For Eng.Kelvin only

chapter_4.pdf

Home >Reading homework help >For Eng.Kelvin only

CHAPTER 4

Prevention through Governments

ational, state, and local governments with the power to tax are responsible for many major prevention measures; but regardless of the political system, they

respond to the wishes of the people—at least some of them. People also act collectively through other entities, formal or ad hoc, many rooted in traditions: villages gathering to clean out the irrigation ditches for example. These organizations play an important though unheralded role in many economies: without them, governments are less effective. The chapter begins by discussing how much governments spend on prevention.

This requires a detailed grasp of budget accounting because prevention is not a specific budget item, and prevention is embedded in infrastructure investments, maintenance, and other spending. In four chosen countries, identified prevention spending was lower than post-disaster spending. But this does not necessarily imply it is “too little,” only that disasters increase spending on relief and that such expenditures remain high for several subsequent years, perhaps for good reasons. The effectiveness of prevention spending is more important than its magnitude, and some indicators can suggest the benefits of reversing the past neglect of maintenance and other types of preparedness spending. The chapter next examines who determines government spending. It is easy to

assert that politicians are shortsighted. But competition in the markets for votes, like other competitions, provides the public with the services they want—with a twist that arises when voters can observe inputs (building a levee), not outputs (protection from floods, also requiring other unobservable actions). So, even if voters want prevention, they could vote against such spending if they doubt that it would result in effective protection. The chapter then discusses how to improve collective decisions. Institutions and

political competition improve collective decision making, and against this backdrop, cost-benefit analysis is a useful guide to spend effectively. For disaster prevention in particular, ignoring the value of life tilts the balance against prevention; but using such values requires ethical considerations and a deeper appreciation of the tool. Cost- benefit analysis is a filter, not a scoop: it can rank alternatives, but the alternatives must be conceived by others. Last, it examines three items that have public-good characteristics related directly

to prevention. An early warning system is one such choice of great benefit to some

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 2017-04-07 06:21:02.

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 .

countries and in some places because warnings save lives and property. They are based on hazard warnings. All countries can benefit from modest but well-allocated spending on such systems and from sharing data among themselves. Critical infrastructure reduces the loss of life and property during and after a

disaster, and what is critical depends on the situation and the hazard. In Bangladesh, safe schools are important shelters during disasters. In Turkey, hospitals may be critical because earthquakes result in injuries. But critical is not a synonym for its importance in normal times, and the choice requires informed judgment. For environmental buffers, it is cheaper to protect than to restore them.

Development, including sustainable development, involves change, and choosing what to protect requires a broader understanding of the forces of nature and their effects. Much of the cost-benefit analysis in this area is flawed, and careful analysis is difficult, but important.

How much do governments spend? Governments do not routinely collect or monitor spending on disaster prevention. Budgets are often allocated by ministries, but even if a “Ministry of Disaster Prevention” existed, it would have little to do. Most preventive measures are embedded in the design and construction of infrastructure (such as the location and height of a dam) or in other spending (such as school buildings that serve as shelters). So, measuring prevention spending requires much effort and considerable judgment to identify spending categories across sectors and levels of government and to collect budgeted amounts. This was attempted for this report in Colombia, Indonesia, Mexico, and Nepal. Local consultants drew on their own knowledge and that of the governments’

disaster management organizations using a common template to separate spending on prevention and relief. Pre-disaster spending includes expenditures on identifying risks (risk mapping and hazard assessments), risk reduction (physical/structural works to withstand damage), risk transfer (insurance), and disaster preparedness (early warning systems and public training and awareness about risks and prevention). Post-disaster spending includes expenditures on emergency response (search and rescue operations, relief), rehabilitation, and reconstruction (repairing and reconstructing houses, commercial establishments, and public buildings). Except in Colombia, pre-disaster spending was generally lower than post-, spending on relief fluctuates far more than on prevention, and relief expenditures rise after a disaster and remain higher than prevention spending for several subsequent years (de la Fuente 2009) (figure 4.1). For instance, Mexico’s relief spending rose after the 2005 hurricanes and the 2007 floods (in the southeastern state of Tabasco) and remained three times higher than prevention spending from 1998 to 2008.

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 .

Figure 4.1 Post-disaster spending fluctuates more than pre-disaster spending

Source: de la Fuente 2009.

Though one cannot conclude from this alone that prevention is “too little” (or that relief is “too much”), this exercise is the first step in systematically estimating how much is spent on pre- and post-disaster management. If and when data are available, these estimates can be further refined by:

Tracking expenditures at subnational levels. With decentralization in many countries, many prevention measures are now undertaken at subnational levels, as in Turkey, where the disaster risk management cycle was highly centralized but is now being decentralized (see Spotlight 2). Accounting for measures indirectly related to prevention. For example, any anti-poverty policy or program, which, even though not disaster specific, reduces vulnerability or exposure. Accounting for those post-disaster expenditures where the reconstruction of buildings and infrastructure includes disaster-resistant measures that lead to future prevention. Doing so would provide a basis for tracking global expenditures on prevention and for related policy implications in hazard- specific and geographically specific contexts. But the data constraints 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 .

resource requirements for doing so should not be underestimated.

What, then, of allocation and effectiveness of spending? Too little goes for intangibles and maintenance. Effective spending has high rates of return but is difficult in practice. A cost-benefit analysis is a useful ex ante guide, and ex post evaluation ensures that lessons are learned. But rarely is either used. So we grope for indicators that seem reasonable (but will not persuade a skeptic that much spending is poorly allocated and ineffective). For example, infrastructure built appropriately to reduce disaster risk may not be

sufficiently maintained, lowering the effectiveness of the original capital spending. About 30 percent of infrastructure assets of a typical African country need rehabilitation (figure 4.2), and just $0.6 billion on road maintenance would yield $2.6 billion in annual benefits (BriceñGarmendia, Smits, and Foster 2008). Government neglect of maintenance is similar to neglecting spending on other

intangible items that yield future benefits, such as environmental protection and education (World Bank 2000; López and Toman 2006). Per capita spending, except for rapidly growing Ireland and the Republic of Korea, is greater for physical capital than for intangibles, which also have high rates of return (figure 4.3).1

Figure 4.2 Underspending on maintenance implies an enormous infrastructure rehabilitation backlog in Sub-Saharan Africa

Note: The rehabilitation index shows the average percentage across countries of each type of infrastructure that is in poor condition and thus in need of rehabilitation. Source: BriceñGarmendia, Smits, and Foster 2008.

Figure 4.3 Per capita spending is greater for physical capital

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: Adapted from López and Toman 2006.

In Vietnam, because of a rule that the growth rate of capital expenditure should be higher than the growth rate of recurrent expenditure, capital budgets grew faster than recurrent budgets. A decline in recurrent expenditure has been particularly acute in transport (figure 4.4) while capital expenditure boomed. Were expenditure to remain at its current level, the percentage of national roads in good condition would fall to just 10 percent of the total network. In its funding request for 2003 to 2005, the Vietnam Roads Agency secured less than half of the finance required to cover all maintenance needs on national highways (World Bank 2007). Poor coordination between capital and maintenance expenditure is common in

countries that operate dual budgeting systems. The introduction of medium-term expenditure frameworks may help address the issue, since a medium-term perspective helps highlight the capital savings offered by adequate maintenance. Implementing such frameworks effectively, however, is fraught with difficulties if the institutional environment remains poor.

Figure 4.4 Vietnam’s recurrent spending is a low and declining share of total transport expenditure

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: World Bank 2007.

The examples are related to disaster prevention since some types of infrastructure and natural barriers help prevention. But a government that misallocates important spending items is also unlikely to spend effectively on prevention.

Who really decides government spending? Economics has made useful contributions to political science since Mancur Olson (1971) examined how interest groups form and influence collective decision making. The benefits and costs of government spending are spread unevenly, providing an incentive for groups to form and influence spending and policies in their favor. This holds for all governments: democracies differ only in that some aspects are voted on periodically. A government is a complex organism, and its inner workings are rarely visible and poorly understood. Yet it provides important services that require funding. So who really decides how much to spend and on what? The public, neither fully informed nor entirely selfless, may prefer the government to spend on what benefits them yet accept spending that benefits others. Similarly, politicians are neither completely corrupt nor wholly idealistic. And civil servants are not always civil or serving the public. Empirical studies complement this analytical strand, but they are limited by what

one can observe and measure: voting patterns, who funds politicians, what legislation officials approve, and so on. Such work requires data that are available in democracies (mainly available for the United States and India), though the same forces operate elsewhere, including under closed dictatorships. Two different groups that influence the adoption of prevention and relief measures

are politicians, voters, and the media on one hand; and foreign donors on the other, especially in poor countries where they may have some influence. The following section examines the first group, and foreign donors are discussed in the next chapter.

Relief spending responds to media attention

Spending on relief increases with media coverage. Besley and Burgess (2002) find 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 2017-04-07 06:21:02.

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 .

that politicians respond with greater alacrity to disasters that the media cover. And though their regressions reflect correlation (and a common underlying cause), a causal direction is plausible. They find that newspaper circulation increases the government’s disaster responsiveness: a 10 percent drop in harvests increases public food distribution by 1 percent in states with median per capita newspaper circulation, but in states at the 75th percentile in circulation, food distribution rises by 2.3 percent for the same drop in harvests. Francken, Minten, and Swinnen (2008) investigate what drove relief to 249

communities affected by cyclone Gafilo in March 2004 in Madagascar. Access to radio increased the probability of government relief by 24 percentage points, consistent with the results of focus group discussions where half the communes believe that the media influence politicians’ decisions and improve responsiveness. And the probability of government relief was 65 percentage points higher in communities where the majority supported the president during the 2001 elections. Such effects are the same in developed countries. For about 5,000 disasters

occurring outside the United States between 1968 and 2002, the U.S. government’s relief response was often crowded out by other noteworthy media events clearly unrelated to disasters (such as the Olympics or the World Series) that coincided with the disaster (Eisensee and Strömberg 2007). For example, disasters are on average 5 percent less likely to receive relief during the Olympics than at other times. A disaster occurring during the Olympics must also have three times as many fatalities than a disaster on an ordinary day to have an equal chance of receiving relief. Nearly half of all Federal Emergency Management Administration (FEMA) disaster

relief payments in the United States were motivated by politics rather than need (Garrett and Sobel 2003). And presidential disaster declarations, often a prerequisite for federal aid, are more frequent in election years, though disasters and electoral cycles themselves are clearly unrelated (Sobel and Leeson 2008). Under the current U.S. system of disaster assistance, a state governor may ask

the president to declare a “major disaster.” The president does not unilaterally determine the amount of aid that follows (the House and Senate must approve, though they usually concur), but is responsible for a necessary step and may benefit politically as a consequence. What drives the declaration, when some states benefit and others share the cost? Many (though not all) of the peaks correspond to presidential election years,

consistent with disaster assistance often being an electoral issue that rewards incumbents (figure 4.5).2

Figure 4.5 Presidential disaster declarations: Peaks often coincide with presidential election years

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: Kunreuther and Michel-Kerjan 2009.

So while disasters may strike at any time, presidents are more likely to declare disasters during re-election campaigns. When analyzing flood-related presidential disaster declarations between 1965 and 1997, the higher number of declarations in re-election years (28.4 versus 19.4) is statistically significant (Downton and Pielke 2001).3

When relief spending rises but prevention measures are ignored, does it suggest politicians’ collective myopia, or is it a result of voter preferences? Analyses of electoral data in the United States and India shed some light on this.

Do voters or politicians prefer relief over prevention?

Voters appear to favor relief over prevention spending. Healy and Malhotra (2009) examine voting patterns, disaster damage, and federal government spending for five U.S. presidential electoral cycles (1988, 1992, 1996, 2000, and 2004) in all 3,141 counties. They disentangle voters’ responses to events outside the incumbents’ control (such as hurricanes) from those that they do control (such as relief and prevention spending). They find evidence of underinvestment in disaster prevention and conclude that a dollar spent on prevention is more than ten times more valuable than a dollar spent on relief in net present value. They are careful in their interpretation: they contrast this finding on disasters with the excessive spending to protect against a repeat of the 2001 attacks using passenger aircraft; so it is just preparedness against natural hazards that is insufficient. Voters appear to behave this way when policies have future benefits and when issues have low salience for political platforms, as with disasters. In India, there are more votes against incumbents from the ruling party after rain-

related disasters—even when the government provides large assistance to farmers (Cole, Healy, and Werker 2008). Voters reward incumbents for relief if they think the losses were from bad luck, not government neglect (presumably the maintenance and operation of dams and irrigation canals)—more evidence of voter sophistication. Voters appreciate relief, but relief spending has only a small effect on re-election prospects: average relief spending reduces the probability of losing the election by

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 .

one-seventh relative to no relief spending. There is also a robust link between public food distribution and disaster relief

(droughts, floods) as well as political measures such as election turnout, political competition, and the timing of elections (Besley and Burgess 2002). For a given fall in food production or increase in crop damage, greater political competition (turnout in state elections in the previous period) is associated with higher public food distribution and calamity relief. The Indian government’s disaster relief measures seem to reflect voter preferences. So, if politicians are responsive in elected periods, are the voters myopic (cannot

see the future benefits) or do they misperceive the risk of disasters (think expected future benefits of prevention are low)? Chapter 2 discussed and put aside the third possibility of a high discount rate for the poor and summarized recent research on risk misperception. That people misperceive risk in experimental settings lends credence to the view that voters may misperceive risks; but insufficient prevention is equally consistent with far-sighted—but skeptical—voters acting in their self-interest. Effective prevention requires myriad measures that work harmoniously together: for

example, flood prevention requires appropriately sited dams and, when there is heavy rainfall or snowmelt upstream, their sluices must be opened and closed at the right time and sequence to hold the rushing waters in the available reservoirs. With enough storage capacity, floods could be prevented. But if reservoirs were already mostly filled, the authorities must quickly decide where to redirect the floods: ideally, where the least value would be lost. Warnings and evacuations must also be coordinated. While voters may not know all the intricate details, they do know (given the history of floods) when they are not protected. If, under these circumstances, the voter must choose between getting cash relief and spending on a levee—just one note in the intricate prevention symphony—they may vote for relief even if everyone wants effective prevention. Voters may be less prone to vote for public goods provision when there is a

substantial ethnic diversity or social fragmentation. Productive public goods—roads, sewers, and trash pickup—in American cities are inversely related to the city’s ethnic fragmentation, which in turn is negatively related to the share of local spending on welfare (Alesina, Baqir, and Easterly 1999). Voters chose lower public goods when a significant fraction of tax revenues collected on any given ethnic group was used to provide public goods shared with other ethnic groups. These results suggest that public good provision requires some sense of community and could increase with a more cohesive society. The empirical findings are striking, but the intelligence of voters, when fully

informed, to look past the labels and promises should not be underestimated. Indeed, voter preferences tend to result in more prevention in countries that have more effective governments and better institutions.

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 .

Incidence and voice: Why they matter

Ignoring incidence—who ultimately bears the burden of an intervention—can also undermine collective prevention measures, particularly for those most affected. It is easy—and all too common—to use country and governments synonymously with victims. But victims are overwhelmingly poor households, most official aid goes to governments, and relief and prevention spending does not always benefit victims. Government actions reflect the preferences of those who influence its decisionmaking; if marginalized sections of society—often the very poor—have little economic clout or political voice, their well-being gets ignored. So incidence is of great concern, particularly if choices on collective prevention do not reflect their preferences. If the poor have little voice, decisions to spend and locate large scale protective infrastructure may either bypass the poor completely or result in their dislocation—with often little or no compensation—if it turns undesirable land where they reside into coveted real estate. Developing this land may well displace poor residents to other risk-prone parts of a city or places far from economic opportunity. Moreover, because they would be dislocated, the poor would not even reap the benefits of protective infrastructure put in place. If the poor therefore do not have the opportunity to influence public goods decision making, spending and allocation of collective prevention measures could be biased against those most at risk. In this illustration, early warning systems rather than protective infrastructure may have served the poor better. What happened in Indian states when the poor were not consulted in the use of anti-drought funds (box 4.1)?

Box 4.1 India and anti-drought funds In his book Everybody Loves a Good Drought: Stories from India’s Poorest Districts, journalist Palagummi Sainath details how measures to manage drought in the mid-1990s in the states of Bihar, Maharashtra, and Orissa were appropriated by the influential at the expense of the poor. The central government Drought Prone Areas Programme (DPAP) was put in place to manage and reduce the effects of drought. But the selection of the DPAP block became politicized because monetary benefits followed. For example, the town of Lonavla— with abundant rainfall (seldom below 1,650 millimeters annually and sometimes as much as 2,000 millimeters) —was designated as a DPAP block. DPAP blocks in Maharashtra grew 73 percent of sugar cane, a highly water-intensive crop, and the irrigated area in DPAP blocks was almost 50 percent higher than the state average. Meanwhile the poor in drought-stricken areas were not consulted and did not participate in the use of anti-drought funds.

Source: World Bank staff.

How to improve collective prevention measures Having examined how much is spent on prevention and who decides on such spending, the chapter turns to how collective prevention could be improved. Prevention need not be only through governments, and Ostrom’s work is a good reminder that alternatives abound, especially in cohesive communities (see Spotlight

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 .

3). But governments do provide collective goods and services. Much of the discussion is on how governments could improve prevention, particularly through institutions and political competition. Because specific actions are often at a country level, we outline a useful tool—the familiar but oft-neglected cost-benefit analysis, which must be used with care and sensitivity, especially when choices require ethical judgments such as putting an explicit value on lives. We then turn to early warning systems, critical infrastructure, and environmental buffers, where all countries can expect to reap large benefits by reducing the deaths and damage from disasters.

Institutions and political competition improve decisions

Well informed citizens are more likely to vote, especially for candidates who further their interests (World Bank 2002). A better informed electorate makes a government more responsive, especially if the information is translated into easily understandable “scores,” as has been done in Bangalore, India.4 So the development of trustworthy entities that “digest” information would improve accountability and thus the effectiveness of government relief spending. But what makes for the emergence of trustworthy entities? Countries that prevent

death and destruction better than others appear to have something—institutions—that work better. What these institutions are and the mechanisms they operate through are unclear, but they manage to inform voters and politicians who approve the needed spending and ensure prevention. Two studies drive home the point: Kahn (2005) finds that rich countries do better, and Keefer, Neumayer, and Plümper (2009) underscore the salubrious effect of competing political interests. Kahn (2005) finds that geography matters: Asia is 28 percent more likely to have a

disaster than Africa.5 But income—which proxies for the quality of institutions—also matters. Richer countries, even though they do not experience fewer disasters, incur fewer fatalities. Deaths are less likely (statistically significant) in countries with higher per capita income: 28 percent less likely in a country with a per capita income of more than $2,000. Less democratic countries and those with greater inequality suffer more deaths. Sen’s (1982) observation that famines became less frequent in India after independence suggests that self-rule and democracy ensure greater government responsiveness to people’s needs. But some states within India do better than others, and similarly some democratic countries do better than others. In background work for the report, Keefer, Neumayer, and Plümper (2009) find that

differences across countries in disaster mortality can be explained by more than just whether political decision makers are chosen in competitive elections (the conventional notion of democracy). Also critical is the degree to which citizens are informed and the ability of politicians to make credible commitments to (most) citizens. Key components of political credibility are political parties that allow citizens to hold

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 .

them accountable for success or failure and that allow individual politicians to make credible promises to pursue public policies in the broad public interest. Across both non-democracies and democracies, the existence of “institutionalized” political parties is significantly associated with reductions in disaster mortality. For example, mortality from earthquakes falls by 6 percent for an additional year of competitive elections and by 2 percent when the average party age rises by a year. Institutionalized party systems are therefore more likely to respond to citizens’ needs, with or without competitive elections. Some non-democracies embrace institutionalized ruling parties, bureaucracies, or militaries that facilitate effective responses to disaster; others do not. The broad finding is consistent with Sen’s (1982) observation that democracy helps

in responding to emergencies and disasters because voters hold governments accountable. But voting alone is neither necessary nor sufficient. A wide variety of political systems can serve the purpose, and “institutions” are needed to inform all concerned about the alternative prevention measures available, their cost, and their effectiveness. Incentives matter, and political competition could drive the spread of information; but some institutions work better than others for reasons not fully understood.

Cost-benefit analysis: A subtle and sensitive scalpel

Information and new technology increases choices, but how to choose among them? Collective choice requires alternatives to be narrowed and, if not ranked, their distributional implications examined. Cost-benefit analysis is especially useful when the issues are complex and there are several competing proposals. An investment whose benefits exceed the costs should be undertaken; and if there

are competing proposals, the one with the highest benefit-cost ratio should be preferred.6 Cost-benefit analysis is a well-known tool, particularly useful for governments seeking to compare alternatives (such as the private sector’s profit measure). Its use has declined over the years, even at the World Bank (Garcia, forthcoming 2010). To arrive at the right choice when prevention saves lives requires valuing them.

Valuing lives may be abhorrent to many and is always controversial. But ignoring it implicitly considers people useless—and it would be unethical and unfortunate if property is protected but lives are not. For example, background work done for the report shows how, if the value of lives saved were ignored, retrofitting buildings in the Turkish district of Atakoy would not be cost-effective, with a benefit-cost ratio lower than 1. Background work done for the report finds that including a value of life of $750,000 in the benefits, however, tips the scales toward retrofitting (IIASA/RMS/Wharton 2009). And only by including the value of lives saved (at $400,000 each) did earthquake strengthening measures for apartment buildings and schools in Turkey pass the cost-benefit test (Smyth and others 2004a, 2004b).

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 .

Box 4.2 Valuing life: Worthless, priceless, or useless statistic? Some consider life priceless—and it is. But people make choices regarding the value of life both for themselves and for others, if implicitly (mandating polio vaccine for children benefits many but a few succumb). “Human capital” is a dehumanizing term but a useful concept. One could measure the “cost” of education

(that parents or “the country” spend); but this is an “input” that together with nutrition, parents’ time, and so on “produces” human capital. Such human capital is combined with its physi-cal (machines) and natural (land) counterparts to produce output. While the cost of education is the usual measure of human capital, one could infer its value from what it produces, and the two measures differ: Bill Gates or Warren Buffet earn far more than their education cost, while scientists like Albert Einstein may not. The point is that such measures, while useful for some purposes, cannot value the sum total of a person’s life or their contributions to society. Individuals often make choices from which one could infer the value they attach to their own lives. For

example, the willingness to take up riskier jobs for higher pay allows one to use the increased risk and reward to calculate the value of statistical life, or VSL. Such estimates are based on revealed preference and not surveys, an alternative but highly flawed technique. Even so, the result is a wide range of values, in part because data and econometric techniques have limitations. Moreover, because the estimation technique assumes a particular functional form, these estimates are valid only within the range of risk observed. VSLs are often used in cost-benefit analysis; but they are not a measure of what is “lost” when a person

dies. The family also values a breadwinner’s companionship and contributions to raising children. How to value these? If injured, valuing pain and suffering is also difficult. In a background paper for the report, Cropper and Sahin (2009) review the literature on valuing death and

injury and suggest how they could be roughly estimated. There are few empirical esti-mates of VSL for developing countries, but estimates from high-income countries could be transferred for use in middle- and low-income countries. When reductions in deaths and injuries are an important part of project benefits, calculating the reduction in injuries and deaths in terms of quality-adjusted life years (or “QALY,” a year of life adjusted for its quality) is reasonable. The costs of the project, minus the non-health benefits of the project, can then be divided by the QALYs saved to calculate a cost per QALY avoided. One advantage of this approach is that it would be easy to compare the cost per QALY across policies—to reduce disaster risks and across health and safety policies in various sectors—to encourage consistency in decisionmaking.

Source: World Bank staff.

Deep ethical and philosophical factors must be considered in attaching a value to life, especially if the decisions affect others (as collective actions do). Economists must be aware of their tools’ limits (box 4.2). Regardless of whether lives are valued and what value is attached to them,

prevention measures do not reduce risk for all. Building an embankment, for example, diverts water from one area to another, and in doing so may reduce death and damage. But some groups are adversely affected, even if these are fewer in number and their possessions have lower values. And prevention measures often twist the damage-probability distribution, not reduce it everywhere: even for those protected by the embankment, there is a lower risk of damage from small floods but far greater in the event the embankment fails—which is why cost-benefit analysis is a useful guide but should not become the sole judge.

Early warning systems: Spending on improving weather forecasting and sharing data have high returns

Even a few minutes of warning gives people time to flee from a flash flood, tornado, or tsunami.7 Local authorities use early warnings of tropical cyclones to evacuate

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 .

large numbers to safer ground. Warnings issued well before an event (lead time) also enable people to protect some property and infrastructure. Reservoir operators could reduce levels gradually to accommodate incoming floodwaters. Local authorities could position equipment for emergency response. People could shutter their windows and reinforce rooftops when warned of severe winds or a cyclone. Chapter 1 showed how deaths and damage from extreme weather events have risen, though more slowly than population and economic activity, largely because of successful prevention measures including better hydro-meteorological forecasts combined with effective emergency preparedness. Several lower income countries with recurrent disasters like Bangladesh and Cuba,

by developing effective early warning systems, experience far less mortality (Golnaraghi 2010). Cuba’s Tropical Cyclone Early Warning System is credited with reducing deaths dramatically for weather-related hazards such as tropical cyclones, storm surges, and related flooding: five successive hurricanes in 2008 left only seven dead. Bangladesh’s similar efforts are described in Spotlight 1. France continually updates all aspects of its Vigilance System developed after the December 1999 winter storm Lothar. After the 2003 heat wave that killed 15,000, the system was upgraded to include heat/health warnings. Flood warnings were added after 2007 when two large cities, Nimes and Montpellier, had major floods.8 Mortality in the United States declined significantly over the years because its early warning systems for recurring hazards such as lightning, floods, storms, and heat waves are continually improved: mortality fell by 45 percent and injuries by 40 percent in 15,000 tornadoes from 1986 to 1999 (Teisberg and Weiher 2009). Yet many countries have not benefited as much as they could have, and this section discusses what is needed for them to do so. Four parts of effective early warning systems require coordination across many

agencies from national to community levels: detecting, monitoring, and forecasting hazards; analyzing risks; issuing timely warnings, which should carry the authority of government; and activating community-based emergency plans to respond to the warnings.9 The focus here is mostly on the first component—also the most technically complex—since the economics of detecting, monitoring, and forecasting hazards plays out at a global scale, unlike the economics of analyzing risks, issuing timely warnings, and requiring emergency evacuations, which are dictated largely by local, social, economic, and cultural circumstances. It is important to emphasize, however, that the strength of a chain is in its weakest link, and all four parts are necessary for an effective early warning system.10

Detecting, monitoring, and forecasting hazards

There is an obvious and important difference in the lead times available for responding to hazards that can be forecasted (or predicted) in advance and those that can be detected and monitored only after they have occurred. Many geological

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 .

hazards can be detected and monitored but not yet forecast, so earthquakes and landslides remain largely unpredictable, though their risks in various zones can be estimated.11 But detecting underwater earthquakes, landslides, or volcanic eruptions using sophisticated ocean monitoring networks and modeling techniques allows issuance of tsunami warnings and evacuations along coastal zones because the lead time varies from a few minutes to several hours (the 2009 tsunami in Samoa). In contrast, meteorological hazards can be forecast with lead times ranging from a

few minutes (enough to save lives) to several days (enough to save lives and protect property, at least to some extent). Weather forecasting is fundamental to an early warning system for meteorological, hydrological, and climate-related hazards, and advances in technology are only making it more accurate (figure 4.6). All countries should be able to benefit from more accurate weather forecasting yet

many do not. Generating forecasts is complex and requires the following elements:

Collecting and sharing data in a systematic and timely manner. Telecommunication systems that allow exchange of information. Numerical weather prediction models, which simulates the physics of the atmosphere. Computational facilities and supercomputers for processing data and models to generate forecasts at different spatial and time scales and resolutions.

Figure 4.6 Increasing the accuracy of weather forecasts

Note: The colored pairs of lines on the top (for the northern hemisphere) and on the bottom (for the southern hemisphere) show that forecasts (3-day, 5-day, 7-day, and 10-day) in the northern hemisphere are generally more accurate than in the southern hemisphere, but that this difference has narrowed over the years. All forecasts are becoming more accurate: the 7-day forecasts today (green) are almost as good as the 3-day forecasts (blue) in the early 1980s. The units of measurement are hectopascal (hPa). Source: World Bank Working Paper No. 151 2008, Washington, DC.

Figure 4.7 Internationally coordinated network of WMO and 189 national meteorological and hydrological services

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: Golnaraghi, Douris, and Migraine 2008.

Underpinning these elements is the need for qualified staff, which continues to be a constraint particularly in lower income countries. Because of its global nature, generating forecasts also requires an enormous

internationally coordinated effort, with many real-time actions that need to be synchronized across countries and time zones. The World Meteorological Organization (WMO) facilitates this massive undertaking through its members’ network (figure 4.7). The data collection system (geostationary weather and polar orbiting satellites,

surface and ocean observing systems) is essentially global and similar in most developed countries. Every day the different national agencies gather and transmit massive amounts of real-time and near real-time data (barometric pressures, temperatures, humidity at various locations and altitudes). They then send the data to the WMO-coordinated Global Forecasting Data Processing and Forecasting System, including three Global Meteorological Data Centers (USA, Australia, and Russia), and 40 Regional Specialized Meteorological Centers. The frequency and scope of observed data vary. For instance, as part of the global network, the United States (NOAA) gathers data from upper-atmosphere soundings (weather balloons) every 12 hours, and complete radar scans are available every eight minutes; data from ships and aircraft are gathered opportunistically. Radiance data from satellite spectrometers, almost continuous, are being used increasingly in weather forecasting.12 But not all regions have adequate data collection services (map 4.1).

Map 4.1 Red dots indicate where few, if any, synoptic weather observations are being received

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: “Synoptic” observations are meteorological observations on the Earth’s surface or in the upper-air made at standard time. The above map refers to synoptic weather observations received at Regional Basic Synoptic Network stations. Source: http://www.wmo.int/pages/prog/www/OSY/Gos-components.html.

Global Telecommunications System

Data are disseminated through the WMO Global Telecommunication System (GTS), which connects all countries through their national meteorological services (figure 4.8 shows just a small section). The data and information that flow through the GTS are used for running highly complex weather models. Other analysis supports the meteorological and climate research community. The GTS also distributes tsunami- related information and warnings, where available, so that every country at risk can receive the information in a timely manner. Global weather forecasts are generated by processing data using various models

that differ in complexity and purpose. For example, global models covering the world are operated by different meteorological centers and use different grids ranging from coarse grids (110 kilometers or 1 degree) to fine grids (20 kilometers or 0.18 degree) producing forecasts of large-scale weather systems. One such model produces 10- day forecasts at a coarse spatial resolution used by 31 participating countries ranging from Norway in the north to Morocco in the south and Ireland in the west to Turkey in the east.

Figure 4.8 Coordinating data collection is complex: A section of the Global Telecommunication System (for Europe) to share data and warnings

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://www.wmo.int/pages/prog/www/OSY/Gos-components.html

Source: WMO 2009, http://www.wmo.int/pages/prog/www/TEM/GTSstatus/R6rmtni.gif;

http://www.wmo.int/pages/prog/www/TEM/GTS/index_en.html.

Every country’s meteorological service has free access to the output13 from global models; but these must be scaled to a finer resolution for local forecasts. Global and regional forecasting models provide boundary conditions for shorter term, geographically focused, and more accurate models that each country’s national weather service could generate. But these require more frequent and accurate local observations and the ability to process them. For example, NOAA’s and the National Weather Service’s national forecasts (for North America) use more frequent data with a 35-square-kilometer grid and an even finer grid for local forecasts to allow greater resolution for densely populated or hazard-prone locations. NOAA also allows direct public access to the output of the models themselves and provides four global forecasts every 24 hours. Some of the smaller countries (Estonia, Netherlands) have merged to make

forecasting more cost-effective. But many countries do not make local forecasts with the accuracy that technology now allows. A 2006-07 WMO survey found that more than 60 percent of their member countries, primarily lower income, have insufficient meteorological capabilities (box 4.3).14 National meteorological and hydrological services often lack basic equipment and instruments; but even when they have them, they are stymied by the absence of modern computing and telecommunication equipment or the lack of qualified staff.

Box 4.3 WMO’s 2006–07 country-level assessment In 2006, WMO surveyed 187 National Meteorological and Hydrological Services (NMHSs) and 139 responded. The survey found:

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://www.wmo.int/pages/prog/www/TEM/GTSstatus/R6rmtni.gif

http://www.wmo.int/pages/prog/www/TEM/GTS/index_en.html

Nearly 60 percent of NMHSs were limited by inadequate training of forecasters. More than 60 percent either did not have adequate observation stations, telecommunication systems, or 24/7 operational forecasting capacities or could not maintain them or the databases. About 90 percent felt the need to improve forecasting and warning capabilities, and half of them wanted better partnerships with other agencies involved in disaster risk reduction. Less than half the countries had combined national meteorological with hydrological services. In 44 other countries, the (separate) National Meteorological Service (NMS) and the National Hydrological Service (NHS) collaborated to some extent, particularly for hazard warnings; but most require better coordination to issue warnings.

Source: World Bank staff based on WMO 2006.

Collecting data on weather and climate and developing forecasts is costly, but the potential benefits can be huge. Weather-related information and forecasts tell farmers and agribusinesses when to plant, sow, fertilize, and harvest; electricity utilities how to cater to demand; and airlines and shipping companies where to plan routes. Benefits exceed costs sometimes more than tenfold:

An estimate in China from 1994–96 found a benefit-cost ratio between 35 and 40 (Guocai and Wang 2003). Meteorological services in Mozambique were estimated to have a benefit- cost ratio of 70 (World Bank 2008).15

The ratio of the economic benefits of improved hydro-meteorological information (calculated as avoided losses) to the costs of national hydro- meteorological services modernization programs vary between 2.1 to 14.4 for some European and Asian countries (World Bank 2008). Benefits of improved weather forecasts estimated for U.S. households exceed the cost of the U.S. National Weather Service modernization program more than threefold (Lazo, Teisberg, and Weiher 2007).

These high benefit-cost ratios suggest that expenditures on improving national hydro-meteorological services are potentially worthwhile. Many governments do not fund their hydro-meteorological services adequately, due to the services’ low visibility or the poor funding of public agencies. Some governments—following countries in Europe—want them to partially finance themselves by selling their data and forecasts. So, data and forecasts may not be shared as willingly as before for fear that the recipient is a potential customer—or would pocket the revenue by selling it to one. Generic weather forecasts and warnings are public goods, and such attempts to generate revenues from sale of data or forecasts inhibit the data sharing essential for good regional and global forecasts. The potential benefits of greater spending on hydro-met services will be realized

only if the spending is well directed and organized. The need for a complete

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 .

meteorological forecasting system must be well identified before spending on expensive technologies, such as Doppler radars, which can run between $1 million to 2 million per unit, and several are needed. A satellite system costs about $380 million. Running it costs about $50 million.16

The United States with frequent tornados reduced the annual death toll by an average of 79 and injuries by 1,052 thanks to more accurate forecasting (from 40 to 75 percent) (Simmons and Sutter 2005). And the use of Doppler Radar’s ability to identify tornados while still in the clouds has led to the longer lead time for tornado warnings (from 5.3 minutes to 10). But these expenditures may not be warranted in other countries if hazards do not occur or are less frequent. The point is not that Doppler radars are unwarranted but that spending on

expensive equipment has to be carefully assessed against the needs and means of a country. Operational and maintenance costs also need to be considered for long-term sustainability. Also, more mundane needs such as estimating and calibrating models, carrying out hazard analysis, and using past data, which in many countries is stored in warehouses on deteriorating paper, may have high returns. Digitizing these data is low tech with high returns. In addition to gains from short-term weather forecasts, seasonal forecasts are also

improving to support medium and longer term socioeconomic decision making. Recurring climatic patterns (like the El Niño Southern Oscillation) can now be forecast with a few months lead time in some places and for some periods of the year. Predicting droughts (a big killer in Africa) requires not only weather forecasts but also data on air temperature, humidity, soil moisture, vegetation, ground, and reservoir levels. National agencies must begin to gather such data and learn to use it effectively if local droughts are to be forecast accurately.

Analyzing risks, issuing timely warnings, and activating responses

Establishing early warning systems requires much information. The spatial distribution of hazards, their severity, timing, and frequency are largely a matter of science. But their economic effects require assembling data that governments already have in some form. These data must be systematically analyzed to determine whether and where early warning systems should be established. Cost-benefit analysis is a good guide. In some cases, one may only need to identify and analyze one major risk (or a few risks) that is (are) sufficient to justify producing warnings, which would then be available to minimize other risks that may not be as easy to quantify.

Box 4.4 Communications to the community Almost all households (98 percent) have access to radio and TV in Cuba, so these are the main communication channels that the national meteorological service (with government authority) uses to issue tropical cyclone and related flood warnings. In Bangladesh, far fewer have televisions and radios, so the Bangladesh Meteorological Department conveys

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 .

cyclone and storm surge warnings through multiple channels (fax, internet, radio, and TV). But the centralized warning center of the Bangladesh Cyclone Preparedness program ensures the warnings reach coastal communities. The center alerts a network of volunteers through HF/VHF radio broadcasts, and they in turn fan out into the communities to warn the people. Shanghai also issues warnings though HF/VHF radio, using a network of community volunteers to warn

those in rural surroundings and TV broadcasts and mobile phone messages (SMS) for those in urban areas.

Source: World Bank staff.

A warning is based on a forecast, but it should carry the authority of the government. So hazard warnings developed by technical agencies should be communicated to the authorities who must then quickly decide whether to warn the public and to activate evacuation and emergency plans. To decide is to weigh their costs and benefits: false alarms are expensive (much of the cost falls on citizens, not the government) and too many false alarms will result in warnings being ignored (box 4.4). Such decisions and their responses require much preparation: pre-positioning

equipment, emergency responders, sandbagging (a low technology effective only if what is being protected is chosen carefully), and redirecting traffic all require not just planning but also periodic drills in communities. Bangladesh shows that the response can be effective even in poor countries.

Critical infrastructure

All infrastructure should be well designed, constructed, and maintained. But it is especially important that some function when most needed. Such “critical” infrastructure must be identified before a disaster to ensure its adequacy. Every sector has parochial advocates (education specialists favor “safe schools”;

doctors, “safe hospitals”), but even jails could be critical because they keep robbers from looting. Governments decide what is critical, but the choice should not be left to officials alone: the government of Myanmar was warned of the intensity and likely path of Cyclone Nargis five days before landfall in 2008, but the military junta did not warn the population, lest it disrupt a referendum under way. The military moved its planes and ships to protect them from damage, but not the people—and 140,000 died. What is critical depends on local conditions and the likely hazard. In quake-prone

Istanbul, hospitals to treat broken bones and crushed bodies may be critical. But in flood-prone Bangladesh, hospitals may be less critical than water treatment plants— or schools, less for the education than to serve as shelters. Critical assets are specific: a particular bridge, not all bridges. An example

illustrates the point. A bridge connecting residential areas with a hospital separated by a river and a bridge that links them to the industrial area must both have sufficiently high economic rates of return. But the bridge to the hospital is “critical” if

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 area is prone to quakes. The “willingness to pay” for the lost service immediately after a disaster would be a good yardstick to measure which asset qualifies as critical. Once selected, the “margins of safety” (the extra strength that engineers build into designs) should be higher than usual. Designing the bridge to higher standards may raise the cost and so reduce its economic rate of return;17 but sensible judgments must prevail. Critical is not a synonym for “socially important.” Conversely, noncritical does not

mean unimportant: it just implies that interruption of service is tolerable. Even the United States encounters difficulties in keeping “critical infrastructure” manageably small, and other governments will undoubtedly discover this as well (box 4.5). Sectoral ministries may know enough to propose the list, but the decision should not be theirs to make. The choice requires judgment, and while collective judgments have their shortcomings, the country’s decision-making structure should be respected.

Box 4.5 The United States tries to identify critical infrastructure In U.S. public policy, the meaning of “critical infrastructure” has evolved over the years, defined only when a presidential commission on Critical Infrastructure Protection was established in the wake of the 1995 Oklahoma City bombing. The commission identified eight sectors as critical: telecommunications; electrical power systems; gas and oil storage and transportation; banking and finance; transportation; water supply systems; emergency services (including medical, police, fire, and rescue); and continuity of government. Since then, successive federal laws, reports, and executive orders have sought to clarify the concept, and

the number of infrastructure sectors and the types of assets considered “critical” expanded. After the September 2001 attacks, President Bush’s new executive orders added nuclear sites, agriculture, and livestock to the list. A year later, the National Strategy for Homeland Security added chemical plants and postal and shipping services as well. The list now has 13 sectors, each including thousands of physical structures in different locations, some privately owned (power stations). The Information Analysis and Infrastructure Protection Directorate in the Department for Home-land Security

is now responsible for identifying critical assets, and there were 1,700 such assets in April 2004. There is much confusion and controversy because of the implications for private owners and because the state governments have their own lists and agenda. Nor are the criteria clear: some electric generating plants, for example, are not in use and others generate little power. And if being on the list attracts resources, potential beneficiaries scramble for the spoils. The amorphous “threat” results in an unclear and changing list of what is critical, and when it includes too

many assets, the costs rise without commensurate benefits. It may well be that the United States now seeks to protect too many facilities, or the wrong ones (or both).

Sources: Motef and Parfomak 2004; Forest 2006.

Figure 4.9 Three modes of operation of the SMART Tunnel

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: Mott MacDonald Group 2009.

Selection does not conclude the issue: all infrastructure needs maintenance: fixing potholes in the road before the winter or the rain; painting steel bridges before they weaken through corrosion; and inspecting and fixing cracks in concrete bridges. All engineers know this, but they do not obtain budget appropriations—even in the United States, where the 2007 bridge collapse in Minneapolis drew attention to such neglect. Public finance theory suggests that spending should go down a list of projects arranged in descending order of (economic) rates of return. But when subject to arbitrary budget spending limits, lumpiness, and interruption costs, dynamic maximization could put some low-return spending ahead of postponable high-return spending. Since maintenance can be postponed, it gets deferred—repeatedly—until the asset crumbles. Multipurpose infrastructure, such as Kuala Lumpur’s Stormwater Management and

Road Tunnel (SMART), is critical infrastructure tailored to the specific hazard. Floods from heavy rains are a hazard, and the 9.7-kilometer-long $514 million tunnel has three levels (figure 4.9), the lowest for drainage and the upper two for road traffic. The drain allows large volumes of flood water to be diverted from the city’s financial district to a storage reservoir, holding pond, and bypass tunnel. Combining the drain with the road has two advantages: it ensures maintenance of a drain that otherwise would be used only sporadically, and it costs less than building the road and drain separately. Critical infrastructure should still pass the cost-benefit criterion, and designs such

as the SMART require imagination and innovation. Maintenance remains neglected, and although economists generally disapprove, earmarks may have merit—some earmarking a fuel tax to fund road maintenance, for example.18 But it may prove ineffective in other settings: the public works department (or its equivalent responsible for roads and bridges) may use the funds to paint stone culverts (which do not rust) rather than steel bridges that do. So many good decisions at many levels of

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 .

government are needed, underscoring the importance of “institutions.” Such institutions accompany and promote economic development—which is why death and destruction are inversely related to a country’s per capita income. But the correlation is not perfect, suggesting that some lower income countries do better than others.

Protecting environmental buffers

A degraded environment exacerbates the effects of disasters, and environments are stressed by growing populations that do not protect them. Natural and manmade prevention measures can complement each other (table 4.1).

Physical limits

While ecosystem buffers offer some protection, they do not prevent all disasters. Forests and wetlands offer little protection from extreme flooding when soils are already saturated. Similarly, mangrove belts a few hundred meters wide can reduce the destruction from a sizeable tsunami but not significant ones, for example, those taller than 10 meters. A narrow swath of trees could do more harm than good if they topple and add to the water borne debris. Many were injured and killed by splintered mangroves in Papua New Guinea floods. But mangroves also trap floating debris (including tsunami victims who would otherwise be swept out to sea during the backflow) and over the long term help to protect against coastal erosion (FAO 2007).

Analytical limits

Several studies report impressive numbers on the value of natural defenses:

As coastal defenses, Mangrove forests in Malaysia have been estimated to have an economic value of $300,000 per kilometer based on comparison with engineered alternatives (ProAct 2008). Since 1994, communities have been planting and protecting mangrove forests in northern Vietnam to buffer against storms. An initial investment of $1.1 million saved an estimated $7.3 million a year in sea dyke maintenance and appeared to significantly reduce losses of life and property from typhoon Wukong in 2000, compared with other areas (WWF 2008). In the Lužnice floodplain—one of the last floodplains in the Czech Republic with an unaltered hydrological regime—470 hectares have monetary values per hectare of $11,788 for flood mitigation (water retention), $15,000 for biodiversity, $144 for carbon sequestration, $78 for hay production, $37 for fish production, and $21 for wood production (ProAct 2008). The economic value of forests for preventing avalanches is estimated at around $100 per hectare per year in open expanses of land in the Swiss Alps and up to more than $170,000 per hectare per year in areas with valuable assets (ProAct 2008).

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 .

A recent study on the role of wetlands in reducing flooding associated with hurricanes in the United States calculated an average value of $8,240 per hectare per year, with coastal wetlands estimated to provide $23.2 billion a year in storm protection services (Costanza and others 2008). The two reserves that form the Muthurajawella Marsh, in Sri Lanka, cover 3,068 hectares near Colombo. The economic value of flood attenuation (converted to 2003 values) has been estimated at $5 million a year (Costanza and others 2008). Benefits from forest protection in the upper watersheds of Mantadia National Park, in Madagascar, in reduced flood damage to crops, have been estimated at $126,700 (Kramer and others 1997).

Table 4.1 Natural hazards and protection

Natural hazard Type of ecological protection Examples

Landslides and avalanches

Dense and deep-rooted vegetation helps to bind soil together, resisting slippage of surface layers. Forests form a physical barrier against upslope avalanches and pin down the snow pack, reducing the chance of a slide beginning.

Reforestation has been used to protect against avalanches in Switzerland, complementing and in some cases substituting for engineered barriers (UNISDR 2009): 17 percent of forests are managed to protect against landslides and avalanches.

Floods

Dense vegetation cover within upper watershed areas increases infiltration of rainfall as opposed to surface run-off, reducing peak flow rates except in the most extreme conditions when soils are already fully saturated. Vegetation also protects against erosion, thereby reducing soil loss and the transport of mud and rock that greatly increase the destructive power of floodwaters. Dense vegetation also protects river banks and adjacent land structures from erosion by floodwaters. Wetlands and floodplain soils absorb water, reducing peak flow rates downstream.

Hurricane Jeanne hit several Caribbean islands, but the number of flood- related deaths was more than 3,000 in Haiti versus only a few dozen in all other affected countries, largely due to Haiti’s highly degraded watersheds (Stolton and others 2008). The pattern was similar during the 2008 hurricane season. In 1992, the World Bank committed $85 million to alleviate poverty in three Indian states (Andhra Pradesh, Orissa, and West Bengal) through employment creation at shrimp and fish farms. Mangrove forests were depleted to make space for shrimp farms. When cyclones hit the newly denuded coast, however, they found little resistance. And a significant part of the investment was lost. Two cyclones, one in Andhra Pradesh in 1997 and one in Orissa in 1999, destroyed the newly constructed sites for shrimp farming (Independent Evaluation Group 2007). A study around Mantadia National Park, Madagascar, concluded that conversion from primary forest to swidden (area cleared for temporary cultivation by cutting and burning the vegetation) can increase downstream storm flow by as much as 4.5 times (Stolton and others 2008). Communities have successfully planted bamboo to protect channel embankments from annual floods in Assam (UNISDR 2009). Canalization and drainage in the Mississippi floodplain were estimated to have reduced flood storage capacity by 80 percent, and have subsequently been linked to subsidence of large areas and to the severity of the impact from Hurricane Katrina (WRI 2005).

Tidal waves (tsunami)

Coral reefs and sand dunes (which in coastal areas typically depend on associated plant communities for maintenance) provide a physical barrier against waves and currents. Salt marshes and lagoons can divert and contain

Modeling for the Seychelles suggests that wave energy has doubled partially as a result of changes in coral reef structure due to bleaching and changes in species composition (Stolton and others 2008). In the Caribbean, as a result of reef degradation, more than 15,000 km of shoreline could experience a 10 to 20 percent reduction in protection from waves and storms by 2050 (Stolton and others 2008). Re-establishment of salt marshes forms part of coastal defense measures in areas of the U.K. (UNISDR 2009). Following the 2004 tsunami, studies in Hikkaduwa, Sri Lanka, where reefs are in a marine park, noted that damage reached only 50 meters

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 .

and storm surges

floodwaters. Mangroves and other coastal forests can absorb wave energy and trap floating debris, greatly reducing the destructive power of waves.

inland and waves were only 2 to 3 meters high. At nearby Peraliya, where reefs have been extensively affected by coral mining, the waves were 10 meters high, and damage and flooding occurred up to 1.5 kilometers inland. In Japan, where accurate historical records exist, the role of forests in limiting the effects of tsunami damage has been demonstrated (Stolton and others 2008). The Black River Lower Morass is the largest freshwater wetland ecosystem in Jamaica. The marsh acts as a natural buffer against river floodwaters and incursions by the sea (Dudley and others 2010).

Hurricanes and storms

Forests, coral reefs, mangroves, and barrier islands buffer against immediate storm damage.

The protected mangrove system known as the Sundarbans in Bangladesh and India helps to stabilize wetland and coastlines and to buffer inland areas from wind and wave surges resulting from cyclones. Mangroves can break up storm waves that exceed 4 meters during cyclones (Dudley and others 2010).

Sources: Dudley and others 2010; Stolton, Dudley, and Randall 2008; Independent Evaluation Group 2007; and UNISDR 2009.

While the figures are impressive, modeling the effectiveness of alternative vegetation types or land uses needs considerable longitudinal data from the specific area, and subjecting such protection to cost-benefit analysis, while possible in principle, is difficult in practice (box 4.6). Moreover, the benefits of protection are the avoided expected damage; but cost-benefit ratios are sometimes stated on the basis of actual damage without multiplying it by the probability of occurrence. Benefits are hard to value, and it is easy to make mistakes, especially when

environmental protection is sought and quality varies. As Dahdouh-Guebas and others (2005) note about mangroves protective role in storm protection:

“Our surveys of villages and post-tsunami observations make it clear that mangroves play a critical role in storm protection, but with the subtle point that this all depends on the quality [emphasis added] of the mangrove forest.”

Box 4.6 Costs and benefits of mangroves or shrimp ponds on the Thai coast Sathirathai and Barbier (2001) calculated the net present value per hectare of Thai mangroves by adding the value of forest products local people collected (around $540), the increase in coastal fishery yields (around $270), and storm protection (around $74,600). Storm protection contributed most of the total net present value (NPV) of more than $75,000 per hectare. It also comprised most of the $1,150 per hectare NPV of converting mangrove to shrimp ponds. The results are reported in several publications. To evaluate habitat and storm protection services, Barbier (2007) developed a new “dynamic” approach that

incorporated the change in wetland area within a multi-period harvesting model of the fishery. The NPV of storm protection was recalculated based on actual storm damage rather than the replacement cost of engineered coastal defenses (the original analysis), yielding a NPV per hectare of around $10,000. Three points are worth noting about the difficulties of evaluating ecosystem protection services:

1. The NPV per hectare of mangroves declined considerably because replacement cost methods, which essentially use a cost to estimate a benefit, generally overestimate storm protection services.

2. Although mangroves still comprise most of the NPV of shrimp farming, without the value of mangroves for storm protection, it would not be worth converting the shrimp ponds back into mangroves.

3. Because of the lack of data, the estimate of expected storm damage due to protection by mangroves could not control for other possible mitigating factors, such as storm intensity, coastal topography, and other natural barriers, such as coral reefs and seagrass beds.

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: World Bank staff.

Timing also complicates the cost-benefit decision of even whether to restore. For example, because mangrove habitats can recover naturally without artificial intervention, restoration projects should be undertaken only if recovery is not happening on its own.19 Determining why natural recovery is not taking place and removing necessary stresses are essential to any successful artificial restoration effort. Costs are often difficult to quantify too: costs of restoring mangroves in the United

States alone fall into three orders of magnitude, ranging from about $225 per hectare to over $200,000 per hectare.20 Costs of restoring natural hydrology by “reconnecting” divided mangroves could be lower but may increase exponentially if large-scale earthworks are needed to re-landscape an area. Maintenance costs are often ignored, though they are considerable, particularly if there are ongoing human pressures: effective management of protected areas costs is high even in low-income countries. It is harder to protect wetland habitats (including coral reefs) that are more vulnerable to non-point pollution and the removal or introduction of particular species. The role of natural ecosystems in reducing the adverse effects of disasters is

recognized, but evaluating both their costs and benefits is difficult. Moreover, governments tend to emphasize physical investments at the expense of intangible assets. The Bangladesh spotlight (Spotlight 1) shows how water management authorities initially favored the construction of embankments, and the Haiti spotlight (Spotlight 3) how deforestation, a major cause of the mudslides, remains unaddressed. Other examples from Argentina and the United States show a similar government preference for physical structures (Gentile 1994; Penning-Roswell 1996; Driever and Vaughn 1988). Such policies put people at risk and distort urban development. Protecting the environment is generally more cost-effective than restoring it, but

successful protection requires the participation of users whose livelihoods depend on the resources in question: fences and policing are rarely effective. Well-functioning communities have long found diverse ways to share and protect the commons. Elinor Ostrom (1990)21 describes a variety of such arrangements among local users, including clearly defined boundaries and effective monitoring by people part of or accountable to the appropriators, complemented by graduated sanctions for resource appropriators who violate community rules. These broad principles underlie successful institutions and have significant ramifications for long-term sustainability for common property regimes (Gibson, Williams, and Ostrom 2005). To summarize, governments can do more to prevent disasters. This does not

always require more spending, but it often requires spending differently. Most important (and this is difficult), preventing disasters requires continual monitoring of

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 effectiveness of such spending. Transparency and disclosure are important for this reason. And when voters are confident that such spending is not wasted, they will be more willing to reward such spending. Three spending items generally have high returns. The first is more funding for

weather forecasting with accompanying oversight to prevent careless spending. This would allow countries to take advantage of greatly improved technology. Early warning systems and evacuation drills and procedures are warranted in some of the more risky areas. The second is ensuring that certain critical infrastructure remains functional after a disaster. And the third is protecting environmental buffers, sensible but difficult to translate into action: better institutions will help.

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 4 on Ethiopia

Deaths from droughts or Derg? Ethiopia is prone to many hazards including earthquakes—the African and Arabian tectonic plates meet in the Rift valley. Although flash floods are more frequent, droughts have been far deadlier. Global data show that almost a million people have died from droughts since 1970, mostly in Africa (chapter 1, figure 1.3). Over the past six decades, Ethiopia has been particularly drought-prone, with one every 3 to 5 years and some lasting over several years. There is rain, averaging 1005 mm annually, but it varies greatly by region and is particularly unpredictable.1

Many of these deaths were avoidable, although droughts are not—because a “slow onset” hazard allows ample time for food to reach those who would otherwise starve, but this did not always happen. Chapter 2 reports on the empirical association between disasters and conflicts that continue to simmer in the region; but conflict is not the only reason for food not to reach the starving. Amartya Sen drew attention to the absence of famines in India after independence when the authorities became more responsive to their people.2 Better institutions, both domestic and external, could prevent deaths from droughts.

Living with unreliable rains

Much of Ethiopia’s agriculture—accounting for half its GDP and sustaining 80 percent of its people—is rainfed, and its 80 million people have long adapted to its unpredictability. Farmers and pastoralists cope with droughts differently, and both groups have great difficulty with extended droughts. Farmers grow multiple crops (drought-resistant teff, maize, and barley) on small, rainfed plots.3 They also keep sheep and goats to sell when needed. Pastoralists’ livestock store protein for lean years, and a tenth of the rural population are nomadic pastoralists in Somali and Afar, the Borana zone of Oromiya region, and Benshagul and Gambela (spotlight map 1).4 Migration helps, but some international borders that now separate related tribes are becoming less porous: Turkana tribesmen’s herds are being blamed for spreading foot and mouth disease across the Kenyan and Sudanese borders.5 Population growth and the settlement of pastoral nomads add to the pressures on the land.

Spotlight map 1 Rainfall in Ethiopia

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: World Bank staff.

Wars and conflicts inhibit and interfere with these traditional coping mechanisms. These conflicts have local origins and have been fed by regional and super powers with their own concerns. Ethiopia shows how famines happen, but it requires an explanation of its politics and ethnic divisions that overlap with ideology and groups seeking support for their cause.

The Emperor’s neglect in the 1972–73 drought

Ethiopia became a nation-state in the 19th century, and its kings managed to keep it from being fully colonized. Italy seized Eritrea in 1889 and occupied parts of present day Ethiopia from 1936 until early World War II, exiling Haile Selassie, the Emperor, after 1930. Selassie was restored to the throne in 1941, though the British administered the territory under a United Nations mandate until 1951 when Eritrea joined the federal state, but its relations with the central government remained difficult. The Amharans and Tigrayans in the north, along with the Agau and Oromo from Wollo, had resisted the shift of power to the south since the late 1800s and, by extension, the Emperor’s rule.6

The Emperor was widely admired internationally, but became increasingly unpopular domestically. After a 1943 revolt in the north, the Emperor confiscated northern and central Wollo lands (former provinces in northeastern Ethiopia). Making farmers into tenants required rents to be collected even when droughts reduced their harvests (Tigray in 1958 and Wollo in 1966). So resentment simmered, and the incipient independence movement began. The Emperor became more isolated and autocratic after the 1960 attempted coup,

instigated by his bodyguard’s commander, was thwarted. During the 1972–73

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 .

droughts, grain was taken from the affected areas in the north and sent to Dessie and Addis Ababa, the provincial and national capitals. The resulting famine killed over 100,000 people (although some claim 200,000 died in Wollo alone). The Emperor’s rule crumbled when his neglect was exposed.7 Students and the middle classes revolted in the capital, the military mutinied, and the Derg (Amharic for committee) took over. The Emperor died in custody in 1975, and the Derg’s Marxist-Leninist ideology initially attracted student support. Agriculture was collectivized, and kebelles (peasant associations) became instruments of central government control (Wolde 1986). The ideological shift lost Ethiopia one superpower’s support and gained it

another’s. But domestic ideological divisions overlapped regional and ethnic differences. Disputes within the Derg pulled Colonel Mengistu Haile Mariam out of the shadows to take control in 1977 and unleash the “Red Terror.” Thousands who opposed the government were killed and separatist movements strengthened: the Tigray People’s Liberation Front (TPLF) sought an independent Tigrayan state, and the Eritrean People’s Liberation Front (EPLF) sought the same for themselves further north. Support and weapons poured in across porous borders. Somalia invaded Ethiopia

in 1977 to annex Ogaden where there was much dissatisfaction with Addis Ababa’s rule. After fierce fighting in 1977–78, Ethiopia repelled the invasion with the help of Cuban troops. Bitter memories and suspicion lingered long after these troops left, and fighting has since renewed—this time on the Somali side of the border. The deadly drought in 1984 shows what could happen again when food and its denial become weapons of war.

Food as a weapon in the 1984 drought

Rains failed in 1983–84, and the Ethiopia Relief and Rehabilitation Commission, a government agency created after the 1972–73 famine, appealed for help. Fighting in Tigray and Eritrea made donor governments understandably suspicious that the drought was exaggerated to garner aid that could then be diverted (Adejumobi 2007). Only after the international media began reporting on the dying thousands did food aid begin to flow in. But the Derg restricted its movements as well as those of migrants and traders while military offensives and aerial bombardment destroyed opponents’ cattle and grain stores (Porter 2008). Some claim that over a million people died in the famine; subsequent studies confirm that mortality, other than direct casualties of the conflict, was greater in areas with more fighting (Kiros and Hogan 2001). Colonel Mengistu remained president after the 1987 non-competitive election, but

fled the country in 1991 after losing both domestic and international support. The TPLF and EPLF movements wrested local control and Eritrea’s independence in 1993 left Ethiopia landlocked. Fighting between them erupted in May 1998 over what seemed a minor border dispute, and the peace since the June 2000 Algiers

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 .

agreement has been intermittent. The Boundary Commission awarded Eritrea the disputed town of Badme, but the transfer remains incomplete, and Eritrea’s foreign policy, especially toward Somalia, lost international support. The tenuous peace makes for intermittent aid. Ethiopia’s National Meteorological

Service Agency forecast poor belg rains (typically from February to May) in January 2000, but donors pledged no aid until April because the war with Eritrea raised the same suspicion of exaggerated need (Broad and Agrawala 2000). The suspicions were mutual, and when food finally became available, the Ethiopian government balked at shipping it through the Eritrean port of Assab that handled three quarters of relief before the conflict. By the time disputes over logistics and control over its distribution were resolved, a localized famine was well underway.

A good beginning: Social safety nets and better preparedness

Widespread starvation was averted during the subsequent and more severe drought of 2002–03 because fighting abated and food aid reached 13.2 million people, although some went hungry and livestock was lost. The government subsequently developed a more permanent safety net and supplemented the emergency food distribution system with the Productive Safety Nets Program (PSNP) in 2005. The PSNP finances public works (such as building terraced fields on hill slopes to reduce soil erosion and increase water retention) paying cash for up to five days a month per household member and six months a year (but for no more than three years to avoid dependency). In addition, about 10 percent of the poorest beneficiaries get unconditional cash or food transfers. The PSNP is also linked to the Other Food Security Program, which provides credit and agricultural extension services and funds irrigation and water harvesting schemes. The PSNP, sub-Saharan Africa’s second largest social safety net (after South

Africa), now reaches over 7 million people (spotlight figure 1). It appears to target households well, although the transfer amounts are often small and distribution remains irregular. A survey after the 2008 drought found that beneficiaries living in households that got at least 10 days of work a month in the 3 previous months consumed 30 percent more calories and held more livestock than non-beneficiaries (0.62 TLU).8 The effectiveness of any single intervention may not be significant, but public works combined with seeds, credit, and irrigation raised wheat and maize yields by about 200 kilograms per hectare.

Spotlight figure 1 Number of PSNP beneficiaries, (millions) 1992–2009

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: The number of PSNP beneficiaries is a subset of the total number of emergency assistance beneficiaries. Source: World Bank (2009) Project Appraisal Document for PSNP Phase 3.

Weather forecasting and early warning systems are being improved, and the National Disaster Prevention and Preparedness Fund was established to finance relief and make assistance more timely and predictable. The World Food Programme (WFP) and the World Bank support software (LEAP, for Livelihoods, Early Assessment, and Protection) that projects food shortfalls and calculates funding requirements by linking weather and crop production/rangeland data. This software was used in 2008 to allocate a $25 million World Bank Contingency grant. Food aid has been sizable, fluctuating between 0.4 and 2.5 percent of GDP

between 1996 and 2001.9 USAID and the WFP now jointly issue monthly early warning bulletins, but the need for food aid sometimes becomes a source of controversy.10 NGOs sometimes report deaths from starvation that the government disputes. The government is understandably eager to be less dependent on donors, and even if they are correct that there is enough food at the national level, it may not always be aware of local shortages. People can starve even if there is food elsewhere in the country, and this is less likely if information and food flowed more freely internally.

The long view: Increasing investment and irrigation

More food could be grown with better policies and investment. Productivity would increase if farmers’ tenures were made secure: there are reports that kebelles in some areas threaten to evict farmers who oppose the government.11 The past neglect of agricultural research could be reversed, and while this has begun, it must be sustained to develop higher yielding drought-resistant crops unique to Ethiopia. Better transport and storage facilities to prevent pockets of shortages would require only modest investments. Irrigation has considerable potential, but requires major investments and potentially difficult international agreements. Only 2 percent (200,000 hectares) of Ethiopia’s cropland is irrigated despite erratic

rainfall, using little water (0.3 billion cubic meters) from its plentiful rivers: the Blue Nile begins in Ethiopia’s Lake Tana (see spotlight map 1) and joins the White Nile in Khartoum (Sudan) before flowing north through Egypt and into the Mediterranean

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 .

Sea. Egypt and Sudan’s large populations have used much of the Nile waters to irrigate crops since the time of the pharaohs, and many dams were built over the centuries. In the 1950s, Egypt raised the height of the dam at Aswan creating the large but shallow Lake Nasser upstream. Egypt and Sudan signed the 1959 Nile Waters Agreement entitling each to 55.5 billion and 18.5 billion cubic meters a year respectively (though Sudan uses only 13.5 billion now because of its internal conflict). International laws generally recognize these claims, though Ethiopia was not part of, and does not recognize, the 1959 agreement. There are large potential economic benefits from using the Nile waters more

efficiently. Evaporation losses from Lake Nasser of around 10 billion cubic meters annually would be reduced substantially if it were in the cooler Ethiopian highlands, where deep valleys allow dams to create lakes with a smaller surface area. Hydroelectricity generates additional gains: a dam on the Abbay sub-basin (spotlight map 1) could produce more power than Ethiopia currently consumes. Exporting the surplus to its power hungry neighbors requires more and better transmission grids.12 All these require major investments and the consent of the other riparian countries. But such consent comes with cooperation, not conflict. Ongoing conflicts in neighboring Somalia and Sudan inevitably engage Ethiopia and

Eritrea on opposing sides, and recent reports are worrying: “This cross-border area [i.e. where Kenya, Ethiopia and Somalia meet] is a conveyer belt that moves arms to and from all three countries, and across the African continent.”13 During its 1998– 2000 war with Ethiopia, Eritrea assisted Somalia’s Ogaden National Liberation Front to relieve pressure on its front by drawing Ethiopian troops southward. Such alliances continue while tactics and location shift: Ethiopia recently withdrew

its troops from Somalia after trying since December 2006 to oust al-Shabab, now labeled a terrorist group, from the government. Prospects of large oil and natural gas reserves in Ogaden and similar finds in Sudan complicate matters (Chinese engineers were killed). Fighting in these areas continues, and the flow of weapons and munitions is being financed while food aid appeals are issued during the periodic droughts.

Preventing deaths from droughts: 2009–2010

After meager 2008 rains, particularly in eastern Ethiopia, the main kiremt rains in 2009 (June to September) were diminished and delayed by four to six weeks.14 Better early warning systems and the organization of the safety net notwithstanding, droughts under these circumstances retain their deadly potency. Donors tried to raise $175 million in the last months of 2009, although some government officials dispute the risks of starvation.15 Even if the appeal for aid is answered, time and transport are necessary for the food to reach the starved. Starvation is easier to prevent than droughts, but it requires that the authorities be

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 .

both concerned and informed about the people’s predicaments.16 The WFP reports that in 2009, violence forced 350,000 people from their homes in southern Sudan, where seasonal rains were meager. Centralized controls do not permit the accurate and timely flow of information and food, and conflicts around the borders make relaxing such controls more difficult. Peace is possible, but has been elusive. The World Bank’s forthcoming 2011 World

Development Report will examine conflicts, fragile states, and the roles played by stress—both internal and external (including external interference)—capability, and expectations. Conflicts become more complex when they involve governments of distant superpowers: the long reach of their military and clandestine services supply sophisticated weapons. Better domestic institutions are undermined when the fighting is financed or instigated by foreign powers whose electorates and representatives are not always fully aware of what is happening in distant and unfamiliar lands—so better external institutions would help. Many scholars have noted that deaths during droughts are associated with

conflicts, and the analyses in chapters 4 and 5 found that death and destruction are lower when there are good institutions (typically also associated with democracy and better governance), and that this link operates through political competition, not just periodic elections. Droughts cause death when food does not reach the starving, and spot shortages could occur despite an adequate harvest and ample food stocks. Earlier deaths stemmed from the Emperor’s neglect (in 1972–73), conflict (in 1984), and disputes with donors (in 2000). Being ill-informed or ill-prepared are some of the many avoidable reasons for starvation. A more liberated flow of information and goods would reduce these dangers.

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 .