WPS7632 Policy Research Working Paper 7632 The Triple Dividend of Resilience Background Paper Higher Losses and Slower Development in the Absence of Disaster Risk Management Investments Stephane Hallegatte Mook Bangalore Marie-Agnes Jouanjean Development Economics Climate Change Cross-Cutting Solutions Area April 2016 Policy Research Working Paper 7632 Abstract Global economic losses from natural disasters continue to even before a disaster strikes. The paper’s main message increase. Yet, investments in disaster risk management are is that disaster risk management investments can provide not universal, as they are traditionally seen as in competi- two dividends: reduced losses when a disaster strikes, and a tion with other development and economic priorities. The shift of investment strategies and perhaps even an increase multitude of benefits from disaster risk management invest- in investment value that would benefit the economy even ments are not traditionally accounted for in cost-benefit before a disaster strikes. Providing evidence to policy analyses. This paper contributes to this discussion by high- makers and investors about the existence of both types of lighting the multiple benefits from disaster risk management dividends can provide a narrative reconciling short-term investments, focusing on the avoided losses when a disaster and long-term objectives, thereby improving the acceptabil- occurs, but also on the impacts on economic development ity and feasibility of disaster risk management investments. This paper is a product of the Climate Change Cross-Cutting Solutions Area, and a background paper to “The Triple Dividend of Resilience” report, a joint initiative by the Global Facility for Disaster Reduction and Recovery and the Overseas Development Institute. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at shallegatte@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Higher Losses and Slower Development in the Absence of Disaster Risk Management Investments Stephane Hallegatte1, Mook Bangalore1, Marie-Agnes Jouanjean2 1 Climate Change Group, World Bank, Washington, DC 2 International Economic Development Group, Overseas Development Institute, London, UK Acknowledgements The authors thank Reinhard Mechler, Junko Mochizuki, Emily Wilkinson, Swenja Surminski, and Adrien Vogt-Schilb for very useful comments on previous drafts of this paper. Keywords Disaster risk management, resilience, development, economic losses JEL Codes Q54, O10, D81, G11 1. Introduction Global economic losses from natural disasters are increasing over time and in 2014 totaled USD $110 billion (Munich RE, 2015), and there are repeated calls to do more to prevent disasters or minimize their consequences on the affected population. Investments in disaster risk management (DRM) however have to compete with many other development and economic priorities, and economic analyses of the cost and benefits of disaster risk investments have been the focus of intense research and discussions. This paper discusses the assessment of the benefits from DRM expenditures, investigating benefits that go beyond the losses of lives and assets that can be avoided thanks to better risk management. These additional benefits include two major categories. First, looking only at direct asset losses would lead to a large underestimation of the welfare impact of natural disasters. There are many reasons why the value of the repair or reconstruction is a bad proxy for the impact on welfare: assets that are not damaged may stop producing (for instance because there is no electricity or no transportation to bring workers); reallocation of resources and investments toward reconstruction may hurt other activities or projects (e.g., governments may cut social assistance to finance infrastructure repair); the loss of assets may disproportionally affect poor people who are living close to the subsidence level and suffer a lot even from small loss of income or assets; and asset losses can prevent asset accumulation and keep people in poverty (Hallegatte et al., 2016). So welfare losses may be much larger than what direct asset and human losses may suggest. But equating the benefits from DRM with avoided losses also misses half the story. In addition to avoided losses, DRM investments can unlock suppressed economic potential. While risk-taking is a pre-requisite for growth and development, households and firms tend to reduce their level of investments when they are aware of the potential occurrence of a disaster and instruments to cope with potential negative outcomes are not available (Eeckhoudt et al., 1996; Elbers et al., 2007). However, we argue in this paper that if this “background risk” from natural disasters can be managed through DRM investments, at the household and firm levels, risk aversion can be reduced and risk-taking fostered – from investing in high- risk but high-return crops in agriculture, to bearing financial risk associated with entrepreneurship and innovation. Critically, these benefits from DRM investments are experienced even before a disaster strikes. This ex-ante, gain-centric view of DRM is known as the second dividend (or “development dividend”) of DRM investments (ODI and GFDRR, 2015). A third dividend, known as “co-benefits” and unrelated to hazards and risks, is also possible (ODI and GFDRR, 2015). Beyond increasing resilience, DRM investments may yield positive economic, social, and environmental side-effects. For instance, a shelter can be used as a community space before a disaster hits. Flood protection infrastructure can be combined with road infrastructure. A zoning policy that prevents development in flood zones can be used to favor transit-oriented development. The hydro- meteorological information created to design a flood early warning system can then be used to manage better the energy system or optimize agricultural practices (Hallegatte, 2012a). The three dividends of resilience are outlined in Figure 1. 2 Figure 1. The three dividends of investing in DRM. Source: (ODI and GFDRR, 2015) This paper draws from the framework of Figure 1 and focuses on developing the narrative and providing evidence for part (iii) of Dividend 1 (reducing economic losses) and on Dividend 2 (development).1 The paper’s main message is that DRM investments can provide two dividends: reduced losses when a disaster strikes, and a shift of investment strategies and perhaps even an increase in investment value that would benefit the economy even before a disaster strikes. Providing evidence about the existence of both those dividends to policy makers and investors can provide a narrative reconciling short-term and long-term objectives, thereby improving the acceptability and feasibility of DRM investments. 2. Higher disaster losses at the macro level Total economic losses2 from a disaster can be much larger than the face value of what is affected (asset damages) and its impacts extend beyond the population and infrastructure directly affected. In this section, we review how and why losses can be higher at the macroeconomic level. Section 3 examines the microeconomic level. If disaster losses only impacted the people and assets directly affected, then managing risks could be considered an individual responsibility. Each individual would be in charge of making sure that the risks 1 Benefits linked to Divided 3 (on co-benefits) are discussed in Vorhies and Wilkinson (2015). 2 This chapter focuses on economic impacts only, leaving aside the critical issue of human lives lost from a disaster, the first priority of disaster risk management. 3 he or she takes are acceptable, and would be responsible for dealing with the consequences when a hazard materializes. In such a world, governments would have a limited role in risk management. In reality, there are three main reasons why risks often need to be managed collectively (World Bank, 2013). First, infrastructure to protect against natural disasters such as dikes and drainage systems is often “lumpy”. It consists of a complex system (such as multiple rings of dikes and pumping stations), has large up-front costs, covers large areas, and has to be built at once. Individuals or firms cannot provide hard protection to their houses or production facilities independently from investments at the collective level. Second, there is an unavoidable moral hazard issue when natural risks are concerned. Even if individuals take preventive action, governments have a duty to support the population in need following a natural disaster. Part of the losses is always socialized at the community, regional, or national level, and it makes economic sense for a government to invest in risk reduction in order to reduce the need for support after disasters. The third reason is that disasters not only affect the people and assets directly impacted by flooding water or high winds. Assets spared from a storm are often indirectly impacted and their use incapacitated as a result of “ripple effects” from disasters. For instance, segments of the population not hit by a storm can experience less income due to lower demand. These indirect impacts can affect long-term prospects of economic growth and development in the geographic area in which the shock occurred, but also elsewhere according to the level and scale of economic integration. 2.1. Indirect losses from the disruption of economic infrastructure and activity Disasters are “macro” events as they can affect all economic actors in the area where they occur – households, government agencies, and firms – even those who do not experience any material or human losses. (Smith and Mccarty, 2009) investigating the impact of the 2004 hurricane season in Florida on household displacement, find that among the 21% of households forced to move out of their homes after a disaster, 50% had to do so because of the loss of utilities (e.g., no running water). Only 37% had to move because of structural damage to their house. (Tierney, 1997) and (Gordon et al., 1998) also found that the loss of utility services and transport heavily impacted firms, following the 1994 Northridge earthquake in Los Angeles, California. According to (Tierney, 1997), 65% of the small businesses investigated closed after the earthquake because of the need to clean up damages. The five other most important reasons mentioned by 40% to 59% of the sample include the loss of electricity, employees’ inability to get to work, the loss of telephones, damages to owner’s home, and the reduction in demand with few or no customers (Table 1). Such issues are not related to direct structural damages to the business itself, but to off-site impacts. (Gordon et al., 1998) find that transportation loss amounted to 39% of total business loss. (Kroll et al., 1991) find comparable results for the Loma Prieta earthquake in San Francisco in 1989: the major problems for small businesses were customers and employees’ access as well as shipping delays, issues that had nothing to do with direct structural damages to businesses facilities. Utilities were of course an issue, but only in the short term,; these services were restored rapidly, while transportation issues had longer lasting consequences. 4 Reason Percentage of Firms Reporting Local (L) or Indirect (I) Needed to Clean-up Damage 65.2 Local Loss of Electricity 58.7 Indirect Employees Unable to Get to Work 56.4 Indirect Loss of Telephones 49.8 Indirect Damage to Owner or Manager's Home 44.4 Indirect Few or No Customers 39.9 Indirect Building Needed Structural Assessment 31.5 Local Could Not Deliver Products or Services 24.0 Indirect Loss of Machinery or Office Equipment 23.7 Local Building Needed Repair 23.4 Local Loss of Inventory or Stock 21.9 Local Loss of Water 18.2 Indirect Could Not Get Supplies or Materials 14.9 Indirect Building Declared Unsafe 10.1 Local Could Not Afford to Pay Employees 9.5 Local Loss of Natural Gas 8.7 Indirect Loss of Sewer or Waster Water 5.3 Indirect Other 15.8 Both Table 1. Reason for business closure following the 1994 Northridge earthquake in Los Angeles. Reasons linked to local damages to the business are highlighted in blue; others are indirect reasons, due to perturbations in infrastructure services such as transport or electricity. Source: (Tierney, 1997) Business activity does not occur in isolation. They are often integrated in a value chain and depend on upstream and downstream activities and stakeholders. Therefore, due to complex economic intricacies, business output losses can be the consequence of a shock on the economic activity both upstream (backward) and downstream (forward), and the creation of bottlenecks within supply chains.3 According to the position of the bottleneck in the value chain, ripple effects can be “backward” or “forward”:  Backward ripple effects arise when a shock propagates from clients to suppliers. For example, if the production of a client is incapacitated, input demand to its suppliers will also reduce. For suppliers, sales will reduce, this despite the absence of direct damages to its production capacity.  Forward ripple effects arise when the impact propagates from suppliers to clients. For example, when a client is open for operation but its supplier is unable to produce or sell inputs needed for production processes. 3 These ripple effects can even take place within a factory, if one segment of the production process is impossible and therefore interrupts the entire production. 5 The output losses from a disaster depend on firm-to-firm network characteristics such as the average number of suppliers, the degree of complementarity, and the shape and structure of connections between firms (Henriet et al., 2011). Modern organization of production, characterized by international production networks,4 a limited number of suppliers, small stocks, and production on demand have created new forms of vulnerabilities to natural disasters, well beyond domestic economy frontiers. The impact of disasters on global value chains are illustrated by the recent Tohoku-Pacific earthquake in Japan in March 2011, and its consequences on domestic industrial production and the resulting decrease in exports of goods used as inputs for instance in the auto industry. The Economic Times, an Indian newspaper, reported that “Japan's Toyota Motor will cut production at its Indian subsidiary by up to 70% between April 25 and June 4 due to disruption of supplies” (The Economic Times, 2011). If an economy’s capital stock consists of a bundle of complementary assets, the destruction of one component reduces the overall productivity of the entire production system with an indirect impact much larger than what could be expected from the analysis of one destroyed component only. One relatively straightforward example illustrating the difference between direct and indirect losses is given by the case of two cities connected by a single road. Destruction of only a segment of this road is enough to disrupt freight connections between those two cities. The loss resulting from the destruction of one segment of this road can therefore not only be estimated based on the value of this segment, but requires an analysis of the entire production system depending on the connection between the two cities. The same is true – to some degree – for the entire economic system: the loss of one asset will have repercussions on others that depend on it. According to asset substitutability within the system,5 this dependence will increase losses to a much higher value than the value of the lost asset. Past disasters provide useful examples. The San Francisco–Oakland Bay Bridge, essential to both cities’ economic activity, was closed for one month after the 1989 Loma Prieta earthquake (Figure 2). This closure impacted almost all small and large business in the Bay Area (Kroll et al., 1991), and while it was difficult to quantify the losses in economic activity, the scale of the output losses was likely an order of magnitude higher than the amount needed to repair the bridge. The health care system in New Orleans is another example. Beyond the immediate economic value of the service it provides, a functioning health care system creates positive externalities acting for example as a pull factor attracting workers to the region. With Katrina’s landfall in 2005, the health care system experienced significant disruption and did not recover to function quickly (Hallegatte, 2008; Rudowitz et al., 2006). Poor health care services made it more difficult to attract construction workers to the region (indeed, construction is a high-risk occupation), slowing down the reconstruction process. As a consequence, the disruption of health care services from the storm and its aftermaths went beyond the loss of its asset value. 4 We alternatively use the expressions “international production networks” and “global value chains”. 5 This problem is removed if one assumes that the capital stock is the result of an optimal process of capital accumulation, but this assumption is not valid in post-disaster contexts. 6 Figure 2. The Oakland-San Francisco Bay Bridge, which was closed for one month following the 1989 Loma Prieta earthquake. Source: Dan Bluestein, Wikimedia Commons. 2.1.1. Potential positive spillovers depending on demand absorption capacity Not all indirect impacts are negative. Disasters reduce production capacity, but also increase demand for outputs from the reconstruction sector. Thus, reconstruction can act as a stimulus. However, the resulting dynamic depends on pre-existing economic conditions, such as the phase of the business cycle and the existence of distortions that lead to under-utilization of production capacities (Hallegatte and Ghil, 2008). If the economy is efficient and in a phase of high growth, in which all resources are fully used, the net effect of a stimulus on the economy will be negative, for instance through diverted resources, production capacity scarcity, and accelerated inflation. If the pre-disaster economy is however depressed, the stimulus effect may in some cases (e.g. when there are substantial aid flows) yield benefits to the economy by mobilizing idle capacities. For instance, the 1999 earthquake in Turkey caused direct destruction amounting to 1.5-3% of Turkey’s GDP, but consequences on growth were limited, as the economy had significant unused production resources (labor and capital) available at the time, with Turkish GDP contracting by 7% in the year preceding the earthquake. The reconstruction demand from the earthquake aftermath allowed to use this under-utilized production capacity and acted as a stimulus, increasing economic activity (Ewing et al., 2004). In 1992, when Hurricane Andrew hit south Florida, the region’s economy was sluggish, with 50% unemployment among construction workers (West and Lenze, 1994). Reconstruction had a large stimulus effect in the economy, which would have been impossible in a better economic situation such as the one in 2004 when four hurricanes hit Florida during a housing construction boom (West and Lenze, 1994). 7 Finally, old and low-quality construction is generally more vulnerable to damages than more recent capital. In the case of a disaster, the destruction of low-quality assets may allow the possibility of “building back better”, improving the situation post-disaster. For instance, an earthquake may destroy old, low-quality, buildings, making it possible to rebuild with improved building norms. For example, after the Christchurch earthquake in New Zealand and 2011, building norms for energy efficiency led to better comfort and lower energy bills (Miles et al., 2014). Further, after the Victoria bushfires in Australia in 2009, measures to build back better including land-use planning and structural design improvements were successfully implemented (Mannakkara et al., 2014). However, experiences from the reconstruction process in Haiti after the 2010 earthquake show that building back better may be much more difficult in practice, due to a lack of adequate funding, and technical expertise and raw materials in the disaster location (Kijewski- Correa and Taflanidis, 2011). More general exploration of this effect, known as the “productivity effect” can be found in (Albala- Bertrand, 1993; Benson and Clay, 2004; Okuyama, 2003; Stewart and Fitzgerald, 2001). This effect is modeled in (Hallegatte and Dumas, 2009), who find that disasters can increase the production level but not the economic growth rate.6 Depending on how reconstruction is carried out (with more or less improvement in technologies and capital), moreover, accounting for the productivity effect can either decrease or increase disaster costs, but this effect cannot turn disasters into positive events, especially in areas lacking the technical capacity and raw materials to build back better. 2.2. Impact on long-term growth and development Natural disasters have economic impacts, which extend beyond the short and medium run, and affect long-run growth. Reconstruction indirectly impacts the economy by crowding-out consumption and investment. Post-disaster, uninsured households divert consumption towards reconstruction or draw down savings, potentially reducing the availability of investments in the economy (Hallegatte, 2014). The same is true for firms, which have to divert investments and profit redistribution to households towards reconstruction spending. This effect can have a broad, economy-wide depressing impact. Ranger et al. (2011) find that the total indirect effect to the economy from the 2005 floods in Mumbai, India, would have been halved (reduced by USD $200 million) if all losses had been paid through insurance instead of letting households use their savings and firms their own resources, as occurred. While such diversion can potentially have negative effects on the economy, so can a lengthy reconstruction process, which depends on the degree and capacity to divert funds away from investments and consumption. While the 10 billion euros spent on reconstruction expenditure following the 2002 floods in Germany only corresponds to the equivalent amount of investments spent over 10 days in the country, reconstruction was spread out over more than 3 years, suggesting only a small fraction of investments can be dedicated towards reconstruction. Therefore, reconstruction processes might become lengthier than expected as consumers, insurance and re-insurance companies, firms and public organizations need time to direct large amounts of money to reconstruction, a constraint especially stringent in developing economies which are already lacking 6 The authors use a simple model with embodied technical change. The finding that disasters cannot influence the economic growth rate is analogous to the way the saving ratio operates in a Solow growth model. 8 financial service infrastructure and lagging behind in investment capacity (Benson and Clay, 2004). Another source of friction is the reconstruction sector’s capacity to absorb the increase in demand following a disaster: skill availability and organizational capacities are adapted to the normal state of affairs and are not always able to face huge increases in demand. One illustration of this issue was the long reconstruction period that followed the French storms in 1999 or the AZF factory explosion in Toulouse in 2001, due to the lack of as roofers and glaziers. Therefore, the extent of the indirect losses due to the destruction of productive assets and infrastructure on economic activity and growth not only depends on the physical intensity of the natural event, but also on the coping capacity of the impacted human system and its ability to rebuild rapidly and efficiently. While investment spillovers are not an asset “loss”, in the absence of tools to better manage risk and reallocate resources post-disaster, economic losses are certainly higher. (Hallegatte, 2008) uses a model7 to quantify the direct and indirect losses from Katrina-like disasters in Louisiana. A non-linear relationship emerges: when direct losses are less than $50bn, reconstruction is rapid and aggregated indirect losses stay close to zero.8 Beyond $50bn of direct losses, the reconstruction period extends over several years and indirect losses increase exponentially. When direct losses exceed $200bn, total losses are twice as large as direct losses (Figure 3). 400 Indirect losses (US$b) 300 200 100 0 0 100 200 300 Direct losses (US$b) Figure 3. Indirect (output) losses as a function of direct (asset) losses, in Louisiana for Katrina-like disasters of increasing magnitude. The red curve signifies indirect losses. Source: (Hallegatte, 2008). Such non-linear relationships lead to large and long-term reductions in growth and lost output and may lead potentially to macro level poverty traps, with entire regions falling into a vicious circle leading the economy toward a lower growth equilibrium and reducing development capacity (Hallegatte et al., 2007). Such poverty traps can be explained by the amplifying feedbacks presented in Figure 4. Many regions have limited capacity to rebuild after a disaster. If the region is regularly affected, it may not have enough time 7 The model is based on input-output tables, considers sector production capacities, forward and backward propagations within the economic systems, and adaptive behaviors. For more information, see (Hallegatte, 2008) 8 Note the aggregation hides important disparities among sectors and social categories. 9 and resources to rebuild its asset base between two events. As a result, it may end up in a permanent state of reconstruction, allocating resources to rebuild rather than investing in new additional infrastructure and equipment, preventing capital accumulation and infrastructure development. Such a cycle, in the absence of external intervention, can lead to a permanent disaster-related under- development. Limited reconstruction capacity Reduced economic Long reconstruction development Amplifying period after disasters Loop Economic cost of disasters Reduced accumulation Ex-post (asset losses) and of capital and infrastructure Ex-ante (reduced investment) Figure 4. Amplifying feedback loop that illustrates how natural disasters could potentially become responsible for macro-level poverty traps. Over the long-run, the effect on economic growth is a balance between negative and positive spillovers. The impact of long-term disaster losses on economic growth has been studied in-depth with ambivalent results. On one hand, early studies, notably (Albala-Bertrand, 1993) and (Skidmore and Toya, 2002) suggest natural disasters have a positive influence on long-term economic growth as a result of both the productivity and the stimulus effects of reconstruction. On the other hand, more recent studies,9 suggest the overall impact of disasters on growth to be negative. At local scale, (Strobl, 2010) investigates the impact of hurricane landfall on county-level economic growth in the US. Growth in a county struck by at least one hurricane over a year is reduced on average by 0.79 percentage points. (Noy and Vu, 2010) investigate the impact of disasters on economic growth at the province level in Vietnam, and find lethal disasters decrease economic production while costly disasters increase short-term growth. Further, disasters have a differential macroeconomic impact, according to the ability a province to elicit transfers from the central government. Macro-impacts also depend on the ability of a country to access international capital (aid or new debt) from international financial institutions (IFIs) or capital from the international reinsurance system (e.g., through claim payments). Mechler, (2004), in a macroeconomic simulation of Honduras, shows that 9 This recent literature includes, in addition to the described studies, (Hochrainer, 2009; Jaramillo, 2010; Noy and Nualsri, 2007; Raddatz, 2009). 10 welfare effects ex-post depend on the willingness of the international community to provide aid, of IFIs to grant additional loans, and the extent to which sovereign insurance is taken on by the country (see also Mechler et al., 2015). The lack of consensus on the impact of disasters on GDP growth may be the consequence of the non- linearity of impacts due to the size of a disaster.10 Large disasters seem to have negative impacts on growth, while smaller disasters may not exert such an effect. For instance, (Felbermayr and Gröschl, 2014) find that disasters in the top decile of magnitude lead on average to a 3% reduction in GDP growth. The loss is only 1.5% for disasters in the top 15%, and 0.8% for disasters in the top 20%. For smaller disasters, no impact can be detected. Along the same lines, (Hsiang and Jina, 2014) find that “income losses arise [from disasters] from a small but persistent suppression of annual growth rates spread across the fifteen years following disaster, generating large and significant cumulative effects: a 90th percentile event reduces per capita incomes by 7.4% two decades later, effectively undoing 3.7 years of average development.” Through interruptions of infrastructure and baseline services, propagation in supply-chain, and diversion of spending by households and firms toward reconstruction, disaster losses go well beyond the direct assets destruction and affect the overall macroeconomics dynamics. These indirect effects, and what moderates them, are summarized in Table 2. Importantly, these macroeconomic indirect effects can be of the same order of magnitude than direct asset losses. (Hallegatte, 2012b; Hallegatte, 2008) estimate that Hurricane Katrina caused indirect losses that amounted up to 10-50% of direct losses. Furthermore, the most recent literature on the impact of disasters on growth suggests that while small disasters may not have long-term macroeconomic consequences, large ones are likely to have measurable long-term negative effects on economic growth (Felbermayr and Gröschl, 2014; Hsiang and Jina, 2014; Loayza et al., 2012). Type of Indirect Effect Moderated by Losses in electricity and transport Infrastructure quality and reliability Supply chain ripple effects Complementarity and size of shock Crowding out investment Level of insurance penetration Stimulus Existing economic situation Capital replacement Type of capital replacement Table 2. Summary of the indirect effects of natural disasters at the macroeconomic scale. 3. Welfare losses at the microeconomic level The previous section shows that focusing the evaluation of macroeconomic losses from a disaster on direct losses can be misleading and leads to underestimating the welfare impact. The total impact – which also includes output losses and forgone investment as a result from asset destruction and reconstruction processes – can be significant in the short-, medium- and long-run. 10 Other potential explanations for the ambivalent results is the low confidence in the data, the fact that not all types of disasters are included, during which part of the the business cycle the event occurs, and that shocks may increase GDP through defensive investments. 11 But underestimation of the welfare impact can also arise from the disregarding of the distributional impacts of disasters. For instance, it seems rather intuitive to think that the impacts of disasters on the livelihoods of poor and marginalized people are more substantial, first because of their higher exposure to physical risks, but also because of the reliance of their livelihood strategy on fewer and more vulnerable assets. While the impact of the disaster can be disastrous for such people, the repercussion on GDP can be invisible, especially if the very poor own close to nothing. Thus, to more precisely examine the impacts of a disaster at the micro level, it is important to examine who is affected and how they are affected. Below we first examine how asset losses are distributed among the population, and then how asset losses translate into welfare losses. 3.1. Asset losses differ depending on who is exposed Before examining how asset losses translate into total losses, it is imperative to assess how asset losses at the microeconomic level are determined and distributed. Asset losses are a function of the hazard, exposure and vulnerability. While a hazard is not determined by socio-economic characteristics, exposure and vulnerability are. One major determinant of asset losses is poverty status. First, poor people may be more exposed to natural disasters due to the role of formal and informal land markets: if natural risks are included in land price valuation (or desirability), poor households should be more likely to live in risky areas where land is cheaper (Fay, 2005). This explains why slums are typically located in floodplains or in areas at risk of mudslides, and why poor people are approximately 70% more likely to be exposed to disasters in cities such as Mumbai (see more in Section 0) (Patankar, 2015). But this may not always be the case. Risky locations may attract richer people: coastal cities are often highly exposed to flood risk, but they host households that are generally richer than those from rural and inland regions, due to sunny weather and amenities. For instance, (Carter et al., 2007) find that Hurricane Mitch in Honduras in 1998 affected only 22% of households in the poorest quintile, as opposed to 68% in the richest quintile. (Hallegatte et al., 2016) proposes a review of case studies of post-disaster contexts examining exposure of poor and non- poor people, see Figure 5, panel a. While the evidence on poverty exposure to disasters is scale- and context- dependent, it is generally well observed that when hit, poor people lose more in relative terms. This “vulnerability bias” is due to poor people having lower quality assets of which a larger portion in material form and thus more vulnerable to disasters. For instance, while (Carter et al., 2007) find that poor people did not have a higher exposure to Hurricane Mitch, they were nonetheless more vulnerable in relative terms: poor people had lost 31% of their assets while the rich only 8%. (Hallegatte et al., 2016) proposes a review of case studies of post- disaster contexts examining the vulnerability of poor and non-poor people, see Figure 5, panel b. 12 (a) (b) Figure 5. Poverty exposure bias and poverty vulnerability bias exhibited in prior case studies of disaster contexts. See sources in Hallegatte et al. (2016). The above studies suggest poor people are often more exposed to disasters, and when hit, lose more. The welfare impacts of disasters can be underestimates by aggregate loss figures, since the value of the assets of poor people is too small to appear in aggregate figures. Therefore, aggregated or averaged asset or output losses do not appear as a metric able to capture the full complexity of disaster outcomes. Instead, welfare losses may be a more appropriate metric. But how to calculate welfare losses from asset losses? 3.2. Welfare losses are different from asset losses Taking asset losses as a starting point, two additional steps are needed to estimate welfare impacts: i) how asset losses translate into income losses, and ii) the coping capacity and social protection offered at the individual and government levels. Figure 6 shows the chain from hazard to welfare impacts. Hazard? Exposed? Damages to Impact on Coping capacity Impact on assets? income? and social welfare? • Flood level • Localization of protection? • Wind speed people and • Housing quality • Diversification of • Marginal utility assets • Livestock and income • Social protection of consumption • Hard and soft other assets • Link between and scaling-up • Income protection • Infrastructure assets and • Smoothing distribution quality income (savings and • Non- • How long will borrowing) consumption the shock last? • Insurance and poverty and remittances welfare impacts Figure 6. The chain from a natural hazard to its impacts on welfare. Source: (Hallegatte et al., 2015). The impact of asset damage on income depends on three parameters. The first is related to the reconstruction duration, as described in Section 2. The second is the link between assets and income (productivity), and the third is the diversification of income. 13 Estimating how asset losses translate into income losses is difficult. Asset inventories are needed not only of assets that people own, but also on the complementary assets such as transportation infrastructure and production facilities that people use to make a living, as well as some information on the vulnerability of assets owned and used. According to the linkages presented in Section 2, the effect of a disaster on a household’s income strategy depends on employment in the firms nearby, and in the case of self- employment, changes in demand for goods. For example, after the 2011 floods in Bangkok, Thailand, (Noy and Patel, 2014) quantify the direct and spillover effects on income. Households who were directly affected lost on average THB 7,600 (approximately USD $220) in income; households not directly affected by the flood lost almost as much due to reduction in demand or business interruption and ripple-effects: THB 6,700. A third component that moderates or magnifies the impact of asset loss on income loss is the diversification of income, including from transfers such as pensions, social protection, and remittances. The impact on households’ income of the loss of local activities can be smoothed from income sources less affected by a disaster. In particular, government transfers such as pensions and social protection are diversified at the country-level, and if a disaster only impacts a small part of the country, transfers be only slightly reduced.11 Given impacts on income after the disaster, access to private and public coping strategies can reduce the welfare impact of an income loss. Strategies include private financial mechanisms as well as government targeted interventions and more generally social protection. Government transfers can be made available for a period of time after the disaster occurs to allow households to recover from the shock. After the 2005 floods in Mumbai, households received compensation from the government amounting to on average 10% of household asset losses (Patankar and Patwardhan, 2014). However, the compensation scheme did not appear to target poor people or those who lost the most (as is also found after the 2011 flood in Thailand, (Noy and Patel, 2014)). More generally, a socio-economic environment with a subsidized health care system and opportunities for employment can reduce welfare losses. Financial inclusion can also help. Savings accounts and insurance can smooth the impact of a shock over time. However, in most developing countries, while access to finance is slowly improving globally (Van Oudheusden et al., 2015), the ability for poor households to access such services remains limited. Another coping strategy at the household level is voluntary migration, of the entire family or one member of the family. However, evidence from rural Bangladesh suggests that poor households do not take advantage of migration opportunities, due to the cost of migration and possibility of not finding a job opportunity in a city (World Bank, 2013). A long reconstruction process, strong linkages between asset destruction and income generation, and few opportunities to smooth the shock through social protection or insurance increase the welfare impact of a disaster. 11 This is true in large countries. However, in small islands where almost all of the population is affected by a disaster, risk-sharing through diversification may not be an option. 14 In an environment with no or little access to social protection and smoothing mechanisms, a shock can potentially lead to poverty traps, especially for asset-poor households. Empirical evidence suggests that poor households may liquidate assets in order to cope with shocks and smooth consumption. If the liquidation of assets is insufficient or if shocks are too frequent to rebuild an asset base, households can fall into persistent poverty (Krishna, 2006). However, it is also shown that extremely poor households might on the contrary choose to smooth assets rather than consumption (Carter et al., 2007). These households choose to forego consumption rather than further liquidating limited assets in the hope of avoiding poverty traps in the current generation (if their asset base becomes too low, the household may be permanently stuck in poverty). But evidence suggests that such strategies may result in inter- generational poverty traps, as reduced consumption leads to in health and educational deficiencies that impact the human capital of children. Evidence suggests acute impacts on health from lower post-disaster consumption, especially after droughts. Following weather shocks in Sub-Saharan Africa, asset-poor households feed children with lower-quality nutrition food (Alderman et al., 2006; Dercon and Porter, 2014; Hoddinott, 2006; Yamano et al., 2005) and are less likely to take sick children for medical consultations (Jensen, 2000). These behaviors have short- and long-term impacts particularly for children younger than two. Among this group, households reducing nutrition lowered growth by 0.9 cm in six months post-disaster (Yamano et al., 2005) and were more likely to suffer recent illness (Dercon and Porter, 2014). In the long term, asset- smoothing households permanently lowered stature by 2.3 cm (Dercon and Porter, 2014) and 3 cm (Alderman et al., 2006). (Hoddinott, 2006) also observes the body mass index (BMI) of women reduced 3%; while this recovered the following year, impacts on children are long-lasting. Furthermore, the literature shows that health shocks are more likely to bring households into poverty where households can only borrow at high interest rates (e.g., Krishna, 2006). It suggests that health shocks create poverty not only by reducing income, but also through health care expenditures and excess borrowing. More generally in reconstruction contexts, the experience after Cyclone Nargis in Myanmar in 2008 suggests high borrowing rates cripple the speed of recovery (World Bank, 2015a). Financial inclusion and universal health care insurance could therefore be powerful instruments to reduce the welfare impacts in disaster contexts. Another impact of lower-post disaster consumption on children’s human capital occurs through education. Jensen (2000) finds enrollment rates declined 20% in exposed regions, (Alderman et al., 2006) find drought-affected households delayed starting school of children on average 3.7 months, and this sample completed 0.4 fewer grades. (Dercon and Porter, 2014) find those younger than 36 months at the apex of the famine were less likely to have completed primary school, and estimate the impact to be equivalent to a 3% income losses per year (Dercon and Porter, 2014).12 Inter-generational impacts may endure: recent research in Uganda suggests that educated household heads are much less likely to choose coping strategies that involved taking their children out of school (Helgeson et al., 2013). 12 Such findings are not restricted to Africa; similar impacts on health and education post-disaster have been found for instance in Asia, Latin America and elsewhere (Baez et al., 2014; Maccini and Yang, 2009). 15 The above discussion first highlights that exposure and asset vulnerability differ based on poverty status, with poor people being more impacted in relative terms from a disaster. Further, we highlight the importance coping capacity and social protection, when examining the impact on welfare. This highlights the potential for DRM investments to help people manage risks, long-term impacts on welfare and prevent potential for poverty traps dynamics to occur after a disaster further degrading poor households’ welfare. 4. Slower development in the absence of DRM investments The take-away from Sections 2 and 3 is that at the macro and micro levels, asset losses do not tell the whole story, and while the impact of a disaster might not be extensive at the macro level, without DRM investments, welfare losses can still be substantial for parts of the population, especially poor people. While this ex-post, loss-centric focus is certainly important, it ignores the ex-ante impacts. In risky environments, evidence suggests that without the proper tools to manage natural risk, risk-averse households and other economic actors tend to spread risk over a large array of lower risk activities, thereby reducing returns to assets and investments. For example, smallholders plant low-return, low-risk crops (Cole et al., 2013), and infrastructure services may not be well provided in risky areas (Benson, 2002). Such phenomenon can be self-sustaining as evidence suggests that after a disaster, economic actors may change behaviors and become more risk averse, with less risks taken in other domains, for example in innovation and entrepreneurship (Cameron and Shah, 2015). In this section, we examine first the benefits of taking “natural” risks when these risks are well managed. Then we highlight the existence of suppressed economic potential in risky areas, an issue that could be partly addressed with DRM investments. 4.1. Development and the exposure to natural hazards Economic losses from natural disasters have been growing as fast or faster than economic growth in many countries (Neumayer and Barthel, 2011; Nordhaus, 2010; Pielke et al., 2008). This evolution can be explained by growth – richer countries have more assets potentially damaged, but also by increasing risk- taking behaviors with an increasing share of assets located in at-risk areas (Field et al., 2012; Neumayer and Barthel, 2011; Schmidt et al., 2009). For instance, the fact that hurricane losses are increasing more rapidly than GDP in the US is explained by migration toward hurricane-prone areas (Pielke et al., 2008). Globally, between 1970 and 2010, the population grew by 87%, but the population living in floodplains increased by 114% and in cyclone prone coastlines by 192%. The share of GDP globally exposed to tropical cyclones increased from 3.6% to 4.3% over the same period (UNISDR, 2011). Many scholars have looked at reasons why individuals may over-invest in risky locations and take excessive risks in terms of flood or storm exposure. (Kunreuther, 1978; Thaler and Sunstein, 2009; Tversky and Kahneman, 1974) investigate behavioral biases in how people perceive and react to risks. (Burby et al., 1991; Laffont, 1995; Michel-Kerjan, 2008) study moral hazard issues linked to post-disaster support and ill-designed insurance schemes. 16 But beyond these well-identified biases in decision-making, taking risks is sometimes an unavoidable (or desirable) consequence of development and economic growth. Investing in risky areas can be a conscious and well-informed choice, justified by economic benefits. For instance, increase exposure to natural hazard can be an unavoidable side-effect from investments to create additional employment and growth from international trade in areas characterized by low transportation costs but exposed to flood risks (e.g., (Gallup et al., 1998)). In China, for instance, (Fleisher and Chen, 1997) find that Total Factor Productivity (TFP) is 85% higher in coastal regions than inland, and that TFP growth is not significantly different in spite of higher investment in inland regions, suggesting a permanent productivity advantage in coastal regions from lower transport costs. As an illustration, landlocked countries have higher transport costs (measured by the shipping costs), and had over the 1965-1990 period a growth rate on average 1 percent lower than coastal countries, which are at risk from coastal floods and storms (Gallup et al., 1998). Cheap waterway transport attracts industrial production close to floodplains, and partly explains why most large cities are located on rivers. In coastal areas, increased exposure to flood can therefore be a deliberate trade-off against higher productivity and economic growth. The same thing may happen in cities. The drivers of economic growth are concentrated in cities, and productivity growth is larger in cities in part because of positive agglomeration and concentration externality. (Ciccone, 2002; Lall and Deichmann, 2012; World Bank, 2008) report urban-rural income ratios between 1.5 for developed countries and up to 3 for developing countries, suggesting higher productivity in cities at all stages of development. And not only are productivity and consumption higher in urban areas, but amenities and infrastructure services are often superior: among low-income countries with urban population shares of less than 25 percent, access to water and sanitation in towns and cities is around 25 percentage points higher than in rural areas (World Bank, 2008). These differences create strong incentives for rapid rural-urban migration. Confronted with land scarcity and high land costs in large cities, this migration has led to construction in at-risk areas (Burby et al., 2006, 2001; Lall and Deichmann, 2012). In the most marginal and risky locations, informal settlements and slums are often present, putting poor and vulnerable populations in a situation of extreme risk (Ranger et al., 2011). An illustrative example of poor people settling in risky areas is Mumbai, prone to high flood risk from the Mithi River. Within the Mithi River basin lie a dense population estimated at 1.5 million and a high concentration of assets. Flood hazard, combined with high socio-economic exposure pose a major risk, evidenced by unprecedented floods in July 2005, which resulted in more than 500 fatalities and $2bn in asset losses (Ranger et al., 2011). Patankar (2015) reports on a surveys of poor households living in Mumbai’s flood areas, and shows that poor people are well aware of this risk, and are making a deliberate decision to live there to benefit from higher-wage jobs, better schools and medical care, and existing social networks. Risk-taking can also increase welfare through environmental amenities (e.g., from sea views) and generate revenues from tourism. One measure of the importance of tourism to an economy is the percentage of international tourism expenditures as a percentage of total imports. As of 2012, in eleven countries (Bahamas, Cabo Verde, Dominica, Grenada, Macao, Maldives, Montenegro, Samoa, Sao Tome & Principle, St. Lucia, St. Vincent and the Grenadines, and Vanuatu) tourism accounted for more than half of total exports (World Bank, 2015b). Most of these countries are island nations exposed to natural risk (mostly 17 hurricanes and sea level rise), yet expenditures from overseas visitors plays a large role in economic output and can hardly be realized without increasing risk. In situations where there is a trade-off between exposure to natural hazards and productivity or economic growth, improved risk management and more resilient development can mitigate this trade-off and accelerate growth and improved productivity (Hallegatte 2014). This issue relates to the opportunity costs of ex-ante natural risk management, both by households and lenders. Uninsured risk exposure as well subjective perception endogenously change behaviors, and thus the conditional expected wealth creation dynamics. Due to failures in financial markets and risk aversion, the risk of weather-related shocks, including disasters, influences household choices of livelihood strategies in order to minimize the consequences of a shock. Households trade off expected gains for the reduced risk of suffering catastrophic losses. Such livelihood strategies often entail diversification of activities and less productive investments, constraining productivity and wealth accumulation: households undertake costly behaviors as a means of reducing their exposure to uninsured risk, resulting in forgone welfare gains. Taking into account the prospective consequences of shocks, poor households may manage risk exposure by selecting low-risk, low-return asset and activity portfolios that reduce the risk of greater suffering but limit growth potential and investment incentives (Carter and Barrett, 2006; Dercon, 2005; Rosenzweig and Stark, 1989). This for instance discourages adoption of new technologies and decreases incentives to invest in productive capital accumulation. An illustration of this effect is provided in an agricultural context in Zimbabwe by (Elbers et al., 2007). They find that farmers exposed to risk exhibit a mean capital stock which is half as large as for farmers who are not exposed. Of this reduction in capital, ex-ante risk accounts for two-thirds of the difference. In that case, therefore, most the welfare impact of risk is through reduced investments and risk-taking, not through the damages and losses when the hazard does materialize into an actual event. 4.2. Lower risk-taking due to “background risk” The previous section highlights that investing in natural risk-prone areas, such as settling in urban flood- prone areas, may be necessary to improve living conditions and access higher-paying jobs and better schools. But households and firms face a wide variety of potential shocks that they have to manage together. As an illustration, the 2014 World Development Report (World Bank, 2013) reports the frequency of occurrence of a variety shocks, from loss of job to health and floods, in a number of developing countries (Table 3). In most countries surveyed, a large proportion of rural households report being affected by two or more shocks, with drought and flood predominant across most countries and regions. 18 Afghanistan India Lao PDR Malawi Peru Uganda Shocks U R R U R U R U R U R One or more 16.4 48.9 61.6 34.4 72.1 40.0 66.8 20.7 34.4 29.7 56.2 Two or more 8.7 39.2 23.4 11.9 36.1 12.7 40.4 1.4 1.9 5.6 15.6 Natural disasters (drought, flood) 10.6 42.2 57.3 5.6 36.0 10.4 47.2 2.6 21.5 19.9 52.1 Price shocks 0.2 3.0 –– 4.4 4.9 21.1 42.0 –– –– 1.7 3.2 Employment shocks 6.4 4.3 –– 9.3 3.1 7.7 3.4 6.4 1.5 1.9 0.7 Health shocks (death, illness) 6.9 14.0 30.2 23.2 33.8 10.1 18.0 9.1 8.9 11.8 14.9 Personal and property crime 1.8 6.6 0.9 5.8 1.9 8.5 8.4 3.2 3.1 6.6 8.7 Family and legal disputes –– –– 1.9 0.0 0.9 1.7 4.3 0.7 0.3 –– –– Table 3. Households in developing countries face many shocks. Percentage of respondents reporting type of shock. Source: (World Bank, 2013) based on data from household surveys, various years 2005-2011. U=Urban, R=Rural. Importantly, the evidence reviewed in this section suggests households consider their vulnerability to natural risks like floods and droughts when making other risk-related decisions in other domains – such as creating a business or migrating to a city. Because these risks interact, the existence of natural risk can reduce the willingness to take these other risks, which are necessary for development and growth. Empirical evidence on innovation and entrepreneurship suggests for instance that increased risk-taking behaviors are associated with higher economic growth and development: - The contribution of risk-taking (e.g. through increased innovation/entrepreneurship) to economic growth is well-established in the economic literature and was grounded on the theory of endogenous technical change (Aghion and Howitt, 1992; Grossman and Helpman, 1991; Romer, 1990). The empirical evidence that has followed has largely supported the theory. For innovation, early reviews find a positive link between innovation and output (Cameron, 1998; Nadiri, 1993). Econometric studies (measuring innovation through patents) provide further support and suggest countries hosting a larger number and higher quality patents also experience higher economic growth (Hasan and Tucci, 2010; LeBel, 2008; Şener and Sarıdoğan, 2011; Yang, 2006). - Regarding entrepreneurship and growth, early studies suggested new business formation promotes employment growth (Birch, 1987; Reynolds, 1999; Wennekers and Thurik, 1999), increased incomes (Carree and Thurik, 2002; Davidsson, 1995; Picot et al., 1998) and greater total factor productivity growth (Aghion et al., 2004; Baumol, 2014). In a review of 57 studies, (van Praag and Versloot, 2007) find that entrepreneurial firms have higher productivity growth and increased innovation. However, employing data on 37 countries, (Wong et al., 2005) suggest that while fast growing new firms lead to increased job creation and higher growth, new firms in general do not. - Furthermore, risk aversion has been linked to lower investment in physical and human capital (Rosenzweig and Stark, 1989), wage growth (Shaw, 1996), and technology adoption (Liu, 2012) thereby reducing growth and economic development potential. If high natural risks lead individuals to become less risk-taking in terms of innovation, education or entrepreneurship, growth and development will suffer. 19 Gollier’s seminal work (Eeckhoudt et al., 1996; Gollier and Pratt, 1996; Gollier and Schlee, 2006) finds, under fairly general conditions, that a higher level of “background risk” (here, the flood/drought risks) makes individuals less willing to take risks in other domains (e.g. innovation/entrepreneurship). In other words, being exposed to one risk increases an individual’s risk aversion regarding other categories of risk. These results suggest that households consider their vulnerability to natural risks like floods and droughts when making other risk-related decisions in other domains – such as creating a business or migrating to a city. Empirical work finds that higher levels of background risk are associated with increased risk-aversion in financial decisions (Guiso and Paiella, 2008; Heaton and Lucas, 2000; Lusk and Coble, 2008). More recent literature also finds evidence of risk vulnerability with regards to land reform (Tella et al., 2007), early life financial experiences (Malmendier and Nagel, 2011), stock-market crises (Guiso et al., 2013), and violent trauma (Callen et al., 2014; Voors et al., 2012). There are two mechanisms through which an increase in the background risk can lead to high risk aversion and lower investment in growth and development. - The first is perfectly rational: there is a possibility that the two independent risks (one related to disasters, the other to risk-taking in general) materialize together (Gollier and Pratt, 1996). This combined risk – and the non-linearity in the utility function – increases risk aversion because a large income shock changes not just an individual’s location on the utility function, but also the shape of that function (Cassar et al., 2015). - The second mechanism is behavioral. A shock such as a flood can lead to an overestimation in an individual’s perceived likelihood of future natural shocks occurring. (Cameron and Shah, 2015) find, after a flood in Indonesia, that an individual’s expectations of future flood occurrence is an order of magnitude higher than the true probability. Emotional responses can lead individuals to have greater fear of any negative event, reducing risk-taking (Cassar et al., 2015). Consequences, either real or perceived, from multiple shocks occurring in close proximity or simultaneously can be devastating. The importance of past events on risk aversion is documented in a number of countries including Indonesia, Bangladesh, Nicaragua and Peru. Indonesia. East Java, Indonesia has a population of 37 million and is particularly prone to natural disasters – with flood and earthquake posing the largest risks. (Cameron and Shah, 2015) examine whether recent experience of floods and earthquakes impacts the level of background risk and risk-taking within the region. In October 2008, the authors ran a series of experimental games in a random sample of 1,550 individuals across 120 villages, and find that individuals in villages that suffered a flood/earthquake in the past three years exhibit higher levels of risk aversion compared to individuals in villages that did not experience a disaster (41% decrease in probability of making a risky choice in the experiment). A year later, the authors conduct a survey asking households to report the probability (or likelihood) that a flood and/or earthquake would occur in their village the following year. For floods (but not earthquakes), individuals who experienced an event are significantly more likely to report a higher probability of flood 20 in the following year (43%) compared to those who did not experience a flood (12%). Given the true probability of around 3%, the findings suggest households with recent flood experience over-weight the probability that a future flood occurs. The same is true with severity: those who experienced a flood also perceive that future floods to be worse. These findings suggest individuals with recent experience perceive the world a riskier place; the authors suggest this causes individuals to take fewer risks. Evidence is further provided that behavior in experiments is correlated with “real life” risk-taking such as entrepreneurship. Bangladesh. Bangladesh is particularly at risk of coastal flooding and cyclones. (Ahsan, 2014) examine risk preferences in three coastal communities in the Bagerhat, a district in southwest Bangladesh, which regularly experiences cyclones. Socio-economically, the communities studied are heavily reliant on aquaculture and agriculture and are low-income, with average household annual income from farming reported at USD $1,400. Through experiments, risk preferences are investigated and compared with exposure to cyclone. The authors find that on average, non-cyclone-affected subjects bet more in a risk game than subjects who have been affected by cyclone (Figure 7). Figure 7. Subjects in a risk game in southwest Bangladesh more exposed to cyclones exhibit more risk-averse behavior. Source: (Ahsan, 2014) Nicaragua & Peru. Nicaragua and Peru are two disaster-prone countries in Latin America, at risk of flood, drought and hurricane. In 2007, (van den Berg et al., 2009) conducted risk experiments on a random sample of 100 individuals across regions within each country (Chinandega in Nicaragua; Ancash, Cajamarca, Piura, Tumbes in Peru), the authors found past experience of a disaster to have a large and significant effect on risk aversion. Across both samples, comparing individuals who lost assets to those who did not, those who lost a home exhibit 30% higher risk aversion; for those who lost animals this measure is 50% and for crops 60%. The authors similarly suggest that such reductions in risk aversion continue in the medium run, 2 years after a disaster. While the authors do not provide evidence on the 21 mechanism through which risk aversion manifests, one plausible conclusion is the increased perception of background risk. In Vietnam, (Reynaud and Nguyen, 2012) find experience of floods to have a significant positive effect on demand for insurance, which may reflect higher levels of risk-aversion. Also in Vietnam, (Dang, 2012), combining historical and contemporary survey data in Vietnam, finds that individuals living in villages that frequently experience a disaster and those who recently experienced a shock show higher levels of risk- aversion. (Abreha, 2007) finds similar results of drought experience and risk-aversion among farmers in Ethiopia. However, some studies find the opposite – that exposure to natural disasters can make people more risk- loving. In Louisiana, Eckel et al. (2009) conducted an experimental test on individuals exposed to Hurricane Katrina in September 2005, a month after the storm. He finds evacuees to be more risk-loving (less risk averse) after the storm, although this effect is not observed 10 months later. Various other analyses provide evidence on such change in perception: (Page et al., 2014) provide evidence of a similar risk-loving effect with households who suffered loss as a result of the 2011 Australian floods in Brisbane;13 (Hanaoka et al., 2015) exhibits similar findings using panel data from Japanese households after the 2011 earthquake and tsunami, but only for men, who gamble and drink more after the event.14 Andrabi and Das (2010) find the 2005 earthquake in Pakistan increased risk aversion; Said et al. (2014) find a similar result for the 2010 flood in Pakistan. But not all studies confirm the finding that risk preferences change. (Bchir and Willinger, 2013), in a field experiment of lahars risk in Peru, find no significant difference of risk aversion between exposed and non- exposed households. (Becchetti et al., 2012) finds a similar result of no significant difference in a sample of 380 Sri Lankan microfinance borrowers. In addition, preliminary findings from an experimental game in Cambodia’s Battambang province actual finds experience with a natural disaster has a positive and significant impact on risk behavior of participants (Fiala, 2015). The contradiction cannot be easily explained by different context, since studies disagree even in one given location. (Cassar et al., 2015), through risk experiments of 334 subjects from Thai villages affected in different degrees by the 2004 event, find that individuals hit hardest by the disaster exhibit strong risk aversion four and a half years after the disaster (in 2009). (Callen, 2011), finds no evidence that risk preferences were changed in an experiment conducted on a sample of 456 wage workers in July 2007. One possibility is that the impact of background risk is more complex than a simple increase in risk aversion. For instance, (Li et al., 2011) finds that individuals exposed to earthquake and snowfall risk in China cannot simply be described as more risk- seeking, but that individuals give more weight to low probabilities after the 2008 China earthquake and snowfall event. 13 One limitation of this study is that relative to the household’s situation at the time of survey, the risk game only presents gain options. 14 The authors find evidence that men become more engaged in gambling and drinking if they were more exposed to the Earthquake. 22 5. Conclusion and implications for policy Most investments in DRM still rely on cost-benefit analyses that estimate the benefits from a project or action through the value of the asset (and/or human losses) it can prevent. But indirect losses can be as substantial as direct asset losses; and indirect losses can lead to human losses (e.g., through undernourishment and children stunting) that need to be added to direct human losses. While these costs are difficult to quantify, and perhaps because so, they are typically excluded from cost- benefit analyses. Nonetheless, the benefits of a DRM policy to reduce indirect losses can be large. Some DRM action even reduce only indirect losses – for instance, insurance and social protection cannot do much to reduce asset losses, but they minimize the welfare impacts of these losses.15 Welfare losses can also be much higher than asset losses, when considering the distribution of these losses, and especially the impact on the poorest. And the development benefits from better-managed risks – for instance through the ability to take other risks linked to entrepreneurship or innovation – could also be significant. Yet, this is also difficult to quantify and include in a cost benefit analysis. For instance, the benefits people gain from settling in risky areas in urban areas are typically not valued. Put simply, at present, the “benefits” of DRM are understated – both in terms of avoided losses and increased development. Considering these benefits in policy design is critical to better manage risk. For example, some actions to reduce risk (or prevent risk generation) may be counter-productive. What is really needed is not risk reduction – that would try to reduce the amount of risk-taking indiscriminately – but risk management – that prevents excessive risk-taking while allowing risk-taking in cases where the benefits (e.g. proximity to job opportunity) are clear. For instance, policies that prevent all investments in flood zones in developing countries cities may be extremely costly. They would reduce migration to cities, thereby potentially preventing individuals to access higher-pay jobs, better services and children’s to access to education. It would be more efficient to implement more detailed zoning policies that distinguish between different types of investments to allow worthy ones but prevent inappropriate ones (e.g., by making a difference between housing and production units). Another option is to invest in safe places. Indeed, it can be rational to experience growing disaster losses only if investments in risky locations are “more” productive than investment in safe places. If investments in transport can make it as desirable to invest in safe places, risk could be reduced without reducing economic growth and output. People in at-risk informal settlements in developing country cities settle there because they face a difficult trade-off between living in risky places with good access to jobs and services or to live in safe places without these opportunities. They would settle in safe places and reduce flood exposure if better transportation infrastructure and options would connect safe living areas to urban opportunities. Similarly, manufacturing plants are created in at-risk coastal areas, but they could be installed in safe areas if transport infrastructure made it possible to ship their production at similar costs. In the broad framework proposed in this paper, transportation investments are risk mitigation 15 Note that well-designed insurance schemes can also create a positive incentive to invest in risk mitigation and prevention. 23 investments, when they connect safe areas to the opportunities and amenities that currently exist in risky areas. Providing a strong and holistic risk management framework in a country, region, or city makes it possible for all actors to take the risks that are desirable, avoiding excessive risk-taking without constraining growth and development. It also makes it possible to deal with the rare but unavoidable cases where a physical hazard is so violent that it exceeds protection capabilities and causes large losses. In other words, the same DRM policy which reduces welfare losses from a disaster can also provide benefits even before a disaster strikes. 6. References Abreha, N.H., 2007. An economic analysis of farmers’ risk attitudes and farm households’ responses to rainfall risk in Tigray Northern Ethiopia (PhD Thesis). Wageningen University. Aghion, P., Blundell, R., Griffith, R., Howitt, P., Prantl, S., 2004. Entry and Productivity Growth: Evidence from Microlevel Panel Data. Journal of the European Economic Association 2, 265–276. doi:10.1162/154247604323067970 Aghion, P., Howitt, P., 1992. A Model of Growth Through Creative Destruction. Econometrica 60, 323– 351. doi:10.2307/2951599 Ahsan, D.A., 2014. Does natural disaster influence people‫׳‬s risk preference and trust? An experiment from cyclone prone coast of Bangladesh. International Journal of Disaster Risk Reduction 9, 48– 57. doi:10.1016/j.ijdrr.2014.02.005 Albala-Bertrand, J.M., 1993. Political economy of large natural disasters: with special reference to developing countries. Clarendon Press ; Oxford University Press, Oxford, [England] : New York. Alderman, H., Hoddinott, J., Kinsey, B., 2006. Long term consequences of early childhood malnutrition. Oxf. Econ. Pap. 58, 450–474. doi:10.1093/oep/gpl008 Andrabi, T., Das, J., 2010. In Aid We Trust: Hearts and Minds and the Pakistan Earthquake of 2005. Baez, J., Lucchetti, L., Salazar, M., Genoni, M., 2014. Gone with the Storm: Rainfall Shocks and Household Wellbeing in Guatemala. Baker, J., Basu, R., Cropper, M., Lall, S.V., Takeuchi, A., 2005. Urban Poverty And Transport : The Case Of Mumbai, Policy Research Working Papers. The World Bank. Banerjee, A., Duflo, E., 2012. Poor Economics: A Radical Rethinking of the Way to Fight Global Poverty, Reprint edition. ed. PublicAffairs, New York, NY. Baumol, W.J., 2014. Stimulating growth amid recession: Entrepreneurship, innovation, and the Keynesian revolution. Journal of Policy Modeling, Rapid Growth or Stagnation in the U.S. and World Economy? 36, 629–635. doi:10.1016/j.jpolmod.2014.03.004 Bchir, M.A., Willinger, M., 2013. Does the exposure to natural hazards affect risk and time preferences? Some insights from a field experiment in Perú (Working Paper No. 13-04). LAMETA, Universtiy of Montpellier. Becchetti, L., Castriota, S., Conzo, P., 2012. Calamity, Aid and Indirect Reciprocity: the Long Run Impact of Tsunami on Altruism (CSEF Working Paper No. 316). Centre for Studies in Economics and Finance (CSEF), University of Naples, Italy. Benson, C., 2002. Disaster Management: Pro-Poor Infrastructure Provision. Benson, C., Clay, E., 2004. Understanding the Economic and Financial Impacts of Natural Disasters, Disaster Risk Management. The World Bank. Birch, D.G.W., 1987. Job Creation in America: How Our Smallest Companies Put the Most People to Work (SSRN Scholarly Paper No. ID 1496185). Social Science Research Network, Rochester, NY. 24 Burby, R.J., Birch, E.L., Wachter, S.M., 2006. The problems of containment and the promise of planning, in: In Rebuilding Urban Places after Disaster: Lessons from Hurricane Katrina. University of Pennsylvania Press. Burby, R.J., Cigler, B., French, S., Kaiser, E., Kartez, J., Roenigk, D., Weist, D., Whittington, D., 1991. Sharing environmental risks : how to control governments’ losses in natural disasters. Westview, Boulder, CO. Burby, R.J., Nelson, A.C., Parker, D., Handmer, J., 2001. Urban Containment Policy and Exposure to Natural Hazards: Is There a Connection? Journal of Environmental Planning and Management 44, 475–490. doi:10.1080/09640560120060911 Callen, M., 2011. Catastrophes and time preference: evidence from the Indian Ocean earthquake. Unpublished Manuscript, University of California at San Diego. Callen, M., Isaqzadeh, M., Long, J.D., Sprenger, C., 2014. Violence and Risk Preference: Experimental Evidence from Afghanistan. American Economic Review 104, 123–48. doi:10.1257/aer.104.1.123 Cameron, G., 1998. Innovation and Growth: a survey of the empirical evidence. Cameron, L., Shah, M., 2015. Risk-Taking Behavior in the Wake of Natural Disasters. J. Human Resources 50, 484–515. doi:10.3368/jhr.50.2.484 Carree, M.A., Thurik, A.R., 2002. The impact of entrepreneurship on economic growth. Carter, M.R., Little, P.D., Mogues, T., Negatu, W., 2007. Poverty Traps and Natural Disasters in Ethiopia and Honduras. World Development 35, 835–856. doi:10.1016/j.worlddev.2006.09.010 Carter, P.M.R., Barrett, C.B., 2006. The economics of poverty traps and persistent poverty: An asset- based approach. The Journal of Development Studies 42, 178–199. doi:10.1080/00220380500405261 Cassar, A., Healy, A., Kessler, C. von, 2015. Trust, Risk and Time Preferences after a Natural Disaster: Experimental Evidence from Thailand. Ciccone, A., 2002. Agglomeration effects in Europe. European Economic Review 46, 213–227. Clarke, D.J., Hill, R.V., 2013. Cost-Benefit Analysis of the African Risk Capacity Facility (SSRN Scholarly Paper No. ID 2343159). Social Science Research Network, Rochester, NY. Cole, S., Gine, X., Tobacman, J., Topalova, P., Townsend, R., Vickery, J., 2013. Barriers to Household Risk Management: Evidence from India. American Economic Journal: Applied Economics 5, 104–35. doi:10.1257/app.5.1.104 Dang, A.D., 2012. On the Sources of Risk Preferences in Rural Vietnam (ANU Working Papers in Economics and Econometrics No. 2012-593). Australian National University, College of Business and Economics, School of Economics. Davidsson, P., 1995. Culture, structure and regional levels of entrepreneurship. Entrepreneurship & Regional Development 7, 41–62. doi:10.1080/08985629500000003 Dercon, S., 2005. Insurance Against Poverty. OUP Oxford. Dercon, S., Porter, C., 2014. Live Aid Revisited: Long-Term Impacts of the 1984 Ethiopian Famine on Children. Journal of the European Economic Association 12, 927–948. doi:10.1111/jeea.12088 Eckel, C.C., El-Gamal, M.A., Wilson, R.K., 2009. Risk loving after the storm: A Bayesian-Network study of Hurricane Katrina evacuees. Journal of Economic Behavior & Organization, Individual Decision- Making, Bayesian Estimation and Market Design: A Festschrift in honor of David Grether 69, 110–124. doi:10.1016/j.jebo.2007.08.012 Eeckhoudt, L., Gollier, C., Schlesinger, H., 1996. Changes in Background Risk and Risk Taking Behavior. Econometrica 64, 683–689. doi:10.2307/2171866 Elabed, G., Carter, M.R., 2015. Compound-risk aversion, ambiguity and the willingness to pay for microinsurance. Journal of Economic Behavior & Organization. doi:10.1016/j.jebo.2015.03.002 Elbers, C., Gunning, J.W., Kinsey, B., 2007. Growth and Risk: Methodology and Micro Evidence. World Bank Econ Rev 21, 1–20. doi:10.1093/wber/lhl008 25 Ewing, B.T., Kruse, J.B., Özdemir, Ö., 2004. Disaster Loses in the Developing World: Evidence from the August 1999 Earthquake in Turkey (No. 2004/19). Discussion Paper, Turkish Economic Association. Fay, M., 2005. The Urban Poor in Latin America, Directions in Development - General. The World Bank. Felbermayr, G., Gröschl, J., 2014. Naturally negative: The growth effects of natural disasters. Journal of Development Economics, Special Issue: Imbalances in Economic Development 111, 92–106. doi:10.1016/j.jdeveco.2014.07.004 Fiala, O., 2015. Experiencing natural disasters: how this influences risk aversion, trust and the demand for microinsurance. Field, C.B., Barros, V., Stocker, T., Qin, D., Dokken, D.J., Ebi, K.L., Mastrandrea, M.D., Mach, K.J., Plattner, G.-K., Allen, S.K., Tignor, M., Midgley, P.M., 2012. Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. A Special Report of Working Groups I and II of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK and New York, NY, USA. Fleisher, B.M., Chen, J., 1997. The Coast–Noncoast Income Gap, Productivity, and Regional Economic Policy in China. Journal of Comparative Economics 25, 220–236. doi:10.1006/jcec.1997.1462 Gallup, J.L., Sachs, J.D., Mellinger, A.D., 1998. Geography and Economic Development (Working Paper No. 6849). National Bureau of Economic Research. Gollier, C., Pratt, J.W., 1996. Risk Vulnerability and the Tempering Effect of Background Risk. Econometrica 64, 1109–1123. doi:10.2307/2171958 Gollier, C., Schlee, E., 2006. Increased Risk-Bearing with Background Risk (Working Paper). Department of Economics, W. P. Carey School of Business, Arizona State University. Gordon, P., Richardson, H.W., Davis, B., 1998. Transport-related impacts of the Northridge earthquake. National Emergency Training Center. Grossman, G.M., Helpman, E., 1991. Quality Ladders in the Theory of Growth. The Review of Economic Studies 58, 43–61. doi:10.2307/2298044 Guiso, L., Paiella, M., 2008. Risk Aversion, Wealth, and Background Risk. Journal of the European Economic Association 6, 1109–1150. Guiso, L., Sapienza, P., Zingales, L., 2013. Time Varying Risk Aversion (Working Paper No. 19284). National Bureau of Economic Research. Hallegatte, S., Bangalore, M., Bonzanigo, L., Fay, M., Kane, T., Narloch, U., Rozenberg, J., Treguer, D., Vogt-Schilb, A. Shock Waves : Managing the Impacts of Climate Change on Poverty. Washington, DC: World Bank. https://openknowledge.worldbank.org/handle/10986/22787 Hallegatte, S., 2014. Natural Disasters and Climate Change. Springer International Publishing, Cham. Hallegatte, S., 2012. A Cost Effective Solution to Reduce Disaster Losses in Developing Countries: Hydro- Meteorological Services, Early Warning, and Evacuation. World Bank Policy Research Working Paper No. 6058. Hallegatte, S., 2012. Economics: The rising costs of hurricanes. Nature Climate Change 2, 148–149. doi:10.1038/nclimate1427 Hallegatte, S., 2008. An Adaptive Regional Input-Output Model and its Application to the Assessment of the Economic Cost of Katrina. Risk Analysis 28, 779–799. Hallegatte, S., Bangalore, M., Vogt-Schilb, A., 2015. Socio-economic resilience to floods in 90 countries. Hallegatte, S., Dumas, P., 2009. Can natural disasters have positive consequences? Investigating the role of embodied technical change. Ecological Economics 68, 777–786. Hallegatte, S., Ghil, M., 2008. Natural disasters impacting a macroeconomic model with endogenous dynamics. Ecological Economics 68, 582–592. Hallegatte, S., Henriet, F., Patwardhan, A., Narayanan, K., Ghosh, S., Karmakar, S., Patnaik, U., Abhayankar, A., Pohit, S., Corfee-Morlot, J., 2010. Flood risks, climate change impacts and 26 adaptation benefits in Mumbai: an initial assessment of socio-economic consequences of present and climate change induced flood risks and of possible adaptation options. OECD Publishing. Hallegatte, S., Hourcade, J.C., Dumas, P., 2007. Why economic dynamics matter in assessing climate change damages: illustration on extreme events. Ecological Economics 62, 330–340. Hallegatte, S., Przyluski, V., 2010. The economics of natural disasters: concepts and methods. World Bank Policy Research Working Paper No. 5507. Hanaoka, C., Shigeoka, H., Watanabe, Y., 2015. Do Risk Preferences Change? Evidence from Panel Data Before and After the Great East Japan Earthquake. NBER Working Paper No. 21400. Hasan, I., Tucci, C.L., 2010. The innovation–economic growth nexus: Global evidence. Research Policy 39, 1264–1276. doi:10.1016/j.respol.2010.07.005 Heaton, J., Lucas, D., 2000. Portfolio Choice in the Presence of Background Risk. The Economic Journal 110, 1–26. Helgeson, J., Dietz, S., Hochrainer, S., 2013. Vulnerability to weather disasters: the choice of coping strategies in rural Uganda. Ecology and Society 18. Henriet, F., Hallegatte, S., Tabourier, L., 2011. Firm-network characteristics and economic robustness to natural disasters. Journal of Economic Dynamics and Control. Hochrainer, S., 2009. Assessing the Macroeconomic Impacts of Natural Disasters: Are there any? World Bank Policy Research Working Paper Series, Vol. Hoddinott, J., 2006. Shocks and their consequences across and within households in Rural Zimbabwe. The Journal of Development Studies 42, 301–321. doi:10.1080/00220380500405501 Hsiang, S.M., Jina, A.S., 2014. The Causal Effect of Environmental Catastrophe on Long-Run Economic Growth: Evidence From 6,700 Cyclones (Working Paper No. 20352). National Bureau of Economic Research. Jaramillo, C., 2010. Do natural disasters have long-term effects on growth? Jensen, R., 2000. Agricultural Volatility and Investments in Children. American Economic Review 90, 399–404. doi:10.1257/aer.90.2.399 Kijewski-Correa, T., Taflanidis, A.A., 2011. The Haitian housing dilemma: can sustainability and hazard- resilience be achieved? Bull Earthquake Eng 10, 765–771. doi:10.1007/s10518-011-9330-y Krishna, A., 2006. Pathways out of and into poverty in 36 villages of Andhra Pradesh, India. World Development, Part Special Issue (pp. 324–404). Corruption and Development: Analysis and Measurement 34, 271–288. doi:10.1016/j.worlddev.2005.08.003 Kroll, C.A., Landis, J.D., Shen, Q., Stryker, S., 1991. Economic impacts of the Loma Prieta earthquake: A focus on small businesses. Kunreuther, H., 1978. Disaster insurance protection: Public policy lessons. John Wiley & Sons. Laffont, J.J., 1995. Regulation, moral hazard and insurance of environmental risks. Journal of Public Economics 58, 319–336. Lall, S.V., Deichmann, U., 2012. Density and disasters: economics of urban hazard risk. The World Bank Research Observer 27, 74–105. LeBel, P., 2008. The role of creative innovation in economic growth: Some international comparisons. Journal of Asian Economics 19, 334–347. doi:10.1016/j.asieco.2008.04.005 Li, J.-Z., Li, S., Wang, W.-Z., Rao, L.-L., Liu, H., 2011. Are people always more risk averse after disasters? Surveys after a heavy snow-hit and a major earthquake in China in 2008. Appl. Cognit. Psychol. 25, 104–111. doi:10.1002/acp.1648 Liu, E.M., 2012. Time to Change What to Sow: Risk Preferences and Technology Adoption Decisions of Cotton Farmers in China. Review of Economics and Statistics 95, 1386–1403. doi:10.1162/REST_a_00295 27 Lusk, J., Coble, K., 2008. Risk aversion in the presence of background risk: Evidence from an economic experiment, in: Risk Aversion in Experiments, Research in Experimental Economics. Emerald Group Publishing Limited, pp. 315–340. Maccini, S., Yang, D., 2009. Under the Weather: Health, Schooling, and Economic Consequences of Early- Life Rainfall. American Economic Review 99, 1006–26. doi:10.1257/aer.99.3.1006 Malmendier, U., Nagel, S., 2011. Depression Babies: Do Macroeconomic Experiences Affect Risk Taking?*. The Quarterly Journal of Economics 126, 373–416. doi:10.1093/qje/qjq004 Mannakkara, S., Wilkinson, S., Potangaroa, R., 2014. Build back better: implementation in Victorian bushfire reconstruction. Disasters 38, 267–290. doi:10.1111/disa.12041 Mechler, R., 2004. Natural Disaster Risk Management and Financing Disaster Losses in Developing Countries. Verlag Versicherungswirtsch. Mechler, R., Mochizuki, J., Hochrainer-Stigler, S., 2015. Disaster risk management and fiscal policy. From assessing fiscal risk to building resilience: narratives, tools and evidences. Forthcoming in World Bank Policy Research Working Papers. Mechler, R., Weichselgartner, J., 2003. Disaster Loss Financing in Germany—The Case of the Elbe River Floods 2002. Michel-Kerjan, E., 2008. Disasters and public policy: Can market lessons help address government failures, in: National Tax Association. Miles, S., Brechwald, D., Davidson, R., Demeter, K., Johnston, D., Pampanin, S., Wilkinson, S., 2014. Building Back Better - Case study of the 2010-2011 Canterbury, New Zealand Earthquake Sequence (A Learning from Earthquakes Report). Munich RE, 2015. Review of natural catastrophes in 2014: Lower losses from weather extremes and earthquakes. Nadiri, M.I., 1993. Innovations and Technological Spillovers (NBER Working Paper No. 4423). National Bureau of Economic Research, Inc. Neumayer, E., Barthel, F., 2011. Normalizing economic loss from natural disasters: A global analysis. Global Environmental Change 21, 13–24. doi:10.1016/j.gloenvcha.2010.10.004 Nordhaus, W.D., 2010. The Economics of Hurricanes and Implications of Global Warming (SSRN Scholarly Paper No. ID 1618685). Social Science Research Network, Rochester, NY. Noy, I., Nualsri, A., 2007. What do exogenous shocks tell us about growth theories? Working Papers, Santa Cruz Center for International Economics. Noy, I., Patel, P., 2014. Floods and spillovers: Households after the 2011 great flood in Thailand (Working Paper Series No. 3609). Victoria University of Wellington, School of Economics and Finance. Noy, I., Vu, T.B., 2010. The economics of natural disasters in a developing country: The case of Vietnam. Journal of Asian Economics 21, 345–354. ODI, GFDRR, 2015. Unlocking the triple dividend of resilience - why investing in DRM pays off [WWW Document]. URL http://www.odi.org/tripledividend (accessed 5.21.15). Okuyama, Y., 2003. Economics of Natural Disasters: A Critical Review. Page, L., Savage, D.A., Torgler, B., 2014. Variation in risk seeking behaviour following large losses: A natural experiment. European Economic Review 71, 121–131. doi:10.1016/j.euroecorev.2014.04.009 Patankar, A., Patwardhan, A., 2014. Estimating the Uninsured Losses due to Extreme Weather Events and Implications for Informal Sector Vulnerability: A Case Study of Mumbai, India. Picot, G., Manser, M., Zhengxi, L., 1998. The role of self-employment in ob creation in Canada and the United States. Pielke, R., Gratz, J., Landsea, C., Collins, D., Saunders, M., Musulin, R., 2008. Normalized Hurricane Damage in the United States: 1900–2005. Natural Hazards Review 9, 29–42. doi:10.1061/(ASCE)1527-6988(2008)9:1(29) 28 Raddatz, C., 2009. The wrath of God: macroeconomic costs of natural disasters. Ranger, N., Hallegatte, S., Bhattacharya, S., Bachu, M., Priya, S., Dhore, K., Rafique, F., Mathur, P., Naville, N., Henriet, F., others, 2011. An assessment of the potential impact of climate change on flood risk in Mumbai. Climatic change 104, 139–167. Reynaud, A., Nguyen, M.-H., 2012. Monetary Valuation of Flood Insurance in Vietnam. Reynolds, P., 1999. Creative Destruction: Source or Symptom of Economic Growth?’, in: Entrepreneurship, Small and Medium-Sized Enterprises and the Macroeconomy. Cambridge University Press, Cambridge, UK, pp. 97–136. Romer, P.M., 1990. Endogenous Technological Change. Journal of Political Economy 98, S71–S102. Rosenzweig, M.R., Stark, O., 1989. Consumption Smoothing, Migration, and Marriage: Evidence from Rural India. Journal of Political Economy 97, 905. doi:10.1086/261633 Rudowitz, R., Rowland, D., Shartzer, A., 2006. Health Care In New Orleans Before And After Hurricane Katrina. Health Aff 25, w393–w406. doi:10.1377/hlthaff.25.w393 Said, F., Afzal, U., Turner, G., 2014. Attitudes Towards Risk in the Wake of a Rare Event: Evidence from Pakistan (CREB Working paper No. 2-2014). Centre for Research in Economics and Business, The Lahore School of Economics. Schmidt, S., Kemfert, C., Höppe, P., 2009. Tropical cyclone losses in the USA and the impact of climate change — A trend analysis based on data from a new approach to adjusting storm losses. Environmental Impact Assessment Review 29, 359–369. doi:10.1016/j.eiar.2009.03.003 Şener, S., Sarıdoğan, E., 2011. The Effects Of Science-Technology-Innovation On Competitiveness And Economic Growth. Procedia - Social and Behavioral Sciences, The Proceedings of 7th International Strategic Management Conference 24, 815–828. doi:10.1016/j.sbspro.2011.09.127 Shaw, K.L., 1996. An Empirical Analysis of Risk Aversion and Income Growth. Journal of Labor Economics 14, 626–653. Skidmore, M., Toya, H., 2002. Do natural disasters promote long-run growth? Economic Inquiry 40, 664– 687. Smith, S.K., Mccarty, C., 2009. Fleeing The Storm(s): An Examination of Evacuation Behavior During Florida’s 2004 Hurricane Season. Demography 46, 127–145. Stewart, F., Fitzgerald, E., 2001. War and Underdevelopment. Oxford University Press, Oxford. Strobl, E., 2010. The Economic Growth Impact of Hurricanes: Evidence from U.S. Coastal Counties. Review of Economics and Statistics 93, 575–589. doi:10.1162/REST_a_00082 Tella, R.D., Galiant, S., Schargrodsky, E., 2007. The Formation of Beliefs: Evidence from the Allocation of Land Titles to Squatters. The Quarterly Journal of Economics 122, 209–241. doi:10.1162/qjec.122.1.209 Thaler, R., Sunstein, C., 2009. Nudge: Improving Decisions about Health, Wealth, and Loss Aversion. Penguin: London. The Economic Times, 2011. Honda to cut production at Indian arm by half. Tierney, K.J., 1997. Business Impacts of the Northridge Earthquake. Journal of Contingencies and Crisis Management 5, 87–97. doi:10.1111/1468-5973.00040 Tversky, A., Kahneman, D., 1974. Judgment under uncertainty: Heuristics and biases. science 185, 1124– 1131. UNISDR, 2011. 2011 Global Assessment Report. van den Berg, M., Fort, R., Burger, K., 2009. Natural Hazards And Risk Aversion: Experimental Evidence From Latin America. Presented at the International Association of Agricultural Economists Conference, Beijing, China. Van Oudheusden, P., Klapper, L., Demirguc-Kunt, A., Singer, D., 2015. The Global Findex Database 2014 : measuring financial inclusion around the world (No. WPS7255). The World Bank. 29 van Praag, C.M., Versloot, P.H., 2007. What is the value of entrepreneurship? A review of recent research. Small Bus Econ 29, 351–382. doi:10.1007/s11187-007-9074-x Voors, M.J., Nillesen, E.E.M., Verwimp, P., Bulte, E.H., Lensink, R., Van Soest, D.P., 2012. Violent Conflict and Behavior: A Field Experiment in Burundi. American Economic Review 102, 941–64. doi:10.1257/aer.102.2.941 Vorhies, F., Wilkinson, E. 2015. Co-benefits of disaster risk management. Forthcoming in World Bank Policy Research Working Papers. Weber, E.U., Johnson, E.J., 2011. Psychology and Behavioral Economics Lessons for the Design of a Green Growth Strategy: White Paper for World Bank Green Growth Knowledge Platform. Wennekers, S., Thurik, R., 1999. Linking Entrepreneurship and Economic Growth. Small Business Economics 13, 27–56. doi:10.1023/A:1008063200484 West, C.T., Lenze, D.G., 1994. Modeling the regional impact of natural disaster and recovery: a general framework and an application to Hurricane Andrew. International regional science review 17, 121–150. Wong, P.K., Ho, Y.P., Autio, E., 2005. Entrepreneurship, Innovation and Economic Growth: Evidence from GEM data. Small Bus Econ 24, 335–350. doi:10.1007/s11187-005-2000-1 World Bank, 2015a. Another Nargis Strikes Everyday: Post-Nargis Social Impacts Monitoring Five Years On. World Bank, 2015b. World Development Indicators, International tourism, receipts. World Bank, 2014. Philippines: A Lesson that Helped Save Lives. World Bank, 2013. Risk and Opportunity: Managing Risk for Development (World Development Report). World Bank, 2008. Reshaping Economic Geography, World Development Report. World Bank, Washington, DC. Yamano, T., Alderman, H., Christiaensen, L., 2005. Child Growth, Shocks, and Food Aid in Rural Ethiopia. Am. J. Agr. Econ. 87, 273–288. doi:10.1111/j.1467-8276.2005.00721.x Yang, C.-H., 2006. Is innovation the story of Taiwan’s economic growth? Journal of Asian Economics 17, 867–878. doi:10.1016/j.asieco.2006.08.007 30