microinsurance and natural disasters: challenges and options
TRANSCRIPT
Microinsurance and natural disasters:Challenges and options
Daniel J. Clarke a,*, Dermot Grenham b
aDepartment of Statistics, University of Oxford, 1 South Parks Road, Oxford OX1 3TG, UKbDepartment of Social Policy, London School of Economics and Political Science, Houghton Street, London WC2A 2AE, UK
e n v i r o n m e n t a l s c i e n c e & p o l i c y 2 7 s ( 2 0 1 3 ) s 8 9 – s 9 8
a r t i c l e i n f o
Article history:
Received 16 December 2011
Received in revised form
15 June 2012
Accepted 18 June 2012
Published on line 20 July 2012
Keywords:
Microinsurance
Index insurance
Indemnity insurance
Migration
Disaster risk financing
Natural disasters
a b s t r a c t
Providing disaster microinsurance to low income individuals is far from easy. Designing and
structuring products so that they can be sold at low cost raises a set of challenges, and even
then the level of voluntary purchase can be disappointingly low. Recent innovations in
providing agricultural insurance could have broad implications for other disaster micro-
insurance products. If a market is to be viable governments will have key roles to play,
including: approving insurance policy small print and product designs; supporting or
leading industry-wide investments in data collection, index design, and loss adjustment
capacity; financing sustained public information campaigns; nudging or compelling pur-
chase; and committing to limits on post-disaster financial assistance to the uninsured. If
disaster microinsurance prices do not accurately reflect the risks faced, the presence of
disaster microinsurance can distort incentives to migrate. However, in the presence of risk-
based pricing, it seems plausible that disaster microinsurance could make it easier for
people to stay where they are even as their environment becomes more fragile, yet increase
post-disaster migration through increasing post-disaster wealth.
# 2012 Elsevier Ltd. All rights reserved.
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/envsci
1. Introduction
Natural disasters can be devastating, particularly when they
affect vulnerable, low income households. Disaster micro-
insurance could in principle substantially increase welfare by
offering people a way to smooth out the worst years. However,
in practice many of the people who need protection the most
are either not offered appropriate products or, if they are, they
do not purchase them.
Whilst low income people are typically exposed to a variety
of shocks, this paper will focus specifically on natural disasters,
providing a conceptual overview of how microinsurance can
and cannot help in mitigating the impact of natural disasters on
low income people in a changing and potentially riskier world.
As disaster microinsurance is already in place in a number of
* Corresponding author. Tel.: +44 1865 272867.E-mail addresses: [email protected] (D.J. Clarke), d.j.grenham@
1462-9011/$ – see front matter # 2012 Elsevier Ltd. All rights reservedhttp://dx.doi.org/10.1016/j.envsci.2012.06.005
countries as a means of protecting low income people from the
adverse effects of disasters, it is natural to ask whether it has the
potential to provide protection to an increased number of low
income people if climate change increases the population
exposed to the adverse effects of natural disasters. We
introduce issues in the supply and demand of disaster
microinsurance, present evidence for broader linkages between
disaster microinsurance and migration decisions, and offer a
critical analysis of policy options that could be used to stimulate
a market for disaster microinsurance in developing countries.
We restrict attention to non-life microinsurance products
which provide, for example, agricultural or building insurance
for a short, typically annual, period. Such short-term disaster
microinsurance offers protection against unforeseen disasters
but does not offer protection against long term changes to the
underlying risks faced.
lse.ac.uk (D. Grenham).
.
e n v i r o n m e n t a l s c i e n c e & p o l i c y 2 7 s ( 2 0 1 3 ) s 8 9 – s 9 8S90
The compelling rationale for disaster microinsurance as a
mechanism for reducing vulnerability, combined with inno-
vations in the design and delivery of products, has led to a
plethora of pilots across developing countries (Hess et al.,
2005). However, many of these pilots have not matured into
sustainable scaled-up programs, and the effects of those that
have scaled up on migration or climate adaption decisions are
not well understood. Cummins and Mahul (2009) offers an
extensive review of disaster insurance programs currently in
place across developing countries, including some disaster
microinsurance programs, but does not consider in any degree
of detail the effects of climate change on disaster micro-
insurance or the potential impact of disaster microinsurance
on migration decisions.
In this paper we will first describe what microinsurance
looks like, the various types of product available, and how
microinsurance could protect against the effect of cata-
strophes. In particular we look at recent innovations in
markets for agricultural insurance, and discuss how these
could be applied more generally to natural disaster insurance.
We will then consider supply and demand issues before
looking at how climate change may affect the risks that
insurance typically protects against and how microinsurance
could respond to these changes. We address links between
migration and microinsurance, and the special case of Asian
megacities before concluding with various options for future
developments. We leave others to discuss the actual extent to
which climate change is causing and will cause more and
worse natural disasters and the direct effects of climate
change on migration (see Overseas Development Institute,
2006 and Coleman, 2009).
2. Issues in the provision of microinsurance
2.1. Defining microinsurance
There are a variety of different definitions of microinsurance.
Churchill (2006) defines microinsurance as ‘the protection of
low-income people against specific perils in exchange for
regular premium payments proportionate to the likelihood
and cost of the risk involved.’ A peril in this context refers to
the type of event (such as health problems, death, and
property damage) that could have a financial impact on a
person. Under this commonly used definition, microinsurance
is simply insurance for low-income people. The Indian
Regulatory and Development Authority instead focuses on
the size of the premium and benefits provided when defining
microinsurance in its microinsurance regulations (IRDA, 2005,
paragraph 2(d)) as ‘‘any health insurance contract, any
contract covering the belongings, such as hut, livestock or
tools or instruments or any personal accident contract, either
on individual or group basis’’ with premium and benefits
within specific bounds. Yet other definitions understand the
‘‘micro’’ in microinsurance as applying to the locus of decision
making and delivery, for example, through local self-help
groups for social insurance which are not part of a national
health scheme (Dror and Jacquier, 1999).
While for many purposes the actual definition of micro-
insurance may not matter, it is important for regulatory
authorities who wish to regulate microinsurance products in a
different way to other insurance products. For this they will
need to be able to differentiate objectively between micro-
insurance and other insurance products.
Irrespective of the actual definition, since the target
consumers of microinsurance products are quite different to
the middle- and high-income people typically targeted by
developed country non-life insurers, one might not be
surprised to learn that successful microinsurance products
are quite different in many respects to insurance products sold
in the developed world.
Many of the perils covered by microinsurers, including life,
health and home fire, are insured in developed countries.
However, microinsurance products target perils that specifi-
cally affect low-income households. For example, with 900
million rural poor dependent on the uncertain business of
agriculture for their livelihoods, agricultural insurance is a
much more important line of business for microinsurers than
for traditional insurers.
2.2. Supply of microinsurance
The traditional risk carriers for microinsurance products are
insurance companies, reinsurance companies including the
large multinational reinsurers, and governments. NGOs and
some mutual organisations have in some places provided
forms of cover but these have been acting as informal insurers
with the risk that, as they are unregulated, they may not be
able to pay contractually agreed claims in the aftermath of a
disaster. Regulators are increasingly bringing in regulations
specifically for the microinsurance market in their countries,
as distinct from mainstream insurance regulation. For exam-
ple, India introduced regulations in 2005 (IRDA, 2005), the
Philippines in 2010 (PIC, 2010), and the International Associa-
tion of Insurance Supervisors is currently drafting revised
guidance on informal insurance.
Only if microinsurance products are profitable will it be
attractive for insurance companies to offer them in significant
volumes. To be profitable the premiums charged, absent any
subsidy, must cover the cost to the company of covering the
insured risk, its marketing and administration costs and the
cost of holding capital.
Increasing the reach of microinsurance will require
insurers to have suitable distribution and administration
capacity provided by themselves or in collaboration with
others (Churchill, 2006, Part 4). By definition, microinsurance
will be for relatively small premiums but potentially for large
volumes of policyholders. Insurers will therefore need to
develop cost effective means of reaching their customer base,
for example by partnering with community-based mutuals,
microfinance organisations, or other intermediaries and by
making use of developments in technology such as the
increasing reach of mobile phones for financial transactions.
Administering large volumes of small policies will, in order to
be profitable, demand low overheads and highly efficient
administration systems. In some cases the development and
implementation of these systems may need to be subsidised.
Insurers could try and make use of existing infrastructure
to distribute and administer their products thus reducing
the cost involved in introducing microinsurance products.
e n v i r o n m e n t a l s c i e n c e & p o l i c y 2 7 s ( 2 0 1 3 ) s 8 9 – s 9 8 S91
However, they will need to be satisfied that they are partnering
with organisations that are trustworthy and financially
secure.
There will also need to be a high level of trust in financial
institutions on the part of the potential purchasers of
insurance. On the one hand, purchasers need to be satisfied
that the insurer will be able to pay their claims without going
insolvent. However, often more importantly, they will need to
be confident that the insurer will not reject a valid claim for
spurious or legalistic reasons. The solvency of insurers can
largely be ensured by having appropriate regulations in place.
Ensuring insurers pay valid claims is more difficult to
guarantee although regulators have an important role to play.
Although microinsurance products may be small in terms
of premium or benefits sometimes they need to be relatively
sophisticated. Local insurers may not therefore have the
expertise to be able to price, administer and manage claims for
such sophisticated products. Immature insurance markets
usually start with car and employer liability and group life
insurance and may focus on urban areas. To sell agricultural
insurance in a rural area or disaster insurance anywhere could
be very challenging for inexperienced companies without a
significant amount of support from others, for example
reinsurers, international insurance companies, and other
organisations which have at least some expertise in these
areas.
Currently most microinsurance policies, including the
small number of existing disaster microinsurance policies,
are annual products where the premium is set at the start of
each policy year. Whilst this may be appropriate at present,
improvements in early warning systems such as those
described in Lenton (2013) could lead to adverse selection
on the part of the insurer, with insurance provided at
affordable premiums only in those years in which claims
are considered to be unlikely. Clarke et al. (2012) report on the
publicly funded weather index insurance program in India in
which private sector insurers, after competing to offer
subsidised contracts in a state, will choose the length of the
sales window depending on up to date forecasts for the
coming season, with shorter sales windows if payouts are
expected and longer sales windows if payouts are not
expected. Farmers and government would together pay prices
based on historical data for a series of years, only for cover to
be effectively withdrawn in years predicted to be bad. Without
proper regulation or the initiation of multi-year contracts, this
form of adverse selection on the part of the insurer could lead
to healthy profits, but offer little protection to policyholders.
Moreover, this type of product may not be sustainable over the
long term as farmers realise that the product is of relatively
little value to them.
2.3. Indemnity and indexed microinsurance products
A recent focus on agricultural insurance, particularly by
development economists, has stimulated innovations of some
significance for the current discussion on protection against
natural disasters. In particular, microinsurers have experi-
mented with offering indexed insurance products to their clients
rather than or in addition to indemnity-based products (Skees
et al., 1999).
Under a traditional indemnity insurance product the claim
payment to the insured depends only on their own loss; if the
insured incurs a large enough insured loss the insurer is
contractually obligated to make a claim payment. It may be
possible to offer indemnity-based insurance for some climat-
ic-related disasters. For example Turkey’s system of earth-
quake insurance for homes and the Philippines’ new scheme
for typhoon insurance for homes are both indemnity-based
(Gurenko et al., 2006; Morsink et al., 2011). However,
indemnity-based approaches to insurance do not seem
possible for some important perils.
Using the example of agricultural insurance, an indemnity-
based insurance policy such as a multiple peril crop insurance
(MPCI) policy would lead to a claim payment to a farmer if that
farmer’s yield for specified crops on specified plots was below
a contractually agreed threshold. Such individual indemnity-
based policies have been sold in many countries but are
plagued by a number of problems. There is a risk of moral
hazard, where insured farmers might take less care over their
farm than uninsured farmers on the expectation that any
reduced yield would be covered by an insurance claim. This
risk could be reduced, although probably not eliminated, by
imposing an excess such that the insured will only be covered
over a certain level of loss. The insurer might also be exposed
to the risk of adverse selection by its potential customers
whereby farmers who consider that they are more likely to
make a claim purchase insurance. This could lead to higher
than expected claims and higher future premiums which
would then further discourage lower risk farmers to take out
insurance. This risk could be mitigated by making the
insurance compulsory, for example, making it a requirement
for farmers taking out microfinance loans to take out a
microinsurance policy to cover the loan. A further potential
problem with indemnity insurance is that the high cost of
detecting and stopping fraudulent behaviour could mean that
insurers prefer to pay all claims and allow the fraudulent
claims to increase premiums. All of these problems with
indemnity insurance are a consequence of information
asymmetries, i.e. one or other party having access to
information not available to the other party. These problems
and the high contracting costs have been so acute in practice
that there have been very few, if any, successful MPCI schemes
for small-scale farmers (Hazell, 1992; Skees et al., 1999).
One exception is the Fondos system in Mexico. Fondos are
organizations of farmers who join together on a non-profit
basis to mutually insure each other against specific named
risks through risk pooling and reinsurance. The Fondos’
insurance operations are managed on normal insurance
principles including proper underwriting of risks, the use of
technical guideline for claims assessment and the use of
actuarial methodologies to calculate premiums. The Fondos
contains a number of automatic moral hazard limiting
controls so that individual farmers do not try and take
advantage of other farmers, and the Fondos as a whole does
not try and take advantage of the reinsurer (Ibarra and Mahul,
2004).
The controls put in place by the Fondos include limiting the
size of their membership so that the farmers have a high level
of knowledge of each other and would know if someone was
trying to gain an unfair advantage. The reinsurer has some
e n v i r o n m e n t a l s c i e n c e & p o l i c y 2 7 s ( 2 0 1 3 ) s 8 9 – s 9 8S92
protection as it can quickly cancel the reinsurance arrange-
ment and the Social Fund mechanism encourages good
behaviour among the farmers. Moreover, the joint reporting
mechanism whereby the Fondos jointly reports all farmer
claims to the reinsurer rather than each farmer reporting
individually relaxes the incentive compatibility constraint,
allowing a reinsurer to audit just a sample of farmers, only
worrying about fraud at the level of the Fondos rather than
at the level of the individual. However, it does not get away
from the other major problem with indemnity based
insurance schemes namely the potentially high cost of
assessing claims.
In contrast to an indemnity insurance policy, under an
indexed insurance product claim payments to the insured
depends only on the realisation of some index. The level of the
index is typically chosen to be objective and correlated with
the insured’s loss. Rainfall indexed insurance is one such
example of an index insurance policy that has been sold to
farmers. Under this type of product, the index is a measure of
the rainfall at a specified contractual weather station. Too
much or too little rainfall at the wrong time might spell
disaster for a farmer and so it is possible to design an indexed
insurance policy which would make claim payments in case
the rainfall index is too high, too low, or either. Since the
product depends only on the amount of rain at the contractual
rainfall station, the same product can be sold to a large
number of farmers working near the contractual weather
station, who will then all receive the same claim payment at
the same time (Gine et al., 2010).
A wide variety of indices can be used and they can be
tailored to specific crops, or other risks that are being insured.
Certain crops are sensitive to hours of sunlight, temperature or
humidity and indices have been developed based on these
parameters (Clarke et al., 2012). Rather than setting an index
based on hazards such as rainfall and sunshine, which act as
inputs to the agricultural process, a special category of indices
may be described as sample output-based indices which are
calculated as the mean from a statistical sample of outputs,
such as the yield on crops or the mortality rate of livestock. For
example, a sample output-based area yield index is the sample
mean yield for a particular crop in a defined geographic area,
where the sample of plots in which the yield assessment is
conducted is chosen to be representative of the area (Mahul
et al., 2012).
By conditioning claim payments on an objective, cheaply
observable index, over which policyholders have little control
or private information, indexed insurance can significantly
reduce the problems of moral hazard, adverse selection and
insurance fraud which can plague indemnity insurance
contracts. However, indexed insurance has one significant
disadvantage; it is only a hedge, and does not offer watertight
protection. Of particular concern, it increases exposure to
downside risk. For example, without any agricultural insur-
ance the worst agricultural outcome that can occur from the
perspective of a farmer is one in which the farmer loses his
entire crop. However, a farmer who has purchased weather
indexed insurance is exposed to the risk of an even worse
outcome: she could pay her insurance premium and then lose
her entire crop due to pestilence, disease or localised weather
conditions, yet receive no claim payment due to the weather
index not having triggered, leaving her worse off by the
amount of the insurance premium (Clarke, 2011).
Attractive indexed insurance policies offer low premiums,
relative to the level of cover, and claim payments that are
highly correlated with the losses of target policyholders
(American Academy of Actuaries, 1999). In particular, a good
index should trigger claim payments when the insured has
incurred a large loss on the insured asset. For example,
conditional on a home being destroyed a good indexed home
insurance policy will have a high probability of making a
material payout, and conditional on a crop being destroyed,
leading to very low yield, a good indexed crop insurance policy
will have a high probability of making a material payout.
In designing an indexed insurance product care has to be
taken over the choice of index. To avoid disputes over when a
claim payment will be made the index has to be well defined
and measured independently of both the insured and insurer.
Even for seemingly manipulation-free automatic weather
stations care must be taken to prevent fraud. Clarke et al.
(2012) reports on cases of farmers placing ice cubes on weather
stations in India to trigger large claim payments for tempera-
ture indexed frost cover. The trigger point of the index needs to
be clear and payouts correlated with the losses which would
be incurred by farmers in the event of the index reaching the
trigger point. To enable the insurer to be able to set actuarially
fair premiums there ideally needs to be sufficient historical
data available for the index in the geographical locations
covered by the indexed insurance product. What constitutes
‘sufficient’ will differ by region and product. When developing
a new product in a new area there will be a lack of relevant
data and use will need to be made of whatever data is
available, such as from similar products developed in similar
areas. Insurers will therefore be taking a greater risk in the
early years that their premiums are incorrect and may
therefore tend to err on the side of setting high premiums
to avoid making a loss, which might make the product
financially unattractive. This could be an area where govern-
ments or donors could consider subsidising premiums so that
policies can be sold and data gathered. They can also invest in
the infrastructure required to collect relevant data.
Despite the growth in indexed insurance policies in
developed countries there has been very little analysis of
whether these products are objectively good for target clients.
Although such analysis has been conducted in developed
countries for indexed insurance policies and other derivatives
purchased by firms (for example see Cummins et al., 2004),
derivatives are typically not sold to individuals in developed
countries. Moreover, the available evidence for hazard-based
indexed insurance policies sold in developing countries is
somewhat negative. For example, Clarke (2011) finds that a
portfolio of weather indexed insurance policies sold to low
income farmers in 2007 comprised mostly of objectively poor
products, due to the combination of high premiums and low
correlation between claims and losses. Banerjee et al. (2011)
finds a correlation of 1% between wind speed indexed
insurance policies and losses incurred by rice farmers in the
Philippines. Clarke (2011) finds that weather indexed insur-
ance policies sold across one state in India do not pay anything
in one third of catastrophic years and the average correlation
between indices and losses over all products is only 13%.
e n v i r o n m e n t a l s c i e n c e & p o l i c y 2 7 s ( 2 0 1 3 ) s 8 9 – s 9 8 S93
Whilst this research is not yet conclusive, these findings are
concerning. There is little evidence on sample output-based
indices in developing countries, but evidence from developed
countries is much more positive. For example, Deng et al.
(2007) finds that sample-based area yield insurance may be
preferable to indemnity insurance, even for large farms in the
US.
One approach that may be appropriate for some perils is to
use indexed insurance contracts as a way of transferring risk
from community groups to formal insurers or, particularly for
the case of disaster risk, from local insurers to international
reinsurers. For the case of agricultural insurance, the pooling of
local agricultural risk through a community indemnity-based
mutual insurance group can provide protection to individual
farmers against individual shocks, whilst formal sector indexed
insurance can provide protection to the mutual against
aggregate shocks. Such a scheme could lead to reliable
protection at an acceptable cost (Dercon et al., 2011). Such an
economic arrangement could also be optimal for other
catastrophic perils. For example, it can take years for insurers
to agree to and pay indemnity claim payments to large number
of homeowners in the aftermath of a large disaster in a low or
middle income country. However, loss adjustment, that is ex
post claims processing including auditing and payment, is
typically much faster and cheaper for indexed products than for
indemnity products. If well designed the transfer from a
reinsurer to a local insurer in the aftermath of a large disaster
could be quite close to the amount that would be payable under
an indemnity-based reinsurance contract, but with reinsurance
claims paid quicker and at a lower cost. One complex technical
challenge is defining indices that accurately capture losses, but
as we propose in Section 3 sample output-based indices could
be a useful tool for the index designer.
2.4. Demand for microinsurance
There is a clear need for insurance for the poor. Having a low
income in a poor country is risky. Self-employment provides
unpredictable income, and the constant threat of health or
mortality shocks leave households vulnerable to serious
hardship (Dercon, 2004; Collins et al., 2009). Although the
financial size of shocks may be small compared to those in
developed countries, there is nothing ‘micro’ about the shocks
faced by the poor in poor countries; it is not unheard of for
households to respond to large income shocks by taking
children out of school and reducing nutritional intake,
particularly for girls and women (Ferreira and Schady, 2009;
Dercon and Krishnan, 2000).
If people wish to reduce the risks they face they can try and
transfer all or part of the risk or take steps to reduce the impact
of these risks. Farmers have traditionally had a number of
informal ways of reducing and transferring risk such as
diversifying the crops they grow, having more than one source
of income, sending family members to work in cities, or taking
part in some form of community based risk pooling process.
Insurance is a form of risk transfer that can supplement or
replace some of these more traditional methods especially
where they cannot provide sufficient protection against those
risks which are not diversifiable at the famer and community
level.
By affecting entire communities and extended families at
the same time, natural disasters have an even larger impact
than idiosyncratic shocks which affect one individual at a
time. Those affected are typically left with little in the way of
assets, and surviving the aftermath often requires selling any
remaining assets, such as jewellery, at fire sale prices. Friends
and families, whilst useful for support in the event of
idiosyncratic shocks, are all likely to be affected at the same
time and cannot offer substantial help. By providing income
from an external source in the event of natural disasters,
insurance can lessen their impact on the vulnerable.
However, need is not the same as demand. If one
extrapolates from decisions made about low impact, high
probability events, people should be willing to pay a large
premium for disaster insurance. However, even people in rich
countries with high levels of financial education consistently
underinsure against high impact, low probability events
(Kunreuther and Michel-Kerjan, 2008). One reason is that
personal experience is normally of little value for making such
decisions; once-in-a-lifetime events typically happen only
once in a lifetime.
For microinsurers, faced with low levels of financial literacy
within their target market, the problem is harder still (Gaurav
et al., 2010). Trust is likely to significantly suppress demand,
since an insurer alone may find it difficult to convince clients
that they will be able to make claim payments in the aftermath
of a large disaster. Even in the absence of trust issues, a
responsible microinsurer must provide information and
training programs, clearly defined products with simple rules
and restrictions, and easily accessible claims documentation
requirements, all without overselling the benefits of insur-
ance. Indexed insurance may be particularly difficult to sell,
since clients may be rightly worried about incurring large
losses but receiving no claim payment due to an inappropri-
ately designed index.
Another reason why effective demand for microinsurance
may be less than expected is that potential purchasers simply
cannot afford to pay the premiums in one go due to a lack of
cashflow or credit. In these cases allowing payments to be
made in instalments or at times of the year when potential
purchasers are not credit constrained may increase actual
demand. One such potential solution is the approach taken by
the Horn of Africa Risk Transfer for Adaptation project under
which famers can pay for the insurance either in cash or by
providing labour for climate change risk mitigation projects
(Oxfam, 2011). This has the dual benefit of providing insurance
and at the same time of reducing the risk of climate change
related disasters.
Another method that can be used to increase effective
demand is to make the purchase of a microinsurance product
compulsory. This could be as a result of national or state
government regulations, in much the same way as third party
motor insurance is a legal requirement for car owners in most
countries. Ensuring compliance with any such regulations will
be costly for an insurer and will lead to an increase in premium
rates as these should be set at a level to cover the insurer’s
expenses. An alternative method of making microinsurance
compulsory is to bundle it up with the purchase of other
products for example including it along with a microfinance
loan or when a farmer buys inputs for his farm. However, this
e n v i r o n m e n t a l s c i e n c e & p o l i c y 2 7 s ( 2 0 1 3 ) s 8 9 – s 9 8S94
approach is not able to guarantee that the microinsurance
product meets the actual needs of the farmer, for example he
may already be insured through some other means and so
could end up being over insured. On the other hand if the level
of disclosure is insufficient the farmer may not realise that he
is insured.
2.5. Insurance and climate change
Microinsurance will only be able to protect against the impact
of climate change if products covering the effects of natural
disasters are available at affordable prices. However, to the
extent that future climate change affects the overall level of
risk and volatility of a catastrophe, microinsurance providers
will respond by increasing premiums or deductibles, broad-
ening exclusions or withdrawing cover. If a microinsurance
provider did not do at least one of these and had insufficient
reinsurance it would be exposing itself to an increased risk of
insolvency and an inability to meet its claims.
Climate change could have a number of different effects on
the risks covered by microinsurance, including shifting the
average outcome and/or changing the frequency or severity of
extreme events. For example, Gosling (2011) reviews evidence
that suggests that tropical cyclone frequency could decrease
with climate change, whilst intensity (e.g. wind speed and per
event total precipitation) could increase. As well as how
climate changes affect the distribution of risk, the speed of the
change will also affect how responsive microinsurers would
be to the change in risk. Climate change could also affect the
demand for insurance, particularly if the beliefs of potential
clients change, although the beliefs of insurers might adapt
more quickly to a changing environment than the beliefs of
individuals.
Insurance premiums are broadly made up of a ‘risk
premium’ to cover the cost of the expected claims plus
loadings for the cost of holding capital, administrative and
marketing expenses, and profit:
Commercial Premium ¼ Risk Premium þ Cost of Capital
þ Expenses þ Expected Profit (1)
The risk premium is generally set using statistical analysis
of past experience, incorporating expert opinion and allowing
for noticeable trends in past experience. However, it will take
some time, or greater reliance on expert opinion, for
microinsurance providers to be able to determine how
changes in the impact of disasters relate to changes in the
underlying frequency or severity with which these disasters
occur. Insurers may be aware that the risks are changing but
not be able to determine by how much and whether the
change is in the frequency of disasters or in the severity, or
both. They may also be faced with different experience in
different areas that will make it even more difficult to assess
how widespread the effects are. This lack of knowledge will
tend to lead them to take a prudent approach in setting
premium rates or to being more restrictive in what they offer.
The cost of holding capital arises as a result of insurers and
reinsurers needing to put aside some capital as a buffer
against worse than expected claims experience. While
insurers may have their own view of how much capital to
hold, regulators set minimum standards. For example,
Solvency II in the EU will require insurers to demonstrate
resilience against a ‘1 in 200 year’ event. This can lead to high
capital requirements for insurance against events with
uncertain odds, such as climatic-related disasters. The cost
of holding this capital in the form of suitable assets is rightly
passed on to policyholders and can be substantial.
Adding loadings for expenses, capital costs and profit can
make the premium charged significantly higher than the risk
premium. For example, it is common for the premium for
catastrophic cover against natural disasters to be more than
four times the risk premium (Perry, 2009), a significant
constraint on the ability of providers to offer value for money
for disaster microinsurance.
One of the possible mechanisms by which climate change
could affect farmers is by causing a slow but continual
deterioration in their environment, for example, a slow
increase or decrease in the amount of rainfall. Such a gradual
change may not be severe enough, in the short term, to trigger
a claim under an insurance policy. By the time the environ-
ment has deteriorated to a level where the farmers could make
a claim, the insurance companies, having already identified
the trend may have changed the trigger point. If the
deterioration continues then even if there is a claim under
an insurance policy, farmers may not be able to remain where
they are as the conditions may be such that they can no longer
adapt and produce sufficient food to support themselves and
their families. Their situation may have come to such a pass
that even with the insurance payout (all of which may be
needed to repay outstanding loans) they may have insufficient
assets to enable them to invest in alternative livelihoods or
even to migrate except as paupers. Whilst longer term
insurance contracts could protect against longer term
‘changes in the odds’, the capital requirements for such
products are likely to be even more significant than for annual
disaster microinsurance products, rendering them even more
expensive.
2.6. Microinsurance and migration
Migration, whether from a rural area to an urban centre or
from one country to another can be a response to changes in
perceived risk or part of a coping strategy in the aftermath of
catastrophic disasters. It is therefore pertinent to ask whether
microinsurance could have a role to play in increasing or
reducing migration before or after disasters.
Migration is relatively easy to define but much harder to
measure. At one extreme a migrant is simply someone who
moves. This could be within a country, typically from rural
areas to urban areas especially in developing countries, or
cross borders. Trying to count migrants is difficult because of a
lack of reliable data on who is moving where, even in countries
where migration is a politically sensitive topic and where one
would therefore expect efforts to be made in trying to monitor
the levels of migration. Many countries lack administrative
registration systems and do not carry out regular censuses
which would give them at least some indication of previous
levels of migration. Even figures on legal cross border
migration, which perhaps ought to be the most reliable, can
be lacking, for example where there is free cross border
e n v i r o n m e n t a l s c i e n c e & p o l i c y 2 7 s ( 2 0 1 3 ) s 8 9 – s 9 8 S95
movement such as within the EU. Trying to quantify the
degree of illegal migration is even more of a challenge.
There are many different types of migrant, from those
moving to another place to study, in their own country or
abroad, after which they intend returning home, to those who
are forced to leave their usual place of residence as a result of
conflict or natural disaster.
There are many theories explaining why and how people
migrate and the effect they have on home and host societies.
Reasons for migrating are often divided into push and pull
factors. The former include poverty (actual or foreseen),
population pressure, lack of services, insecurity, and discrim-
ination. The latter include seeking the ‘bright lights’ of the city,
potential for improved standard of living, freedom, family
reunification, return migration as a result of nostalgia or
homesickness. Migrants often move following well-trodden
routes, using networks that have been established by earlier
pioneers. This would be the case whether migration is well
planned or is a response to a disaster.
As microinsurance products have only been developed
over the past couple of decades and have as yet only achieved
low levels of market penetration in developing countries,
there is a lack of empirical analysis carried out on the direct
effect of the availability of microinsurance on migration
decisions.
If disaster microinsurance premiums do not accurately
reflect the actual risks faced then the availability of disaster
microinsurance could distort decisions to migrate. For example,
government caps on insurance premiums may provide incen-
tives for people to move to, and stay living in, high risk areas.
In the presence of risk-based pricing, it seems plausible
that disaster microinsurance would make it easier for people
to stay where they are even as their environment becomes
more fragile. This is because affordable insurance makes
outcomes more predictable, effectively reducing the welfare
cost of risk and making staying in a risky environment less
unattractive. However, if the effects of climate change on
livelihoods become so severe such that people cannot adapt
and sustain themselves and their families then no amount of
short-term insurance protection will enable them to remain
where they are and they are likely to migrate elsewhere.
Whether people in this situation will migrate within or across
borders will depend on many factors such as immigration
policies in potential host countries and the ability of the
affected people to afford the high cost of international or
intercontinental migration.
There is a body of evidence that suggests that migration is
costly, and positively associated with wealth and social capital
(Skeldon, 1997). This could imply that if microinsurance claims
are settled promptly, an individual would have more wealth
post-disaster if they had previously purchased disaster micro-
insurance than if they had not and so, at the margin, the
purchase of microinsurance could actually increase post-
disaster migration through an increase in post-disaster wealth.
Finally, the effect of macro or meso-level insurance for
governments and businesses could reduce post-disaster
migration. This is because individuals affected by a disaster
are probably more likely to stay if governments are able to
rebuild key infrastructure and continue to offer basic services
and businesses can continue to offer employment. This is
more likely where these governments and business had taken
out disaster insurance.
The continual increase in urbanisation is leading to the
growth of a number of extremely large ‘‘megacities’’, most of the
largest of which are in Asia. In this new context it is again a valid
question to ask whether there is a role for microinsurance in
such places, given the many challenges that these cities have to
provide services to their inhabitants. These ‘‘megacities’’ have
different exposures to natural disasters (for examples, earth-
quakes, storms, tsunamis) and to the impact of climate change.
They will also have different adaptive capacities. What they do
share is the large number of people at risk.
To the extent that climate change increases the frequency
or severity of storms, hurricanes and tornadoes which hit
urban areas, megacities, will be at increased risk of natural
disasters happening. For example, those megacities which are
coastal will potentially be at an increased risk from flooding
due to increased average sea levels. Of China’s estimated
urban population of 400 million, an estimated 130 million live
in coastal cities that are vulnerable to sea-level rise (Prasad
et al., 2009).
These large cities have the potential to have their population
increased to the extent that climate change leads to migration
from rural areas. While this migration is not itself an insurable
event, the influx of large numbers of migrants could have a
serious negative impact on the cities’ infrastructure which may
in turn increase the incidence of insurable events. For example,
if the cities’ sewage system is overwhelmed by the influx of
migrants this could increase the incidence of disease.
Some of the risks described above, which could become
more pronounced as a result of climate change, are potentially
insurable using traditional or microinsurance products.
However, whoever is providing the insurance would need to
ensure that they are not overly exposed to the potentially large
numbers of claims from a single event. This can be achieved by
rationing, limiting the amount of insurance they provide in a
single city, by selling insurance across diverse geographical or
climatic regions or by accessing, through reinsurance, the
greater level of diversification available to international
reinsurers. Alternatively, governments or supranational orga-
nisations could act effectively as a reinsurer of last resort. But
even so, calculating premium rates for products that will cover
the impact of climate change related disasters will be
challenging due to a lack of data and appropriate models.
Innovative marketing and distribution methods as well as
premium collection and claim payment systems would be
required to reach low income households who will often be
living in the slum or areas of the city. As with crop insurance,
indexed products could be designed which would reduce the
cost of claim assessment, such as the property indexed
insurance product currently being sold in Jakarta which pays
claims when the water level rises above a certain fixed point
on the Manggarai Flood Gate in the city.
3. Policy options
Parties interested in providing or supporting the provision of
insurance for climatic-related disasters would do well to bear
the following points in mind.
e n v i r o n m e n t a l s c i e n c e & p o l i c y 2 7 s ( 2 0 1 3 ) s 8 9 – s 9 8S96
First, it seems unlikely that any disaster insurance scheme
would be successful without significant government involve-
ment. As already mentioned, people seem to consistently
underinsure against low frequency, high impact events. At
one extreme, governments have the power to compel disaster
insurance purchase, as is the case for farmers who take out
loans for agricultural purposes in India (Mahul et al., 2012). As
a softer alternative, governments could nudge individuals
towards purchasing disaster insurance by, for example,
requiring that standard building insurance policies or stan-
dard loans for agricultural investments included insurance
cover for disaster, even if opt out was possible if explicitly
chosen by the client. Nudging or weak compulsion is common
for homeowners earthquake insurance; in California every
insurance broker selling homeowners insurance is compelled
to offer earthquake insurance through the publicly managed
California Earthquake Authority (CEA), and in Turkey home-
owners are compelled to purchase earthquake cover through
the Turkish Catastrophic Insurance Pool (TCIP), albeit with
limits to enforcement. Governments can also nudge through
publically funded promotion of disaster insurance purchase,
as exemplified by the high profile long-term marketing
campaign of Turkey’s TCIP (Gurenko et al., 2006). Such
government intervention might be justified on public goods
grounds, since basic education about the value of insurance
against disasters is likely to be undersupplied by private sector
organisations due to concerns about free riding from compe-
titors.
Governments should also have a technical role in approv-
ing disaster insurance products based on international best
practice, to protect policyholders against inappropriate small
print and could require standard information disclosure. In
many developed countries lenders must publicise the Annual
Percentage Rate (APR) of their products to allow a simple, if
somewhat crude, comparison between products; govern-
ments could require insurers to provide objective estimates
of the insurance multiple, the ratio of the premium to the
expected claim payment, and a reasonable measure of the
actuarial cost of the product, or the historic claim payments
that would have been made from a given indexed product.
Insurance indices are a public good in the economic sense,
requiring a coordinated long term investment. Governments
are likely to have an important role to play in the collection
and validation of long term data series to form the basis for
trustworthy indices, such as the weather or yield data
underlying weather or area-yield indices.
Second, thinking more broadly, if insurance cover for
climatic-related disasters is to be available at reasonable
prices, there may also be a role for an appropriately designed
supranational body in offering reinsurance for specific
climatic-related perils. As described above, sensible insurance
regulation naturally leads to high capital costs, and therefore
high premiums, for insurance against catastrophic events
with uncertain odds. The way in which the private sector
packages this risk, for example through cat(astrophe) bonds or
other risk transfer instruments, does not affect this capital
cost very much: behind all the financial wizardry regulators
require someone to hold substantial assets that could be sold
quickly to pay claims as they fall due, and this liquidity is
costly. To substantially reduce capital costs for catastrophic
risks with uncertain odds, one needs access to liquidity
without having to hold large amounts of liquid assets.
Creditworthy governments, or groups of governments, could
provide liquidity without having to hold large amounts of
liquid assets, since they can offer lines of credit secured
against future tax revenues, and could therefore facilitate
disaster insurance at a much lower cost than sensibly
regulated private insurers or reinsurers.
There are many practical challenges to creating a ‘World
Re’, or expanding the remit of existing supranational institu-
tions to offer reinsurance for specified climatic-related perils,
but such a facility may need be investigated if climate change
increases uncertainty for economically important perils. Note
that the present suggestion is fundamentally different to a
‘‘Climate Change Insurance Mechanism’’ as proposed at the
2010 United Nations Climate Change Conference (COP 16), for
which governments would provide reserves, not guarantees
(see Adamson and Sagar, 2002).
Third, whilst indexed approaches to disaster insurance offer
speedy claim payment and (relatively) low premiums, designers
should take the risk of mismatch between losses and claim
payments seriously. One option would be to move from hazard
based indices, such as weather indices, to sample output-based
indices. For example, in the aftermath of an earthquake, before
and after satellite images could be compared to classify damage
to buildings in a predetermined objective manner. Loss
adjusters, hired by insurance providers and reinsurers but
audited by government, could then assess the damage incurred
by a statistical sample of buildings, with the sample weighted
towards buildings classified as severely damaged by satellite
images. The reinsurance claim payments to local insurers
would then be determined based only on the estimate of the
average building damage arising from this sampling procedure.
Local insurers would be responsible for paying indemnity-
based claims to insured individuals, with their exposure well
hedged by a sample output-based index insurance product
purchased from reinsurers. Such approaches are already being
implemented for agricultural insurance, where crop cutting
experiments for India’s area yield indexed insurance scheme,
the modified National Agricultural Insurance Scheme, are being
targeted on a pilot basis based on remote sensing data (Mahul
et al., 2012). Whilst, to the authors’ knowledge, it has not yet
been attempted for other natural disaster perils, the sample
output-based index approach could be used for catastrophic
hurricane, windstorm or flooding insurance to allow inexpen-
sive, timely risk transfer from local insurers to international
reinsurers with claim payments highly correlated to losses.
Such schemes would require coordination and long term
investments in high quality indices and teams of loss adjusters,
but could facilitate the inexpensive transfer of catastrophic risk
from local insurers, allowing them to offer catastrophic cover to
their clients.
Finally, insurance for the poor does not necessarily require
microinsurance products to be delivered directly to the poor.
Regardless of how personal assets are affected, in the
aftermath of a large natural disaster individuals are much
better off if governments and firms are able to continue to
offer services and jobs. For example, tax revenues typically
fall significantly immediately following a large natural
disaster, and can leave governments unable to pay key
e n v i r o n m e n t a l s c i e n c e & p o l i c y 2 7 s ( 2 0 1 3 ) s 8 9 – s 9 8 S97
public sector employees offering critical services. Reinsur-
ance or a line of contingent credit could protect govern-
ments against such a short term liquidity crunch and
provide financing for the rebuilding of public assets, thereby
reducing the effect of the disaster on those affected
(Cummins and Mahul, 2009). By insuring assets or purchas-
ing business interruption insurance, firms can also offer
some degree of protection to employees and customers
from the financial repercussions of natural disasters. Low
income people therefore have much to gain not only from
personal microinsurance products but also from improved
risk management for developing country governments and
firms (Cummins and Mahul, 2009).
4. Conclusions
As the effect of natural disasters may be disproportionately
felt by people on low incomes it is a reasonable task to assess
the extent to which disaster microinsurance, that is insurance
against natural disasters for people on low incomes, could be
used to lessen the negative effects of disasters. In theory,
disaster microinsurance could play a useful role as part of the
risk management strategies of the poor. However, for its
potential to be realised, it will need support and coordination
by government and the private sector.
Just as insurance everywhere in the world has to be bought
rather than sold, demand for disaster microinsurance is likely
to be low due to a lack of financial understanding, low trust
that insurance companies will pay claims in the aftermath of a
disaster, and fears that purchasing insurance will crowd out
public support for disaster relief and reconstruction. Increas-
ing demand may require compulsion, nudging, or premium
subsidies, as well as government committing to limits on post-
disaster financial assistance to the uninsured.
The supply of disaster microinsurance may not be a
binding constraint as an increasing number of reinsurance
and insurance companies are keen to enter this market and
there are other types of organisation, more or less formally,
offering microinsurance products. However, the key issue is
ensuring that the right products are being supplied within a
regulatory environment which is fit for purpose, and that
local insurers can transfer risks to international markets at
low cost to avoid being overexposed to adverse experience in
the area where they operate. Regulators will have to assess
the extent to which they will allow exemptions from certain
regulatory requirements on the basis that they impose a
disproportionate burden on microinsurance products and
providers. We have also proposed two possible policy options
to support the supply of microinsurance, the first is the
sample output-based indexed insurance products and the
second is the creation of a supranational level reinsurance
pool guaranteed by national governments. We have also
noted that as early warning systems for natural disasters
improve, insurance contracts will need to be multi-year or
insurance will be unavailable or only available at very high
premiums in years in which catastrophic disasters are
expected.
As natural disasters have widespread financial and
infrastructural effects, governments have a key role to play
both before and in the aftermath of such disasters. There are a
large number of challenges in providing disaster microinsur-
ance products at affordable premiums. However, paraphras-
ing Sen (1995), it is important that insurance products
designed for the poor do not end up being poor products.
Disaster related insurance schemes implemented by the
private sector and supported by regulators, national govern-
ments and supranational organisations may be a cost effective
way of providing this protection.
There is little evidence available on the direct effect of
microinsurance availability on migration patterns. Nonethe-
less, it is possible to make some informed predictions that
could be empirically tested. In particular, if premiums reflect
the underlying risk it seems plausible that by enabling people
to protect themselves against natural disasters, availability of
disaster insurance could support their living in increasingly
fragile environments for longer. However, the availability of
disaster microinsurance which pays out in a timely manner
could increase post-disaster migration by increasing post-
disaster policyholder wealth.
Acknowledgements
This paper was originally commissioned as part of the
Foresight Global Environmental Migration Project by the UK
Government Office of Science. Without implicating them in
the shortcomings of the work, the authors would like to thank
Nora Ferm, three referees and the editor of the Foresight
Project for very useful comments on aspects of the paper.
Views expressed in this paper are the authors’ and should not
be attributed to the World Bank or the UK Government
Actuary’s Department.
r e f e r e n c e s
Adamson, S., Sagar, A., 2002. Managing climate risks using atradable contingent security approach. Energy Policy 30 (1),43–51.
American Academy of Actuaries, 1999. Evaluating theeffectiveness of index based insurance derivatives inhedging property/casualty insurance transactions: report ofthe index securitization task force. Washington, DC.
Banerjee, C., Berg, E., Wassmann, R., 2011 Efficiency of WindIndexed Typhoon Insurance for Rice, presented at EuropeanAssociation of Agricultural Economists 2011 InternationalCongress. Available from http://purl.umn.edu/114240 (lastaccessed 15.12.12.).
Churchill, C., 2006. Protecting the Poor: A MicroinsuranceCompendium. International Labour Organization, Geneva.
Clarke, D.J., 2011. A Theory of Rational Demand for IndexInsurance. Department of Economics Discussion Paper 572,University of Oxford.
Clarke, D.J., Mahul, O., Rao, K.N., Verma, N., (2012). WeatherBased Crop Insurance in India, World Bank Policy ResearchWorking Paper 5985.
Coleman, D., 2009. Divergent patterns in the ethnictransformation of societies. Population and DevelopmentReview 35 (3), 449–478.
Collins, D., Morduch, J., Rutherford, S., Ruthven, O., 2009.Portfolios of the Poor: How the Worlds Poor Live on $2 a Day.Princeton University Press, Princeton.
e n v i r o n m e n t a l s c i e n c e & p o l i c y 2 7 s ( 2 0 1 3 ) s 8 9 – s 9 8S98
Cummins, J.D., Lalonde, D., Phillips, R., 2004. The basis risk ofcatastrophic-loss index securities. Journal of FinancialEconomics 71, 77–111.
Cummins, J.D., Mahul, O., 2009. Catastrophe Risk Financing inDeveloping Countries: Principles for Public Intervention.World Bank Publications, New York.
Deng, X., Barnett, B.J., Vedenov, D.V., 2007. Is there a viablemarket for area-based crop insurance? American Journal ofAgricultural Economics 89 (2), 508–519.
Dercon, S., 2004. Risk, insurance, and poverty: a review. In:Dercon, S. (Ed.), Insurance Against Poverty. OxfordUniversity Press, UK.
Dercon, S., Hill, R.V., Outes-Leon, I., Bayrou, A., Clarke, D.,Seyoum-Taffesse, A., 2011 Unpublished results. OfferingRainfall Insurance to Informal Insurance Groups: evidencefrom a field experiment in Ethiopia.
Dercon, S., Krishnan, P., 2000. In sickness and in health: risksharing within households in rural Ethiopia. The Journal ofPolitical Economy 108 (4), 688–727.
Dror, D., Jacquier, C., 1999. Extending health insurance to theexcluded. International Social Security Review 52 (1), 71–97.
Ferreira, F., Schady, N., 2009. Aggregate economic shocks, childschooling and child health. World Bank Research Observer24 (2), 147–181.
Gaurav, S., Cole, S., Tobacman, J., 2010. Marketing ComplexFinancial Products in Emerging Markets: Evidence FromRainfall Insurance in India, Research paper 1.Microinsurance Innovation Facility, International LabourOrganisation, Geneva.
Gine, X., Menand, L., Townsend, R., Vickery, J., 2010.Microinsurance: a case study of the Indian rainfall indexinsurance market. Policy Research Working Paper Series5459, The World Bank, New York.
Gosling, S., 2011. What is the current state of scientificknowledge with regard to the likelihood and potentialimpact on ecosystem services of future change in the large-scale climate system? Science review SR6, UK Department ofBusiness Innovation & Skills Foresight project on migrationand global environmental change.
Gurenko, E., Lester, R., Mahul, O., Oguz Gonulal, S., 2006.Earthquake Insurance in Turkey: History of the TurkishCatastrophe Insurance Pool. World Bank Publications, NewYork.
Hazell, P., 1992. The appropriate role of agricultural insurance indeveloping countries. Journal of International Development4 (6), 567–581.
Hess, U., Skees, J., Barnett, B., Stoppa, A., Nash, J., 2005.Managing Agricultural Production Risk: Innovations inDeveloping Countries. The World Bank, Agriculture andRural Development Department, Report.
Ibarra, H., Mahul, O., 2004. Self-Insurance Funds as AgricultureInsurance Providers: The Case of Fondos in Mexico. WorldBank, Washington, DC.
IRDA, 2005. Micro-insurance Regulations, 2005. IndianRegulatory and Development Authority, New Delhi.
Kunreuther, H., Michel-Kerjan, E., 2008. Managing Large-ScaleRisks in a New Era of Catastrophes, Extreme Events Project.
Wharton Risk Management and Decision Processes Center,Wharton.
Lenton, T.M., 2013. What early-warning systems are there forenvironmental shocks? Environmental Science and Policy 27(S1), S60–S75.
Mahul, O., Verma, N., Clarke, D., (2012). Improving famers’access to agricultural insurance in India, World Bank PolicyResearch Working Paper 5987.
Morsink, K., Geurts, P., Kooijman-van Dijk, A., 2011. Impact ofmicro insurance on vulnerability of low income householdsin the Philippines: the case of typhoon re-housing insurance.In: Presented at the Second European Research Conferenceon Microfinance.
Overseas Development Institute, 2006. Disasters 30 (1) 1–150Special issue: climate change and disasters.
Oxfam, 2011. Horn of Africa Risk Transfer for AdaptationHARITA quarterly report: April 2011–June 2011.
Perry, G., 2009. Beyond Lending: How Multilateral Banks CanHelp Developing Countries Manage Volatility. Center forGlobal Development, Washington, DC.
PIC, 2010. Joint IC-CDA-SEC Memorandum Circular No. 01–2010Defining Government’s Policy on Informal InsuranceActivities.
Prasad, N., Ranghieri, F., Shah, F., Trohanis, Z., Kessler, E.,Sinha, R., 2009. Climate resilient cities. In: A Primer onReducing Vulnerabilities to Disasters, World BankPublications, New York.
Sen, A., 1995. The political economy of targeting. In: van de Walle,D., Nead, K. (Eds.), Public Spending and the Poor: Theory andEvidence. Johns Hopkins University Press, New York.
Skees, J., Hazell, P., Miranda, M., 1999. New Approaches toPublic/Private Crop Yield Insurance in Developing Countries,Environment and Production Technology Discussion Paper55. IFPRI, Washington, DC.
Skeldon, R., 1997. Migration and Development: A GlobalPerspective. Longman, Harlow.
Dermot Grenham lectures on demography and development atthe London School of Economics. He is also a Chief Actuary (socialsecurity, demography and overseas pensions) at the UK’s Govern-ment Actuary’s Department. He has a doctorate in mathematicsfrom Oxford University and a MSc in population and developmentfrom the LSE. He was a member of the Council of the Institute ofActuaries in the UK and is currently a member of the UK ActuarialProfession’s and the International Actuarial Association’s micro-insurance working parties.
Daniel Clarke is an actuary and development economist whoworks on disaster risk financing and insurance both for peopleon low incomes and for low and middle income country govern-ments. He is based at the University of Oxford and has workedwith the World Bank and the International Food Policy ResearchInstitute on agricultural and disaster insurance in India, Ethiopia,Bangladesh, Mexico, Peru, Indonesia, the Caribbean region and thePacific region, and was one of the architects of the actuarial designand ratemaking methodology for the Government of India’s Mod-ified National Agricultural Insurance Scheme.