CHAPTER 9

Managing Economic
Crises and Natural
Disasters
There is nowhere to work. We get sick and we don’t dren may rise during economic crises and natural
have the money to get cured, we don’t have medicines disasters. Poor households are often forced to sell
because they are expensive. The government makes their meager assets at depressed prices. These re-
everything expensive. . . . We don’t have money to buy sponses perpetuate chronic poverty, possibly reducing
fertilizers, seeds, everything is in dollars. We don’t future economic growth because of the irreversible
have anything to eat. Everything is so expensive. losses in human and physical capital. That is why
—From a discussion group of adult women at preventing economic crises and natural disasters is so
the time of a banking crisis, Juncal, Ecuador crucial. And that is why, when they occur, among the
top priorities should be to protect poor people. Re-
[Security is] . . . the ability of persons to cope with quired for that protection are not only resources but
disasters. also the instruments (safety net programs) to channel
—From a discussion group, Little Bay, Jamaica those resources to poor households. While develop-
ing countries and transition economies in general are

E
vulnerable to crises and natural disasters, small states
are especially vulnerable to adverse external events
conomic crises and natural disasters can bring because of their remoteness and isolation, high degree
deep and sudden collapses in national output—and sharp of openness, susceptibility to natural disasters, and lim-
increases in income poverty. Together with violent con- ited diversification.1
flicts (see box 3.2 in chapter 3), they are great sources of
vulnerability and insecurity. Worse, because of the collateral Preventing and coping with
damage they cause, such as irreversible loss of human cap- economic crises
ital, they affect not only the current living standards of
poor people but their ability to escape from poverty as well. Even our limited access to schools and health is now
Malnutrition and dropout rates among poor chil- beginning to disappear. We fear for our children’s
161
     ⁄ 

future. . . . What is the justice in sending our children to

the garbage site every day to support the family? external shocks (terms of trade shocks, higher international
—Mother and father commenting on need to pull their chil- interest rates, sudden movements in capital flows as a re-
dren from school in the wake of economic crisis, Thailand sult of contagion). In such circumstances restrictive fis-
cal and monetary policies are inevitable and less costly than
Economywide crises entail sharply falling output, de- the alternative of delaying such measures, which could
clining incomes, and rising unemployment. Pervasive in lead to a larger crash.
the 1990s, they came in different forms: fiscal crises, Once adjustment policies are accepted as inevitable,
balance of payments crises, terms of trade shocks, currency the way governments introduce fiscal austerity can
crises, banking crises, hyperinflation. The economic crises worsen the adverse effects on the living standards of the
in Mexico in 1995, in East Asia in 1997, and in Brazil poor and near-poor. For example, removing food or
and Russia in 1998 received wide media coverage. But fuel subsidies would exacerbate the effects on poor peo-
they were not the only episodes of economic distress.2 ple—unless compensatory measures are taken (chapter
Most crises have been brought on by varying combina- 4). So would increasing the rates and sometimes the cov-
tions of policy mismanagement and such external factors erage of indirect taxes on food and other products that
as terms of trade shocks, volatile capital flows, and con- figure large in the consumption basket of poor people.
tagion in international capital markets. Net government transfers may decline as governments
Economic crises hurt both the poor and the nonpoor, cut social assistance as part of a fiscal austerity program.
but they are far more devastating for those already in Reducing the quantity and quality of public services
poverty or nearly poor, even if they are not hurt dis- used by the poor and near-poor would also worsen their
proportionately. The welfare losses are larger for poor situation.
households and those who fall into poverty than for the But government actions can also mitigate the impact
rest of the population. Poor people are unlikely to have of crises on poor people. The task of the policymaker
enough savings or self-insurance to see them through is to implement the combination of macroeconomic
bad times, and they have little or no access to insur- measures that results in the lowest cost in forgone out-
ance schemes, whether social or market based (chap- put and affords the greatest protection to the living
ter 8). standards of poor people. A key element of a poverty-
An economic crisis affects the living standards of poor sensitive response is the right composition of revenue-
people and those living close to poverty through differ- raising measures and fiscal cuts. A poverty-sensitive
ent channels: response should also allow for the expansion of safety
■ Typically, real wages fall and unemployment rises, nets targeted to poor people (the “social insurance”
driving down labor earnings. component of social spending) during periods of macro-
■ Nonlabor incomes fall as economic activity slows, economic adjustment.
and the prices of the goods and services produced by
poor people may fall relative to other prices. Social impact of crises
■ Private transfers, particularly from family members, are There is a strong link between macroeconomic down-
likely to shrink as living standards fall across the nation. turns and rising income poverty (table 9.1; see also fig-
■ The meager assets of poor people are exposed to in- ure 2.1).3 During crises many people become temporarily
flation or a collapse in prices. poor, and social indicators tend to worsen or to im-
■ Macroeconomic crises slow the accumulation of prove more slowly. Data suggest that the human capi-
human, financial, and physical capital, weakening the tal of poor people, particularly poor children, can
ability of poor people to escape poverty. deteriorate. The damage can be irreversible, affecting the
Is the observed fall in incomes during crises made ability of these children to escape poverty when they reach
worse by the policies to respond to the crises? The debate adulthood.
on this is long-standing. That rising poverty coincides with In most countries in East Asia poverty rose as a result
the policy responses does not mean that the policies of the financial crises of the late 1990s: it is estimated that
caused the rise. Crises can occur because of past unsus- it rose almost 50 percent in Indonesia and that urban
tainable macroeconomic policies or inability to adjust to poverty doubled in the Republic of Korea.4 In both
       

Table 9.1
Effect of economic crises on incidence of poverty in selected countries
Percent

Before Year After

Country and type of crisis crisis of crisis crisis
Argentina, hyperinflation and currency 25.2 47.3 33.7
(1987) (1989) (1990)
Argentina, contagion 16.8 24.8 26.0
(1993) (1995) (1997)
Indonesia, contagion and financial 11.3 18.9 11.7
(1996) (1998) (1999)
Jordan, currency and terms of trade 3.0 .. 14.9
(1986–87) (1989) (1992)
Mexico, currency and financial 36.0 .. 43.0
(1994) (1995) (1996)
Russian Federation, financial 21.9 32.7 ..
(1996) (1998)
Thailand, currency and financial 11.4 12.9 a ..
(1996) (1998)

.. Not available.
Note: Based on national poverty lines and per capita household income except for Indonesia (per capita expenditure), Mexico (household
income), and Russia (household expenditure per equivalent adult). Data for Argentina refer to Greater Buenos Aires. For Indonesia poverty
estimates before and during the crisis are based on the full SUSENAS (the national socioeconomic survey) conducted in February 1996 and
1999; estimates after the crisis are based on a smaller sample. Figures are not comparable across countries because poverty lines differ.
a. Based on the socioeconomic survey conducted between February 1998 and January 1999, which does not fully reflect the impact of the
crisis. Estimates from a smaller survey conducted during June–September 1999 put the poverty incidence at 15.9 percent.
Source: Ministerio de Economía de Argentina 1998; World Bank 1994c, 1999dd; ECLAC 1999b; Lokshin and Ravallion 2000b.

countries, however, poverty fell as the economies recov- rose, sometimes substantially.6
ered. In Russia the incidence of poverty rose from 21.9 The impact of economic crises on living standards is
percent to 32.7 percent between 1996 and 1998. In not fully captured by measures of inequality and income
every crisis in Latin America and the Caribbean the in- poverty. Economic crises are characterized by extensive
cidence of poverty increased and several years later re- mobility: previously nonpoor people may fall into poverty,
mained higher than it had been before the crisis. and previously poor people may escape it. Evidence of
Inequality may rise, fall, or remain unchanged during sharp downward and upward mobility was found after
a crisis. In Latin America inequality (as measured by the the 1998 crisis in Russia, for example.7 Mean expendi-
Gini coefficient) rose in 15 of 20 crisis episodes for which tures of people classified as poor in 1996 actually rose,
there are data. In East Asia during the recent crisis, how- and 42 percent of them escaped poverty after the crisis.
ever, inequality remained practically unchanged, and in By contrast, 61 percent of those who were poor after the
Mexico following the peso crisis in 1995 it fell. When crisis had not been poor in 1996. Put another way, 20
crises are accompanied by increases in inequality, economic percent of the population fell into poverty as a result of
contractions can more than reverse previous gains in the economic downturn. Even though overall inequality
poverty reduction. In Latin America the poverty reduc- fell and a large share of the poor escaped poverty after the
tion from a 3.7 percent increase in per capita income for crisis, there was substantial downward mobility for many
urban areas and a 2 percent increase for rural areas in the who were not previously poor and for some who were al-
1970s was reversed by just a 1 percent decline in per capita ready poor. Those who become poor during economic
income in the 1980s.5 Even if inequality increases, the crises often have different characteristics than the chron-
poorest fifth of the population is not always hurt dis- ically poor. For example, they may be better educated. A
proportionately. In Latin America the income share of the study in the Philippines found that households with
middle fifths of the population often fell most during the more education are more vulnerable to wage and em-
1980s debt crisis, but the share of the top tenth always ployment shocks.8
     ⁄ 

Table 9.2
Social impacts of economic crises in selected countries

Main crisis indicators Health indicators Education indicators

Argentina ■ Per capita GDP fell 4.1%. ■ Per capita daily protein intake ■ Growth in gross primary
1995 ■ Per capita private consumption fell 3.8% in 1995, but enrollment declined from
fell 5.6%. increased 1.9% in 1996. 2.2% in 1993 to 0.8% in
1996.

Mexico ■ Per capita GDP fell 7.8%. ■ Among children under age ■ Gross primary enrollment
1995 ■ Per capita private consumption 1, mortality from anemia increased 0.44% in 1994,
fell 11.1%. increased from 6.3 deaths but fell 0.09% in 1995.
per 100,000 live births in
1993 to 7.9 in 1995.
■ Among children ages 1–4,
the mortality rate from anemia
rose from 1.7 to 2.2 per 100,000.

Indonesia ■ Per capita GDP fell 14.6%. ■ The share of women whose ■ The dropout rate for children
1998 ■ Per capita private consumption body mass index is below the in the poorest fourth of the
fell 5.1%. level at which risks of illness population rose from 1.3% in 1997
and death increase rose 25%. to 7.5% in 1998 for those ages
■ Most indicators of child 7–12 and from 14.2% to 25.5%
nutritional status remained for those ages 13–19. In both
stable. The exception may cohorts the poorest fifth
be the weight (conditional on experienced the largest
height) of children under age increase.
3, suggesting that families may ■ The share of children
be investing in some members in the poorest fourth of the
at the expense of others. population not enrolled in
school rose from 4.9% in 1997
to 10.7% in 1998 for those
ages 7–12 and from 42.5% to
58.4% for those ages 13–19. In
both cohorts the poorest fourth
had the largest increase.

Note: Gross enrollment ratios are used because net ratios were not available. These data should be used with caution.
Source: World Bank 1999cc; IDB Statistical and Social Database; PAHO 1998; Thomas 1999; Frankenberg, Thomas, and Beegle 1999.

Most social indicators either deteriorate or improve at increase rose by a quarter in 1998, and the average weight
a slower pace during a macroeconomic crisis (table 9.2). of children under age three declined.
Social indicators such as infant mortality rates continued School attendance and literacy also take hits during
to improve in Latin America in the 1980s, though more crises. In the Philippines secondary school enrollments
slowly than in the previous decade. But health indicators increased only 0.9 percent between the 1997/98 and
more sensitive to consumption or income downturns 1998/99 academic years, after growing at an average an-
worsened. In Chile the share of low-birthweight infants nual rate of 2.6 percent in the previous five years. In Mex-
and undernourished children rose as the economy de- ico the proportion of each graduating class that enrolled
clined. In Mexico infant and preschool mortality caused in the next education level declined during the 1980s debt
by nutritional deficiency rose in the 1980s, reversing the crisis, particularly among high school and university stu-
trend of the previous decade, and rose again with the eco- dents. The percentage of age-appropriate children entering
nomic crisis of 1995. In Argentina and Venezuela the daily primary school also declined. In rural areas the dropout
per capita intake of protein declined as per capita GDP rate rose by 40 percent. In Argentina and Mexico growth
fell. In Indonesia the share of women whose body mass in gross primary enrollment slowed in 1995. A study for
index is below the level at which risks of illness and death South India found that children are often taken out of
       

school in response to adverse shocks.9 example, ending hyperinflation brought about a signif-
There is no question that economic crises increase tran- icant one-time drop in the incidence of poverty: in
sitory poverty.10 They can also increase persistent or chronic Greater Buenos Aires the incidence of poverty dropped
poverty because of hard-to-reverse effects on the human from 34.6 percent in 1989 to 22.6 percent in 1991.15
capital of poor people. While the trends cited for malnu- The 1990s saw various types of crises, triggered by
trition, infant mortality, and enrollment are national av- weak banking systems and weak financial regulation in
erages, they most likely reflect a deterioration in these a world of large and volatile international capital flows.
indicators among poor people. For Indonesia, information Liberalizing the financial sector was expected to put
by income group shows that the dropout rate in the low- economies on a more stable footing. But the transition
est fourth of the income distribution rose from 1.3 per- from more repressed to more open financial systems in
cent in 1997 to 7.5 percent in 1998 among children ages the developing world has been difficult to manage.
7–12 and from 14.2 percent to 25.5 percent among those Banking crises have been more numerous in the past two
ages 13–19. The proportion of poor children not enrolled decades, when stroke-of-the-pen financial liberalization
in school increased from 4.9 percent to 10.7 percent. became popular.16
Recent research shows a link between macroeco- Some of the reforms introduced in the financial sec-
nomic downturns and education indicators. The aver- tor backfired because the institutional rules allowed ex-
age annual increase in years of schooling in 18 Latin cessively risky behavior while the costs of that behavior
American countries fell from 1.9 years in the 1950s and had to be paid by society as a whole. A vivid example is
1960s to 1.2 in the 1970s and 1980s. Worsening macro- the Mexican financial crisis of 1995.17 At the root of the
economic conditions (short-term GDP shocks, volatil- crisis was a weak banking system, its fragility traced to
ity, and adverse trade shocks) explain 80 percent of the the privatization process used for the banks, some aspects
decline, according to one study.11 As evidence from of the financial liberalization program, and weak regu-
Mexico shows, the negative “income effect” of falling in- latory institutions. Rescuing the banking sector will cost
come tends to outweigh the positive “price effect” of the Mexican taxpayers an amount equal to about 20 percent
lower opportunity cost of attending school.12 Simulation of GDP (in present value terms).
results suggest that the gross secondary enrollment rate To prevent financial crises, governments need to im-
in Mexico would have been 11 percentage points higher prove the prudential regulation and supervision of financial
in 1991 if the economy had grown during the 1980s at intermediaries, introduce new standards for data dis-
half the rate of the 1970s. semination, and implement corporate bankruptcy re-
form.18 These measures are already under way in many
Avoiding crises developing countries, but there is still a long way to go.
Clearly, avoiding crises should be a top priority in any anti- At the same time, a cautious approach should be taken
poverty strategy. There is wide agreement on the kind of to capital account liberalization. Controls on capital
macroeconomic and financial policies governments need inflows—such as those Chile used until recently—can be
to reduce vulnerability to policy-induced crises or adverse an appropriate instrument for tempering the volatility of
external shocks.13 They should avoid profligate fiscal capital flows. There is evidence that capital controls can
and monetary policies, overvalued exchange rates, and un- shift the composition of capital flows toward longer-
sustainable current account deficits—all problems in the maturing investments.19
1970s and 1980s. Other initiatives and measures are also important for
Many parts of the world have made great progress in avoiding crises, such as mechanisms to diversify and in-
steering away from irresponsible fiscal policy. Leading ex- sure against risk. Some governments, such as Chile, self-
amples are the large economies in Latin America and some insure using fiscal stabilization funds. Others, such as
of the transition economies, where the ensuing fall in in- Argentina, negotiate contingent credit lines between the
flation rates has helped build investor confidence and re- central bank and private international financial institu-
duced, if not eliminated, the potential long-term effects tions to ensure access to foreign currency in the event of
of inflation on efficiency and growth.14 Lower inflation a sudden slowdown in capital inflows.20
has also helped reduce poverty, since high inflation often However, actions at the national level may not be
hurts the poor more than the nonpoor. In Argentina, for enough to prevent economywide crises. Domestic actions
     ⁄ 

will have to be complemented by actions at the interna- omy crashed, with per capita GNP falling by 13.4 per-
tional level to foster global financial stability (chapter 10) cent and real wages by 40.6 percent. Altogether, real
and help countries, particularly the poorest and the small- wages fell by 67 percent between 1988 and 1990.21
est, manage commodity price shocks. The 1997 crisis in Thailand shows what happens when
there are no corrective measures to address the buildup of
Formulating a crisis response that protects vulnerability.22 True, the financial panic of domestic and
poor people international investors suddenly concerned about the fate
No matter how skillful the economic management, crises of their portfolios lit the fuse for the explosion. But the
are likely to affect the developing world and transition buildup of structural vulnerabilities provided the
economies for some time to come. That is why articu- dynamite—sharply rising short-term debt that far ex-
lating a response to crises must take into consideration ceeded international reserves, a financial sector that had done
its impact on poor people. A poverty-sensitive response a poor job of intermediating capital inflows and found it-
to crisis should steer toward: self saddled with hugely mismatched assets and liabilities,
■ Helping poor households maintain their consumption. and corporations that were massively overleveraged and ex-
■ Ensuring that poor people do not lose whatever ac- posed to interest and exchange rate fluctuations.
cess they have to basic social services. Not all problems arise from a failure to adjust to an
■ Preventing permanent reversals in the accumulation adverse shock or from unsound macroeconomic policies.
of human and physical capital. In some cases the policy response errs in the direction of
■ Averting self-defeating behavior, such as criminal ac- too much adjustment, with fiscal and monetary policy
tivity, prostitution, and exploitative forms of child more restrictive than necessary to restore equilibrium in
labor. the currency market, the current account, or the capital
A poverty-sensitive response should also provide mech- account. Overreaction can cause more pain than neces-
anisms for those at risk of becoming poor as a result of sary and in some circumstances can be self-defeating. An
the crisis. initial overreaction on the fiscal front can lead to a higher
What does it take to protect those who are already poor fiscal deficit down the road because the larger-than-
and those at risk of becoming so from sharp declines in expected recession lowers government revenues, defeat-
short-term income? Appropriate macroeconomic re- ing the purpose of the initial austerity measures. The
sponses and well-functioning safety nets can enhance reason for overshooting often is that cautious policy-
equity and result in better growth outcomes. Some of the makers prefer to err on the side of excessive adjustment,
recommendations here are already being incorporated in since timid adjustment can be far more devastating.
the standard approach for dealing with crises. The Re- Although it may be hard to tell whether a policy pack-
public of Korea, for example, introduced or expanded age is excessively restrictive, there are some indications
safety nets relatively quickly in the wake of the 1997 fi- that those in place in East Asia during the recent crisis were
nancial crisis (box 9.1). But in general the response con- just that. In Thailand the tax increase in September 1997
tinues to be ad hoc—with measures thrown together in made the ensuing recession worse. In Korea the restrictive
the heat of a crisis. fiscal policy initially made room for the expected costs of
Adopting the right macroeconomic policy mix. Re- bank restructuring. But the fiscal target was subsequently
sponding with the right macroeconomic policy mix after relaxed as both the authorities and the international financial
an adverse shock is one of the biggest challenges policy- institutions recognized that it was unrealistic in light of the
makers face. Driven by political considerations, policy- larger-than-expected slowdown in growth. Aiming toward
makers may postpone needed adjustment and stabilization the original target in the face of worsening economic con-
measures because they are painful—making the situation ditions would have been self-defeating. And for Malaysia
far worse. Peru was an extreme case in the 1980s. The gov- and the Philippines the trend of cyclically adjusted deficits
ernment refused to implement an adjustment program (for both revenues and expenditures) suggests that they did
and in July 1985 announced a cap on external debt pay- not relax their fiscal policy, even though the actual deficit
ments (a de facto unilateral moratorium) equal to 10 per- made it look as though they had.
cent of exports. Peru did well for a while, but the Even if excessively restrictive policies are later cor-
disequilibria continued to mount and in 1988 the econ- rected, the short-term costs can be significant, particu-
       

Box 9.1
Providing social protection in response to crisis in the Republic of Korea

The Republic of Korea was making sustained progress in reduc- in January 1998 to 174,000 in March 1999, still only 10 per-
ing poverty in the 1990s: the urban poverty rate fell an average cent of the unemployed workforce.
20 percent a year during 1990–97, and there were no increases ■ Public work. Since most of Korea’s jobless did not benefit from
in inequality. But then economic crisis struck, sharply increasing the expansion of unemployment insurance, the government in-
unemployment and poverty. The incidence of poverty in urban areas troduced a temporary public work program in May 1998, en-
doubled from 9 percent in 1997 to 19.2 percent in 1998.1 Unem- rolling 76,000 workers. By January 1999 the program was
ployment rose from 2.6 percent in the second quarter of 1997 to providing 437,000 jobs, though the number of applicants was
a peak of 8.7 percent in early 1999. Real wages declined 20.7 per- higher still, at 650,000. By the first quarter of 1999 the public
cent. Most of the newly unemployed were low-paid workers: in work program was benefiting around 2.5 times as many peo-
December 1998 three-quarters were temporary, daily, self- ple as the unemployment insurance program.
employed, or unpaid family workers, and about 20 percent were ■ Livelihood protection. In May 1998 the government introduced
the head of a household with no other income earners. a temporary livelihood protection program, with funding to
Expansionary fiscal policies in 1998 and 1999 were critical in cover 750,000 beneficiaries. It also introduced a means-tested
stemming the economic downturn. Social protection spending noncontributory social pension for 600,000 elderly people.
was increased threefold—from 0.6 percent in 1997 to 2.0 per- Although the government’s social protection response was quite
cent in 1999. The government used three main instruments of exemplary, public spending on health and education did not increase
social protection to help the unemployed, the poor, and the in line with the overall budget, and real spending either fell or re-
elderly: mained constant. But even within the smaller envelope for health,
■ Unemployment insurance. Korea expanded its nascent un- spending on primary care was protected.
employment insurance program—the only such program The government is now focusing on consolidating social safety
among the East Asian crisis countries—from firms with more nets, reducing income disparities, and creating the basis for a com-
than 30 employees to all firms. It also included temporary and petitive and knowledge-based economy. Policies to achieve these
daily workers, shortened the contribution period required for objectives include a law guaranteeing a minimum standard of liv-
eligibility, and extended the duration of unemployment bene- ing, to take effect in October 2000. The law will entitle all Koreans
fits. This expanded the eligible workforce from 5.7 million living under the poverty line to receive income support for living, ed-
workers at the beginning of 1998 to 8.7 million at the end of ucation, and housing expenses. Nearly 2 million poor people are ex-
the year. Beneficiaries increased tenfold—from around 18,000 pected to benefit, four times the current number.

1. The poverty rates were calculated using seasonally adjusted expenditure data and a national poverty line equivalent to about $8 a day (in 1993 PPP dollars).
Source: World Bank 1999w, 2000d.

larly for poor people. If there are vicious cycles of poverty, higher for poor people. Moreover, because poor people live
low education, and poor health, a recession can cause per- close to the subsistence level, their preference may reflect
manent damage for the poor. application of the safety principle (minimizing the prob-
Do the macroeconomic responses to crises that are best ability that their income will fall below a certain level). Or
for the overall output levels of the economy differ from the poor and nonpoor may discount future consumption
those that would be best for the incomes of poor people? differently, with the poor putting a larger premium on pre-
Perhaps. Different policy combinations imply different sent consumption than the nonpoor.
costs for the poor than for the nonpoor because of the The distributive and intertemporal implications of al-
way the reduction in per capita output is distributed.23 ternative adjustment policies are important, but policy-
But even if distributive outcomes were the same, the makers rarely have the luxury of choosing among different
poor and the nonpoor could well prefer different policy adjustment paths. In general, the optimal combination of
packages.24 policies—to achieve the necessary balance of payments ad-
Poor people are more likely to prefer an adjustment that justment with the smallest decline in output—depends
leads to the smallest drop in GDP at any point in time on initial conditions.25 When a currency is under specu-
even if it implies a slower recovery. Nonpoor people are lative attack, a spike in interest rates will in most cases be
more likely to prefer a program that reduces income more needed to stop the attack. But when a country intro-
severely in the short run but yields higher growth in the duces adjustment measures early on, the government may
medium run. This difference results simply from the fact have more freedom to choose among different policy
that the welfare losses from an economic downturn are combinations and thus be more likely to manage a soft
     ⁄ 

landing. Unfortunately, macroeconomic analysis in its to poor people should be protected if not expanded.
current state can offer little guidance in assessing the dis- It may seem obvious that governments should protect
tributive and intertemporal implications of alternative spending that benefits poor people and expand the safety
policy packages, clearly an area in need of far more ana- net programs targeted to them. But this does not neces-
lytical and applied research. sarily happen in practice. Recent research in some coun-
Protecting spending that benefits poor people. How gov- tries in Latin America has found that a 1 percent decline
ernments raise revenues and cut public (nondebt) spend- in per capita GDP leads to an estimated 2–3 percent de-
ing has important policy implications for who bears the cline in targeted public spending per poor person.27 And
burden of adjustment and whether poor people are pro- a study on the Argentine employment program Trabajar
tected.26 To design a poverty-sensitive fiscal adjustment found that its performance in reaching poor people de-
to avoid or respond to a crisis, policymakers need to as- teriorated sharply with cuts to its budget.28
sess the distributional effects of spending programs. A use- There may be several reasons for such “antipoor” pat-
ful tool for this is the public expenditure review (box 9.2). terns in fiscal adjustment. Without budgetary guidelines
As a general rule, areas important for poor people— to direct fiscal austerity, governments may go for pro-
basic education, preventive health care, water and sani- portional cuts to minimize bureaucratic infighting and ease
tation, rural infrastructure—should be protected from acceptance by the legislature. Another reason may be that
budget cuts to ensure that services are adequate. That governments lack the instruments to target resources to
means ensuring that schools and health posts in poor areas the poor—instruments that are difficult to put in place
have at least the basic minimum of supplies. General sub- in the heat of a crisis. Even if the instruments exist, po-
sidies on food staples might need to be maintained in the litical forces may be such that the resources going to poor
short run—even if the benefits leak to the nonpoor— people are cut more than proportionately. In some coun-
unless they can be effectively replaced by targeted pro- tries information can be the major constraint: governments
grams. Safety nets and social assistance programs targeted may lack reliable records of their budget or programs.
What can be done to counter these factors? One way
Box 9.2 to protect spending that benefits poor households is for
Public expenditure reviews to assess the impact the government and legislature to rank current programs
of fiscal retrenchment on poor people by their importance as part of the budget approval process.
When spending cuts are needed, the order of the cuts
Public expenditure reviews—assessments of public sector would be determined by the priority assigned to each pro-
issues that focus on the efficiency and rationale of the public
budget—could be useful tools for evaluating the impact of
gram. Government agencies could be required to evalu-
fiscal adjustment programs and public sector reforms on ate social programs to help policymakers identify those
social programs and safety nets. In economywide crises that are most cost-effective in reducing poverty and there-
that lead to spending cuts, these reviews could help estab- fore should be protected during a crisis.
lish a transparent budget mechanism for rationalizing, allo-
cating, executing, and managing public spending to protect
Peru has introduced guidelines for protecting pro-
poor people and ensure private sector efficiency. grams that benefit poor people as part of its public finance
Public expenditure reviews typically analyze and project reform law (box 9.3). The guidelines combine fiscal rules
public revenues and determine the level and composition of with measures to increase fiscal transparency and ac-
public spending, assessing the allocation of resources among
and within sectors. When planning fiscal retrenchment, a
countability. The program creates a stabilization fund with
short review should be done, focusing on the sectors that ac- the proviso that programs benefiting poor people should
count for the bulk of the public budget (agriculture, education, be protected. Although such budget protocols may not
health, infrastructure). The review should rank expenditures be classified as antipoverty programs, they can have an
on social programs, considering the tradeoff between these
important effect on poverty by protecting pro-poor
programs and other nonessential spending (such as military
spending) that could be minimized during a crisis. This type spending during fiscal retrenchment.
of adjustment is clearly more efficient in protecting vulnera- If benefits targeted to poor people are cut for politi-
ble groups and maintaining private sector efficiency than the cal economy reasons, a third party—such as the multi-
typical uniform spending cut.
lateral lending organizations—could advocate for the
Source: World Bank 1999v.
poor and help governments implementing austerity mea-
sures design a viable way to protect programs and spend-
       

ing that benefit the poor. This happened to some degree One way to avoid this political economy constraint is to
in several countries in the 1990s. design programs with low marginal benefits or high mar-
Changes in the incentive system embedded in targeted ginal costs for the nonpoor.29
programs could also facilitate cuts for nonpoor benefi- Evaluating different types of spending can be difficult
ciaries during periods of austerity. The argument is this: when data are poor, the case in most developing coun-
it is often said that for political economy reasons some tries. Efficiency indicators are almost nonexistent, and data
of the benefits of targeted programs have to go to the non- on actual spending, as opposed to budgeted amounts, are
poor—through “leakage”—to ensure continuing sup- available only after long lags. Usually an evaluation should
port for programs. The same forces will presumably act take the available intermediate information and com-
to limit the welfare losses to the nonpoor from cuts. plement it to determine whether public resources reach
the intended beneficiaries effectively. A social monitor-
Box 9.3 ing and early response unit, such as the one set up in In-
Protecting poor people during fiscal adjustment: donesia during its recent crisis, can help ensure quick and
Peru’s Fiscal Prudence and Transparency Law reliable information for evaluating spending in specific
social programs.30 Where field surveys are infeasible (be-
Peru’s Fiscal Prudence and Transparency Law, overwhelm- cause of budget or time constraints), recent household sur-
ingly approved by the national congress in 1999, does much
to ensure that social protection is maintained during a fiscal
veys can be used to try to determine an efficient and
adjustment. rational allocation of government resources among social
First, the law established fiscal rules on the maximum an- programs and safety nets.
nual deficit of the consolidated public sector, capping it at 2 Putting safety nets in place before a crisis. If the problem
percent of GDP in 2000, 1.5 percent in 2001, and 1 percent
thereafter. (The consolidated public sector includes the cen-
is a lack of instruments for protecting poor people, the so-
tral and regional governments, decentralized agencies, and lution is to introduce, during normal times, safety net pro-
national public enterprises; it excludes local governments and grams that can operate as insurance in times of economic
their agencies and enterprises.) In the event of a national distress. Safety nets are important for several reasons.
emergency, international crisis, or fall in GDP, the fiscal
They can play a crucial role in mitigating the effects of crises
deficit can increase to 2 percent of GDP. The law also set
limits on increases in public spending and debt. The maxi- on the poor and protecting the near-poor from falling into
mum annual growth of nonfinancial public spending is equiv- poverty. A study estimated that if the targeted program
alent to the inflation rate plus 2 percentage points, implying Progresa (see box 5.5 in chapter 5) had existed when the
a future reduction in the relative size of the public sector.
1995 crisis hit Mexico, the poverty gap index in rural areas
Second, the law created a fiscal stabilization fund, to be
funded from three sources: the revenues above the average and the squared poverty gap index (which gives greater
collected during the previous three years, three-fourths of weight to the poverty of the poorest)31 would have declined
future privatization proceeds, and half of all revenues from fu- by 17 percent and 23 percent in the year after the crisis.32
ture concessions. (Savings accumulated in the fund in excess Safety nets can also help prevent irreversible damage to the
of 3 percent of GDP will be transferred to the public pension
fund or used to reduce public debt.) Up to 40 percent of the
human capital of poor people. And they can aid political
fund’s resources can be used in a given year if current rev- acceptance of stabilization and reforms, preventing con-
enues fall below the average collected over the previous flicts over resource distribution that can create stalemates,
three years. Fund resources can also be used in emergencies, deepen economic crises, even cause governments to fall.
such as an economic crisis or a natural disaster.
Third, the law mandates that the fund’s spending on tar-
Recent work has shown that institutional weaknesses, in-
geted poverty reduction programs be given priority over cluding lack of safety nets, have been responsible for
spending on other programs. many crises over the past 25 years.33
To enhance fiscal transparency, the law introduced a Most developing countries lack effective safety nets that
three-year fiscal framework to be developed, approved, and
published by the government. And to improve fiscal ac-
protect poor people from the output, employment, and
countability, it requires that the finance minister submit to price risks associated with systemic adverse shocks. When
congress and publish annual reports assessing the execution these mechanisms are not in place before a crisis occurs,
of the fiscal goals in the multiyear framework. policymakers are often forced to improvise or to use pro-
grams designed for other purposes and other beneficia-
Source: Ruprah 1999.
ries. Emergency responses to emergency situations are
often prepared without technical analysis to identify the
     ⁄ 

groups most vulnerable to the shocks and to evaluate the ures are, the full human and economic costs are even
cost-effectiveness of different social protection options. greater. Human costs include injuries and temporary and
Programs put in place and operating—even on a small permanent disabilities, temporary and permanent dis-
scale—before crises hit do better at protecting poor peo- placement of people, the breakup of families and social
ple than ad hoc emergency measures. networks, increased poverty and disease, and psycholog-
To be effective, safety nets should include a wide range ical scars. Economic costs, based largely on direct physi-
of programs—public work programs, scholarships for poor cal impacts or losses of fixed capital and inventory, are also
children, cash transfers, food-related transfers, food subsi- underestimated. Many indirect and secondary effects on
dies, social funds, and fee waivers for essential services economic activity—such as changes in fiscal policies, the
(chapter 8). Social programs that focus on long-term de- long-term consequences of the reallocation of investment
velopment (for example, such targeted human development resources, or the losses in human capital—go unrecorded.
programs as Mexico’s Progresa) can also perform a safety net Over the past 10 years the incidence of natural disas-
function during economic downturns. The appropriate ters has increased.36 This could be due in part to social fac-
mix of safety net programs will depend on the characteris- tors, as settlements have sprung up in hazardous areas. The
tics of the poor and vulnerable, the type of crisis, and the urban poor in megacities—for example, in Rio de Janeiro
government’s institutional and administrative capacity. and its favelas—are often forced to build on steep, mar-
The international community can play an important ginal land prone to landslides that kill or leave homeless
part by providing policy advice, contributing financial sup- thousands of people every year. But there are also natural
port, and helping policymakers design and fund safety factors. The El Niño events, associated with anomalous
nets. International financial institutions can help coun- floods, droughts, and storms, are getting larger and more
tries design pro-poor fiscal adjustment programs and frequent.37 And warming of the surface of the Atlantic is
safety nets and, for countries too poor to fund a safety increasing the frequency and severity of hurricanes.38
net during a crisis, can provide financing.34 Still, it is often asked whether it would be more correct
to label many of these disasters as “human-made” rather
Reducing vulnerability to natural than “natural.” They are probably both.
disasters
Impact of natural disasters on poor countries
The biggest shock we ever had was Hurricane Gilbert; . . . and poor people
all what we found after Gilbert was one wooden chair.
—Woman, Millbank, Jamaica Unfortunately for me, the land on which I made my farm
was a swampy area and when it rained the whole farm
Economic development is repeatedly interrupted by nat- submerged with water and was destroyed.
ural disasters—by earthquakes, droughts, floods, land- —Elderly man, Atonsu Bokro, Ghana
slides, volcanic eruptions, windstorms, forest fires. Like
economic crises, natural disasters can cause sharp in- Developing countries, especially their most densely pop-
creases in poverty and slow the pace of human develop- ulated regions, suffer the brunt of natural disasters. Be-
ment. And like economic crises, they hurt poor people tween 1990 and 1998, 94 percent of the world’s 568 major
in the short run and diminish their chances of escaping natural disasters and more than 97 percent of all natural
poverty in the longer run. disaster–related deaths were in developing countries (fig-
The damage to agriculture and infrastructure varies by ure 9.1). In Bangladesh alone three storms, four floods,
type and intensity of natural disasters, as do the impli- one tsunami, and two cyclones killed more than 400,000
cations for their indirect and secondary impacts. Droughts, people and affected another 42 million. In southern
for example, can result in heavy crop and livestock losses Africa in 1991–92, Malawi, South Africa, Zambia, and
while leaving infrastructure and productive capacity Zimbabwe experienced severe droughts.39 In Latin Amer-
largely unaffected. ica and the Caribbean major natural disasters associated
Between 1988 and 1997 natural disasters claimed an with El Niño, Hurricane Mitch, Hurricane Georges, and
estimated 50,000 lives a year and caused damage valued the Quindio earthquake in Colombia claimed thousands
at more than $60 billion a year.35 Dramatic as these fig- of lives and caused billions of dollars of damage between
       

Figure 9.1
Developing countries bore the brunt of natural disasters in 1990–98
Percent
100

80
Disaster-related
deaths

60

40

20 Major disasters

0
East Asia Europe and Latin America Middle East South Asia Sub-Saharan Developing High-income
and Pacific Central Asia and the and North Africa Africa countries countries
Caribbean

Note: A disaster is classified as major if it caused more than 50 deaths or affected more than 100,000 people.
Source: USAID, OFDA 1999.

1995 and 1998.40 In 1998 severe flooding of the Yangtze ity of infrastructure in poor communities increases their
River caused devastation in China, and a large earthquake vulnerability.
occurred in Armenia. Another long series of disasters While natural disasters hurt everyone affected by
struck in 1999—a major earthquake in Turkey, a cy- them, poor families are hit particularly hard because in-
clone in Orissa, India, floods in central Vietnam, torrential jury, disability, and loss of life directly affect their main
rains and catastrophic mudslides in parts of Venezuela, asset, their labor. Disasters also destroy poor households’
floods in Mozambique. The list goes on. natural, physical, and social assets, and disrupt social as-
Poverty and lagging development amplify the ad- sistance programs.45 Long-term disabilities and the de-
verse effects of natural disasters. Developing countries struction of assets can trap families in chronic poverty.
are particularly vulnerable, because they have limited ca- Malnutrition impairs children’s ability to learn.
pacity to prevent and absorb these effects. People in The few studies that have analyzed the impact of nat-
low-income countries are four times as likely as people ural disasters on poverty show that the harm to current
in high-income countries to die in a natural disaster.41 and future living standards can be significant. In Ecuador
Despite similar patterns of natural disasters in Peru and El Niño may have increased the incidence of poverty in
Japan, fatalities average 2,900 a year in Peru but just 63 affected areas by more than 10 percentage points.46 In
in Japan.42 Average costs as a proportion of GDP are 20 Honduras Hurricane Mitch caused an estimated 7 per-
percent higher in developing countries than in industrial cent decline in agricultural output in 1998.47 Loss of crops
economies.43 was extensive, affecting a quarter to a half of households.
Poor people and poor communities are frequently Rural households, most dependent on agriculture, lost the
the primary victims of natural disasters, in part because most.48
they are priced out of the more disaster-proof areas and In the 1984 drought in Burkina Faso the income of
live in crowded, makeshift houses.44 The incidence of dis- the poorest third of the rural population fell 50 percent
asters tends to be higher in poor communities, which are in the Sahelian zone, the poorest agroclimate, and 7 per-
more likely to be in areas vulnerable to bad weather or cent in the Sudanian zone.49 There was also evidence that
seismic activity. And there is evidence that the low qual- poor people sold livestock out of desperation. Because they
     ⁄ 

had very small stocks of animals to begin with, these dis- The need to replace damaged infrastructure in disaster-
tress sales may have dangerously depleted their buffer stricken countries diverts government resources from
stocks, leaving them extremely vulnerable to future longer-term development objectives and consumes a sig-
drought and other shocks and possibly trapping them per- nificant share of multinational lending resources. In
manently in dire poverty.50 Mexico as much as 30 percent of the funds approved by
Studies of the impact of the 1994–95 drought in Zim- the World Bank for improving rural water supply over
babwe found that women and young children were the the past decade have been diverted to postdisaster
most affected. For women, the drought’s effect on health rehabilitation.58
(as measured by body mass) was temporary. With good
rains the following year, they regained much of the lost Risk reduction and mitigation:
body mass. But for children ages 12–24 months the lessening vulnerability to disasters
drought will probably have a permanent effect. These Cumulative experience with natural disasters points to an
young children lost an average 1.5–2.0 centimeters of urgent need to move from fatalism to prevention, from
linear growth in the aftermath of the drought. The im- response to preparation, from mobilizing resources after
pact was most severe among children in households with the fact to reducing and transferring risk before the fact.
little livestock, the principal asset of these households for There is a distinct difference in approach to emergency
smoothing consumption.51 The drought had no impact management between many developing and developed
on men’s health. countries. Developing countries emphasize preparedness
On balance, female-headed households fare worse and response—making sure that the resources to respond
than male-headed households following a natural disas- to emergencies are available and ready for dispatch and
ter, in part because of their smaller average resource then that they are dispatched quickly and used efficiently
base.52 Customary or formal laws can make this worse. after an emergency has occurred. Developed countries in-
Among the Tonga of Zambia, for example, a widow has creasingly emphasize reducing or mitigating the impacts
no entitlement to any of the household’s possessions.53 of disasters (box 9.4).
The effect of a natural disaster on poverty can go well Disaster reduction and mitigation can lessen the
beyond the households directly affected. Research on disruption caused by natural disasters, save lives, and
Sub-Saharan Africa suggests that both agricultural and protect property. From a purely economic point of
overall GDP are sensitive to downward fluctuations in view, investing in risk reduction pays off. For example,
rainfall. The 1991–92 drought in southern Africa slowed a cost-benefit analysis for eight cities in the Argentina
growth in agricultural and total output in Malawi, South Flood Rehabilitation Project found an internal economic
Africa, Zambia, and Zimbabwe.54 The impact of drought rate of return of 35 percent. The estimated $187 mil-
shocks on GDP and the recovery time depend in part on lion (1993 dollars) in avoided damages from the 1997
the economic importance of the agricultural sector and flood more than covered the $153 million in investment.
its integration and links with industry. The second-round By installing flood control dams and improving
and subsequent effects are more pronounced in more in- drainage, the Rio Flood Reconstruction and Prevention
tegrated economies. In Senegal and Zimbabwe the effect Project reduced total floodable areas by 40 percent,
of droughts spilled over from agriculture to manufac- achieving an estimated 6.5 benefit-cost ratio for seven
turing.55 The value of Zimbabwe’s manufacturing out- subbasins of the Iguaçu and Sarapui Rivers.59 Com-
put declined 9.5 percent in 1992, largely as a result of the prehensive disaster risk management can be integrated
1991–92 drought, and export receipts from manufactures into development investment decisions. In Turkey in-
declined 6 percent.56 ternational lenders and donors worked with the gov-
The destruction of infrastructure by catastrophic nat- ernment to develop a new disaster management
ural disasters also has both immediate implications and framework in the aftermath of the 1999 earthquakes
longer-term, second-round poverty effects. In Asia, for (box 9.5).
example, where 70 percent of the world’s floods occur, Resettlement—tailored to the needs of poor people—
the average annual cost of floods over the past decade was is often the appropriate risk reduction strategy in flood-
estimated at $15 billion, with infrastructure losses ac- prone or volcanic areas. Where resettlement is not feasible
counting for 65 percent.57 or desirable, neighborhood improvement programs are an
       

Box 9.4 Box 9.5

Mitigation is the cornerstone of emergency Mitigating the risks of natural catastrophes:
management in the United States lessons from the 1999 earthquakes in Turkey

Mitigation—the ongoing effort to lessen the impact disasters A powerful earthquake shook northwestern Turkey in the
have on people and property—is the cornerstone of emergency summer of 1999, killing more than 17,000 people, injuring
management in the United States. It involves keeping homes tens of thousands, and razing several population centers.
away from floodplains, engineering bridges to withstand Three months later a second quake hit, raising the number
earthquakes, creating and enforcing effective building codes of victims and the social and economic losses. Industry and
to protect property from hurricanes, and more. businesses in the areas hit by the quakes had contributed
Over the past 10 years the U.S. Federal Emergency Man- more than 35 percent of the country’s GDP. Their destruc-
agement Agency (FEMA) has spent $25 billion to help peo- tion is likely to affect growth in Turkey for many years to come.
ple repair and rebuild communities after natural disasters. The international community assisted Turkey in relief
Other government agencies and insurance companies have and immediate recovery efforts. In partnership with the Turk-
responded with billions of dollars more. Beyond this, the costs ish government, the European Investment Bank, the Coun-
of emergencies also include lost lives, jobs, and business op- cil of Europe’s Social Development Fund, and other donors,
portunities. A big emergency can reduce local GDP by as the World Bank coordinated the preparation of a framework
much as 10 percent. for a $1.7 billion reconstruction program. A crucial part of the
In 1995 the high and escalating costs of emergencies led framework is a disaster management and response system
FEMA to adopt a national mitigation strategy, with two goals: to prevent similar losses in the future.
to protect people and structures from disasters and to min- Disaster and land development laws will be reviewed and
imize the costs of disaster response and recovery. FEMA es- modified, and the capacity of municipalities to regulate,
timates that every dollar spent on mitigation saves two in plan, and implement disaster-resistant development will be
response and recovery. strengthened. Pilot projects in selected municipalities will
The strategy promotes a community-based approach to help planning and building departments develop risk-based
reducing vulnerability to natural hazards: municipal master plans, means for effective implementation
■ Altering the hazard (seeding clouds during a drought). of building codes, municipal regulations to ensure that
■ Averting the hazard (building dams to control floodwaters). builders follow appropriate licensing procedures, and pro-
■ Avoiding the hazard (moving parts of communities out of grams for evaluating existing buildings.
floodplains). The government’s earthquake insurance program will
■ Adapting to the hazard (constructing earthquake-proof expand its catastrophic risk management and risk transfer
buildings). capabilities. The program will create an insurance mecha-
In February 2000 FEMA announced Project Impact: Buil- nism to make funds readily available to owners (those pay-
ding Disaster-Resistant Communities, a project to provide ex- ing real estate taxes) who need to repair or replace a
pertise and technical assistance to about 200 communities dwelling destroyed or damaged by an earthquake. It will also
striving to become disaster resistant. Three principles drive ensure the financial solvency of the insurance pool after all
the project: Preventive actions must be decided at the local but the most catastrophic events and reduce the govern-
level. Private sector participation is vital. Long-term efforts ment’s financial dependence on donors following major
and investments in prevention are essential. earthquakes.

Source: Olsson 2000. Source: Kreimer 1999.

alternative. In these programs residents of low-income Low-cost local initiatives can also reduce the vulner-
urban areas improve their houses themselves or with com- ability of communities’ income to natural disasters. In
munity help. The programs reduce building code viola- rural areas such initiatives might focus on environmen-
tions by training informal sector construction workers in tal conservation and reforestation. For places prone to
mitigation techniques and by providing finance for low- droughts and floods, community food banks can help.
cost improvements that bring housing to stipulated stan- In Burkina Faso local cereal banks were introduced to
dards. International assistance, channeled through local improve storage, lower food prices, and stabilize them
NGOs, has often helped turn housing reconstruction ef- over the year, including during the drought season.60
forts into low-cost opportunities for mitigating risks in fu- Community agricultural cooperatives can help small
ture disasters (box 9.6). Other important neighborhood farmers obtain credit or crop insurance. And various
upgrading activities include constructing drainage works strategies can help diversify the economic activities
and reducing the risk of flooding and mudslides. within a community.
     ⁄ 

Box 9.6 Box 9.7

Turning reconstruction into risk mitigation with Mitigating risk with catastrophe bonds
the help of a local NGO
Catastrophe bonds—or cat bonds—offer an alternative to in-
In a poor area of Peru partly destroyed by an earthquake in surance in countries lacking active private insurance markets.
1990, Caritas, a local NGO, initiated a reconstruction program A before-the-fact risk transfer mechanism, cat bonds provide
that was also designed to mitigate earthquake-related risks. financial protection against disaster losses.
After consulting with the community, Caritas decided to Consider a government that wants protection against
construct housing from quincha, a local material capable of the risk of flood damage to one of its water treatment plants
withstanding earthquakes. To directly assist the neediest fam- in the next year. Experts estimate the chance of a flood at
ilies, such as households headed by women, Caritas provided 1 in 100, a risk low enough to induce an institutional in-
materials in exchange for participation in communal work. An vestor to purchase a cat bond whose payoff is tied to flood
earthquake in 1991 showed the advantage of using quincha: damage to the treatment plant. The investor buys the bond
most houses resisted the earthquake, which registered 6.2 at the beginning of the risk period at par. At the end of the
on the Richter scale. risk period the investor loses the entire principal if the water
treatment plant is damaged. But if no damage occurs, the
Source: Schilderman 1993. investor recovers the principal plus interest, normally above
the market rate to reflect the risk of losing the principal.
The government invests the funds, which will be used
Reducing economic vulnerability also involves only if a catastrophe occurs, in risk-free securities. The cost
to the government is equal to the difference between the
encouraging—or mandating—the purchase of private interest rate it receives from the risk-free securities and the
insurance for those who can afford it and identifying interest rate it pays the bondholder—a cost analogous to pay-
mechanisms for transferring risk, such as catastrophic rein- ing an insurance premium. The value of the bond—and the
surance and catastrophe bonds (box 9.7). While risk government’s interest payments—would be lower if the
government flood-proofs the treatment plant. So, besides per-
transfer mechanisms can efficiently cover much of the cost
forming an insurance function, the cat bond gives the gov-
of repairing and rebuilding infrastructure, freeing up ernment an incentive to invest in mitigation efforts.
scarce government resources, they may not be easy to apply A potential problem with catastrophe bonds is the diffi-
in poor countries. For one thing, they require systems for culty of verifying damage. The public agency operating the
water treatment plant might exaggerate damage to ensure
verifying damage that cannot be easily manipulated by
that the bondholder pays. One way to deal with this moral
those (governments, for example) who would collect the hazard is to tie payouts to an objective index (such as flood
insurance benefits. To deal with this problem in floods, height) rather than to actual damage.
for example, a country could establish a high-quality
Source: Kunreuther 1999.
measuring and reporting system. This would facilitate in-
surance contracts that link payment schedules to a rain-
fall index.61
of human capital and perhaps to provide shelter for dis-
Coping with natural disasters placed people. Cash transfers to poor families reduce the
In the emergency phase following a disaster, efforts should likelihood that they will need to pull their children out
focus on providing food, water, shelter, and medicine. That of school. Where children need to participate in recov-
makes temporary repair of such infrastructure as roads and ery efforts, schools can adopt flexible schedules.
water supply critical. Priorities need to be based on the Following a widespread natural disaster, national
magnitude of damages and level of vulnerability. The most and local governments need to establish a macroeco-
vulnerable groups—women, children, and the elderly— nomic management scheme to tackle fiscal and current
need special attention. Involving women in the man- account effects—lower tax revenues and higher public
agement of shelters, establishing workfare programs spending, lower exports and higher imports. A calamity
adapted to women’s needs, and ensuring gender neu- fund like that in Mexico can improve governments’
trality in housing acquisition can improve recovery for ability to cover the costs of coping with natural disas-
women and households headed by women. Expanding ters (box 9.8). Calamity funds should focus on ab-
early childhood development programs, particularly sorbing the catastrophic risks that cannot be absorbed
mother and child feeding programs, is also important. Re- by third parties, such as disaster-related damage affect-
building schools should be a top priority—to avoid loss ing farmers and urban dwellers unable to afford private
       

Box 9.8 insurance, and providing social assistance to poor dis-

Sharing the costs of catastrophes: the Mexican aster victims.
fund for natural disasters In the aftermath of a natural disaster targeted inter-
national assistance can help maintain macroeconomic sta-
With tremendous diversity in geography and climate, Mex- bility, accelerate recovery, and protect poor people. But
ico is susceptible to a wide range of natural disasters—
floods, droughts, earthquakes, wildfires, tropical cyclones,
to enable countries to accept financial assistance, inter-
volcanic eruptions. Since 1980 direct damage from natural national financial institutions may have to relax some ad-
disasters has totaled some $6.5 billion, and about 7,000 justment targets during crises. In Zambia in 1992–93 the
people have lost their lives. tight public spending policy being implemented as part
In 1996, to help reduce the country’s vulnerability to
natural disasters, the government established Fonden
of adjustment restricted the government’s ability to raise
(Fondo para desastres naturales, or Fund for Natural Dis- external financing because of the lack of counterpart re-
asters). This federal fund was to be financier of last resort sources in the local currency. As a result, international
for emergency response equipment, disaster relief activ- funds for drought relief were not fully used.62
ities, and reconstruction of public infrastructure and pro-
tected areas.
The reconstruction period provides an opportunity to
In 1998, following a period of particularly heavy losses from reduce vulnerability to natural events (see box 9.6). Targeted
natural disasters, the government decided to use Fonden assistance locally for the most affected populations (with the
more strategically, to provide incentives for insurance use and poor the top priority) and consultation with affected com-
disaster mitigation. After broad consultation with stake-
munities and households should be a key strategy. Includ-
holders, in March 1999 the government changed Fonden’s
operating guidelines to: ing local people in reconstruction activities can foster
■ Increase clarity and transparency in the decision rules for leadership and promote solidarity, helping reduce the psy-
granting access to the fund and in loss assessment chological trauma caused by natural disasters (box 9.9).63
processes.
Countries that have social or rural infrastructure in-
■ Limit moral hazard by encouraging greater use of private
insurance by Fonden’s beneficiaries and establishing vestment funds can use them to channel resources effi-
clear cost-sharing formulas for financing disaster losses ciently.64 Given their experience in building infrastructure
falling under the responsibility of state and municipal and providing social services to communities, these funds
governments. can quickly identify local spending priorities and help min-
■ Encourage mitigation in the reconstruction programs fi-
nanced by Fonden and in beneficiaries’ regular investment
imize corruption. The Honduras Social Investment Fund
programs. performed this role in the aftermath of Hurricane Mitch.
■ Refinance disaster response activities initially financed It financed the opening of several important secondary
through emergency liquidity facilities to speed disaster roads, began the rehabilitation of some water systems in
recovery.
These changes are being formalized through voluntary
secondary towns, and undertook cleanup activities. The
agreements between the federal government and the state fund rapidly put in place a decentralized operating struc-
governments that set out the parties’ rights and responsi- ture and responded to the pressing needs of local munic-
bilities, Fonden’s rules, and agreed cost-sharing formulas for ipalities and communities. Procedures established by the
disaster relief and reconstruction activities. The agreements
will also lead to the establishment of trusts between the fed-
fund immediately after the disaster simplified project
eral government and each state. Under the terms of each preparation and authorization and expedited contracting
trust, spending decisions and contracting of eligible emer- and disbursement of funds. Close monitoring by local res-
gency activities will be carried out by a technical committee idents of private contractor equipment and employment
consisting of state and municipal representatives, acting on
of local residents in cleanup activities helped ensure that
advice from federal entities.
If successful, these measures will increase trans- funds were well accounted for.65
parency, accountability, and efficiency in the use of The experience produced valuable lessons on how to
Fonden’s resources and redistribute the costs of natural improve the capacity of such funds to cope with disaster:
disasters between government and the private sector.
providing adequate financial resources to meet post-
Over time they will also reduce the share of costs borne
by the federal government for mitigating and coping with disaster needs, ensuring that investment projects go be-
disasters. yond rehabilitation and cleanup to include expansion of
existing facilities and new construction, and expanding
Source: Barham 2000.
the fund’s mandate to allow direct social assistance to vul-
nerable people.66
     ⁄ 

Box 9.9
Involving communities in postdisaster reconstruction: lessons from the Maharashtra Emergency
Earthquake Rehabilitation Program

On 30 September 1993 an earthquake struck the Indian state of effective tool for dealing with problems that arise during
Maharashtra, killing some 8,000 people and damaging 230,000 implementation.
houses in Latur, Osmanabad, and 11 other districts. With the help Participation also had a positive psychological effect on com-
of the World Bank, the government of Maharashtra created the munities. Involving local people in the reconstruction helped them
Maharashtra Emergency Earthquake Rehabilitation Program. The overcome the trauma caused by the earthquake. Recognizing this,
program institutionalized community participation and formal con- the government began reconstruction in small villages even before
sultation with beneficiaries at all stages. the rehabilitation program began, appealing to donors, corporations,
The program divided communities into two categories: those NGOs, and religious organizations to “adopt” villages for recon-
that needed to be relocated—the 52 villages that sustained the struction. Some organizations also worked on social issues, such
worst damage—and those that needed to be reconstructed, re- as schooling for children.
paired, or strengthened. The Tata Institute of Social Sciences Information on the program, its processes, and the mecha-
worked in the 52 relocation villages, which had some 28,000 fam- nisms for redress was accessible—and awareness was high. The
ilies. The Society for Promotion of Area Resource Centers orga- participatory process opened many informal channels of commu-
nized community participation in the 1,500 villages—with some nication between the people and the government, helping to nar-
190,000 families—in which rebuilding or repair was to take place. row the gap between them. Beneficiaries learned of their
Over time the program became a people’s project. As re- entitlements and worked hard to secure them. People who felt that
sults materialized, community participation received greater ac- their grievances were not addressed appropriately in the village or
ceptance. Initially skeptical, officials in the project management taluka (an administrative unit that includes several villages) could take
unit later came to acknowledge community participation as an them to the district authorities and the government in Mumbai.

Source: Vatsa 1999.

Workfare programs can usefully be introduced or ex- • • •

panded in disaster areas in conjunction with reconstruc-
tion operations, providing a livelihood to people who Large adverse shocks—economic crises and natural
can no longer support themselves (chapter 8). They can disasters—cause poor people to suffer not only in the short
also help people affected by the less visible impacts of a run. They undercut the ability of the poor to move out of
disaster, such as the poor fishers in Ecuador and Peru who poverty in the long run as well, by depleting their human
fell deeper into poverty as fish fled the waters warmed by and physical assets. Particularly harmful are the effects on
El Niño. In Northeast Brazil the program Frente de Tra- poor children, who may suffer irreversible damage if a cri-
balho (Work Front) provided similar employment op- sis or natural disaster increases malnutrition or forces them
portunities in periods of drought. During the 1979–84 to drop out of school. Integral to any poverty reduction strat-
drought it employed up to 3 million workers in con- egy should thus be measures to prevent and manage economic
struction and drought-related jobs.67 Public work programs crises and natural disasters—and to establish safety nets, with
that build social or community infrastructure or help in ensured financing, to help poor people cope when these ad-
cleanup and reconstruction can also be a good option. verse shocks do occur.