumption over time. Yet institu- former would be the costs associated with
tions that accept savings deposits (e.g., banks migration or the movement of livestock. An
and post oces) are quite sparse in rural areas example of the latter would be the opportunity
of many low-income countries. Further, high cost of holding savings in a relatively liquid
state so they can be used for consumption transfer risk and smooth temporal consump-
smoothing should a shock occur. The opportu- tion, adverse shocks can dramatically reduce
nity cost of keeping funds in such a liquid state the households stock of productive assets ex
is the higher rate of return that could be real- post, either through direct destruction of assets
ized on less liquid investments. Further, some or through distress liquidation. Recognizing
self-insurance strategies can generate adverse this danger, households often choose low-risk,
external eects. Among these are pecuniary low-return, strategies that mitigate risk expo-
externalities as in the case of distress asset sales sure but also lead to low expected returns and
following covariate shocks or environmental thereby a poverty trap.
degradation when common natural resources
are used more intensively (Barrett & Swallow, (c) Credit markets
2006).
A variety of informal risk transfer mecha- In the absence of formal insurance markets,
nisms are utilized to smooth consumption in credit can sometimes be used to temporally
rural areas of low-income countries (Besley, smooth consumption following the occurrence
1995). These mechanisms vary from socially- of a major shock (Binswanger & Rosenzweig,
constructed reciprocity obligations within fam- 1986; Eswaran & Kotwal, 1989). However,
ily, village, religious community, or occupation there is an important dierence between insur-
(Coate & Ravallion, 1993; Fafchamps & Lund, ance and credit markets. Insurance is an ex
2003; Grimard, 1997; Rosenzweig, 1988; Town- ante mechanism that requires only the payment
send, 1994, 1995) to semi-formal micronance, of a premium. Credit is an ex post response
rotating savings and credit, or state-contingent that often requires either a previous history
loan entities (Ho & Stiglitz, 1990; Udry, 1994). of repayment and/or assets that can be used
These family and community oriented mecha- as collateral.
nisms may be better able to address the asym- In rural areas of low-income countries, for-
metric information and transaction costs mal credit markets also tend to be very limited
problems that plague formal insurance and and for exactly the same reasons that limit
credit markets (Arnott & Stiglitz, 1991; Rosen- insurance markets. Contract enforcement is
zweig, 1988; Stiglitz, 1990; Udry, 1994). How- problematic. Asymmetric information prob-
ever, social factors can prevent reciprocity lems make it dicult both to accurately classify
obligations from functioning as eective mu- borrowers prior to making loans and to moni-
tual insurance (Platteau, 1997). Moreover, tor their behavior afterward (Binswanger &
these informal mechanisms tend to fail in the Rosenzweig, 1986; Braverman & Guasch,
presence of large covariate risks (Dercon, 1986; Freedman & Click, 2006; Ho & Stiglitz,
1996; Rosenzweig, 1988; Rosenzweig & Binsw- 1990). Transaction costs are very high, particu-
anger, 1993; Townsend, 1994; Zimmerman & larly as a percentage of funds loaned (Carter,
Carter, 2003) and can be compromised by the 1988), and the lender is exposed to potentially
existence of threshold-based poverty traps high levels of covariate risk exposure (Rosen-
(Santos & Barrett, 2006b). zweig, 1988). The lack of insurance markets
Informal risk transfer mechanisms must further hampers credit markets since lenders
tradeo asymmetric information and transac- may be unwilling to accept uninsured assets
tion costs problems against covariate risk expo- as collateral.
sure. The more (less) geographically proximate
the participants, the fewer (more) the asymmet-
ric information and transaction costs problems 4. INDEX-BASED RISK TRANSFER
but the higher (lower) the exposure to spatially PRODUCTS
covariate risk (Grimard, 1997). There is also
evidence that access to these informal mecha- The literatures on poverty traps and nancial
nisms is positively related to existing wealth market failures in low-income rural settings
(Jalan & Ravallion, 1999; Santos & Barrett, point to a strong potential role for risk transfer
2006b). This is not surprising since the poorest mechanisms, both to help facilitate the develop-
of the poor would have little to oer family- or ment of insurance and credit markets and to
community-based mutual-aid institutions. provide a mechanism for pre-nancing safety
Limited access to insurance and credit, either net and disaster assistance programs. Much of
formal or informal, contributes to the existence this potential is now being directed towards
of poverty traps. Without eective means to nascent applications of IBRTPs.
IBRTPs are nancial instruments that make If an IBRTP is to be eective, the underlying
payments based on realizations of an underly- index must meet several conditions. It must be
ing index relative to a pre-specied threshold. highly correlated with the loss being insured
The underlying index is a transparent and objec- against over a relatively large geographic area.
tively measured random variable. Examples in- Sucient historical data must exist from which
clude area average crop yields, area average to estimate the probability distribution of the
crop revenues, cumulative rainfall, cumulative index. The process that generates random real-
temperature, ood levels, sustained wind izations of the index must be either inherently
speeds, and Richter-scale measures. IBRTPs stationary and homoskedastic (as is true for
can take on any number of forms including some climatic variables) or else one must be
insurance policies, option contracts, cata- able to manipulate the historical data using sta-
strophic bonds, or derivatives. 3 Some highly tistical trend adjustment and heteroskedasticity
standardized IBRTPs are actively traded in sec- correction procedures to generate an accurate
ondary markets. However the focus here is on probability distribution of the index (as is often
IBRTPs that are customized to t the specic done with area yield indices). The index must
risk management needs of the purchaser. These be measured and reported in a timely manner
IBRTPs are typically sold by international rein- by an objective third party. It must be observa-
surers and held by the purchaser until they ex- ble, transparent, secure, and independently ver-
pire. iable (Hazell & Skees, 2006).
Traditional insurance products pay indemni-
ties when realized losses exceed a given thresh- (a) Advantages and limitations
old. Thus, traditional crop yield insurance pays
an indemnity when realized farm-level crop IBRTPs have a number of advantages rela-
losses exceed a stated percentage of the ex- tive to traditional farm-level yield or revenue
pected yield (the deductible). IBRTPs make insurance. Since realizations of the index are
payments when the realized value of the under- exogenous to policy-holders, index insurance
lying index either exceeds or falls short of a gi- is not subject to the asymmetric information
ven threshold. 4 The index is exogenous to the problems that plague traditional insurance
policyholder but correlated with the policy- products. Transaction costs are much lower
holders realized losses. An IBRTP that pro- since the insurer does not have to verify farm-
tects against crop losses would be based on an level expected yields or conduct farm-level loss
index that is presumed to be highly correlated assessment. This is particularly important in
with farm-level yields. Examples include the low-income countries where farmers often do
Group Risk Plan (GRP) area yield and Group not have records of historical yields.
Risk Income Protection (GRIP) area revenue IBRTPs also have one signicant limitation
insurance products currently sold in the United relative to traditional insuranceit is possible
States (Miranda, 1991; Skees, Black, & Barnett, for a household to experience a loss and yet
1997). IBRTPs with indices based on cumula- not receive a payment from an IBRTP. It is also
tive rainfall, cumulative temperature, area live- possible that the household will not experience
stock mortality, and satellite imagery have also a loss and yet receive a payment. This basis
been proposed for agricultural producers risk occurs because the index is not perfectly
(Deng, Barnett, Vedenov, & West, 2007; correlated with farm-level losses. Of course, ba-
Mahul, 2001; Martin, Barnett, & Coble, 2001; sis risk exists with many risk-management
Miranda & Vedenov, 2001; Skees & Enkh- instruments (e.g., hedging using futures or op-
Amgalan, 2002; Turvey, 2001). tions contracts). If the basis risk is relatively
Much recent attention has focused on the po- small, the instrument can still be a highly eec-
tential for using IBRTPs in low-income coun- tive risk management tool. If the basis risk is
tries to protect agricultural assets from losses quite large, the instrument will likely not be
caused by various climatic perils (Hess, Richter, very eective. Various studies have empirically
& Stoppa, 2002; Hess, Skees, Stoppa, Barnett, examined the eectiveness of IBRTPs in the
& Nash, 2005; Sakurai & Reardon, 1997; Skees, presence of basis risk (Barnett, Black, Hu, &
1999a, 2000; Skees & Enkh-Amgalan, 2002; Skees, 2005; Black, Barnett, & Hu, 1999; Deng
Skees, Barnett, & Hartell, 2005; Skees, Gober, et al., 2007; Martin et al., 2001; Turvey, 2001;
Varangis, Lester, & Kalavakonda, 2001; Skees, Vedenov & Barnett, 2004; Wang, Hanson,
Hazell, & Miranda, 1999; Varangis, Skees, & Myers, & Black, 1998). The ndings from these
Barnett, 2002). studies are mixed. In some cases, basis risk can
be reduced by carefully choosing the IBRTP members of a smallholder farmer association.

parameters (e.g., the underlying index, the time The IBRTP is bundled with credit for the pur-
period, and the indemnity function) so that chase of improved groundnut varieties. In
indemnities are more highly correlated with Mongolia an IBRTP pilot product based on
actual losses incurred. However, for heteroge- aggregate levels of livestock mortality is being
neous regions, such as those with many micro- oered to individual herders. Some lenders,
climates, basis risk may be too high for IBRTPs who believe that the IBRTP can reduce loan
to be eective. It is important to remember that default risk, have also oered lower interest
the very characteristic that causes basis risk in rates to borrowers who purchase the IBRTP.
IBRTPs is also what eliminates asymmetric Alternatively, IBRTPs can be used to rein-
information problemsnamely, that payments sure portfolios of either index-based or tradi-
are based on realizations of the exogenous index tional insurance policies. In OECD countries,
rather than actual losses experienced by the IBRTPs are increasingly being used to reinsure
household. portfolios of traditional property and casualty
While farm-level insurance often fails in the insurance policies against covariate risks associ-
presence of covariate risk, IBRTPs will not be ated with hurricanes and earthquakes. IBRTPs
eective unless risks are somewhat covariate. facilitate the transfer of such covariate risks
If, in a given region, farm-level risk exposure into international nancial markets. Large
is completely idiosyncratic, no single index will investors are attracted to IBRTPs for their
provide eective risk transfer for farmers in diversication value since returns on IBRTPs
that region. But if risk exposure is completely are largely uncorrelated with returns on tradi-
idiosyncratic, these risks could be easily pooled tional debt and equity investments.
using formal or informal means. If farm-level
risk exposure is completely covariate, there will (b) Risk layering
be no opportunity for risk pooling at the local
level but an IBRTP should provide eective In capital and reinsurance markets it is com-
risk transfer for all farmers in the region. 5 In mon for risk exposure to be described accord-
reality, farm-level yield risks are generally nei- ing to layers of losses. Likewise, when
ther completely idiosyncratic nor completely considering possible applications of IBRTPs
covariate. Instead, they are what Skees and in low-income countries, it is important to iden-
Barnett (1999) call in-between risks. Losses tify dierent layers of risk exposure. For exam-
are suciently covariate to limit risk reduction ple, consider an index based on cumulative
through pooling but they are also suciently rainfall during the month of August for a coast-
idiosyncratic to cause basis risk with IBRTPs. al weather station in Andhra Pradesh, India
IBRTPs can be sold as retail insurance for where weather-based IBRTPs are currently
households or businesses. For example, in Ma- being oered. Figure 1 presents the probability
lawi a rainfall-based IBRTP is being oered to distribution of the index. The central tendency
Figure 1. Probability distribution of august cumulative rainfall for a coastal weather station in Andra Pradesh, India.
is approximately 1,300 mm of rainfall. Excess The market insurance layer includes loss
rainfall during August causes losses for farmers events that are, at least in principle, insurable
and those working in economic sectors related using IBRTPs. For this example, the market
to agriculture. insurance layer is assumed to include rainfall
As a conceptual tool, three layers of losses events between 1,750 and 2,300 mm. However,
are designated in Figure 1. The exact bound- a number of critical implementation issues must
aries on these layers would vary depending on be addressed before the potential for IBRTPs
the nature of the risk and the circumstances becomes reality. While many of these issues
of the end-user. The risk retention layer is char- are discussed later, we focus next on one very
acterized by high probability, but relatively low important implementation issuewho is the
magnitude loss events. These events are best target market for IBRTPs?
thought of as business expenses that are sto-
chastic but not unexpected. The transaction (c) Target market
costs would be very high to insure against these
high frequency events. Thus, they are most e- IBRTPs may be targeted to micro/house-
ciently handled through self-insurance or infor- hold, meso, or macro level users. The target
mal insurance mechanisms. For this example market has important implications for the
the risk retention layer is assumed to include choice of the underlying index. In choosing an
rainfall events greater than 1,300 mm but less appropriate target market and associated in-
than or equal to 1,750 mm. dex, tradeos generally exist between transac-
The market failure layer is characterized by tion costs and basis risk. For example,
very low probability, but high magnitude, loss separate rainfall IBRTPs could be oered based
events. For this example the market failure on each of several local weather stations. Alter-
layer is assumed to include rainfall events great- natively, a single rainfall IBRTP could be of-
er than 2,300 mm. Individuals nd it very di- fered where the index is a weighted average
cult to correctly process information about over all of the individual weather stations. If
such low probability events (Kunreuther, separate IBRTPs are oered for each weather
1976; Rossi, Wright, & Weber-Burdin, 1982). station, transaction costs will be high but basis
Beyond some threshold, individuals tend to risk may be low relative to the single weighted
treat low probability as though it is zero prob- average index. The single weighted average in-
ability (Kunreuther, 1996; Kunreuther & Slo- dex will have lower transaction costs but may
vic, 1978; Tversky & Kahneman, 1973). This subject micro-level users to high basis risk,
cognitive failure reduces demand for insurance especially if rainfall events tend to be highly
that protects against loss events in the extreme localized.
tail of the distribution. In many cases, households are not the appro-
At the same time, IBRTP suppliers are priate target for IBRTPs. The transaction costs
aware that some density exists in the tail of of servicing many small, household-level insur-
the distribution. When calculating the selling ance policies are quite high. Further, at the
price for an IBRTP, they must estimate the household level, idiosyncratic risk may be a
density in the tail of the distribution based major component of overall risk exposure
on very sparse data that causes insurers to (Dercon, 2005; Lybbert, Barrett, Desta, & Cop-
add ambiguity loads to the cost of IBRTPs. pock, 2004; McPeak & Barrett, 2001; Morduch,
Because of cognitive failure on the part of 2005; Townsend, 1995). This suggests both that
insurance purchasers and ambiguity loading basis risk for IBRTPs might be quite high at the
on the part of IBRTP suppliers, markets for household level and that opportunities exist for
protection against events in this layer tend pooling of idiosyncratic risks through local
to clear at less than socially optimal quanti- (commonly informal) mechanisms.
ties of risk transfer (Skees & Barnett, 1999). Meso-level commercial enterprises, such as
This market failure can be addressed through agricultural input suppliers, micronance insti-
public provision of coverage for this layer or tutions, marketing cooperatives, transportation
through premium subsidies (Hess et al., 2005). providers, agricultural commodity processors,
Both of these responses however, can intro- and retail insurance suppliers, may be better tar-
duce political economy concerns and generate gets for IBRTPs. These institutions can, at least
perverse behavioral incentives that must be to some degree, pool their exposure to house-
carefully considered. These matters are dis- hold-level idiosyncratic risks but often remain
cussed more fully below. heavily exposed to covariate risks (Hess et al.,
2005; Skees, Varangis, Larson, & Siegel, 2005; (d) Pre-nancing safety nets and disaster
Varangis et al., 2002). In addition, decision-mak- assistance
ers within meso-level commercial enterprises are
more likely to have some prior familiarity with IBRTPs can also be used to pre-nance
contingent claims instruments than are house- safety net or disaster assistance programs
hold decision-makers (Platteau, 1997). (Alderman & Haque, 2006; Hess et al., 2005;
Consider the case of micronance institutions Hess & Syroka, 2005). Properly conceptualized
(MFIs) or other rural lenders. When the losses and implemented for environments character-
experienced by borrowers are highly correlated, ized by poverty traps, safety nets are not de-
loan defaults are also likely to be highly corre- signed as income transfer programs to the
lated (Skees & Barnett, 2006). Returning to poorest of the pooras the term is sometimes
the earlier Andhra Pradesh example, a MFI in usedbut rather to protect productive assets
the region could purchase an IBRTP based on of those who might otherwise fall below the
August cumulative rainfall to protect its portfo- critical threshold and thereby fall onto a decu-
lio against the risk of increased loan defaults mulation path towards destitution (Barrett &
caused by excessive rainfall. For rainfall less Maxwell, 2005; Barrett & McPeak, 2005; Der-
than 1,750 mm, the MFI would retain the risk. con, 2005). Safety nets are intended to keep
The IBRTP would make a payment for rainfall those who experience transitory poverty follow-
occurrences between 1,750 and 2,300 mm. For ing a negative shock from becoming chronically
example, suppose that the micronance institu- poor. However, many low-income countries
tion purchased US$ 550,000 of protection. nd it dicult to nance safety net programs.
Since there are 550 mm in this layer, a simple International assistance tends to focus on
payout structure would pay US$ 1,000 for each acute, emergency needs rather than on funding
millimeter of rainfall beyond the 1,750 mm safety net programs designed to keep house-
threshold. The full payout would occur when holds from falling into a vicious cycle of asset
cumulative rainfall equals or exceeds 2,300 mm. decumulation. When international assistance
To further stimulate the availability of rural is provided for safety net programs it tends to
credit, the government or the international do- be too little, too late, and in the form of food
nor community could be involved in oering rather than cash (Barrett & Maxwell, 2005).
protection against extreme losses beyond Government or donor agencies could pur-
2,300 mm (Mahul & Skees, 2006; Skees, Har- chase index instruments to pre-nance safety
tell, & Hao, 2006). If governments wish to be net programs. For example, since 2002 the Na-
involved in subsidizing the cost of IBRTPs, tional Fund for Natural Disasters (FONDEN)
those subsidies should be focused on the mar- in Mexico, in collaboration with the govern-
ket failure layer. Subsidies for other layers are ment agricultural insurer (Agrosemex), has
likely to generate perverse behavioral incentives been purchasing IBRTPs to pre-nance natural
that cause even greater exposure to adverse disaster assistance (Agroasemex, 2006). Of
shocks. course, eective, broad-based, safety net poli-
Local governments also have limited ability cies must be able to respond to all negative
to withstand covariate shocks. Locally pro- shocks, not just those that can be eectively tied
vided public goods (e.g., law enforcement, to an underlying index. Thus, the extent to
maintenance of road and water infrastructure, which IBRTPs can be used to pre-nance a
health clinics, schools) may suer when public broad-based safety net policy will depend on lo-
assets are destroyed by covariate shocks and/ cal conditions. In areas where negative eco-
or public resources are diverted to relief eorts nomic shocks are often caused or amplied by
(Goes & Skees, 2003). Shocks that aect critical measurable risk factors (e.g., drought, ooding,
public goods can reduce spatial market integra- hurricanes), IBRTPs could play an important
tion, thus increasing local price volatility and role in pre-nancing a broad-based safety net
reducing incentives for households to adopt policy. In other areas, it may be less important.
production-increasing technologies (Gabre- Government or donor agencies may also be
Madhin, Barrett, & Dorosh, 2002). Local gov- interested in purchasing IBRTPs to pre--
ernments could use IBRTPs to transfer some nance emergency food aid and other disaster
of their exposure to covariate risks. Alterna- relief eorts (Skees, Varangis et al., 2005).
tively, national governments or donor agencies Some covariate shocks, such as an extended,
could purchase IBRTPs on behalf of local gov- widespread, drought, do not occur suddenly.
ernments. Instead they develop over time. After early
warning systems are triggered, months may It is important to note however, that many
pass before the impact is seen in reduced food humanitarian crises are either caused, or at
availability, incomes and anthropometric mea- least amplied, by factors other than climatic
sures of nutritional status. Unlike with rapid variability (e.g., conict and lack of security,
onset emergencies such as earthquakes or hur- poor governance, lack of market integration,
ricanes, there is time to prepare for slow onset etc.). Thus, while IBRTPs are a valuable tool
emergencies before the full force of the shock that can be used to pre-nance rapid responses
hits. Unfortunately, the lead time available to catastrophes caused by some climatic or nat-
for preparing for slow onset emergencies, ural events, they are certainly not capable of
whether seasonal or regular, is not always well addressing all causes of humanitarian crises.
used. International political and nancial sup- As with safety net policies, emergency response
port for humanitarian assistance often does policies should never be tied exclusively to
not develop until the situation becomes quite IBRTPs. However, under certain circum-
dire (Barrett, 2006; Barrett & Maxwell, stances, IBRTPs can play an important role
2005). Thus, as demonstrated by the 200506 within a broader portfolio of emergency re-
famine in Niger, even after accurate early sponse tools.
warning of a looming disaster, many months
may pass before assistance arrives in the af-
fected areas. During this time, households (e) Experience to date
have to decide between distress sales of
productive tangible assets or disinvestment in While IBRTP programs are either in place
human assets (e.g., malnutrition, removing or under development in several middle-
children from school, forgone needed health- or low-income countries (see Table 1), there
care, etc.). Either decision leads to asset is not yet sucient experience to draw
decumulation that may have long-term reper- denitive conclusions about the long-run
cussions in the presence of threshold-based sustainability of these programs. 6 Except for
poverty traps. And even after the delays, re- India, existing IBRTP programs are in
sponse is often insucient to provide adequate pilot stages so the volume of sales has been
cover for losses experiences. The Consolidated limited.
Appeals Process established by the United Na- In India, ICICI Lombard General Insur-
tions in 1991 to mobilize resources in response ance Company has sold IBRTPs to farmers
to emergencies has largely proved ineective. since 2003. ICICI Lombard partners with lo-
Former UN Secretary General Ko Annan re- cal nancial institutions to market the policies
ported in October 2005 that ash appeals had to farmers. The IBRTPs were rst oered in
generated on average only 16% of the re- the state of Andra Pradesh and marketed
quested funds (Barrett, 2006). through the micronance institution BASIX
This underscores the possibilities associated (Hess, 2003). In Andra Pradesh, the IBRTPs
with IBRTPs that trigger based on an early currently protect against insucient rainfall
indicator of food insecurity (e.g., rainfall mea- during the sowing and crop growth phases
sures or measures from drought early warning of the Kharif (monsoon season) and against
systems). IBRTPs could fund more timely excessive rainfall during the harvest phase
humanitarian response eorts thus reducing (Gine et al., 2007b). ICICI Lombards rain-
the need for households that are already as- fall-based IBRTP oerings have continued
set-poor to engage in asset decumulation cop- to expand such that in 2005, more than
ing strategies. For example, in March 2006, 7,600 policies were sold across six Indian
the United Nations World Food Program an- states (Manuamorn, 2007). In 2004, the In-
nounced that it paid the French insurance com- dian parastatal insurance company AICI also
pany AXA Re US$930,000 for an IBRTP that began selling weather-based IBRTPs to farm-
would pay up to $7.1 million to help up to ers. In 200506, AICI sold IBRTP policies to
67,000 Ethiopian households in the event of more than 125,000 farmers, however most of
inadequate rainfall during the critical March the policies were sold in only one state
October period (Hess, Wiseman, & Robertson, Maharashtra (Barnett & Mahul, 2007). Early
2006; Insurance Journal, 2006; New York analyses of IBRTP purchasing behavior in In-
Times, 2006). This particular IBRTP was india have been conducted by Gine, Townsend,
tended to provide ready cash to fund early and Vickery (2007a), Lilleor, Gine, Town-
interventions as a major drought is developing. send, and Vickery (2005).
Table 1. Summary of index based risk transfer products in middle- and low-income countries
Country Risk event Contract structure Index measure Target user Status
Bangladesh Flood Index insurance for In development
disaster relief
POVERTY TRAPS AND INDEX-BASED RISK TRANSFER PRODUCTS

Bangladesh Drought Index insurance linked Rainfall Smallholder rice farmers In development. Pilot
to lending launch planned for 2008
Caribbean Hurricanes and Index insurance Indexed data from Caribbean country Implemented in 2007
Catastrophe Risk earthquakes contracts with risk NOAA and USGS governments
Insurance Facility pooling for reinsurance
coverage
Ethiopia Drought Index insurance Rainfall World Food Programme $7 million insured for
operations in Ethiopia 2006. Policy not renewed
for 2007
Honduras Drought Rainfall In development
India Drought and ood Index insurance linked Rainfall Smallholder farmers Began with pilot in 2003.
to lending and oered Now index insurance
direct to farmers products are being
oered by the private
sector and the parastatal
insurer with an estimated
300,000 policies sold in
2006
Kazakhstan Drought Index insurance linked Rainfall Medium and large farms In development
to Multiple Peril Crop
Insurance program
Malawi Drought Index insurance linked Rainfall Groundnut farmers who are Pilot began in 2005.
to lending members of National 2,500 policies sold in
Smallholder Farmers 2006 pilot season. $7,000
Association of Malawi in premium volume
Mexico Natural disasters, Index insurance Rainfall, wind speed, State governments for Began in 2001. Available
primarily drought temperature disaster relief. Supports the in 26 of 32 states.
Fondo por Desastres Currently 28% (2.3
Naturales program million Ha) of dryland
cropland is covered.
Expansion limited by
data availability
1777
(continued on next page)
Table 1continued
1778
Country Risk event Contract structure Index measure Target user Status
Mexico Major earthquakes Cat bond and index Richter scale readings Mexican government to Introduced in 2006. Cat
insurance contracts support Fondo por Desastres bond provides up to
Naturales program $160 million. Index
insurance provides
additional funding up to
$290 million
Mexico Insucient irrigation Index insurance Reservoir levels Water user groups Proposed
supply in the Rio Mayo area
Mongolia Large livestock losses Index insurance with Area livestock mortality Nomadic herders Second sales season of
due to severe weather direct sales to herders rate pilot completed in 2007.
Oered in 3 provinces.
14% of eligible herders
are participating
WORLD DEVELOPMENT
Morocco Drought Rainfall No interest from market
due to declining trend in
rainfall
Nicaragua Drought and excess rain Index insurance Rainfall Groundnut farmers Launched in 3
during production, departments in 2006
excess rain during
harvest period
Peru Flooding, torrential Index insurance ENSO anomalies in Rural nancial institutions Proposed
rainfall from El Nino Pacic Ocean
Peru Drought Index insurance linked Area-yield production Cotton farmers Proposed
to lending index
Senegal Drought Index insurance linked Rainfall and crop yield Smallholder farmers Proposed
to area yield insurance
Tanzania Drought Index insurance linked Rainfall Smallholder maize farmers Pilot implementation in
to lending 2007
Thailand Drought Index insurance linked Rainfall Smallholder farmers Pilot implementation in
to lending 2007
Ukraine Drought Index insurance Rainfall Large farms Pilot launched in 2005,
discontinued due to
insucient sales
Vietnam Flooding during rice Index insurance linked River level Smallholder rice farmers In development
harvest to lending
5. IMPLEMENTATION ISSUES bution of the index may be conditioned on

various factors. As an example, some climatic
Index instruments, if they are to be eective events exhibit serial correlation such as the im-
and sustainable, must be designed with careful pacts of El NinoSouthern Oscillation in
consideration of several critical implementation some parts of the Western hemisphere. If po-
issues. 7 tential purchasers have access to relevant
information but premium rates are not condi-
(a) Dening the index tioned on this same information, intertemporal
adverse selection will likely occur. In designing
What measurable stochastic variable or com- index instruments it is critical to establish sales
bination of stochastic variables will be used as closing dates early enough that potential pur-
the index? At what location or locations will chasers do not have access to information that
the stochastic variables be measured? Is the can be used to intertemporally adversely se-
measurement process secure, objective, and lect.
transparent? Over what time period will the in-
dex be measured? Do adequate historical data (d) Legal and regulatory issues
exist to estimate the probability distribution
of the index? As discussed previously, the an- Most low-income countries have a govern-
swers to these questions often involve a tradeo ment agency that regulates the insurance indus-
between transaction costs and basis risk. try. To reduce delivery costs, IBRTPs are
sometimes sold through existing insurance
(b) Estimating the probability distribution companies. However, the legal and regulatory
of the index environment that exists for the insurance indus-
try may not be suciently exible to accommo-
Assuming that adequate historical data are date IBRTPs. Thus, recent eorts to introduce
available, what procedures should one use to IBRTPs in low-income countries have generally
estimate the probability distribution of the in- required changes in the legal and regulatory
dex? Generally, sucient data will not be avail- environment to accommodate risk-transfer
able to simply use empirical distributions instruments other than traditional insurance
especially for estimating the all-important tail (Carpenter & Skees, 2005).
density of the distributions. Various parametric
and non-parametric statistical procedures can (e) Delivery system
be used to estimate the distribution. The choice
of procedure often depends, in part, on the ex- What delivery mechanisms are available and
tent of available historical data. Do the histor- sustainable? Index instruments can take many
ical data suggest that the distribution of the forms and can be delivered through various
index is non-stationary and heteroskedastic? If mechanisms. In some countries, insurance
so, statistical correction procedures must be markets are suciently developed that IBRTPs
used to estimate the probability distribution can be marketed directly via existing insurance
of the index. While a number of common cor- institutions. However, in many countries, -
rection procedures are available, decisions nance, output processing, and/or input sup-
regarding which procedure to employ and ply sectors are better developed than the
how the procedure is applied can have nontriv- insurance sector. Firms in these sectors may
ial implications for premium rating (Hess et al., be willing to oer state-contingent contracts
2005). to clients if the rm can, in turn, transfer its
state-contingent risk using IBRTPs (Skees &
(c) Intertemporal adverse selection Barnett, 2006).
Index instruments do not require insurers to (f) Subsidies

classify potential purchasers according to their
risk exposure. Thus, they are not susceptible Should governments subsidize IBRTPs pre-
to the type of cross-sectional adverse selection miums? Signicant premium subsidies are likely
problems that plague many traditional insur- to crowd out new or existing formal and infor-
ance products. Index instruments can how- mal risk transfer mechanisms. Premium subsi-
ever, be susceptible to intertemporal adverse dies that are denominated as a percentage of
selection. Within any given period, the distri- the unsubsidized premium will also create
perverse behavioral incentives since those who 6. CONCLUSION

take the greatest risks will receive the most in
government subsidies (Barnett, 2003; Skees, As researchers and policymakers increasingly
1999b, 2001). It is likely that the least distorting focus on the tragedy of poverty traps, there is
type of government premium subsidy would correspondingly increasing awareness of the
apply only to the market failure layer (charac- central role that insurance and credit market
terized by cognitive failure and ambiguity load- failures play in perpetuating poverty in many
ing). This layer could be nanced by low-income rural areas. Those market failures
government or the broader international donor are caused by poor contract enforcement mech-
community. anisms, information asymmetries, high transac-
Consider, for example, a World Bank-funded tion costs, and covariate risk exposure.
project on Index-Based Livestock Insurance in IBRTPs can help address these market fail-
Mongolia. Insurance companies oer herders ures. Relative to traditional insurance, IBRTPs
an IBRTP based on annual livestock mortality have simpler contract designs, fewer asymmet-
rates. For each species and county, mortality ric information problems, and lower transac-
rates between 7% and 30% constitute the mar- tion costs. Further, IBRTPs are designed to
ket insurance layer within which insurance transfer covariate risks out of a country or re-
companies make payments. The market failure gion and into international nancial markets.
layer is dened as species-specic mortality For these reasons, IBRTPs may be a valuable
rates that exceed 30%. The government pays instrument for addressing some of the insur-
for those losses. In addition, the government ance and credit market failures that contribute
has formed a reinsurance pool that covers to chronic poverty in low-income countries.
insurance companies against losses in excess This realization has motivated many recent ef-
of 105% of herder premiums. If needed, the forts to develop IBRTPs applications for low-
World Bank will provide the government with income countries. Some of these eorts focus
a loan to intertemporally smooth losses beyond on making market-based IBRTPs available to
those that are covered by the reinsurance pool businesses or households that are highly ex-
and to pay for disaster response costs for mor- posed to specic covariate risks. Others are
tality rates that exceed 30% (Mahul & Skees, attempting to use IBRTPs as the basis for
2006). pre-nancing safety net or disaster assistance
programs.
(g) Other government or donor roles While IBRTPs have many promising fea-
tures, they also have limitations that must be
What else can government or donor agencies recognized and addressed, to the extent possi-
do to facilitate the development of IBRTPs? ble, through careful product design. Holders
Government and donor interventions should of IBRTPs are subject to basis risk that may
focus on eorts that crowd-in, or at least leave them uncompensated for some signicant
do not crowd-out, new or existing formal losses caused by idiosyncratic rather than
and informal risk transfer mechanisms. Invest- covariate perils. In design and implementation
ments in the collection, warehousing, and dis- of IBRTPs, it is therefore essential that the in-
semination of weather data would be an dex used is highly correlated with realized
example. Signicant research and development losses. A related issue is the availability, qual-
costs are required to design an eective ity, security and transparency of the data re-
IBRTP. However, once the IBRTP is devel- quired to establish and maintain the index.
oped, it can be easily copied by competitors. The form of the IBRTP and the choice of deliv-
Government or donor agencies can facilitate ery mechanism will vary depending on local cir-
the development of IBRTPs by absorbing cumstances and existing institutions.
some of these initial research and development Signicant investment in research and devel-
costs. The World Bank is currently engaged in opment is required to address these various
several such eorts (Hess et al., 2002, 2005; location- and application-specic implementa-
Mahul & Skees, 2006; Skees et al., 2001; Skees tion issues. Once an IBRTP is developed, it is
& Enkh-Amgalan, 2002). Government or do- relatively easy for other rms to copy the de-
nor agencies can also absorb some of the sign and sell competing products. International
transactions costs of bundling IBRTPs for reinsurers may therefore be unwilling to make
transfer into international markets (Hess such investments in relatively small, low-in-
et al., 2005). come markets. Recognizing this, the World
Bank and other donor organizations have be- programs based on IBRTPs have an opportu-
gun underwriting some of the research and nity cost of resources not allocated to alterna-
development costs for low-income country tive investments, it remains unclear as to the
applications of IBRTPs. conditions under which IBRTPs are appropri-
IBRTPs show considerable potential for ate investments, when they might complement
addressing certain covariate risk-related nan- other development interventions, and how these
cial market failures that contribute to some peo- instruments t into broader development strat-
ples persistent poverty. However, because the egies. These are key topics for future policy-ori-
development and provision of risk management ented research on poverty traps and IBRTPs.
NOTES
1. See, for example, the collections of papers in Baulch 5. A reviewer notes that, for traditional insurance
and Hoddinott (2000) or Barrett, Carter, and Little products, risk pooling at the local level is not necessary if
(2006). the risk can be transferred into international reinsurance
markets. While, in principle, this is true, pricing for
reinsurance against covariate natural disaster risks has
2. For this reason, a major crop insurer in Mexico proven to be highly cyclical (Jaee & Russell, 1997).
recently announced that it will only insure parcels of at Further, traditional reinsurance contracts are tailored
least 25 ha. instruments that generally have high transaction costs
for legal fees and due diligence (Doherty, 1997). Thus, if
the risk of concern is highly covariate, IBRTPs will often
3. Weather-based index insurance is a specic type of
be a lower cost mechanism for transferring the risk.
IBRTP that has been discussed in the agricultural
economics literature (e.g., Deng, Barnett, & Vedenov,
2007; Barnett & Mahul, 2007; Chantarat, Barrett, Mude, 6. See the following sources for additional information:
& Turvey, 2007; Gine, Townsend, & Vickery, 2007b). Ethiopia (Hess et al., 2006); Malawi (Hess & Syroka,
2005); Mongolia (Mahul & Skees, 2006, 2007); Peru
(Khalil, Kwon, Lall, Miranda, & Skees, 2007; Skees,
4. In a general sense, IBRTPs are conceptually analo- Hartell, & Murphy, 2007); Ukraine (Shynkarenko,
gous to European options on the underlying index 2007); and Vietnam (Skees et al., 2007).
(Barnett, 1999, 2000; Skees & Barnett, 1999). The
instruments can be constructed as calls (a payment 7. More detailed discussions of implementation issues
is made when the realized index value exceeds the are found in Barnett and Mahul (2007), Bryla and Syroka
threshold) or puts (a payment is made when the (2007), Carpenter and Skees (2005), GlobalAgRisk
realized index value falls short of the threshold). (2006), Hess et al. (2005) and Skees, Varangis et al. (2005).
REFERENCES
AGROASEMEX (2006). La Experiencia Mexicana en el Handbook of income distribution. Amsterdam: North-

Desarrollo y Operacion de Seguros parametricos Holland.
Aplicados a la Agricultura. Working paper, April 19, Barnett, B. J. (1995). The federal crop insurance
2006. program: Past, present, and future. Journal of
Alderman, H., & Haque, T. (2006). Countercyclical Insurance Issues, 18, 75100.
safety nets for the poor and vulnerable. Food Policy, Barnett, B. J. (1999). US government natural
31, 372383. disaster assistance: Historical analysis and a
Arnott, R., & Stiglitz, J. E. (1991). Moral hazard and proposal for the future. Disasters: The Journal
nonmarket institutions: Dysfunctional crowding out of Disaster Studies, Policy and Management,
or peer monitoring?. American Economic Review, 81, 23, 139155.
179190. Barnett, B. J. (2000). The US Federal crop insurance
Azariadis, C., & Stachurski, J. (2005). Poverty traps. In program. Canadian Journal of Agricultural Econom-
P. Aghion, & S. Durlauf (Eds.). Handbook of ics, 48, 539551.
economic growth (Vol. 1A). Amsterdam: North- Barnett, B. J. (2003). The federal crop insurance
Holland. program: Opportunities and challenges. Paper pre-
Bardhan, P., Bowles, S., & Gintis, H. (2000). Wealth sented at the US Department of Agricultures Agri-
inequality, credit constraints, and economic perfor- cultural Outlook Forum, Arlington, Virginia,
mance. In A. Atkinson, & F. Bourguignon (Eds.), February 2003.
Barnett, B. J., Black, J. R., Hu, Y., & Skees, J. R. (2005). lished paper prepared for the World Bank,
Is area yield insurance competitive with farm yield September 2005.
insurance? Journal of Agricultural and Resource Carter, M. R. (1988). Equilibrium credit rationing of
Economics, 30, 285301. small farm agriculture. Journal of Development Eco-
Barnett, B. J., & Mahul, O. (2007). Weather index nomics, 28, 83103.
insurance for agriculture and rural areas in lower- Carter, M. R., & Barrett, C. B. (2006). The economics of
income countries. American Journal of Agricultural poverty traps and persistent poverty: An asset based
Economics, 89, 12411247. approach. Journal of Development Studies, 42,
Barrett, C. B. (2006). Food aid in response to acute food 178199.
insecurity. Background Paper prepared for the Food Carter, M. R., Little, P. D., Mogues, T., & Negatu, W.
and Agriculture Organization of the United Nations (2005). The long-term impacts of short-term shocks:
State of Food and Agriculture 2006 report. Poverty traps and environmental disasters in ethiopia
Barrett, C. B., Bezuneh, M., & Aboud, A. (2001). and honduras. BASIS Brief 28, May 2005, Madison:
Income diversication, poverty traps and policy University of Wisconsin.
shocks in Cote dIvoire and Kenya. Food Policy, Carter, M. R., Little, P. D., Mogues, T., & Negatu, W.
26, 367384. (2006). Shocks, sensitivity and resilience: Tracking
Barrett, C. B., Bezuneh, M., Clay, D. C., & Reardon, T. the economic impacts of environmental disaster on
(2005). Heterogeneous constraints, incentives and assets in Ethiopia and Honduras. International Food
income diversication strategies in rural Africa. Quar- Policy Research Institute, Development Strategy and
terly Journal of International Agriculture, 44, 3760. Governance Division, Discussion Paper No. 32.
Barrett, C. B., Carter, M. R., & Little, P. D. (Eds.) Chambers, R. G. (1989). Insurability and moral hazard
. Understanding and reducing persistent poverty in in agricultural insurance markets. American Journal
Africa. London: Routledge. of Agricultural Economics, 71, 604616.
Barrett, C. B., Marenya, P. P., McPeak, J., Minten, B., Chantarat, S., Barrett, C. B., Mude, A. G., & Turvey, C.
Murithi, F., Oluoch-Kosura, W., et al. (2006). G. (2007). Using weather index insurance to
Welfare dynamics in rural Kenya and Madagascar. improve drought response for famine prevention.
Journal of Development Studies, 42, 248277. American Journal of Agricultural Economics, 89,
Barrett, C. B., & Maxwell, D. G. (2005). Food aid after 12621268.
fty years: Recasting its role. London: Routledge. Coate, S., & Ravallion, M. (1993). Reciprocity without
Barrett, C. B., & McPeak, J. G. (2005). Poverty traps commitment: Characterization and performance of
and safety nets. In A. de Janvry, & R. Kanbur (Eds.). informal insurance arrangements. Journal of Devel-
Poverty, inequality, and development: Essays in honor opment Economics, 40, 124.
of Erik Thorbecke. Norwell MA: Kluwer Academic Coble, K. H., Knight, T. O., Pope, R. D., & Williams, J. R.
Publishers. (1997). An expected-indemnity approach to the mea-
Barrett, C. B., & Swallow, B. M. (2006). Fractal poverty surement of moral hazard in crop insurance. American
traps. World Development, 34, 115. Journal of Agricultural Economics, 79, 216226.
Baulch, B., & Hoddinott, J. (Eds.) (2000). Economic Dasgupta, P. (1993). An inquiry into well-being and
mobility and poverty dynamics in developing countries. destitution. Oxford: Clarendon Press.
London: Frank Cass. Dasgupta, P. (1997). Nutritional status, the capacity for
Besley, T. (1995). Nonmarket institutions for credit and work, and poverty traps. Journal of Econometrics, 77,
risk sharing in low-income countries. Journal of 537.
Economic Perspectives, 9, 115127. Deng, X., Barnett, B. J., & Vedenov, D. V. (2007). Is
Binswanger, H. P., & Rosenzweig, M. R. (1986). there a viable market for area-based crop insurance.
Behavioural and material determinants of produc- American Journal of Agricultural Economics, 89,
tion relations in agriculture. Journal of Development 508519.
Studies, 22, 503539. Deng, X., Barnett, B. J., Vedenov, D. V., & West, J. W.
Black, J. R., Barnett, B. J., & Hu, Y. (1999). Cooper- (2007). Using weather-based index insurance to
atives and capital markets: The case of Minnesota protect against dairy production losses caused by
Dakota sugar cooperatives. American Journal of heat stress. Agricultural Economics, 36, 271280.
Agricultural Economics, 81, 12401246. Dercon, S. (1996). Risk, crop choice, and savings:
Bowles, S., Durlauf, S. N., & Ho, K. (Eds.) (2006). Evidence from Tanzania. Economic Development
Poverty traps. Princeton: Princeton University Press. and Cultural Change, 44, 485513.
Braverman, A., & Guasch, J. L. (1986). Rural credit Dercon, S. (1998). Wealth, risk and activity choice:
markets and institutions in developing countries: Cattle in Western Tanzania. Journal of Development
Lessons for policy analysis from practice and modern Economics, 55, 142.
theory. World Development, 14, 12531267. Dercon, S. (2004). Growth and shocks: Evidence from
Bryla, E., & Syroka, J. (2007). Developing index based rural Ethiopia. Journal of Development Economics,
insurance for agriculture in developing countries. 74, 309329.
Sustainable development innovation briefs. New York: Dercon, S. (Ed.) (2005). Insurance against poverty.
United Nations Department of Economic and Social Oxford: Oxford University Press.
Aairs. Dercon, S., & Hoddinott, J. (2005). Health, shocks, and
Carpenter, R., & Skees, J. (2005). Index-based insur- poverty persistence. In S. Dercon (Ed.), Insurance
ance products: Legal and regulatory issues. Unpub- against poverty. Oxford: Oxford University Press.
Doherty, N. A. (1997). Innovations in managing catas- vations in developing countries. The World Bank,
trophe risk. Journal of Risk and Insurance, 64, Agriculture and Rural Development Department,
713718. Report No. 32727-GLB, June 2005.
Eswaran, M., & Kotwal, A. (1989). Credit as insurance Hess, U., & Syroka, J. (2005). Weather-based insurance
in agrarian economies. Journal of Development Eco- in Southern Africa: The case of Malawi. The World
nomics, 31, 3753. Bank, Agriculture and Rural Development Depart-
Eswaran, M., & Kotwal, A. (1990). Implications of ment, Discussion Paper 13.
credit constraints for risk behavior in less developed Hess, U., Wiseman, W., & Robertson, T. (2006).
economies. Oxford Economic Papers, 42, 473482. Ethiopia: Integrated risk nancing to protect livlihoods
Fafchamps, M., & Lund, S. (2003). Risk-sharing and foster development. <http://www.lead.virtualcen-
networks in rural Philippines. Journal of Develop- ter.org/en/ele/econf_03_alive/download/s1_10_Ethi-
ment Economics, 71, 261287. opia.pdf> October 2006.
Foster, A. D. (1995). Prices, credit markets and child Hoddinott, J. (2006). Shocks and their consequences
growth in low-income rural areas. Economic Journal, across and within households in rural Zimbabwe.
105, 551570. Journal of Development Studies, 42, 301321.
Freedman, P. L., & Click, R. W. (2006). Banks that Hoddinott, J., & Kinsey, B. (2001). Child growth in the
dont lend? Unlocking credit to spur growth in time of drought. Oxford Bulletin of Economics and
developing countries. Development Policy Review, 24, Statistics, 63, 409436.
279302. Ho, K., & Stiglitz, J. E. (1990). Introduction: Imperfect
Gabre-Madhin, E., Barrett, C. B., & Dorosh, P. (2002). information and rural credit marketsPuzzles and
Technological change and price eects in agriculture: policy perspectives. The World Bank Economic
Conceptual and comparative perspectives. In T. Bon- Review, 4, 235250.
ger, E. Gabre-Madhin, & S. Babu (Eds.), Agriculture Insurance Journal. (2006). UNs World food program
technology diusion and price policy. Washington: insures against ethiopia famine with AXA Re. March
Ethiopian Development Research Institute and the 9, 2006. <http://www.insurancejournal.com/news/
International Food Policy Research Institute. international/2006/03/09/66294.htm>.
Gine, X., Townsend, R., & Vickery, J. (2007a). Patterns Jacoby, H. G., & Skouas, E. (1997). Risk, nancial
of rainfall insurance participation in rural India. The markets, and human capital in a developing country.
World Bank working paper, May 18, 2007. Review of Economic Studies, 64, 311335.
Gine, X., Townsend, R., & Vickery, J. (2007b). Statis- Jaee, D. M., & Russell, T. (1997). Catastrophe insur-
tical analysis of rainfall insurance payouts in south- ance, capital markets, and uninsurable risks. Journal
ern India. American Journal of Agricultural of Risk and Insurance, 64, 205230.
Economics, 89, 12481254. Jalan, J., & Ravallion, M. (1999). Are the poor less well
GlobalAgRisk (2006). Index insurance for weather risk in insured? Evidence on vulnerability to income risk in
lower-income countries. Washington, DC: United rural China. Journal of Development Economics, 58,
States Agency for International Development, 2006. 6181.
Goes, A., & Skees, J. (2003). Financing natural disaster Just, R. E., Calvin, L., & Quiggin, J. (1999). Adverse
risk using charity contributions and ex ante index selection in crop insurance: Actuarial and asymmet-
insurance. Paper presented at the annual meeting of ric information incentives. American Journal of
the American Agricultural Economics Association, Agricultural Economics, 81, 834849.
Montreal, July 2003. Kazianga, H., & Udry, C. (2006). Consumption smooth-
Goodwin, B. K. (2001). Problems with market insurance ing? Livestock, insurance and drought in rural
in agriculture. American Journal of Agricultural Burkina Faso. Journal of Development Economics,
Economics, 83, 643649. 79, 413446.
Grimard, F. (1997). Household consumption smoothing Khalil, A. F., Kwon, H. H., Lall, U., Miranda, M. J., &
through ethnic ties: Evidence from Cote dIvoire. Skees, J. R., (2007). ENSO-based index insurance for
Journal of Development Economics, 53, 391421. oods: Statistical risk analyses and application to
Hazell, P., & Skees, J. R. (2006). Insuring against bad Peru. Water Resource Research, 43(10), Manuscript
weather: Recent thinking. In R. Radhakrishna, S. K. 2006WR005281.
Rao, S. Mahendra Dev, & K. Subbarao (Eds.). India Krishna, A. (2006). For reducing poverty faster: Target
in a globalising world: Some aspects of macroecono- reasons before people. Duke University working
my, agriculture, and poverty. New Delhi: Academic paper.
Foundation. Kunreuther, H. (1976). Limited knowledge and insur-
Hess, U. (2003). Innovative nancial services for rural ance protection. Public Policy, 24, 227261.
India: Monsoon-indexed lending and insurance for Kunreuther, H. (1996). Mitigating disaster losses
smallholders. The World Bank, Agriculture and through insurance. Journal of Risk and Uncertainty,
Rural Development working paper 9, 2003. 12, 171187.
Hess, U., Richter, K., & Stoppa, A. (2002). Weather risk Kunreuther, H., & Slovic, P. (1978). Economics, psy-
management for agriculture and agri-business in chology, and protective behavior. American Eco-
developing countries. In R. S. Dischel (Ed.), Climate nomic Review, 68, 6469.
risk and the weather market. London: Risk Books. Lilleor, H. B., Gine, X., Townsend, R., & Vickery, J.
Hess, U., Skees, J., Stoppa, A., Barnett, B., & Nash, J. (2005). Weather insurance in semi-arid India. The
(2005). Managing agricultural production risk: Inno- World Bank working paper, March 23, 2005.
Little, P. D., Smith, K., Cellarius, B. A., Coppock, D. Quiggin, J., Karagiannis, G., & Stanton, J. (1994). Crop
L., & Barrett, C. B. (2001). Avoiding disaster: insurance and crop production: An empirical study of
Diversication and risk management among east moral hazard and adverse selection. In D. L. Hueth, &
African Herders. Development and Change, 32, W. H. Furtran (Eds.), Economics of agricultural crop
387419. insurance. Norwell MA: Kluwer Academic Publishers.
Lybbert, T. J., Barrett, C. B., Desta, S., & Coppock, D. Reardon, T. (1997). Using evidence of household income
L. (2004). Stochastic wealth dynamics and risk diversication to inform study of the rural nonfarm
management among a poor population. The Eco- labor market in Africa. World Development, 25,
nomic Journal, 114, 750777. 735747.
Mahul, O. (2001). Optimal insurance against climatic Reardon, T., Delgado, C., & Matlon, P. (1992). Deter-
experience. American Journal of Agricultural Eco- minants and eects of income diversication
nomics, 83, 593604. amongst farm households in Burkina Faso. Journal
Mahul, O., & Skees, J. R. (2006). Piloting index-based of Development Studies, 28, 264296.
livestock insurance in Mongolia. AccessFinance: A Reardon, T., & Taylor, J. E. (1996). Agroclimatic shock,
newsletter published by the nancial sector vice income inequality, and poverty: Evidence from
presidency, The World Bank Group, Issue No. 10, Burkina Faso. World Development, 24, 901914.
March 2006. Rosenzweig, M. R. (1988). Risk, implicit contracts and
Mahul, O., & Skees, J. R. (2007). Managing agricultural the family in rural areas of low-income countries.
risk at the country level: The case of index-based Economic Journal, 98, 11481170.
livestock insurance in Mongolia. The World Bank Rosenzweig, M. R., & Binswanger, H. P. (1993). Wealth,
Policy Research Working Paper WPS4325, August 1, weather risk and the composition and protability of
2007. agricultural investments. Economic Journal, 103,
Manuamorn, O. P. (2007). Scaling up microinsurance: 5678.
The case of weather insurance for smallholders in Rosenzweig, M. R., & Wolpin, K. I. (1993). Credit
India. The World Bank, Agriculture and Rural market constraints, consumption smoothing, and the
Development Discussion Paper 36. accumulation of durable production assets in low-
Martin, S. W., Barnett, B. J., & Coble, K. H. (2001). income countries: Investments in bullocks in India.
Developing and pricing precipitation insurance. Journal of Political Economy, 101, 223244.
Journal of Agricultural and Resource Economics, 26, Rossi, P., Wright, J., & Weber-Burdin, E. (1982). Natural
261274. hazards and public choice: The state and local politics of
McPeak, J. G., & Barrett, C. B. (2001). Dierential risk hazard mitigation. New York: Academic Press.
exposure and stochastic poverty traps among East Rothschild, M., & Stiglitz, J. (1976). Equilibrium in
African pastoralists. American Journal of Agricul- competitive insurance markets: An essay on the
tural Economics, 83, 674679. economics of imperfect information. Quarterly Jour-
Mehlum, H., Moene, K., & Torvik, R. (2005). Crime nal of Economics, 90, 629649.
induced poverty traps. Journal of Development Eco- Sakurai, T., & Reardon, T. (1997). Potential demand for
nomics, 77, 325340. drought insurance in Burkina Faso and its determi-
Miranda, M. J. (1991). Area yield insurance reconsid- nants. American Journal of Agricultural Economics,
ered. American Journal of Agricultural Economics, 73, 79, 11931207.
233242. Santos, P., & Barrett, C. B. (2006a). Heterogeneous
Miranda, M. J., & Glauber, J. W. (1997). Systemic risk, wealth dynamics: On the roles of risk and ability.
reinsurance, and the failure of crop insurance mar- Cornell University working paper.
kets. American Journal of Agricultural Economics, 79, Santos, P., & Barrett, C. B. (2006b). Informal insurance in
206215. the presence of poverty traps: Evidence from Southern
Miranda, M. J., & Vedenov, D. V. (2001). Innovations Ethiopia, Cornell University working paper.
in agricultural and natural disaster insurance. Amer- Shynkarenko, R. (2007). Introduction of weather index
ican Journal of Agricultural Economics, 83, 650655. insurance in UkraineObstacles and opportunities.
Morduch, J. (1995). Income smoothing and consump- Presentation at the 101st EAAE seminar on man-
tion smoothing. Journal of Economic Perspectives, 9, agement of climate risks in agriculture, Berlin,
103114. Germany, July 56, 2007.
Morduch, J. (2005). Consumption smoothing across Skees, J. R. (1999a). Opportunities for improved e-
space: Testing theories of risk-sharing in the ICRI- ciency in risk sharing using capital markets. Amer-
SAT study region of South India. In S. Dercon (Ed.), ican Journal of Agricultural Economics, 81,
Insurance against poverty. Oxford: Oxford University 12281233.
Press. Skees, J. R. (1999b). Agricultural risk management or
New York Times (2006). The drought derivative. May 1, income enhancement?. Regulation, 22, 3543.
p. 18. Skees, J. R. (2000). A role for capital markets in natural
Platteau, J. (1997). Mutual insurance as an elusive disasters: A piece of the food security puzzle. Food
concept in traditional rural communities. Journal of Policy, 25, 365378.
Development Studies, 33, 764796. Skees, J. R. (2001). The bad harvest. Regulation, 24,
Priest, G. L. (1996). The government, the market, and 1621.
the problem of catastrophic loss. Journal of Risk and Skees, J. R., & Enkh-Amgalan, Ayurzana (2002).
Uncertainty, 12, 219237. Examining the feasibility of livestock insurance in
Mongolia. The World Bank, Policy Research Work- Smith, V. H., & Goodwin, B. K. (1996). Crop
ing Paper 2886, September 2002. insurance, moral hazard, and agricultural chemical
Skees, J. R., & Barnett, B. J. (1999). Conceptual and practical use. American Journal of Agricultural Economics,
considerations for sharing catastrophic/systemic risks. 78, 428438.
Review of Agricultural Economics, 21, 424441. Stiglitz, J. E. (1990). Peer monitoring and credit markets.
Skees, J. R., & Barnett, B. J. (2006). Enhancing micro World Bank Economic Review, 4, 351366.
nance using index-based risk transfer products. Thomas, D., Beegle, K., Frankenberg, E., Sikoki, B.,
Agricultural Finance Review, 66, 235249. Strauss, J., & Teruel, G. (2004). Education in crisis.
Skees, J. R., Barnett, B. J., & Hartell, J. (2005). Journal of Development Economics, 74, 5385.
Innovations in government responses to catastrophic Townsend, R. M. (1994). Risk and insurance in village
risk sharing for agriculture in developing countries. India. Econometrica, 62, 539591.
Paper presented at the workshop Innovations in Townsend, R. M. (1995). Consumption insurance: An
Agricultural Production Risk Management in Cen- evaluation of risk-bearing systems in low-income
tral America: Challenges and Opportunities to Reach economies. Journal of Economic Perspectives, 9,
the Rural Poor, Antigua, Guatemala, May 2005. 83102.
Skees, J. R., Black, J. R., & Barnett, B. J. (1997). Turvey, C. G. (2001). Weather derivatives for specic
Designing and rating an area yield crop insurance event risks in agriculture. Review of Agricultural
contract. American Journal of Agricultural Econom- Economics, 23, 333351.
ics, 79, 430438. Tversky, A., & Kahneman, D. (1973). Availability: A
Skees, J. R., Gober, S., Varangis, P., Lester, R., & heuristic for judging frequency and probability.
Kalavakonda, V. (2001). Developing rainfall-based Cognitive Psychology, 5, 207232.
index insurance in Morocco. The World Bank, Policy Udry, C. (1994). Risk and insurance in a rural
Research Working Paper 2577, April 2001. credit market: An empirical investigation in
Skees, J. R., Hartell, J., & Hao, J. (2006). Weather and northern Nigeria. Review of Economic Studies,
index-based insurance for developing countries: 61, 495526.
Experience and possibilities. In A. Sarris, & D. United Nations Millennium Project (2005). Investing in
Hallam (Eds.), Agricultural commodity markets and development: A practical plan to achieve the millen-
trade: New approaches to analyzing market structure nium development goals. New York: United Nations
and instability. Northampton, MA: Edward Elgar. Development Program.
Skees, J. R., Hartell, J., & Murphy, A. (2007). Using Varangis, P., Skees, J. R., & Barnett, B. J. (2002).
index-based risk transfer products to facilitate micro Weather indexes for developing countries. In R. S.
lending in Peru and Vietnam. American Journal of Dischel (Ed.), Climate risk and the weather market.
Agricultural Economics, 89, 12551261. London: Risk Books.
Skees, J. R., Hazell, P., Miranda, M. (1999). New Vedenov, D. V., & Barnett, B. J. (2004). Eciency of
approaches to crop insurance in developing coun- weather derivatives as primary crop insurance instru-
tries. EPTD Discussion Paper No. 55, International ments. Journal of Agricultural and Resource Econom-
Food Policy Research Institute, Washington DC. ics, 29, 387403.
Skees, J. R., & Reed, M. R. (1986). Rate-making and Wang, H. H., Hanson, S. D., Myers, R. J., & Black, J.
farm-level crop insurance: Implications for adverse R. (1998). The eects of crop yield insurance designs
selection. American Journal of Agricultural Econom- on farmer participation and welfare. American Jour-
ics, 68, 653659. nal of Agricultural Economics, 80, 806820.
Skees, J. R., Varangis, P., Larson, D. F., & Siegel, P. Zimmerman, F. J., & Carter, M. R. (2003). Asset
(2005). Can nancial markets be tapped to help poor smoothing, consumption smoothing and the repro-
people cope with weather risks?. In S. Dercon (Ed.), duction of inequality under risk and subsistence
Insurance against poverty. Oxford: Oxford University constraints. Journal of Development Economics, 71,
Press. 233260.
Available online at www.sciencedirect.com

Poverty Traps and Index-Based Risk Transfer Products

Uploaded by

Copyright:

Available Formats

Poverty Traps and Index-Based Risk Transfer Products

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Poverty Traps and Index-Based Risk Transfer Products

Uploaded by

Copyright:

Available Formats

World Development Vol. 36, No. 10, pp.

Poverty Traps and Index-Based Risk

1. INTRODUCTION This article reviews relevant threads of the

be reduced by carefully choosing the IBRTP members of a smallholder farmer association.

POVERTY TRAPS AND INDEX-BASED RISK TRANSFER PRODUCTS

5. IMPLEMENTATION ISSUES bution of the index may be conditioned on

Index instruments do not require insurers to (f) Subsidies

perverse behavioral incentives since those who 6. CONCLUSION

AGROASEMEX (2006). La Experiencia Mexicana en el Handbook of income distribution. Amsterdam: North-

Available online at www.sciencedirect.com

You might also like