76634

    Between the State and the Market:
    Can Informal Insurance Patch the
               Safety Net?


                                         Jonathan Morduch

Most households in low-income countries deal with economic hardships through informal
insurance arrangements between individuals and communities rather than through pub-
licly managedprograms or market-provided insurance schemes. Households may, for ex-
ample, draw on savings, sell physical assets, rely on reciprocal gift exchanges, or diversify
into alternative income-generating activities. These mechanisms can be highly effective in
the right circumstances, but most recent studies show that informal insurance arrange-
ments are often weak. Poor households, in particular, have substantial difficulties coping
with even local, idiosyncratic risks. Public policy can help reduce vulnerability by encour-
agingprivate, flexible coping mechanisms while discouraging those that arefragileor that
hinder economic and social mobility. Promising policies include creating self-regulating
workfare programs and providing a supportive settingforinstitutions working to improve
access to credit, crop and health insurance, and safe and convenient saving opportunities.



Many low-income countries, from Sub-Saharan Africa to Southeast Asia, have suf-
fered major natural disasters and political upheavals in the 1990s. These events re-
mind observers of a reality hidden in official poverty statistics: that the condition of
poverty is linked closely to vulnerability. Many poor households are exposed regu-
larly to risks from illness, harsh weather, political instability, and economic misman-
agement. Concern with vulnerability may be bodi intrinsic and tied to implications
for income generation as well as to the longer-term consequences for the health and
education of children (Jacoby and Skoufias 1997; Hoddinott and Kinsey 1998; Rose
1999).1 Fear of risk can lead poor households to forgo potentially valuable new tech-
nologies and profitable production choices. Rosenzweig and Binswanger (1993), for
example, use data from rural South India to show that an increase in risk (as mea-
sured by an increase of one standard deviation in the coefficient of variation of the

The World Bank Research Observer, voL 14, no. 2 (August 1999), pp. 187-207.
© 1999 The International Bank for Reconstruction and Development / THE WORLD BANK       187
date of the onset of the monsoon) leads to a 35 percent reduction in farm profits for
the poorest quarter of households but has no effect on the wealthiest farmers. Vul-
nerable households may also spiral downward into poverty following adverse eco-
nomic or climatic shocks as productive assets are depleted to protect consumption
levels. Addressing risk can thus be an important complement to redistributive efforts
and antipoverty strategies focused on increasing economic growth and employment.
   Yet even with holes in both public safety nets and private insurance markets, poor
households are not completely exposed to risk. Most have developed coping strate-
gies to deal with the harshest blows. Most of these mechanisms are provided neither
by the market nor by the state but instead are private "informal insurance" arrange-
ments. They include individual and community actions, such as drawing down sav-
ings, selling of physical assets, reciprocal exchanges of gifts and loans, diversifying
crops, and expanding income-generating activities. (For recent surveys, see Alder-
man and Paxson 1994; Besley 1995; Morduch 1995; and Haddad and Zeller 1996.)
Some, like ritualized gift giving, have roots going back generations or even centuries,
while others are newer responses to difficult situations (Mauss 1967).
   Recent studies warn that some public policies may do little more than crowd out
these informal mechanisms, but most evidence shows that crowding out is unlikely
to be a substantial problem. Most informal insurance mechanisms are typically weak
and often provide only inadequate protection to poor households. Studies from re-
gions as diverse as rural India, China, and Sub-Saharan Africa suggest that despite
informal insurance arrangements, households are exposed to considerable risk from
adverse shocks—even idiosyncratic shocks that do not simultaneously affect their
neighbors. Moreover, private informal mechanisms that are effective in reducing
vulnerability can retard economic growth and social mobility. Thus, even where
informal insurance is well developed, public actions that displace informal mecha-
nisms can yield net benefits.
   The emerging evidence suggests that policymakers need to be concerned about
more than providing disaster relief in the wake of large, aggregate shocks such as
floods, earthquakes, droughts, and other natural disasters. It is equally important to
consider the needs of households that are facing losses due to adverse personal, eco-
nomic, or other crises such as illness, poor (local) harvests, and temporary unem-
ployment. Policy options include creating a supportive environment for institutions
that offer safe and reliable means for poor households to borrow and, in particular,
to save. Recent experience shows that it is also possible to offer limited life insurance
and protection against other basic exigencies in a simple, low-cost manner. Much
more speculatively, it may be possible to improve on existing insurance arrange-
ments for poor households by drawing lessons from the emerging microfinance
movement and by relying on nongovernmental organizations and profit-making
commercial enterprises to take key roles. Public workfare programs that offer tempo-
rary employment at wages that are too low to attract those who already have work,

1 «8                                        The World Bank Research Observer, voL 14, no. 2 (August 1999)
such as India's Employment Guarantee Scheme, can also provide households with a
flexible means for self-insurance in times of particular need.



Evidence on Risk Sharing
New tests of informal insurance mechanisms relate the variability of total house-
hold consumption to income variability. If households can use coping mechanisms
to "smooth" consumption somewhat, income should be more variable than con-
sumption. Sharper testable implications can be drawn with respect to perfect
consumption-smoothing arrangements. If communities perfecdy pool their incomes
to share risks (and any given household's income is a small pan of the total), the
consumption level of a given household relative to its neighbors should be a function
only of total community income and the household's assigned share of the total. The
household's own income realization should then not affect consumption patterns,
and all idiosyncratic risk (relative to village shocks) should be eliminated.2
   Townsend (1994) first tested the idea using data from the International Crop
Research Institute for the Semi-Arid Tropics (iCRlSAT) studies of villages in rural
South India. He finds that the evidence does not fully support the proposition with
respect to perfect risk pooling but that it comes surprisingly close: having controlled
for community resources, Townsend finds that the marginal propensity to consume
out of a household's own income is nowhere greater than 0.14, while the theory of
perfect risk sharing predicts it should be zero. Morduch (1991), Ligon, Thomas, and
Worall (1997), and Ravallion and Chaudhuri (1997) find weaker evidence using the
same data, however, with Ravallion and Chaudhuri's estimates of the marginal pro-
pensity to consume falling between 0.12 and 0.46. These results suggest that infor-
mal insurance exists but that it is not nearly perfect. More critically, the studies do
not reveal the specific mechanisms that drive the results: they are consistent with
both gift exchange within communities (but not with perfect risk sharing) and with
self-insurance activities such as borrowing and saving (but not with the perfect abil-
ity to smooth consumption). In practice, borrowing and saving are typically far more
important coping mechanisms than the exchange of transfers (see, for example, Lim
and Townsend 1998).
   Similar studies from other countries also find evidence of highly imperfect infor-
mal insurance. Deaton (1997), for example, finds little evidence of strong risk shar-
ing in samples from Cote d'lvoire, and Townsend (1995a, b) reports a mixed record
of risk sharing in a sample of Thai villages. Jalan and Ravallion (1997) note that the
poorest 10 percent of households in rural China can protect themselves from just 60
percent of an adverse income shock—although the richest 10 percent can cope with
90 percent on average. This measure echoes my evidence from rural South India that
households with large landholdings have little difficulty coping with idiosyncratic

Jonathan Morduch                                                                    189
income shocks but diat the consumption levels of landless households and smallholders
decline sharply as income falls (Morduch 1993).
    Gertler and Gruber (1997) take a different cut on tests of risk sharing. In evaluat-
ing the ability to cope with the costs of illness in Indonesia, they find that house-
holds insure adequately against about 70 percent of common health shocks but can
protect their consumption levels against only about 30 percent of the illnesses that
seriously impair long-term performance. Their findings mirror Cochrane's (1991)
evidence using data from the United States and Lund and Fafchamps's (1997) data
from the rural Philippines. Lund and Fafchamps, in particular, find that informal
insurance arrangements are effective only in the case of young adults who are acutely
ill; older adults who fall ill are far less likely to be helped. Lund and Fafchamps's data
show that informal insurance also helps with funerals but not with crop failures,
mild illnesses, or unemployment (other than that of the head of the family and his
spouse). The household's social network also matters: households with more friends
(especially richer friends) have a greater ability to use informal insurance. House-
holds that are not so well connected fare much worse.
   These results, based on disaggregations by class and type of shock, reveal weak-
nesses in informal insurance that would not otherwise be evident. And they imply
that there is ample scope for potentially beneficial interventions that go beyond di-
saster relief.



Informal Insurance Mechanisms
Reciprocal gift giving is a common way to solidify social and economic relationships
—and one potentially important form of informal insurance. In North China, for
example, gifts are given to mark births, deaths, and weddings, as well as to help the
elderly, the ill, and women who have just given birth (Kipnis 1997). Anthropologists
have tended to downplay gift giving as a product of a rational calculus associated
with informal insurance systems, instead highlighting its role in securing social sta-
tus and signaling commitment to the community (Malinowski 1922). But, given
the sense of obligation engendered by gift exchange and the great potential for gift
exchange to address risk, it is only natural that economists have taken the gift-
insurance relationship seriously. Economists have thus tended to view gifts as they
do other transfers such as public aid (Cox and Jimenez 1991).
   Private transfers of cash, food, and clothing are large and frequent in some coun-
tries. For example, 40 percent of black South Africans reported either receiving or
giving cash transfers (Cox and Jimenez 1997). Cash transfers reach large numbers of
urban residents in Colombia (Cox and Jimenez 1998), Thailand (Paulson 1995),
and the Philippines, where 82 percent of urban and 89 percent of rural households
report receiving transfers (Cox and Jimenez 1995).

19O                                         The WorU Bank Research Observer, vcLN.no.2   (August 1999)
   Remittances from migrants—whether migrants to anodier country or to die city
from the country—can also be substantial (see Lucas and Stark 1985 on Botswana
and Paulson 1995 on Thailand). Roughly two-thirds of all transfer inflows in Paki-
stan between 1985 and 1988 came from migrants who sent money home to their
families (Foster and Rosenzweig 1999). In the Philippines, 26 percent of urban and
 13 percent of rural households received remittances from migrant parents or chil-
dren (Cox and Jimenez 1995).
   But elsewhere, especially where migration is limited, reported transfers are of mi-
nor consequence. In a data set from India fewer than 400 of 4,000 households re-
ported receiving net transfers in 1968-71 (Foster and Rosenzweig 1999). The lack
of transfers is especially notable in contexts in which they are expected to be most
valuable. In the smaller ICRISAT survey of poor villages in rural South India, Rosenzweig
(1988) finds that transfers respond to risk but that they cover less than 10 percent of
the typical shortfalls in income. Studies of exchange in Sub-Saharan Africa by Reardon,
Madon, and Delgado (1988) and Czukas, Fafchamps, and Udry (1998) reinforce
this picture. Data from rural China are only slightly more optimistic. In a four-year
study of 16 villages in the north of the country, Morduch and Sicular (1999) find
that no more than a quarter of households report receipts of transfers from their
neighbors; 10 percent report receiving gifts from outside their village. The transfers
were modest, moreover, averaging about 10—20 percent of average household
income.
   Lim and Townsend (1998), who have examined ICRISATfiles,conclude that house-
holds bridge the gap between income and desired consumption levels through sav-
ing, borrowings and the use of buffer stocks of grain. The importance of these mecha-
nisms is evident in most low-income economies (Deaton 1992; Alderman 1996).
Households might also take actions to smooth income after an economic shock by
working longer hours, for example, or taking an exua job (Kochar 1999). And they
may take precautions beforehand to reduce the probability or extent of loss (Morduch
1995). These mechanisms can be relatively effective in the right circumstances, lead-
ing to concerns that publicly provided resources would simply replace—or crowd
out—these informal activities.



Reexamining the Costs of "Crowding Out"
To the extent that informal mechanisms are limited, the concern about crowding
out should be small. Still, some evidence is disturbing. Cox and Jimenez (1995) use
household-level data from the urban Philippines to give a particularly striking ex-
ample. They estimate that the receipt of net transfers from other households is par-
ticularly sensitive to whether the recipient is unemployed. The magnitude of the
sensitivity is such that they conclude that if the government were to institute a simple

Jonathan Morduch
unemployment insurance scheme, net private transfers to the unemployed would
fall by 92 pesos for every 100 pesos offered by the public program. In the end, the
average unemployed worker would be better off by only 8 pesos.
    Exercises based on extrapolations like this yield provocative—but not definitive—
results. First, the research is not based on the effects of an actual program. Second,
researchers typically have data from just one side of any given transfer: who makes it
or who gets it—but not both. Without complete information, it is difficult to tease
out the exact reasons for the transfers. A transfer that looks like informal insurance
against a bout of illness, for example, might instead reflect a correlation arising for
other reasons, such as a gift from a child to provide a parent with old age support. In
the evidence from the Philippines, Cox and Jiminez (1995) concluded that the criti-
cal retirement income variables reflected the desire of givers to help retirees. Is that in
fact the case? Or do the variables instead signal a type of household (one with little or
no retirement income) that is more likely to include migrants—and thus to receive
remittances—rather than one without migrants? In the latter case, the receipt of net
transfers could look like informal social insurance but instead could simply reflect
steps taken to maximize household income. Disentangling the explanations requires
richer data.
    Some of the sharpest evidence on crowding out is from South Africa. When the
apartheid system was falling apart, the government extended basic pension benefits
to black South Africans on terms similar to those that had been available to whites
who had no private pensions. The program, which was fully implemented by early
1993, provided a state pension equal to about $3 a day to all women over age 60 and
to all men over age 65, subject to a means test (Case and Deaton 1998). The means
test excluded nearly all whites and only the richest blacks. In the past, blacks had had
to rely mainly on their own means to cope with aging and with economic down-
turns, and for the most part the new benefits were not expected. How were tradi-
tional private mechanisms affected by the postapartheid public pension reform? Jensen
(1998) estimates that for those households receiving private transfers, every publicly
provided rand led to a reduction of 0.2-0.4 rand in private transfers to the elderly.
Migration by children was also reduced slightly.
   A similar degree of crowding out is predicted on the basis of studies of transfers to
urban residents in Peru and the Philippines: a displacement of 17 percent and 37
percent, respectively, for each unit transferred as a retirement benefit (Cox and Jimenez
1995, 1998). Without the chance to evaluate the introduction of an actual new
program (as in Jensen 1998), the predictions derive from estimates of the sensitivity
of net transfer receipts to existing social security or private pension arrangements.
Cox and Jimenez then use coefficients from this exercise to predict the consequences
of the introduction of a broad state pension system.
    Even with displacement rates as high as 20-40 percent, are the social losses pro-
portionately high? Not necessarily, because leakage does not imply pure wastage.

192                                          The WoHd Bank Research Observer, voL 14, no. 2 (August 1999)
The degree to which displacement undercuts policy objectives depends on the spe-
cific objectives of the program in question: Is it poverty reduction? Old-age support?
Enhancing economic efficiency? Reducing vulnerability?
   In die case of South Africa, older citizens received less in total than supposed, but
others gained—and those gains were socially valuable. In richer contexts, private
transfers tend to be from the old to the young. In a sample of white South Africans,
for example, the net recipients were five-and-a-half years younger than the net givers
(Cox and Jimenez 1997). But in poorer contexts, the reverse is most often true,
largely because parents invest in children with the expectation that the children will
support them in their old age. In line with that hypodiesis, net recipients among
black South Africans are on average eight years older than net givers. The displaced
transfers thus tended to return to young households, many of which are as poor as
older households, yielding litde leakage as far as poverty reduction is concerned. In
addition, keeping the funds in the hands of younger households is more likely to
encourage investment in human capital accumulation and other productive activi-
ties. Second, public transfer systems may be more efficiently delivered than private
transfers, yielding a net gain to society through displacement. For example, public
transfer schemes may be able to pool resources more efficiendy than local private
arrangements (Cox and Jimenez 1997). Third, some displacement, even if it consti-
tutes an unwanted leakage, may be a required cost of strengthening and widening
the safety net to include particularly vulnerable households. In South Africa, for
example, just under half of pension recipients do not receive private transfers at all
(Jensen 1998). Putting the other arguments aside, tolerating some crowding out of
the transfers received by half of the elderly black population can be seen as a cost of
extending the safety net to the other half.



Tensions in Informal Insurance
Why is evidence of risk sharing so weak in places where it is expected to be strong?
For example, consider the highly risk-prone semiarid tropics of South India, where
two out of every ten years on average bring drought. Despite the importance of
bad weather, most of the variation in measured household incomes over time is
idiosyncratic to particular households. Morduch (1991) shows that 75-96 percent
of the variance of the logarithm of household income remains after removing varia-
tion due to changes in average village income and average household income over
the study period (1976—82). Some of this idiosyncratic, residual variation is surely
measurement error, but even if half of it is error, substantial idiosyncratic variation
still remains. As a result, within-village gift exchanges, designed so that no net
redistribution takes place over the period, could in principle reduce the variability
of household after-gift income by as much as 90 percent in one of the villages

Jonathan Morduch                                                                     193
under study. In practice, reported transfers are not nearly big enough to do so
(Rosenzweig 1988).


Tensions in Systems of Reciprocal Transfers
To explain why systems of reciprocal transfers are weak, researchers point first to
problems in enforcing understandings. Household A will help household B today,
with the expectation that B will eventually reciprocate. But what will keep B from
reneging? If A and B are related by blood or marriage, altruism may hold them
together. But without altruism or enforceable contracts, self-interest is needed to
keep incentives In line. The repeated nature of the interaction over time allows for
self-interested reciprocity. If A can credibly commit to end all future insurance rela-
tionships with B in the event that B reneges, B may well see fit to fulfill obligations.
B's decision will depend on whether the gain from reneging today is smaller than the
flow of future benefits from continued participation.
   As Coate and Ravallion (1993) suggest, however, the degree of effective insurance
that is provided will adjust so as not to tip the balance toward reneging. In practice,
tensions are heightened when both parties are down on their luck (during a drought,
say) or when a partner's luck is particularly bad. When an individual is pushed close
to the subsistence constraint, holding onto whatever one has may be especially tempt-
ingj despite the agreement to share with others. As a result, reciprocal exchange tends
to fall apart (or to offer less of a return) when insurance is most needed. In general, it
works best when participants have a cushion against poverty. Consistent with the
evidence from China, the Philippines, and South India, theory suggests that systems
of reciprocal transfers will be more effective for slightly richer households and those
in less dire contexts (Coate and Ravallion 1993; Ligon, Thomas, and Wo rail 1997;
Kletzer and Wright 1998).
   Moral hazard is also likely to limit group-based informal insurance, just as it un-
dermines standard insurance markets. When insurers cannot adequately observe and
enforce that insurees are taking all due precautions, incentives can be enhanced by
providing only partial insurance coverage. This is one reason that informal insurance
is most prevalent among relatives or neighbors in similar professions, that is, those
with good flows of information.
   Transfer-based systems can also run into trouble when households have opportu-
nities to accumulate savings because they then have a degree of insurance free of
obligation to neighbors and kin. Similarly, when incomes of participants grow at
different rates, richer households tend to opt out rather than face the possibility of
systematically redistributing to others. Richer participants find themselves giving
relatively more than they get back on average, and at a point they will leave, either to
form a new group with other richer households or to fend for themselves individu-
ally. Evidence of these sorts of bifurcations is given by Platteau and Abraham's (1987)

194                                         The World Bank Research Observer, ml 14, no. 2 (August 1999)
study of reciprocity in fishing villages in Kerala, India, and by Lund and Fafchamps's
(1997) study of risk-coping mechanisms in the rural Philippines. In a reverse ex-
ample, Platteau (forthcoming) argues that one reason for low saving rates in Sub-
Saharan Africa is that rural communities and families discourage saving in order to
avoid eventual cleavages.
   These tensions explain how common mechanisms can solidify economic and social
barriers along ethnic, gender, generational, and class lines (Fafchamps 1992; La Ferrara
 1997) and contribute to "poverty traps" (HofF 1997; Platteau forthcoming). Fafchamps
(1992) draws on African experiences to suggest that instead of leading to cleavages,
reciprocal exchange may instead lead to voluntary patron-client relationships. Rather
than being asked to give more than poorer households, relatively rich households may
find themselves in a position to extract surpluses from poorer households. Rich house-
holds, with their stocks of wealth, can offer a great deal to poor, vulnerable households.
But the poor may have to offer labor at concessional rates to obtain protection from
their patrons in hard times. The terms of reciprocal exchange may thus gready favor
the rich, although the terms are to everyone's absolute advantage. This seemingly feu-
dal scenario may play out in subtle forms throughout poor economies.
   Fafchamps's model shows how informal insurance may adapt to particular eco-
nomic conditions, but observers suggest that despite the ability to adapt, these recip-
rocal mechanisms have started falling apart in recent years in Africa. The blame is
put on economic and political upheavals, reinforced by increasing mobility and ur-
banization. In principle, urbanization and the increasing ease of mobility can both
help and hinder the functioning of informal insurance. The negative is straightfor-
ward: by moving away, households are able to "default" on their obligations to rela-
tives and neighbors. This may explain why until recently migrants from Kenya and
elsewhere often moved as a family. Now, prohibitive costs and risks make that less
prevalent, and workers often move on their own, adding to the likelihood of default-
ing on obligations to their ex-neighbors.
   The positive aspects rely on continued links. Migration allows geographic diversi-
fication of incomes, increasing the value of reciprocal relationships. Paulson (1995)
shows evidence from Thailand that some family members migrate pardy to diversify
the family's "portfolio" of earnings sources; Lucas and Stark (1985) and Rosenzweig
and Stark (1989) make similar claims based on data from Botswana and rural South
India, respectively. Because links must remain unsevered for informal insurance to
work, only insurance among family-based groups can typically survive mobility.
   In a recent theoretical contribution, Banerjee and Newman (1998) embed these
ideas in a more general model of structural change. They suggest that the lack of
insurance mechanisms in urban areas can inhibit mobility from villages. In the vil-
lage, a worker can count on some security through group-based insurance mecha-
nisms but will have relatively low earning opportunities. The city offers greater earn-
ing opportunities but weaker insurance mechanisms. The result is that only the

Jonathan Morduch                                                                       193
relatively rich (who can cope better without group-based insurance) and the rela-
tively poor (who never had much group-based insurance to start with) will migrate.
People in the large middle segment of the population will stay put, even though it
may be economically beneficial for them to break their ties with the village and join
the modern sector. The presence of informal insurance in villages can then be a drag
on economic development. Drawing on data from Indian villages, Das Gupta (1987)
provides evidence along these lines. In parts of Sub-Saharan Africa the greatest prob-
lems tend to be caused by excessive rural-urban migration rather than by insufficient
mobility, but even here the Banerjee-Newman model can still provide useful in-
sights. The basic ideas can be applied to explain inefficient mobility between eco-
nomic sectors, for example, rather than just inefficient geographic mobility.
    The final set of tensions centers on the role of the family. The family has been
hovering in the background in this discussion because people generally turn to their
relatives first—and often again as a last resort—in times of need. On the one hand,
the institution of the family, stretching over generations and bearing well-under-
stood protocols, greatly facilitates informal insurance. Most important, information
and enforcement problems are mitigated. On the other hand, the family tends to
have a much more limited pool of resources on which to draw relative to the broader
community.
    The most important tensions arise when the demographic structure of households
is shaped to meet the purposes of informal insurance. For example, the old age secu-
rity theory suggests that children are produced partly to provide informal social secu-
rity. In situations with overcrowding and in cases in which parents do not take into
account the negative externalities imposed by their children (through congestion
and environmental degradation, for instance), social welfare may be enhanced by
shifting to alternative social security mechanisms (Dasgupta 1993; Anand and
Morduch 1999). For example, establishing secure, convenient savings programs may
allow households to reduce the number of children they have without undermining
their ability to cope with less income in old age and can provide a second round
of benefits to the community through reductions in negative population-related
externalities.
   A number of other insurance-demographic links lead to similar tensions, includ-
ing social pressure to migrate and to select marriage partners in order to provide the
family with insurance (Rosenzweig and Stark 1989; Paulson 1995); family "churn-
ing" (that is, the turnover of responsibility) as a response to the death of a head of
household; and the practice of taking in a foster child. Child fostering is common in
Cote d'lvoire, Ghana, and Sierra Leone, although explanations differ. Bledsoe and
Isiugo-Abanihe (1989) discuss insurance-related motives for fostering; Ainsworth
(1996), however, in a survey from Cote d'lvoire, finds that the need for labor is a
more important factor.



196                                        The WoHd Bank Research Obsenvr, ivL 14, no. 2 (August 1999)
Tensions in Other Forms of Informal Insurance
Other insurance mechanisms also tend to be least effective just when they are most
needed. In principle, buying and selling assets provides a good hedge against idio-
syncratic risks (at a minimum). Rosenzweig and Wolpin (1993), for example, find
that buying and selling bullocks is an important consumption-smoothing device in
semiarid India. But even there, the correlation between poor harvests and the price
of bullocks (the covariation of risk) can raise problems. In fact, Lim and Townsend
(1998) suggest that covariant shocks and the nature of bullock transactions instead
^/volatility to cash holdings rather than protecting them, and after carefully sifting
through the same data, they uncover little evidence that is consistent with the
Rosenzweig-Wolpin findings. The Lim-Townsend finding is consistent with the ten-
sions that are introduced when risks covary. Thus, asset prices may fluctuate widely
when every household wants to buy goods—or dump goods—at the same time. As a
result, it may not be surprising that Czukas, Fafchamps, and Udry (1998) find that
in Burkina Faso, selling livestock protected households against only 20-30 percent
of the income shortfalls suffered as a result of a drought. (It is possible, however, that
effective insurance would have been stronger had the drought been less widespread.)
   In addition, informal mechanisms are typically weak against repeated shocks. Simu-
lations by Deaton (1992) show that the efficacy of using buffer stocks or savings
accounts to smooth consumption is conditioned largely on the degree to which bad
shocks follow one another over time. When harsh conditions are likely to persist for
several years in a row, households would have to have very large stores of assets to
achieve adequate protection. This is one reason that the consequences of droughts
and floods may be especially bad: because they frequently entail adverse environ-
mental changes (runoff; desertification; poor soil conditioning), they play out even
after the climate has returned to normal.
   Even where mechanisms work well in a narrow sense, they may do so only at large
long-run social costs. First, many mechanisms are inherendy costly. In risk-prone
areas of India, for example, households may sacrifice as much as 25 percent of aver-
age income to reduce exposure to shocks (Walker and Ryan 1990). In principle,
then, improving safety nets can increase average incomes by reducing reliance on
these costly measures (Platteau 1991; Morduch 1993, 1994). Perhaps more impor-
tant, the desire to stay with tried and true technologies limits experimentation and
innovation, creating ongoing problems for households.
   Improving insurance may also mitigate social inequalities. Many informal insur-
ance mechanisms have a gender dimension as well. Women often bear the brunt of
arranged marriages, migration, and child fostering. Women may also lose out more
than men during downturns. In India, for example, Rose (1999) finds that child
mortality rates increase during periods of very low rainfall and are significandy higher



Jonathan Morduch
for girls than for boys. Reducing households' vulnerability and instituting more flex-
ible insurance instruments may thus have broader social implications.


Policy Implications
A first set of policy priorities includes actions to reduce risk itself. For example,
improving governance can sharply reduce the vulnerability of households to down-
turns resulting from economic mismanagement. Increasing macroeconomic stabil-
ity, reining in inflation, securing property rights, improving transport and commu-
nications, and creating a stable political environment can go a long way toward
reducing the frequency and size of downturns and creating a supportive environ-
ment to facilitate private risk-reducing activities. Similarly, risk can be reduced through
public health campaigns for immunization and sanitation, civil works projects (dams,
retaining walls, irrigation), and, in some cases, price stabilization. Higher incomes
and stable employment opportunities further enhance the ability to cope with risk.
But these are all policy areas that are on the table for other reasons and are best
judged by other criteria.
   In richer countries, households typically prepare for income declines by acquir-
ing savings accounts, lines of credit, pensions, insurance, and annuities. Where
these actions run into limits, governments typically provide means-tested poverty
alleviation programs, unemployment benefits, health insurance, and social secu-
rity (Subbarao and others 1997, ch. 3). But neither the administrative capacity nor
the funding exists in most low-income countries to build similar public safety nets.
Public action can, however, help to address smaller, local hardships by providing
regulatory and institutional frameworks that expand households' access to insur-
ance, credit, employment opportunities, and convenient ways to save. Limiting
the government's role conserves scarce administrative resources and avoids poten-
tial conflicts of interest between short-term political exigencies and requirements
for longer-term institutional sustainability. These policies provide ways to strengthen
informal coping mechanisms and broaden their accessibility rather than displace
private actions.

Promoting Savings
It had long been thought that most poor households had little desire to save in
banks, but the experience of Indonesia's Bank Rakyat Indonesia (BRl) and similar
microfinance programs are turning that view around. After BRI established a safe,
convenient savings vehicle, consumers responded enthusiastically. BRI now has more
than 16 million low-income depositors (compared with 2 million borrowers), gready
aiding the bank's profitability. Although there is no systematic evidence on the in-

19«                                         Use World Bank Research Observer, voL 14, HO. 2 (August 1999)
come levels of depositors, bank staff note that they tend to be poorer on average than
borrowers and from diverse socioeconomic backgrounds. Partly as a result, savings
mobilization efforts are now being renewed in microfinance programs in Africa, Asia,
and Latin America. Public policy can aid by ensuring an appropriate regulatory envi-
ronment and helping to keep inflation in check.
   One promising program has shown the surprising demand for savings deposits
among poor households in the slums of Dhaka, Bangladesh. SafeSave, a nongovern-
mental organization, patterned its savings program on that of the local rotating sav-
ing and credit associations, which operate by taking in small sums from participants
(Rutherford 1999). The response to SafeSave has been much greater than expected,
and depositors have been able to slowly build up usefully large sums of money. As a
consequence, the program is being replicated by other nongovernmental organiza-
tions in South Asia.
   Without easy saving opportunities, households are tempted to squander surpluses
or are susceptible to calls for short-term help from family members or neighbors—
often at the expense of long-term progress (Platteau forthcoming). In this way, sav-
ings instruments may well be much more important than the provision of credit in
raising incomes and reducing risk—and easier to accomplish.
   Suchfinancialdeposits can be particularly effective in helping households weather
the difficult scenarios that undermine gift exchanges. Consider the various forms of
shocks that households encounter. Events that occur infrequendy—old age, death in
the family, and chronic disability—can hit households hard and may require a con-
tinuing flow of transfers. Given tliat such transfers are not guaranteed, savings de-
posits can be critical in ensuring that people have enough income to satisfy basic
needs. (Note, however, that if interest rates fall below inflation rates, the purchasing
power of these deposits can erode quickly.) Savings also allow households to avoid
borrowing from moneylenders at interest rates as high as 5-10 percent a month
when emergency funds are needed and can be especially valuable in a crisis (Von
Pischke 1991). A regional drought, for example, will lead to a decline in the price
of assets as affected individuals simultaneously try to sell their holdings. Financial
savings, however, will generally hold greater value (and could increase in value as
prices fall).
   Public policy that leads to better integrated savings programs can help to contain
risk, allowing thefinancialsystem to handle shocks more easily. The theoretical rela-
tionship of deposit mobilization, efficiency enhancement, and die generation of eco-
nomic growth is described by Bencivenga and Smith (1991). The keys to a successful
savings program are providing long-term security and convenience, finding a way to
hedge against inflation, minimizing costs, and exploiting opportunities to relend
deposits safely but profitably.3 Existing banks and nongovernmental organizations
may not be up to all of these tasks, and designing effective (but not overly intrusive)
prudential regulations is a critical first step.

Jonathan Mordtuh                                                                    199
Microcredit
Microcredit programs have succeeded by creating hybrid institutions that channel
formal-sector funds to poor households. The programs are not perfect. In Africa, for
example, the challenge is to create mechanisms that work well in semiarid and arid
rural regions where households tend to have less diversified income bases and where
low population densities mean higher transactions costs for financial institutions.
More generally, most poverty-focused programs face high costs that undermine at-
tempts at profitability. A recent survey shows that such programs are able to cover an
average of only 70 percent of their total costs {MicroBanking Bulletin 1998).
   The benefits, though, may be considerable. Using a recent survey of 1,800 house-
holds in Bangladesh, I find that access to microcredit programs yields no appreciable
increase in average consumption levels in the short term (Morduch 1998). For those
with access, however, the volatility of consumption over the three main cropping
seasons is roughly half that of control groups (after controlling for unobservable
variables at the village level). This reduction in consumption variability turns out to
be mainly a product of reduced income variability across seasons, which is made
possible by the employment diversification that credit affords. Helping rural house-
holds to reduce risk further by diversifying into nonfarm labor is an overlooked, but
important, return to microcredit, and more research needs to be done along these
lines to inform discussions of the costs and benefits of supporting credit-based ap-
proaches (Khandker 1998; Morduch forthcoming).4


Insurance
Crop insurance programs have been a disaster nearly everywhere—not unlike tar-
geted credit programs in the 1970s (Yaron, Benjamin, and Piprek 1997). Imperfect
information and high transactions costs have proven to be destabilizing, and there
are no easy solutions. Although reform currently looks unpromising, in principle the
problems of insurance markets are not much more intractable than those of credit
markets—and microfinance programs have shown effective ways around some of
the largest hurdles there. Some microfinance programs have introduced insurance
successfully on a limited scale, offering term life insurance at very low rates (with
benefits large enough to clear debts and provide for a burial but not much more). In
addition, the Grameen Bank of Bangladesh, for example, appears to have had suc-
cess with its "emergency fund" for borrowers. The fund aids with loan repayment
and provides general help in the event of illness and other emergencies. Information
and transactions costs are reduced by coupling these mechanisms with credit provi-
sion. Experimentation will be necessary to determine whether these types of insur-
ance mechanisms can be provided separately from other microfinance services.



iOO                                        The World Bank Research Observer, 10L 14. no. 2 (August 1999)
   As Morduch and Sicular (1999) suggest, there may be ways to insure poor house-
holds by drawing broader lessons from microfinance. In a study of northern China,
we describe an insurance company that has found some success in selling crop insur-
ance to groups (a whole village, say), rather than to individuals. At the moment, this
group insurance is used just to lower transactions costs for insurers and is a poor
analogue to the group-lending practices used by microfinance institutions. But if
future premiums were tied to the history of losses, a group-based contract could
provide incentives for peer monitoring along the lines that microcredit programs
have found successful in addressing moral hazard. This is an area open to specula-
tion, and many roadblocks remain—for example, how should a program discourage
collusion by the entire group?
   One important lesson from microfinance is that programs operated direcdy by
governments tend to have inherent difficulties in generating compliance by partici-
pants; borrowers are far more likely to default on loans from government sources,
and governments are more likely to tolerate defaults in the name of political expedi-
ency. This has proved disastrous for the long-term sustainability of public credit
programs. There is a parallel in the case of insurance; insurees appear less likely to
take due precautions when governments are the insurers. Facilitating insurance pro-
vision by nonprofits, nongovernmental organizations, and for-profit companies may
thus be an important step forward.


Employment Guarantee Schemes
Direct public interventions can also help to reduce vulnerability, especially for the
poorest households. Among the most promising are rural public works programs
such as India's Employment Guarantee Scheme in Maharashtra State, mentioned
earlier, and the Food for Work program in Bangladesh, both of which are described
by Ravallion (1991). These programs provide wage employment in return for work
on constructing and maintaining public infrastructure. Ravallion (1991) reports that
India's scheme provided about 100 million person-years of employment between
1975 and 1989. On average, 500,000 people participated per month (of a total of 20
million rural workers). Walker, Singh, and Asokan (1986) find that the coefficient
of variation of income among landless laborers in two villages in Maharashtra with
access to the employment scheme was half that of a similar village without access.
   The work requirement and low wage rates provide a way to target the aid to
truly needy households, allowing the programs to avoid instituting costly means
tests. Participants take advantage of the program only when needed, often in lean
seasons before harvests. During peak seasons, alternative employment opportuni-
ties are generally more attractive. Thus the programs avoid long-term dependency
and ever-growing lists of participants. It also means that the programs are set up to



Jonathan Morduch                                                                   2O1
help households cope with temporary hardships, not mainly as an answer to chronic
poverty.
   The success of such programs will depend on government budgets, eligibility cri-
teria, and wage rates. According to Ravallion (1991:171-72), there should be "as
few restrictions on eligibility as feasible, and wage schedules and die rights of partici-
pants should be well defined, well known, and nondiscriminatory. Ideally, all who
want work at die going wage rate should be able to get it." The principle of inclusive-
ness is a key to reducing vulnerability because households (or at least those with
available workers) are reassured by knowing that they will have a place to turn when
they fall on hard times.


Conclusion
Poor households throughout the world face twin disadvantages. The first is difficulty
in generating income. The second is vulnerability to economic, political, and physi-
cal downturns. Inflation, recession, drought, flood, illness, and civil war hit hardest
those households that are least well equipped to handle the shocks. Harder still, the
two disadvantages reinforce each other. Poverty is a source of vulnerability, and re-
peated exposure to downturns reinforces poverty.
   The circular nature of poverty and vulnerability does not, however, preclude ef-
fective responses. The evidence to date suggests several broad directions to pursue. In
addition to helping households cope with large natural disasters, governments need
to encourageflexibleprivate interventions. Concern has arisen about whether public
action will crowd out private informal insurance mechanisms. To the contrary, well-
designed public action can strengthen and broaden the capacity of households to act
independendy through informal mechanisms.
   Making saving safer and more convenient, helping to expand credit access, and
fostering basic insurance programs are all promising ways to help households help
themselves in the face of adversity. The possibility of crowding out existing informal
arrangements should not be ignored, but in most low-income countries, it is un-
likely to substantially undermine steps to help poor households. First, informal in-
surance is often very limited, and second, the crowding out of some private actions
can have valuable social benefits.


Notes
Jonathan Morduch is a MacArthur Foundation Research Fellow at Princeton University. The author
is grateful for comments from Harold Alderman, Angus Deaton, and the journal's referees. He has
also benefited from discussions with Marcel Fafchamps, Trina Haque, John Hoddinott, Emmanuel



SOS                                            The World B*nk Research Observer, voL 14, no. 2 (August 1999)
Jimenez, Helena Ribe, and participants in a seminar at the World Bank in July 1997. This work was
supported by the Africa Department of the World Bank. The first draft was completed while a Na-
tional Fellow at the Hoover Institution, Stanford University, and the revision was completed with
support from the John D. and Catherine T. MacArthur Foundation.
    1. Economists' calculations of the welfare costs of vulnerability typically lead to relatively small
 numbers for the benefits of risk reduction (see, for example, Newbery and Stiglitz 1981). But con-
sumer riots in the face of price swings, evidence on the lengths to which households and governments
go to avoid volatility, and participatory assessments of poverty (such as World Bank 1994) suggest
 that approaches grounded in simple microeconomic theory have been too narrow to capture the full
extent of concern.
   2. The specific tests of consumption insurance are rooted in the theory of optimal social alloca-
tions, and they begin with the assumption that each household has a fixed social weight, 9, in a social
planner's problem. If household A's utility from consumption is «(C\) and household B's utility is
 u(CJ, then marginal utilities are simply u'(CJ and u'(C^). In the benchmark theory without enforce-
 ment or information problems, transfers should be made in every period so that 8^ "'(C,) = &B u'(C^).
That is, redistribution from household A to B or from B to A should occur until there is no possible
way to increase the weighted sum of their utilities. The burden of idiosyncratic shocks should thus be
shared by the participants. And, in principle, the relationship should also hold for households A and
C, A and D, A and E, B and C, and so forth. With data in two periods, the relationships continue to
hold, so that 0^ u\C^ = Qg u'(Cm) and 8^ "'(C^) = 0fl u'iC^). Putting the two relationships together
shows diat marginal utilities must grow over rime at the same rate for all households: u'iC^lu'iC^) =
«'(C )/«'(Cai). Under assumptions commonly made about die shape of utility functions—for ex-
ample, u(c) = ^"'/(l — p), where/) > 0 is the coefficient of relative risk aversion—die relationship also
holds for the growth of consumption itself (or its logaridim). This relationship has provided the basis
for a test of die basic theory on risk sharing. If the proposition and die assumptions about the form of
preferences are correct, once the consumption growth of any single household (or of a region in
aggregate) is known, the consumption growth of everyone should be known. Moreover, no other
variables (such as income or income growdi) should have influence. Given die assumptions, die test
boils down to whedier or not coefficients on income and income growth are statistically significant in
explaining patterns of household-level consumption once regional consumption aggregates are con-
trolled for. If die proposition holds exacdy, die marginal propensity to consume out of idiosyncratic
income changes should be zero. Townsend (1994) and Cochrane (1991) give a more explicit presen-
tation. The presentation here implicidy assumes that households have identical preferences for con-
sumption over time, that consumption and leisure are separable in utility, diat utility is additively
separable over time, and diat utility is a function only of consumption levels. If instead utility also
depends on household characteristics (a reasonable view), the testable implication is diat consump-
tion growdi depends only on preference parameters, not on budget parameters—but this is a harder
proposition to test.
   3. How and where deposits are invested appears to be far less important than diat deposits are
mobilized from poor households. In principle, there is no reason not to invest die money abroad, for
example, if domestic options prove difficult and returns are unattractive. Where it is costly to set up
savings bank branches, simple mechanisms such as post office savings plans (or innovations based on
die African susu collector) may offer appealing options.
   4. Reardon, Matlon, and Delgado (1988), for example, show that nonfarm income accounted for
30—40 percent of total income in drought-affected Burkina Faso and that it was only imperfecdy
correlated widi crop income, providing protection against the drought. Some of the nonfarm income,
though, may come as an ex post response to downturns in farm income, leading to negative correla-
tions. Czukas, Fafchamps, and Udry (1998), however, find diat nonfarm income was positively corre-
lated with farm income during a similar circumstance in Burkina Faso, so diat nonfarm income was
not an important offset to crop income.




Jonathan Morduch                                                                                    203
References
The word "processed" describes informally reproduced works that may not be commonly available
through library systems.
Ainsworth, Martha. 1996. "Economic Aspects of Child Fostering in Coted'Ivoire." InT. PaulSchultz,
   ed., Research in Population Economics, vol. 8, pp. 25—62. Greenwich, Conn.: JAI Press.
Alderman, Harold. 1996. "Saving and Economic Shocks in Rural Pakistan." Journal of Development
   Economics 5l(2):343-66.
Alderman, Harold, and Christina Paxson. 1994. "Do the Poor Insure? A Synthesis of the Literature
   on Risk Sharing Institutions in Developing Countries." Economics in a Changing World: Proceed-
   ings ofthe Tenth World Congress ofthe International Economics Association, Moscow. Vol 4. London:
   Macmillan.
Anand, Sudhir, and Jonathan Morduch. 1999. "Poverty and the 'Population Problem.'" In Massimo
   Livi-Bacci and Gustavo dc Santis, eds., Population and Poverty in the Developing World. Oxford:
   Oxford University Press.
Banerjee, Abhijit, and Andrew Newman. 1998. "Information, the Dual Economy, and Develop-
   ment." Review ofEconomic Studies 65(4):631—54.
Bencivenga, Valerie, and Bruce D. Smith. 1991. "Financial Intermediation and Endogenous Growth."
   Review ofEconomic Studies 58(2): 195-209.
Besley, Timothy. 1995. "Non-Market Institutions for Credit and Risk-Sharing in Low-Income Coun-
   tries." Journal ofEconomic Perspectives 9(3): 115-27.
Bledsoc, C. H., and U. C. Isiugo-Abanihc. 1989. "Strategies of Child Fosterage among Mende Gran-
   nies in Sierra Leone." In R. Lesthaeghe, ed., African Reproduction and Social Organization. Berke-
   ley, Calif: University of California Press.
Case, Anne, and Angus Deaton. 1998. "Large Cash Transfers to the Elderly in South Africa." Eco-
   nomic Journal 108(450): 1330-61.
Coate, Stephen, and Martin RavaJlion. 1993. "Reciprocity without Commitment: Characterization
   and Performance of Informal Insurance Arrangements." Journal ofDevelopment Economics 40(Feb-
   ruary):l—24.
Cochranc, John. 1991. "A Simple Test of Consumption Insurance." Journal of Political Economy
   99(5):957-76.
Cox, Donald, and Emmanuel Jimenez. 1991. "Achieving Social Objectives through Private Trans-
   fers: A Review." World Bank Research Observer 5(2):205-18.
         . 1995. "Private Transfers and the Effectiveness of Public Income Redistribution in the Phil-
   ippines." In Dominique van de WaJle and Kimberly Nead, eds., Public Spending and the Poor:
   Theory and Evidence. Baltimore: Johns Hopkins University Press.
        -. 1997. "Coping with Apartheid: Inter-Household Transfers over the Life-Cycle in South
   Africa." Working paper. Boston College and World Bank. Processed.
        -. 1998. "Risk-Sharing and Private Transfers: What about Urban Households?" Economic
   Development and Cultural Change 46(3):621 -39.
Czukas, Katherine, Marcel Fafchamps, and Christopher Udry. 1998. "Drought and Saving in West
   Africa: Are Livestock a Buffer Stock?" Journal of Development Economics 55(2):273—305.
Das Gupta, Monica. 1987. "Informal Security Mechanisms and Population Retention in Rural In-
   dia." Economic Development and Cultural Change 13(March): 101—20.
Dasgupta, Partha. 1993. An Inquiry into Well-Being and Destitution. Oxford: Clarendon Press.
Deaton, Angus. 1992. Understanding Consumption. Oxford: Clarendon Press.


2O4                                           The U arid Bank Raearth Obsmrr. i oL 14. no. 2 'August 1999)
         . 1997. The Analysis of Household Surveys: A Microeconomic Approach to Development Policy.
    Baltimore: Johns Hopkins University Press.
Fafchamps, Marcel. 1992. "Solidarity Networks in Preindustrial Societies: Rational Peasants with a
    Moral Economy." Economic Development and Cultural Change 41 (October): 147-73.
Foster, Andrew, and Mark Rosenzweig. 1999. "Imperfect Commitment, Altruism, and the Family:
    Evidence from Transfer Behavior in Low-Income Areas." University of Pennsylvania, Depart-
    ment of Economics, Philadelphia. Processed.
Gertler, Paul, and Jonathan Gruber. 1997. "Insuring Consumption against Illness." NBER Working
    Paper 6035. National Bureau of Economic Research, Cambridge, Mass.
Haddad, Lawrence, and Manfred Zeller. 1996. "How Can Safety Nets Do More with Less? General
    Issues with Some Evidence from Southern Africa." FCND (Food, Consumption and Nutrition
    Department) Discussion Paper 16. International Food Policy Research Institute, Washington,
    D.C. Processed.
Hoddinott, John, and Bill Kinsey. 1998. "Child Growth in the Time of Drought." Draft paper.
    International Food Policy Research Institute, Washington, D.C. Processed.
HofiF, Karla. 1997. "Informal Insurance and the PovertyTrap." Draft paper. University of Maryland,
    Economics Department, College Park, Md. Processed.
Jacoby, Hanan, and Emmanuel Skoufias. 1997. "Risk, Financial Markets, and Human Capital in a
    Developing Country." Review ofEconomic Studies 64(3):311-36.
Jalan, Jyorsna, and Martin Ravallion. 1997. "Are the Poor Less Well Insured? Evidence on Vulner-
   ability to Income Risk in Rural China." Policy Research Working Paper 1863. World Bank, De-
   velopment Research Group, Washington, D.C. Processed.
Jensen, Robert. 1998. "Public Transfers, Private Transfers, and the 'Crowding Out' Hypothesis:
    Evidence from South Africa." John F. Kennedy School of Government Faculty Working Paper.
   Harvard University, Cambridge, Mass. Processed.
Khandker, Shahidur. 1998. Fighting Poverty with Microcredit: Experience in Bangladesh. New York:
   Oxford University Press.
Kipnis, Andrew B. 1997. Producing Guanxi: Sentiment, Self, and Subculture in a North China Village.
    Durham, N.C.: Duke University Press.
Kletzer, Kenneth, and Brian Wright. 1998. "Sovereign Debt as Intertemporal Barter." Working Pa-
   per C98-100. University of California, Berkeley, Center for International and Development Eco-
   nomics Research, Berkeley, Calif. Processed.
Kochar, Anjini. 1999. "Smoothing Consumption by Smoothing Income: Hours-of-Work Responses
   to Idiosyncratic Agricultural Shocks in Rural India." Review of Economics and Statistics 81(Febru-
   ary):50-6l.
La Ferrara, EliaM. 1997. "Ethnicity and Reciprocity: An Analysis of Credit Transactions in Ghana."
   Working paper. Harvard University, Economics Department, Cambridge, Mass. Processed.
Ligon, Ethan, Jonathan Thomas, and Tim Worall. 1997. "Informal Insurance Arrangements in Vil-
   lage Economies." University of California, Berkeley, Department of Agricultural and Resource
   Economics. January. Processed.
Lim, Youngjae, and Robert M. Townsend. 1998. "General Equilibrium Models of Financial Sys-
   tems: Theory and Measurement in Village Economies." Review of Economic Dynamics 1 (Janu-
   ary):59-118.
Lucas, Robert E. B., and Oded Stark. 1985. "Motivations to Remit: Evidence from Botswana." Jour-
   nal ofPolitical Economy 93(5) :901-18.
Lund, Susan, and Marcel Fafchamps. 1997. "Risk-Sharing Networks in Rural Philippines." Stanford
   University, Department of Economics, Stanford, Calif. Processed.


Jonathan Morduch                                                                         205
Malinowski, Bronislaw. 1922. Argonauts ofthe Western Pacific. London: Roudedgc.
Mauss, Marcel. 1967. The Gift. New York; W. W. Norton.
MicroBanking Bulletin. 1998. Boulder, Colo.: Economics Institute. July.
Morduch, Jonathan. 1991. "Consumption Smoothing across Space." Harvard University, Depart-
   ment of Economics, Cambridge, Mass. Processed.
        . 1993. "Risk, Production, and Savings: Theory and Evidence from Indian Households."
   Harvard University, Department of Economics, Cambridge, Mass. Processed.
        . 1994. "Poverty and Vulnerability." American Economic Review 84(May):221-25.
       -. 1995. "Income Smoothing and Consumption Smoothing." Journal of Economic Perspectives
   9(Summer): 103-14.
       -. 1998. "Does Microfinance Really Help the Poor? Evidence from Flagship Programs in
   Bangladesh." Stanford University, Hoover Institution, Stanford, Calif. Processed.
        . Forthcoming. "The Microfinance Promise." Journal of Economic Literature.
Morduch, Jonadian, and Terry Sicular. 1999. "Risk and Insurance in Transition: Perspectives from
   Zouping County, China." In Masahiko Aoki and Yujiro Hayami, eds., Market and Community in
   Economic Development. New York: Oxford University Press.
Newbery, David, and Joseph Stiglitz. 1981. The Theory of Commodity Price Stabilization. Oxford:
   Oxford University Press.
Paulson, Anna. 1995. "Insurance Motives for Migration: Evidence from Thailand." Princeton Uni-
   versity, Princeton, N.J. Processed.
Platteau, Jean-Philippe. 1991."Traditional Systems of Social Security and Hunger Insurance." In E.
   Ahmad, J. Dreze,J. Hills, and A. Sen, eds., Social Security in Developing Countries. Oxford: Clarendon
   Press.
        . Forthcoming. "Traditional Sharing Norms as an Obstacle to Economic Growth in Tribal
   Society." In Institutions, Social Norms and Economic Development. Chur, Switzerland: Harwood
   Academic Publishers.
Platteau, Jean-Philippe, and Anita Abraham. 1987. "An Inquiry into Quasi-Credit Contracts: The
   Role of Reciprocal Credit and Interlinked Deals in Small-Scale Fishing Communities."/o«r7w/o/'
   Development Studies 23(4):46l-90.
Ravallion, Martin. 1991. "Reaching the Rural Poor through Public Employment: Arguments, Evi-
   dence, and Lessons from South Asia." World Bank Research Observer 6(2): 153-75.
Ravallion, Martin, and Shubham Chaudhuri. 1997. "Risk and Insurance in Village India: Com-
   ment." Econometrica 65(0:171-84.
Reardon, Thomas, Peter Marion, and Christopher Delgado. 1988. "Coping with Household-Level
   Food Insecurity in Drought-Affected Areas of Burkina Faso." World Development 16(9): 1065—74.
Rose, Elaina. 1999. "Consumption Smoothing and Excess Female Mortality in Rural India." Review
   ofEconomics and Statistics 81 (February):4l-49.
Rosenzweig, Mark. 1988. "Risk, Implicit Contracts, and the Family in Rural Areas of Low-Income
   Countries." Economic Journal 98(,December):\ 148-70.
Rosenzweig, Mark, and Hans Binswanger. 1993. "Wealth, Weather Risk, and the Composition and
   Profitability of Agricultural Investments." Economic Journal 103(January):56-78.
Rosenzweig, Mark, and Oded Stark. 1989. "Consumption Smoothing, Migration, and Marriage:
   Evidence from India." Journal ofPolitical Economy 97(4):905-26.
Rosenzweig, Mark, and Kenneth Wolpin. 1993. "Credit Market Constraints, Consumption Smooth-
   ing, and the Accumulation of Durable Production Assets in Low-Income Countries: Investment
   in Bullocks in India." Journal of Political Economy 101(April):223-44.


9O6                                              The World Bank Ratmrch Observer, voL 14, no. 2 (August 1999)
Rutherford, Stuart. 1999. The Poor and Their Money. Finance and Development Research Programme
  Working Paper Series 3/1999. Manchester, U.K.: University of Manchester, Institute for Devel-
  opment Policy and Management.
Subbarao, K., Aniruddha Bonnerjee, Jeanine Braithwaite, Soniya Carvalho, Kene Esemenari, Carol
  Graham, and Alan Thompson. 1997. Safety Net Programs and Poverty Reduction: Lessons from
   Cross-Country Experience. Washington, D.C.: World Bank.
Townsend, Robert. 1994. "Risk and Insurance in Village India." Econometrica 62:539-92.
        . 1995a. "Consumption Insurance: An Evaluation of Risk-Bearing Systems in Low-Income
  Countries." Journal of Economic Perspectives 9:83—102.
       -. 1995b. "Financial Systems in Northern Thai Villages." QuarterlyJournal of Economics 110(No-
  vember): 1011-46.
Von Pischke, J. D. 1991. Finance at the Frontier: Debt Capacity and the Role of Credit in the Private
  Economy. EDI Development Study. Washington, D.C.: World Bank.
Walker, Thomas, and James Ryan. 1990. Village and Household Economies in India's Semi-Arid Trop-
   ics. Baltimore: Johns Hopkins University Press.
Walker, Thomas, R. P. Singh, and M. Asokan. 1986. "Risk Benefits, Crop Insurance, and Dryland
  Agriculture." Economic and Political Weekly 21(June):A81-A88.
World Bank. 1994. "Zambia Poverty Assessment," vol. 5 (Participatory Poverty Assessment). Wash-
  ington, D.C.
Yaron, Jacob, McDonald P. Benjamin, and Gerda L. Piprek. 1997. Rural Finance: Issues, Design, and
  Best Practices. Environmentally and Socially Sustainable Development Studies and Monograph
  Series 14. Washington, D.C: World Bank




Jonathan Morduch                                                                           307