For Eng.Kelvin only

chapter_5.pdf

Home >English homework help >For Eng.Kelvin only

CHAPTER 5

Insurance and Coping

he earlier chapters showed that individuals generally take prevention measures within their choice set. But full prevention is neither attainable nor desirable, and

residual disaster risks will remain. Insurance and complementary measures to have funds when needed (such as borrowing or setting aside reserve funds), remittances, and relief “soften the blow,” and this chapter examines their roles in turn. The chapter begins with the basics of the insurance business: the advantages of

pooling and transferring risk to those willing to bear it and how insurers deal with the many complications that arise from adverse selection and moral hazard. Insurance clearly increases a person’s choice and thus well-being: the contract specifies the resources transferred from one person to another when the event (such as a disaster) occurs. In doing so, it shifts the risk from the individual to the pool of the insured. But softening a disaster’s blow concomitantly dilutes the incentive to prevent —unless the premium reflects the risk and the prevention measures a person undertakes. Commercial insurance companies calculate the premia using detailed data on the

frequencies and intensities of hazards and how they affect exposed assets. The premia must also cover the considerable costs of administration, marketing, and monitoring. Many people may forgo insurance if the premia are too high. And while parametric insurance—a type of insurance that specifies the payout based on a parameter related to the hazard but unrelated to actual damages incurred—reduces some of the monitoring costs, such schemes have low penetration rates in developing countries where they have been introduced. When an insurance industry does develop, it invariably draws the government in as

regulator, as provider (in many countries), or as reinsurer. Governments inevitably add a political dimension, and pressures to subsidize the premia may increase. The U.S. experience with flood insurance shows that this is not just an issue in developing countries. Too low a premium encourages construction in hazard-prone areas, thereby increasing exposure and vulnerability. The chapter next turns to whether governments should buy insurance to have funds

to spend after a disaster, simply borrow, or set aside funds in reserve. Many are already indebted and even those with low debts may find it difficult to borrow when they most need to. Politicians who want to spend on worthy programs tend to deplete funds set aside in a reserve fund. To avoid this “honey pot syndrome,” governments

The, World Bank, Bank World, and Nations (UN) United. Natural Hazards, UnNatural Disasters, edited by World Bank The, et al., World Bank Publications, 2010. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/apus/detail.action?docID=606027. Created from apus on 2017-04-10 14:36:39.

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

may purchase insurance. The World Bank’s Catastrophe Risk Deferred Drawdown Option and other such facilities can help countries. While individuals are risk-averse, there are good reasons for some governments

acting on their behalf to be risk-neutral. A risk-neutral entity would buy insurance only if the premium were lower than the probability times the expected loss (which leaves nothing to cover the insurer’s costs). But the likelihood of a disaster that is large relative to an economy’s size (as in the Caribbean, where the main unknown is which island will be hit) may make some governments risk-averse, especially when rapid access to funds after disasters could be difficult or costly. Such governments, and those seeking to avoid the “honey pot syndrome,” would benefit from buying insurance. The Caribbean Catastrophe Risk Insurance Facility pools disaster risks regionally, helping countries purchase insurance less expensively than otherwise. Comparing prices offered by insurance firms against those in capital markets, as Mexico did when issuing catastrophe bonds, is also advantageous. The chapter then examines remittances sent by private individuals and groups

abroad to help people cope with a disaster. Remittances are directed to victims and their survivors, even when the disaster does not attract any media publicity. The funds arrive quickly without the involvement of governments or other organizations. But sometimes unnecessary government policies (controls on capital flows, dual exchange rates) impede arrival of the funds. Remittances that arrive before a disaster also help with prevention. Although remittances augment consumption, particularly consumer durables, they are also used to improve the quality of housing. Mud and straw huts give way to houses built of brick and cement. Private remittances also help develop banking and money transfer facilities, which in turn strengthen the area’s commercial ties with other parts of the country and the world. Last, the chapter examines the role of aid in prevention. Post-disaster aid can also

be double-edged: while some aid is warranted, it can also give rise to the Samaritan’s dilemma—the inability to credibly deny help following a disaster to those who have not taken sufficient prevention measures. Some new but not very strong evidence shows that post-disaster aid could reduce prevention. Donors should therefore be aware of the disincentives they may create, and concern for the victims should be moderated by the effect on incentives.

Insurance: Useful if the premium is priced right As with any voluntary transaction, insurance benefits all parties to the contract: the insurer benefits from the business and the insured reduce the adversity of the worst states of nature by giving up some of the benefits in the good states.

The basics of the business

Insurers take pride in covering unique risks like an opera singer catching a cold, or a The, World Bank, Bank World, and Nations (UN) United. Natural Hazards, UnNatural Disasters, edited by World Bank The, et al., World Bank Publications, 2010. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/apus/detail.action?docID=606027. Created from apus on 2017-04-10 14:36:39.

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

racehorse breaking a leg. But such insurance is a side show, and the bulk of their business covers more mundane, predictable, and diversifiable risks (such as life and property insurance). Consider insuring houses against fire: one cannot tell when and if a particular house will catch fire, but data on past fires allow the number of house fires in an area to be reliably predicted with probabilities attached. Making the average loss more predictable allows firms to insure individual houses against fire: the insurer collects an annual payment (premium) from risk-averse home owners (the insured) and promises to pay (the insured amount could be actual damages or a specified sum) if their house burns down (the trigger). The aggregate annual premia collected from the insured must cover the insurer’s operating costs and the likely payouts. And if unexpectedly fewer houses burn down, the insurer has a surplus beyond its normal profits. There is always a chance that an unexpectedly large number of houses will catch

fire (as in a particularly dry year), so the insurer has a buffer, which is the owners’ capital and surpluses accumulated from earlier years. This buffer is invested and the ensuing earnings (dividends or interest) augment the premia the firm collects. If the surpluses accumulate over time, competitive pressures would prompt the insurer to lower the premia; if they are depleted, the premia would be raised. This is the basic principle of insurance; but complexities quickly multiply. Adverse selection arises when a person buys insurance knowing that his risk

exceeds that of the larger pool that is the basis for determining the premium. If only those knowing their risk is greater buy such insurance, the insurer’s surplus will fall as the risk of the pool rises. Moral hazard arises when the insured take additional risks because they are insured (not repairing the building sprinklers that extinguish fires if the building is insured against fire). Co-payments (where the insured bears a specified fraction of the loss) and deductibles (where the insured bear losses up to a specified amount) reduce but do not eliminate these difficulties. Contracts become complex and the costs of administering claims, resolving disputes, and increasing monitoring mount. Insurers continually seek observable proxies of the risks they insure, link their premia to these risks, and continually test the insured’s price sensitivity to premia that must cover these costs. The costs result in a premium that greatly exceeds expected losses, but sufficiently

risk-averse people buy insurance all the same because it protects them from the devastating financial implications of a disaster. Insurance does not “shift the loss” collectively: the insured pay for the losses through their premia, and those who do not make a claim, essentially pay for others who do. Insurance can be a competitive industry, but some economists find that there is

little competition (insurers are exempt from antitrust laws in the United States) or that costs are not driven down. Administrative and marketing costs are about 35 percent of aggregate premia in the United Kingdom (the insured get roughly 50 percent in

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

payouts, a figure that is broadly similar in the United States) where private insurance firms compete, in contrast to 10 percent in Spain where a state-owned monopoly provides coverage (Von Ungern-Sternberg 2004). Germany privatized its provincial monopoly providers under a European Union directive only to find that operating and administrative costs rose as a consequence and insurance premia were raised between 35 and 75 percent in five years. Regardless of who owns insurers, governments invariably get involved as

regulators if not as providers because buyers “get the product” (the promised payout) only after a disaster; and the insurer may find some reason to refuse payment, reduce coverage, or go out of business.

The government’s inevitable involvement

Insurance is limited in the developing world, but a large industry in many developed countries.1 These countries’ governments are involved in each of them, though in different ways. Courts, not governments, enforce contracts, but the payouts to the insured may be delayed or denied. Insurers invariably write and interpret the clauses to their advantage (“the fine print”), and seemingly fair clauses are not always so. British insurers and the insured could cancel a policy with seven days’ notice, and insurers cancelled coverage in 1997 when it became apparent that the erupting volcano on Montserrat would destroy every building on the island, akin to cancelling coverage after a fire has started (Von Ungern-Sternberg 2004). In the 1800s, insurance was for named perils, with covered losses stemming only

from the specified risk. In the 1930s, all peril property insurance became more common. Whatever the coverage, governments try to ensure that insurers honor their contracts (consumer protection), and when this requires insurers to have adequate funds (solvency), regulations often extend to approving the premia. The premia are sometimes high, and to ensure coverage, property insurance is sometimes mandatory (as in Germany and many Swiss cantons); but sometimes populist pressures cause premia to be too low (as in the United States with flood insurance), needlessly increasing exposure in hazardous areas. The fertile lands in the flood plains attract farmers, and many settlements in the

United States are periodically inundated. Sympathy for the victims would prompt public assistance, and settlers would rebuild in the same area. After several major floods in the 1950s and 1960s, private insurers were no longer willing to cover floods (which became an “uninsurable” risk), and the U.S. government, recognizing that it was unable or unwilling to deny assistance to those affected, established the National Flood Insurance Program (NFIP) in 1968. The premia were set low to induce homeowners to buy the NFIP insurance, but

very few people voluntarily purchased coverage (Kunreuther and Michel-Kerjan 2009). The federal government then required this coverage as a condition for federally

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

insured mortgages, but the mandate was poorly enforced and many people canceled their policies, especially if there was no flood for several years, and others purchased insurance just after a flood (Michel-Kerjan and Kousky 2010). They examine more than five million insurance policies, the largest flood insurance sample ever studied, and find that of the one million residential NFIP flood insurance policies in place in Florida in 2000, a third were cancelled by 2002 and about two-thirds were cancelled by 2005. There was no effective mechanism to prevent or discourage more people from settling in the areas known to be hazardous: the NFIP is a federal program, while zoning and insurance regulation are state issues, and local politicians reflected the settlers’ desires. The number of policies nationwide managed by the NFIP increased from 2.5 million in 1992 to 5.6 million in 2007 and, in nominal terms, the property value covered rose from $237 billion to $1,100 billion during the same period. The NFIP’s other shortcomings were exposed after Hurricane Katrina flooded much

of New Orleans in 2005. The NFIP covers floods, but private insurance covers wind damage. Many disputes arose over who should pay when damage from wind could not be easily separated from that by floods (Kunreuther and Michel-Kerjan 2009). Victims were given the runaround and payouts were delayed. In a background paper for this report, Kunreuther and Michel-Kerjan note how

multihazard insurance can address insurer-insured disputes by having homeowners’ coverage move from the traditional one-year insurance contract to multiyear contracts (say 10 or 15 years) tied to the property (not the owner as is the case today). The premia would reflect insurers’ best estimate of the risk over that period and would assure policy holders of coverage. The possible denial of coverage was a major concern in hazard-prone areas because insurers canceled policies following the 1992 and 2005 hurricane seasons. Following Hurricane Andrew, Florida passed a law in 1992 limiting the cancellation of policies by insurers to 5 percent a year at the state level and to 10 percent at the county level (Jametti and von Ungern-Sternberg 2009). Both insurers and home owners cancel policies for different reasons, and the premia are subject to political pressures. These major changes in government policy require appropriate regulatory authority and decisions (Kunreuther and Michel Kerjan 2008). Comprehensive, multihazard insurance will entail higher premia. Some policyholders may think they are being charged for coverage they do not need (a person in an earthquake area not prone to hurricanes and floods may only have quake insurance), but they would not be overcharged if premia reflect risk accurately. Whether premia accurately reflect risks becomes all the more important. Government involvement inevitably brings political pressures; and vested interests

and populist pressures exist in all countries, though they manifest themselves differently. Insurance subsidies are usually regressive: those with assets to insure are generally better off than the poorer segments that often pay indirect taxes that pay

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

for the subsidies. But underpriced insurance is not always the result of government pressures. Insurers sometimes make mistakes or may take unwarranted risks and then discover that the risks were greater than they assumed. To compensate for these mistakes, insurers often find reasons to deny payments, redefine the risks that are covered (terrorism was made a separate risk that got excluded), and raise deductibles and premia.2

Pricing the premium

The premium is an important price: too low, and excessive construction in exposed areas and insufficient prevention result; too high, and few buy insurance. Calculating the appropriate premium is not trivial: probability distributions and loss functions must be estimated, and the relevant pool and observable characteristics that correlate well with the underlying risk (an unobservable) must be identified. These estimates are a firm’s “proprietary information.” And while competition may drive insurers to continually improve these correlates and hence their contract terms and prices, the European experience (showing the lower operating costs of monopoly providers) suggests that this may not always follow. Some additional complexities arise with infrequent hazards: diversification among

many policy holders (contemporaneous) may not suffice, and diversification over time (intertemporal) is more difficult (box 5.1). Examining data from the largest U.S. catastrophic risk reinsurer for 1970 to 1998, Froot (2001) finds that catastrophe insurance premia are far higher than expected losses (up to seven times greater). The most likely reasons are reinsurance market imperfections (such as government intervention in insurance markets) and the market power exerted by traditional reinsurers. As noted in several parts of this report, governments can do much to improve data

quality and accessibility. Hurricanes are more frequent than earthquakes, but consider what it takes to set the hurricane insurance premium: several sets of detailed data are needed including the frequency, the likely paths and severity of hurricanes, the value and type of construction of all structures in their path (so accurate property records are essential), and how much damage each structure would likely suffer at various wind speeds (so local universities and engineering associations must know and test the strength of materials and designs of existing buildings). Climate science models estimate the forces (such as wind speed and air pressure) and engineering determines how buildings withstand them; allowing estimates of loss exceedance curves (insurers use this combination of cumulative probability distribution function with values at risk). Even with good data, it is far from clear whether the frequency and severity of

hurricanes has changed (chapter 6 discusses how frequency and severity might be affected in the future because of climate change). Insurers in the United States had taken note of the dangers to property after 1992’s Hurricane Andrew, but were

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

nevertheless caught unprepared for the string of storms and major hurricanes (Katrina, Rita, and Wilma) in 2004 and 2005. Insurers incurred large payouts and raised the premia; but one cannot tell if the string of hurricanes were a low probability drawing from an unchanged distribution (making the premia increase unjustified)—or a shift in the distribution itself.

Box 5.1 Catastrophe risk in insurance and financial markets Pooling risks reduces aggregate variance; so losses that are large and unpredictable for a victim become small and predictable in the aggregate pool. Risk pooling could be contemporaneous or intertemporal, but the latter requires the insurer to have enough capital to make the payouts and replenish it over time with annual premia. Catastrophes are infrequent, and insuring against them illustrates the issues with intertemporal diversification. The risk that an event could generate a large loss for a country (1998 Hurricane Mitch for many Caribbean

countries) could be small if that risk were shared worldwide. Reinsurance permits this worldwide sharing of risk through multiple transactions; but some catastrophes may be large enough to exhaust the collective buffer. When payouts are significant relative to the global capital and surpluses of all insurance companies, it raises

two issues. First, it questions the probability distribution that insurers use to set the premia: it is hard to know if a string of large losses are “several heads in a row” or evidence of the probability distribution changing. Second, even if the insurance actuaries were confident that the probability distribution was unchanged, investors in the equity market may be nervous: they drive down the prices of the insurer’s equity on the stock exchange, thereby raising its cost of capital. In either case, pressures would build to raise the premium, a tendency exacerbated by any market power that insurers may have. So insurers raise the premium for catastrophe coverage as the likelihood of exhausting their buffer increases. They could augment this buffer by raising more capital (issuing equity), but their cost of capital would reflect the greater perception of risk. But while insurance and financial markets are linked, they are not entirely integrated, and the price of risk in

the two markets could differ substantially. Integration is easier when liabilities of insurers are traded on financial exchanges (some insurers are organized as mutuals, not corporations), and investors can more easily ascertain the insurers’ exposure. A premium for nondiversifiable risk in financial markets may differ from that in insurance markets: there are potential gains from placing such risk in capital markets because many natural hazards are uncorrelated with the business cycle. Those who straddle both insurance and capital markets— hedge funds that underwrite insurance, CAT-bond issuers and buyers—could profit from these differences. Those seeking to place such risks (reinsurers, governments buying insurance) would benefit from

comparing prices in the insurance and financial markets. Governments must have the ability to assess their risks comprehensively and independently. Small mistakes can result in huge losses (more precisely, large transfers between the insured, the insurers, the reinsurers and CAT-bond holders).

Source: World Bank staff.

In a background paper for the report, Seo and Mahul (2009) found that property at risk rose as much in one decade of coastal development in the United States as from five decades of greater hurricane activity. The premia also rose substantially (an average of 76 percent) after 2005. More reliable information on probability distributions of hazards and the values and strength of structures at risk would reduce the premia regardless of whether coverage is from insurers or capital markets. The insurance premia almost double when probability and loss estimates are ambiguous as opposed to specific (Kunreuther, Hogarth, and Meszaros 1993). These complexities notwithstanding, calculating the insurance premium should be commercial, not political. The government is typically involved (even if only as

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

regulator), and political pressures get transmitted regardless of a country’s institutional arrangement. Governments also gather relevant data (on weather, property values, location, and the like) and not all governments make them readily accessible.

Parametric insurance

Parametric insurance is a type of insurance that makes the payment of claims conditional on a triggering event (wind speeds exceeding a certain threshold; earthquakes exceeding a particular intensity). Because assessing whether the parameter has been triggered is easy, such insurance obviates the need for detailed loss assessments. Insurers therefore avoid some costs (such as monitoring to reduce fraudulent claims, valuing the structures and their strength) and the insured pay a lower premium as a consequence. Although the premium is not tied to prevention measures (because the payouts are specified and unrelated to damage), the insured retains the incentive to prevent because lower damage is to the insured’s benefit alone. There are now some 20 schemes in low- and middle-income countries including

China, Ethiopia, India, Malawi, Nicaragua, Peru, Ukraine, and Thailand (World Bank 2009a). Evaluation was built into the design of two weather-based crop insurance pilots (India in 2003 and Malawi in 2005), and their experiences have been carefully studied. Published evaluations of these schemes find that despite much effort, market penetration has been low. Fewer than 5 percent of eligible households in India and 17 percent of farmers offered insurance/credit in Malawi used these schemes. Subsequent surveys in India found that most farmers did not understand complex contracts or trust those selling insurance (Giné, Townsend, and Vickery 2008; Cole and others 2008). Parametric insurance has had greater success at the wholesale level. In countries

where there is some commercial insurance, domestic insurers reduce their risks by purchasing parametric reinsurance contracts from others. Governments also partake in such schemes directly.

Should governments borrow, set aside funds, or buy insurance? Government revenues may fall after a disaster, especially if output declines and relief spending rises in the immediate aftermath and later, to rebuild damaged and destroyed government property.3 Governments could borrow domestically and internationally to finance the wider budget deficit, but some may find it difficult. Lenders may be unfamiliar with small countries who do not normally borrow, and countries that do may have large existing debts and may not be able to raise additional sums. In addition to the probability and severity of the hazard, the choice between borrowing, setting funds aside, or buying insurance depends on the country’s

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

circumstances (figure 5.1).

Figure 5.1 Managing and transferring financial risks to the market

Note: The figure focuses only on risk transfer/retention schemes and not on risk reduction measures (preventive measures that reduce the risk of death, injury, and damage, such as early warning systems; risk identification and measurement, and safer buildings and structures). Source: World Bank staff.

Specific examples of contingent loans, insurance pools, and insurance-linked securities are discussed in the remainder of the section.

The World Bank’s CAT DDO

The World Bank lends to governments, often after a disaster. Projects and loans take time to process, and those already approved with undisbursed funds are often “restructured” after a disaster to allow quicker disbursements (Independent Evaluation Group 2006). Recognizing this was happening often, the World Bank recently made available the Catastrophe Risk Deferred Drawdown Option (technically called a Development Policy Loan with CAT DDO) that encourages the country to manage natural hazard risks (box 5.2).

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

The disaster risk management requirement encourages governments to consider preventive measures and view disaster risk in a comprehensive manner. There is merit to considering risk comprehensively because, as chapter 2 discussed, the economic effect of a disaster depends not just on the physical damage but also on the country’s economic structure. The destination and composition of its trade, reliance on tourism, sources of tax revenues, reserve holdings, and extent of borrowing all affect risk. The CAT DDO and other World Bank lending facilities help governments borrow. A

MultiCat program allows participants (countries and regions) to buy insurance for multiple perils using documents and legal help developed for this purpose (World Bank 2009b). In 2009 the government of Mexico used this facility to issue a $290 million series of three-year notes with parametric triggers that replaced those that were maturing.

Box 5.2 The World Bank’s catastrophe risk deferred drawdown option (CAT DDO) The CAT DDO, a World Bank loan to middle-income countries exposed to natural hazards, is approved before a disaster and disburses quickly if and when the borrowing government declares an emergency. The loan amount is limited to $500 million, or 0.25 percent of GDP (whichever is smaller), because the CAT DDO provides short-term liquidity (rather than reconstruction financing) following the disaster. It does not preclude other borrowings. The CAT DDO, available for three years, can be renewed up to four times. There is a single front-end fee of

0.5 percent of the approved amount, and each subsequent renewal entails a fee of 0.25 percent. The interest is set at the IBRD rate prevailing when the funds are disbursed. The funds could be repaid at any time before the closing date, and this amount would still be available for subsequent borrowing. Borrowers must, however, have an adequate macroeconomic framework in place when the loan is approved and a disaster risk management program monitored by the World Bank. The World Bank estimates that the CAT DDO is 25 percent less expensive than insurance for the equivalent

risk, so it is attractive, though the approved (though undrawn) amount counts when determining the country’s borrowing limit from the World Bank. Costa Rica, the second most exposed country to multiple natural hazards (Natural Disaster Hot-spots 2005),

was the first to have a CAT DDO approved in September 2008 for $65 million. Domestic politics delayed the payment of the front end fee until after a 6.2 magnitude earthquake on Janu-ary 8, 2009 struck causing damage estimated at $100 million; but Costa Rica drew down a portion of the loan when the fee was paid. As of December 2009, CAT DDOs had been approved for Costa Rica, Colombia, and Guatemala.

Sources: World Bank 2009a. Catastrophe Risk Financing in Middle- and Low-Income Countries: Review of the World Bank Group Operations. April 1, 2009, document prepared for a Technical Briefing to the Board of Executive Directors. http://treasury.worldbank.org: Jose Molina Jr. “Overview of DDO and CAT DDO,” World Bank Treasury (power point presentation). Costa Rica: Earthquake OCHA Situation Report No. 2 printed from www.relief web.int on 1/15/2009.

Should governments purchase insurance? Unlike individuals who arerisk-averse, there are good reasons for some governments to be risk-neutral (meaning they should not buy insurance if the premium exceeds the expected loss) (Arrow and Lind 1970).4 Box 5.1 explained why catastrophe insurance premia exceed expected losses, suggesting that governments should self-insure only by setting aside sufficient reserves—for example, in a contingency fund (holding reserves in such funds has an

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

http://treasury.worldbank.org

http://www.relief web.int

opportunity cost though)—or have access to ready borrowing. But some governments may be risk-averse, not risk-neutral, and buy insurance

even when the premium exceeds expected losses as with small Caribbean countries. The World Bank and other organizations have been encouraging governments to shop carefully, to consider risks comprehensively, and to think about prevention.

The Caribbean Catastrophe Risk Insurance Facility

Hurricanes form in the Eastern Atlantic Ocean and strengthen as they move west and then north (map 5.1). One or more of the many islands spread across the Caribbean is hit almost every year: the main unknowns are which island will be hit and how severe the damage will be.

Map 5.1 The Caribbean region—in harm’s way

Source: Risk Management Solutions (RMS).

The Caribbean Catastrophe Risk Insurance Facility (CCRIF) was set up in June 2007 for Caribbean island governments to collectively purchase parametric insurance. Commercial insurance is available in the Caribbean region, but the total premia paid by businesses averaged about 1.5 percent of GDP between 1970–99 while losses (insured and uninsured) amounted to only about 0.5 percent of GDP (Auffret 2003).5

Donors provided $67 million in start-up capital, and 16 member governments contributed $22 million. Governments purchased parametric insurance paying CCRIF about $20 million in premia for parametric insurance coverage totaling roughly $450 million (Cummins and Mahul 2009). The Facility retains responsibility for the first $20 million of payout (backed by its capital) and transfers the remaining exposure through reinsurance and catastrophe swaps that the World Bank intermediates. Donors expect that its capital and reserves will grow and that it will be self-sustaining.

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

The Facility paid about $1 million to St. Lucia and Dominica within two weeks of the November 2007 earthquake, the most severe in the eastern Caribbean in 30 years. It paid $6.3 million to the Turks and Caicos Islands after Hurricane Ike hit in September 2008.6 There have also been disasters that did not trigger the set parameters: Hurricane Dean in 2007 caused considerable damage in Jamaica because of rain, but there was no payout because wind speed was the parametric trigger. Similarly, the cumulative effect of the 2008 hurricanes in Haiti was devastating, but the winds were not strong enough to trigger a payout. These parametric triggers may be readjusted in future insurance contracts (Simmons 2008). The 7.0 magnitude earthquake that struck Haiti on January 12, 2010, was of

sufficient magnitude to trigger the full policy limit for Haiti’s earthquake coverage purchased under the Facility. Based on calculations from the earthquake location and magnitude data, Haiti has received $7.8 million, the maximum payout under its earthquake policy. This is about 20 times its premium for earthquake coverage of $385,500. Although shaking was felt in Jamaica, another CCRIF-covered country, it was insufficient to generate any loss under the parametric index. Pooling risks among the Caribbean countries and buying residual risk coverage has

some merit because insurance firms cannot count on such diversification: other Caribbean countries may not buy coverage. CCRIF’s premia are thought to be 40 percent lower than commercial coverage (World Bank 2007). Donors are also more confident that if there is a disaster, CCRIF would make funds readily available (if the parameter is triggered).

Mexico’s CAT-bonds: Insuring directly in financial markets

Mexico City, with 18 million people, was devastated by an 8.1 magnitude earthquake in September 1985: some 10,000 people were killed, 412 buildings collapsed, and 3,124 were damaged, including hospitals. The city is not on, or near, a fault line—the epicenter was 400 kilometers away—but Mexico City sits on a drained lake bed in an old volcanic crater with soft clay and ash that amplifies ground movement. So even distant quakes cause damage, and high rises are particularly vulnerable because their natural vibrating frequencies resonate with the seismic shocks. Consequently, many older (low-rise) buildings survived while many modern (high-rise) and well-constructed ones did not. Designing new buildings and retrofitting old ones for earthquake risk requires unusually complex engineering skills. The earthquake came shortly after a humiliating debt default to foreign creditors.

The foreign minister spurned international aid, especially from neighboring United States, and the president refused to suspend payments on recently restructured debts to help with the recovery. Consequently, foreign exchange reserves plummeted and economic management became difficult. To avoid a recurrence, laws were enacted in 1994 requiring federal, state, and

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

municipal public assets to be “insured” through a government entity, FONDEN, created in 1996 with a catastrophe reserve fund in it. FONDEN allowed funds to be spent after a disaster without having to borrow. Reserves were built from nothing in 1999 to about $863 million in 2001 (in 2008 prices) but were almost depleted following subsequent hurricane damage. The government realized that the amounts needed after an earthquake would have

been too large to set aside untouched (“the honey pot syndrome”). Mexico City accounts for 60 percent of the country’s GDP and the 1985 earthquake raised the fiscal deficit $1.9 billion over the next four years (Cardenas and others 2007). In 2006, the Mexican government decided to transfer part of its public sector natural catastrophe risk to the international reinsurance and capital markets. Officials estimated that FONDEN could handle disasters up to $500 million (one standard deviation above average annual spending). FONDEN calculated its expected expenses after an earthquake and compared the amount with the insurance premia. The insurance premia had risen substantially since 2001, so it issued a catastrophe bond through a special purpose vehicle.7

The details are complex, but the concept of a catastrophe bond is simple: a special purpose vehicle issues the bond, and the proceeds are held in escrow. Bond holders (typically hedge funds or money managers) receive a higher interest rate (235 basis points above LIBOR [London interbank offered rate] as in the case of Mexico) than what the escrow earns. This difference amounts to $26 million on three bonds outstanding that total $450 million.8

When triggered by an event (an earthquake of magnitude equal to or greater than 7.5 or 8.0 on the Richter scale depending on specified points in and around Mexico City), the escrowed funds are released to the government and investors get nothing further. So when the earthquake strikes, investors lose financially, not the Mexican government. Meanwhile, Mexico City experienced many significant tremors (a 6.5 magnitude earthquake in Oaxaca in 2008, a 6.0 in April 2009, a 5.7 in May 2009) with little damage. But should a major one strike, the government would have funds to spend on relief and reconstruction as it sees fit. Mexico was the first government to issue a CAT bond in 2006. While new issues

have declined following the financial crisis of 2008, some $9 billion of face value are outstanding (Cummins and Mahul 2009). Issuing a bond has large fixed costs, many hidden, that small countries may find disproportionately expensive.9 Such governments may find it advantageous to reinsure through regional pools, but it is also important to design and build structures well so damage can be reduced. Many new buildings in Mexico now have dampeners, but earthquakes differ in their forces and the resonating effect of the lake bed adds to the complexity and expense of retrofitting existing buildings.

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

Quick and direct help for families What cannot be prevented or insured against must be borne, and a variety of coping mechanisms (“informal insurance,” as distinct from market insurance) have developed over the centuries, many embedded in tradition and custom. People often help their friends and neighbors who suffer a broken leg or the death of an ox, and distant friends and relatives send remittances (some as loans). The main sources of help are:

Remittances and vibrant communities Public safety nets Foreign aid.

Remittances and vibrant communities

Relatives or friends who live outside the affected community can send food, credit, or transfers from unaffected to affected areas. For example, marriages in six South Indian rural villages appear to have been arranged expressly to help households cope with droughts (Rosenzweig and Stark 1989). But transfers cover less than 10 percent of the shortfalls in income: so while they may avert starvation, they are unlikely to prevent consumption from declining substantially (Rosenzweig 1988). Remittances are private financial flows to friends and family. Numerous migrant

workers send small amounts that quickly add up. Most important, they flow directly to the victims—quickly and without fuss or fanfare, though not to all victims. Not all poor families have relatives working abroad: migration requires a large initial expense to buy tickets and work permits. But they may indirectly benefit from what their neighbors receive if they work for them. Many studies of remittances and disasters find credit and transfers from relatives in

distant areas (inside countries and overseas) to affected areas. Lucas and Stark (1985) find remittances increased from urban areas into rural areas in Botswana during the droughts of 1978–79. Miller and Paulson (2007) find that in 1988 Thai households living in a province with below average rainfall the previous year got about 118 baht ($4.72) more remittances. Yang and Choi (2007) find that Philippine households with relatives abroad get remittances of 60 cents for every dollar decline in income between 1997 and 1998. In a background paper for the report, Mohapatra, Joseph, and Ratha (2009)

examine the effect of remittances on both the response and the preparedness. They estimate the effect of disasters (damage, numbers killed and affected) on remittances as a percentage of GDP and control for overall and immigrant population and persistence (this year’s remittance depends on last year’s). Migrant remittance data for 129 developing countries (part of the World Bank’s Development Indicators) are augmented by disaster data from EM-DAT for 1970–2006. For a country with 10

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

percent of its population abroad, remittances rise by $0.50 for every $1 damage in the same year and by $1 in the year following (or $1.50 in the two years). These effects also register as a proportion of GDP: remittances rise by 0.5 percent of GDP in the same year and an additional 0.5 percent in the year following for an additional 1 percent of population affected by a disaster. Remittances are not sensitive to the number of fatalities. The striking results are consistent with what is known about migrant motives and

behavior: many work abroad to augment the income of family left behind. So when their family’s assets are destroyed or their livelihoods threatened, migrants abroad send funds to help. While the migrants’ earnings and savings over their entire stay abroad may not change, the timing of their remittances responds to the needs of the family left behind. The remittances are put to different uses: studies find that a large part goes to

buying consumer durables (refrigerators, radios, televisions) and much of what is invested is for building homes or adding masonry structures to them (Adams 1991). Houses made sturdier could be considered a prevention measure, though the situation varies. In Turkey, 13 years after the 1970 Gediz earthquake, the reconstructed area was peppered with improperly reinforced concrete houses—mostly paid for by the earnings of family members in Germany (Aysan and Oliver 1987). Better building practices (described in chapter 3) are needed to ensure building safety. Mohapatra, Joseph, and Ratha (2009) use household survey data for Burkina Faso

(2003), Ghana (2005), and Bangladesh (1998–99) to separate consumption increases attributable to remittances from other factors. The remittances allowed households to consume more than otherwise identical nonrecipient households after the 1998 flood in Bangladesh. Recipients in Ghana, especially those receiving remittances from higher income countries, had better housing and were more likely to have fixed and mobile telephones (figure 5.2). International remittance recipients in Ethiopia were less likely to sell their livestock during droughts (when prices may be disadvantageous) because they have cash to buy food (figure 5.3). Remittances are not only from family members: expatriate “communities” get

organized as NGOs to raise and send funds after disasters (not classified as remittances in the data). These flows, also well directed, are often spent on relief, not prevention. Private inflows can also take other forms and even factor in longer term reconstruction (box 5.3).

Figure 5.2 OECD remittance recipients in Ghana have sturdier housing and better communication amenities

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

Source: Mohapatra, Joseph, and Ratha 2009.

Figure 5.3 International remittance recipients in Ethiopia are less likely to depend on food aid or sell productive assets during food shortages

Source: Mohapatra, Joseph, and Ratha 2009.

Remittances do not involve governments, but governments affect their flows: dual or parallel exchange rate markets reflect restrictions that the government would do well to remove. Remittances—direct private flows from unaffected to affected areas —are an extension of social networks. Some communities have been known to quickly recover largely on their own. What makes some communities “vibrant”— recovering faster than others—is open to interpretation, but two defining characteristics of such communities are social cohesiveness and leadership. Their members help each other and, under able leadership, demand (and get) public services to which they are entitled. A brief account of how the Vietnamese community recovered quickly after Hurricane Katrina, in contrast to other victims, illustrates this point (Chamlee-Wright and Storr 2009).

Box 5.3 Mobilizing Haiti’s diaspora According to official statistics about a million Haitians live overseas, about half of that million in the United

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

States. Unofficial estimates tend to be significantly larger, with newspapers reporting a million Haitians in the neighboring Dominican Republic alone. This diaspora is important for Haiti’s economy. Before the January 2010 earthquake, it sent an estimated

$1.5 to $1.8 billion in remittances each year, an amount expected to increase following the earthquake. Developed countries can take measures to support this process. Following the earthquake, the United States

granted temporary protected status (TPS) for 18 months to Haitians already in the United States. The TPS allows more than 200,000 Haitians currently residing in the United States without proper documents to live and work in the United States legally, without fear of deportation. It also allows them to send money home quickly and efficiently through formal remittance channels. Preliminary calculations indicate that if the TPS resulted in a 20-percent increase in the average remittance

per migrant, an additional $360 million in remittance would flow to Haiti in 2010. And if the TPS were to be extended once beyond the currently stipulated 18 months (TPS extensions have been granted before for immigrants from El Salvador, Honduras, Nicaragua, Somalia, and Sudan), additional flows to Haiti would exceed $1 billion over three years. That would be a billion dollars of financial help coupled with goodwill and advice, tailored to the needs of the recipient. Other mechanisms can mobilize the Haiti diaspora and bring additional resources to the country to assist in

recovery and reconstruction. Diaspora bonds could be issued to directly tap the substantial wealth of the Haitian diaspora. Israel and India diaspora bonds have been used to raise more than $35 billion of development financing. Ethiopia, Nepal, the Philippines, Rwanda, and Sri Lanka have issued (or are considering) diaspora bonds to bridge financing gaps. Diaspora members are usually—though not exclusively—more interested than foreign investors in investing

in the home country. Offering a reasonable interest rate—a 5-percent tax-free dollar interest rate, for example— could attract a large number of Haitian investors who are getting close to zero interest on their deposits. If 200,000 Haitians in the United States, Canada, and France were to invest $1,000 each in diaspora bonds, it would add up to $200 million. If these bonds were opened to friends of Haiti, including private charitable organizations, much larger sums could be raised. If the bond rating were enhanced to investment grade rating through guarantees from the multi-lateral and

bilateral donors, such bonds could even attract institutional investors. Credit enhancement from creditworthy donors would help this process. Support to tax exemption of such bonds, or a public guarantee, could make such a bond more attractive. Recent estimates are that a $100 million grant from official or private donors to guarantee such bonds (say, for 10 years, on an annual rolling basis) could generate $600 million of additional funding for Haiti.

Sources: World Bank staff; Ratha 2010.

The Vietnamese community lived mainly in the New Orleans east area, which was severely flooded (from 5 to more than 12 feet), but it recovered faster than both the poorer and wealthier (Lakeview) areas equally devastated. They returned to rebuild within weeks of the storm, and by summer 2007, 90 percent of the 4,000 residents living within a mile of The Mary Queen of Vietnam Catholic Church, the physical and spiritual center of the community, were back. And 70 of the 75 Vietnamese-owned businesses in the vicinity were up and running again. In contrast, only 10 percent of wealthier Lakeview residents had returned 16 months after the storm. Similarly, only 28 percent of the ethnically diverse low- and middle-income residents of Broadmoor (in the uptown area) had returned by 2008. The Vietnamese community’s social cohesiveness accounts for its resilience. Many

had come in the mid-1970s after the fall of Saigon, and others arrived later with the help of friends and family. They helped each other evacuate their homes when Katrina

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

struck, remaining in touch with each other when they were displaced. When city officials did not help the elderly repair their homes, other members of the community did. Loans from relatives, labor exchanges, child care services, and rentals of tools and equipment were all organized, spurring the recovery. The community organized petitions to have public services restored. Father Vien

Nguyen was the senior pastor of the church that remained the hub of the community. When municipal officials rebuffed a request to restore electricity in the area, Father Vien Nguyen gave pictures of Mass attendance to Entergy, a local power company, and gathered people’s names and addresses to show that enough paying customers had returned. Power was restored by the first week of November 2005, enabling the return of non-Vietnamese residents as well. While the government decided on the many complex issues of relocation,

strengthening levees, and redirecting river flows, people rebuilt their lives and livelihoods, underscoring the major role of local communities in recovery.

Public safety nets

The term “safety net” encompasses a wide range of public transfer schemes. Some governments use an existing system to help victims of a disaster, while others begin from scratch. As chapter 2 discussed, disasters can lead to permanent effects on victims, especially on children, where malnutrition in early ages can impair cognition, reducing productivity and lifetime earnings. This suggests a critical role for safety nets: timely assistance—in food aid or cash transfers—can prevent adverse effects from becoming permanent. The need to make food available quickly may require that pre-existing stocks, plans, and systems that quickly disburse food aid are in place, such as food relief outlets in Ethiopia and the World Food Program warehouses in many countries. Timely food aid can be effective. During Ethiopia’s 1995–96 drought, food aid offset

the expected increase in child malnutrition (aged between 6 months and 2 years). In contrast, in communities that experienced the drought but did not receive food aid, a 10 percent increase in crop damage reduced child growth (also aged between 6 months and 2 years) by 0.12 centimeters (Yamano, Alderman, and Christiansen 2005). And as Spotlight 4 discusses, Ethiopian households affected by the 2008 drought that received transfers from the Productive Safety Net Program consumed 30 percent more calories than non-beneficiaries. In Bangladesh, free food relief by the government (through the Vulnerable Group Feeding and Gratuitous Relief programs) to those affected by the 1998 floods contributed between 64 and 133 kilocalories a day per person (Pelham, Clay, and Braunholz 2009). Maintaining adequate food reserves is important, but experiences with cash

transfers are becoming increasingly popular (Alderman 2010). Cash transfers, unlike food aid, give users more choice and flexibility and can stimulate domestic supply

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

where local markets exist. Assistance in cash does not, of course, increase the availability of food, cement, or other items that may be needed; but if transport links are functioning, local merchants will ensure availability. Indeed, if they do not, the infusion of cash would raise prices (Spotlight 5). Providing aid in-kind does not require local merchants, but it runs the risk of transporting at great expense some items the victims do not need or like. The challenges of designing effective safety nets should not be underestimated.

The administrative and technical capacities at various levels of government, the size of the affected population, and the depth and liquidity of (food) markets are some factors that determine the appropriate mix of responses in cash and food. Even if well designed, the practical and logistical difficulties of delivering food aid should not be misjudged. In conflict and in fragile situations, food aid can also be a weapon, exacerbating timely provision to those most in need (Spotlight 4). Support sometimes can be inadequate. In Bangladesh, food assistance delivered after the 1998 floods had a positive but limited long-term impact given the small amounts distributed because of delays in delivery (Quisumbing 2005). And bilateral and multilateral aid flows, because they rely on appeals triggered after field assessments, are often slow (figure 5.4). The central point is that to achieve a quick and organized response, safety nets

need to be in place before hazards strike. Trying to put in place safety nets after a hazard strikes is often impractical and ad hoc (table 5.1). There are exceptions—for example, if the affected population is very small, as in Maldives during the tsunami. The Maldives government devised a cash transfer system from scratch and

delivered it to some 53,000 people, about a fifth of the population, within one month of the 2004 tsunami. Teams visited all the affected islands, visually confirmed that the house was damaged, gathered all the people the next day and paid the victims in cash (between $39 and $117 equivalent, depending on the damage). A pre- and post- disaster panel survey found the targeting to be adequate and that few victims were excluded (Maldives Ministry of Planning and National Development 2006).

Figure 5.4 Timeline of key events in the Horn of Africa drought in 2005–06

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

Source: Humanitarian Policy Group 2006.

For Pakistan’s housing reconstruction, a new government agency with no experience in cash transfers rose to the challenge of targeting, disbursing, monitoring, and delivering them to remote locations during the winter. It took four months to identify beneficiaries, and cash grants began disbursing in six monthly tranches in April 2006, six months after the earthquake (Heltberg 2007). In Sri Lanka, local governments disbursed cash in four rounds, each worth about

$50, to the affected households. The first round in March 2005 covered 250,000 households, and while few victims were excluded, many who were unaffected also got the benefits. A survey found that 81 percent of unaffected households received grants (Pelham, Clay, and Braunholz 2009). Subsequent rounds sought to narrow targeting households with damaged houses or with lost income earners. In Turkey, after the earthquake in 1999, 85 percent of survey respondents were

satisfied by the authorities’ treatment and grateful for the assistance.10 Of those who asked for the accommodation and repair allowances, 95 percent got them. The Social Solidarity Fund, with a nationwide network of 900 offices, administered the bulk of the project, quickly mobilizing temporary workers, computers, and other needed resources from around the country. These varied experiences bring out the challenges in deploying safety nets (Grosh

and others 2008). Victims of a disaster may not be exactly those that an existing safety net is designed to catch. Ensuring that only the intended beneficiaries get the benefits requires administrative checks. While normal times allow a wide discussion of who the intended beneficiaries should be (rural poor, all poor, not indolent poor), the choice after a disaster requires speed to help the victims. Responsive governments can provide relief quickly, and while small size is an advantage (Maldives), large size need not be a deterrent (Pakistan).

Table 5.1 Post-disaster safety nets are common

Country Population covered Components Amounts

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

Maldives (2004 tsunami)

All affected households (one-fifth of population)

Cash grant $39–$117 per person depending on damage (equivalent to 2 to 6 weeks of average consumption)

In-kind

Pakistan (2005 earthquake)

250,000 households (30% of affected)

Cash grant $300 per household for livelihood support

Payment for death and injury

$1,660 to next of kin

Payment for housing

$2,900 per house destroyed. $1,250 per damaged house

Sri Lanka (2004 tsunami)

250,000 households in first round (all affected households covered)

Cash grant $200 per affected household, plus grants for housing reconstruction

Turkey (1999 earthquake)

206,145 households

Accommodation $4,000 for accommodation aid and $1,430 per house for damage repairs

Repair

Death and disability

$1,790 to next of kin and $950 (on average) for disability Source: World Bank staff, based on Heltberg 2007.

One aspect of safety nets germane to disasters is that traditional safety nets can fail to reach vulnerable groups (children, women, elderly). Why? Because they tend to target beneficiaries by income or indicators not wholly relevant to vulnerable groups, such as land or asset ownership (Baez, de la Fuente, and Santos 2009). More relevant indicators—such as nutritional screening and disabilities—could overcome this, though collecting some of these data for short-term use can be expensive. Keeping in mind the practical difficulties of providing timely in-kind food assistance, better targeting of children may be achieved by adding ready-to-use-foods (RUFs) to general food distribution. RUFs do not require water (compared to powdered milk- based supplements which do), and are increasingly locally produced and do not spoil that easily.

Aid and the Samaritan’s dilemma

Buchanan (1975) used the analogy with a Good Samaritan who attempts to assist those in need. But if the Samaritan cannot credibly commit to denying help to the negligent, that could encourage carelessness (Gibson and others 2005). Post- disaster humanitanitarian aid, whether from bilateral governments, multilateral agencies, or NGOs, is caught in this dilemma. Several studies have examined the determinants and efficacy of foreign aid in general, but few examine post-disaster aid specifically. Some theoretical models suggest that ex-post aid reduces ex-ante prevention (Raschky and Weck-Hannemann 2007 and Cohen and Werker 2008). In a background paper for this report, Raschky and Schwindt (2009a) empirically

examine this link and find weak evidence that increases in the level of past foreign aid imply higher death tolls resulting from disasters. They construct a model where aid flows could increase collective prevention but predictable ex-post relief could reduce it. They then examine empirically which effect dominates by regressing mortality in 1,763 disasters (divided into three subsamples for storms, floods, and earthquakes)

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

on aid inflow the previous year, controlling for the effects of other factors (hazard exposure, population size, institutional quality, colonial past). The statistically significant positive coefficient of the main variable of interest,

humanitarian aid per capita, implies that more aid in the past is associated with additional deaths from storms. But the results for floods and earthquakes are not significant. There is no apparent reason why the storm results are statistically significant but the flood and earthquake results are not. More research is needed to understand this divergence, and the results need to be interpreted with caution. In a companion background paper, Raschky and Schwindt (2009b) extend their first

study by considering the type and channel of aid. A donor could provide bilateral aid or contribute to multilateral assistance, and could do so in cash or in kind. They examine aid after 228 disasters over eight years (2000–2007). Oil and trade access are two major motives of aid (despite the humanitarian label) that are distinguished. They find that the number of people affected, but not the number of fatalities, are related to the choice between bilateral and multilateral aid. More distant countries get aid multilaterally while those with a higher fraction of fuel exports and better governance indicators get more bilateral aid, perhaps because giving directly bolsters the donor’s influence. Multilateral aid may be distributed on a “needs” basis and so the recipient’s income, governance indicators—and fatalities—may matter. Bilateral donors also favor more open recipient countries, so a larger fraction of fuel exports makes aid more likely. These findings echo those of Fink and Redaelli (2009), who analyzed 400 recent disasters and found that while needs influence relief aid, so do geographic proximity, cultural and colonial connections, and oil exports. The findings suggest that donor self-interest matters (Olsen, Carstensen, and Hoyen 2003).11

Some observers have noted the disincentives of donor programs. For example Nicaragua declined to pursue a weather indexing program after it had been priced in the global reinsurance market: it cited international assistance following Hurricane Mitch in 1998 as an indication of dependable alternatives (Alderman 2010). It may be unfair, though, to blame countries for neglecting prevent-ion:

Mozambique, anticipating major floods in 2002, asked donors for $2.7 million to prepare and got only half the amount, but $100 million were received in emergency assistance following the floods, with another $450 million pledged for rehabilitation and reconstruction (Revkin 2005).

Figure 5.5 Disasters receive about a fifth of total humanitarian assistance

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

Note: Humanitarian aid is “an intervention to help people who are victims of a natural disaster or conflict meet their basic needs and rights,” while official development assistance (ODA) is “money spent on development (education, health, water supply and sanitation, agriculture, and so on) and humanitarian assistance by members of the OECD Development Assistance Committee.” Source: World Bank staff based on data from the Financial Tracking System (FTS) of the UN Office for the Coordination of Humanitarian Affairs.

The report’s overarching theme is that not enough is being done to prevent disasters. Donors usually respond to them only after they strike. About a fifth of total humanitarian aid between 2000 and 2008 was devoted to spending on disaster relief/response (figure 5.5). The share of humanitarian funding going to prevention is small but increasing—from

about 0.1 percent in 2001 to 0.7 percent in 2008 (Harmer and others 2009). But prevention endeavors often imply long-term development expenditures whereas the focus of humanitarian aid—already a tiny part of official development assistance—is immediate relief and response. Donors concerned with prevention could specifically earmark development aid (rather than humanitarian aid) for prevention-related activities. And such aid, if used effectively, could reduce issues arising from the Samaritan’s dilemma: the inability to deny help following a disaster to those who have not taken sufficient prevention measures.

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

Spotlight 5 on the 2004 Tsunami

Warnings: The most effective prevention? Underwater earthquakes occur where tectonic plates meet around the Pacific rim, and severe quakes trigger tsunamis (spotlight figure 1).1 Tsunamis are rarer in the Indian Ocean than in the Pacific. But a devastating one occurred on December 26, 2004, triggered by a massive 9.3 Richter scale2 underwater earthquake that released the energy equivalent of 32,000 Hiroshima-size atomic bombs in the first waves. Its epicenter was just northwest of Indonesia’s island of Sumatra, and its waves spread at 700 kilometers per hour, making landfall at different times (spotlight map 1).

spotlight figure 1 Tectonic plates slipping at fault lines generate a tsunami

Source: U.S. Geological Survey Web site. http://walrus.wr.usgs.gov/tsunami/basics.html.

Spotlight map 1 Indian Ocean tsunami travel time in hours

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

http://walrus.wr.usgs.gov/tsunami/basics.html

Source: U.S. Geological Survey Web site http://staff.aist.go.jp/kenji.satake/Sumatra-E.html.

Spotlight figure 2 Banda Aceh’s shore, before and after the tsunami

Source: DigitalGlobe.

The force of the waves determines the devastation: distance from the epicenter moderates the force, but the contours of the coast and sea bottom may concentrate it. The slope of the land underwater near the shore slows the waves but raises their height: the first wave that struck Aceh, Indonesia, 15 minutes after the quake was over 20 meters high. In some low lying areas, sea water penetrated three kilometers inland, destroying the crops and ruining the land with salinity. The tsunami reached the Thai coast 40 minutes later, devastating some major tourist resorts but sparing others partially protected by the bays. Waves later reached India’s Andaman Islands, villages along India’s southeastern coast (Tamil Nadu state), and western parts of Sri Lanka. The severity of damage varied. Some 230,000 people died, mostly in Indonesia (73 percent) and Sri Lanka (18

percent), many more were bereaved, and 1.7 million were displaced. Damage was

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

http://staff.aist.go.jp/kenji.satake/Sumatra-E.html

particularly severe in Indonesia, especially in Aceh, the poorest of its provinces (spotlight figure 2). Many survivors lost much of what enabled them to live: coastal fishermen in Aceh,

Tamil Nadu, and Sri Lanka lost their boats and nets, and some farmers lost their fields to salinity and permanent inundation.3 And even though many of Thailand’s tourist resorts were physically spared, tourists fled and were reluctant to return. GDP in tourism-dependent Maldives fell 80 percent the following year.

Unprecedented response

News of the devastation spread quickly. Thailand’s resorts were filled with camera- toting European tourists enjoying their Christmas break. Although Europeans were fewer than 1.5 percent of the dead, the pictures they took and their heart-rending stories flashed around the world, producing an outpouring of aid. Government and official aid agencies planned their response in the days following

the tsunami, but it became increasingly clear that coordinating private attempts to help would be a challenge considering the amounts. Aid commitments of $13.5 billion greatly exceeded total damage estimated at $9.9 billion, mostly to private property. Most aid went to Indonesia (over $7billion), and the bulk was through nongovernmental entities. The Indonesian government created a special agency, the BRR, which managed roughly a third of the total contributions. The World Bank managed a multi-donor fund for official bilateral aid.

Unintended effects: Some waste, but avoidable?

A multitude of small nongovernmental organizations (NGOs) eager to help poured in: 435 in Aceh alone. Each brought funds for a specific purpose, but the typical NGO project was small, and “brick and mortar” projects were preferred so that they could “show” contributors how their funds were spent. This preference meant that some spending on intangibles was neglected unless the government did so. While many projects were well managed, some NGOs bought land and built houses

—often before the government decided where they would build roads or provide drinking water, sewerage, and other public infrastructure. So, some newly built houses were in areas unsuitable for townships; some houses that were ceremoniously handed over to victims had no infrastructure and remain unoccupied. The waste is difficult to estimate: the government, not being involved, keeps no

statistics. The report of the Tsunami Evaluation Coalition provides numerous examples of waste in relief spending.4Items that were not useful—canned pork to Muslim Indonesia, 75 metric tons of expired drugs, inappropriate western clothes—took up precious cargo space. Masyrafah and McKeon (2008) have other examples: an NGO provided boats to local fishermen that were left to rust because they were poorly constructed and of unfamiliar design.

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

Waste results from cultural unfamiliarity, especially with well-intentioned help in kind, but the army of aid workers also puts huge demands on the local economy, producing unintended effects. Hiring local workers at “fair” wages so distorted the local labor market that the able- bodied stopped fishing or farming to wait and cook for foreign aid workers on whose continuing presence they had come to depend. Even “well targeted” aid has unintended effects and can benefit those who may not

need help: the incidence of aid (who benefits) is not always obvious. The infusion of cash in the local economy, both as direct grants and spending on services, resulted in inflation (which has distributional effects), as with the spurt of inflation in Banda Aceh (the provincial capital and largest city of Aceh) and in the region’s second city, Lhokseumawe (spotlight figure 3). These effects occur in many post-disaster settings, but the scale of the aid after the tsunami enabled the effects to be readily observed.

Spotlight figure 3 Inflation in Aceh well above the national average

Source: Masyrafah and McKeon 2008.

While some of the waste and unintended effects could have been avoided, many victims and survivors did benefit. Furthermore, the Indonesian government’s considerate and efficient response even stilled the conflict with the Aceh provincial government that had been simmering for 30 years. But such humanitarian impulses do not always overcome animosities.

Warnings and preparedness save lives

The tsunami caught most by surprise, but not all. Scientists at the Pacific Tsunami Warning Center in Hawaii frantically called counterparts in the vulnerable countries to alert them of the inevitable tsunami after detecting and locating the massive quake. Those they reached lacked the authority to act; so public warnings were not always issued or acted on. But some were prepared. A 10-year-old English schoolgirl, having just studied

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

tsunamis in school, deduced from the sudden draining of the sea that a tsunami was imminent. Her family holidaying in the Thai resort of Maikhao Beach fled to safety along with some 100 other tourists. Similarly, the Onge tribe in the Little Andaman Island retreated to higher ground when they felt the quake, as did those on the Indonesian islands of Simeulue, showing that knowledge embedded in culture usefully augments the direct experience of any generation. Simply asking people to go a short distance inland to higher ground would have

saved many lives because the waters quickly abate. Only one person died in Kenya, where radio and television stations discovered news reports on satellite television and broadcast warnings three hours ahead of the tsunami. But 300 died eight hours after the earthquake in neighboring Somalia, because such news was not monitored. Countries are making efforts now to improve warnings, and many are taking

advantage of newer technologies as well. Short message services (SMS) may reach more people quicker, especially with the huge increase in the use of cellular telephones.5In the wake of the 2004 tsunami, Sri Lankan authorities sent out a text message for village chiefs and media outlets that could spread the word to people who did not have mobile phones.6

Warnings alone are not enough: people must know where to go when they receive one. Before the tsunami in 2004, preparedness plans had been completed for only five of Sri Lanka’s 25 districts (de Mel, McKenzie, and Woodruff 2008). But even after the tsunami, an Oxfam survey found that only 14 percent of Sri Lankan survivors knew what to do next time (Oxfam America 2006, p.11). Clearly much remains to be done: evacuation routes and locations must be designated and regular evacuation drills held. Such preparations may also guard against other more frequent hazards.

Zoning for safety: Easier said than done

The tsunami destroyed about 100,000 houses and severely damaged almost 50,000 more in Sri Lanka. A majority of the damaged or destroyed houses were within 100– 200 meters of the sea. Residents were told they would not be allowed to reconstruct in the coastal zone. And while UN Habitat and numerous NGOs announced housing projects away from the coast, many refused to move because they fished and the move would disrupt their livelihood. The law requiring them to move was so unpopular that it was later rescinded. Meanwhile, many reconstructed or repaired their houses in the coastal zone without government assistance.7

Sri Lanka’s experience illustrates the importance of dealing with the social and economic context of communities, not just housing in safe locations. While the speedy rebuilding of shelters is understandable, they may remain unused unless communities move away from unsafe areas—and this requires community involvement and sustained efforts in finding the most appropriate remedy for each situation.

Prediction difficult, detection easy, but cooperation essential The, World Bank, Bank World, and Nations (UN) United. Natural Hazards, UnNatural Disasters, edited by World Bank The, et al., World Bank Publications, 2010. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/apus/detail.action?docID=606027. Created from apus on 2017-04-10 14:36:39.

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .

Earthquakes are difficult to predict, though this will undoubtedly improve with technology and our understanding. The ability of some animals to sense impending quakes (notably elephants who move to higher ground) suggests that instruments may be able to measure what these animals can detect in advance. But a tsunami follows a major quake, so its path and likely destructive force could be charted—and warnings issued—if the period between earthquake detection and tsunami prediction could be shortened. Underwater monitoring instruments and instruments to monitor ocean surface

movements help. But while there are several in the Pacific (where the “rim of fire” makes earthquakes and tsunamis common), there were none in the Indian Ocean. Such instruments are now being installed, and this expense may be worthwhile if they also collect data useful for purposes other than the rare tsunami. Not all countries have good seismic facilities, and it would be helpful if those that do shared real-time data. Some countries are reluctant to do so, especially in real time, because they could also be used for other purposes (such as monitoring its nuclear tests). So, saving lives and reducing damage require an organized response to the

warning but this expense is warranted only when exposure and hazard frequency are sufficiently high (chapter 5).

C o p yr

ig h t ©

2 0 1 0 . W

o rl d B

a n k

P u b lic

a tio

n s.

A ll

ri g h ts

r e se

rv e d .