5.2. Revealing risk and identifying development trade-offs

In nationally reported disasters in the 21 countries in Africa, Asia and Latin America analysed in Chapter 2, there were 63,667 schools and 4,873 health facilities damaged or destroyed since 1989. During this period, 73,000 kilometres of roads were also damaged, and 3,605 municipal water systems, 4,400 sewer systems and 6,980 power installations were reported damaged and destroyed. Of these total losses, 46 percent of the schools, 54 percent of the health facilities, 80 percent of the roads and more than 90 percent of the water, sewer and power installations were damaged or destroyed in frequently occurring extensive disasters rather than in occasional and intensive catastrophes.⁵ 

These losses are massive, and they indicate how public investments in social and economic development are in practice often investments in risk construction and contingent liability (Cardona, 2009). This large loss of publicly owned assets remains largely invisible and unaccounted for (Gall et al., 2009), and impacts are transferred to affected low-income households and communities. This invisibility represents a major political barrier to investing in DRM. Revealing these impacts will not automatically lead to greater investment, but if governments were to account for these recurrent losses and for their future liabilities, they may begin to make more considered decisions based on assessments of the costs, benefits and trade-offs internalized in public investment.

As explained in Chapter 1, a country’s stock of risk comprises a combination of high-severity, low-frequency intensive risks, and low-severity, high-frequency extensive risks. Normally, neither conventional catastrophe risk models nor risk models based on historical disaster-loss data are able to comprehensively estimate both of these risk strata. Fortunately, national disaster loss reporting (see Box 2.5 in Chapter 2) and the growing availability of open-source probabilistic models, such as CAPRA,⁶  have facilitated the development of innovative hybrid models that can estimate both extensive and intensive risks. One such hybrid model (Box 5.3), which combines historical disaster loss data and probabilistic catastrophe risk modelling, has been piloted in Colombia, Mexico and Nepal (ERN-AL, 2011). By integrating assessments of both extensive and intensive risks, the real scale of recurrent loss and future risk begins to emerge.


Estimates of disaster impacts are normally made only after large events,⁸  meaning that recurrent disaster loss is often unaccounted for. The evidence produced by applying the hybrid model in Colombia, Mexico and Nepal indicates that the scale of recurrent losses may be far higher than most governments realize (Figure 5.3). Estimated average annual disaster losses reach US$2.24 billion in Mexico, US$490 million in Colombia and US$253 million in Nepal.


Analysis based on the hybrid loss exceedance calculation for Colombia shows that the government may have to address losses in publicly owned assets as well as the uninsured private assets of low-income groups, ranging from US$100,000 (100 times per year) to US$1 billion (at least once every 30 years). In Mexico, and excluding impacts from drought and in the agriculture sector, the government is likely to incur weather-related disaster losses of over US$1 million at least 50 times a year, of more than US$15 million at least 10 times a year, more than US$300 million at least once a year, and more than US$1 billion at least once every 6 years. In Nepal, the government is implicitly liable for losses amounting to US$1 million almost 10 times each year, and of US$100 million almost every second year.

This is the real scale of disaster loss in these countries. It gives an idea of the magnitude of the public funds required if a government were to compensate for and replace public assets, and support the recovery of low-income households and communities. Also, it is not just recurrent losses that governments are ill prepared to deal with. With some notable exceptions, governments are rarely adequately prepared, by either contingent financing or insurance, to cover the probable maximum losses from a low-probability intensive event. Taken by surprise by liabilities that they have never assessed, governments are then forced to rely on slow and often unreliable international assistance for recovery and reconstruction.

To put these losses into a political perspective, Figure 5.4 shows the value of nationally recorded losses for five successive presidential periods in Mexico from 1982 to 2009. All Mexican governments since 1982 have had to absorb disaster losses of over US$10 billion during their period in power, rising to almost US$20 billion in the new millennium. This is the scale of loss any incoming government is likely to have to deal with unless serious investments are made in DRM.



From an economic perspective, the losses are significant in all three countries studied. In Colombia, for example, as Figure 5.5 shows, the estimated annual loss from disasters represents approximately 1 percent of GDP. Although this is less than the cost of cyclical unemployment, disaster losses are higher than the cost of 5 percent inflation, and are comparable to the cost of armed conflict, which was estimated at 1.1 percent of GDP for the period of 1991 to 1996. Furthermore, the maximum probable disaster loss with return periods of 500 and 1,000 years, represent costs of 2.3 percent and 2.9 percent of GDP, respectively, equivalent to the losses caused by the financial crises of the 1980s and 1990s (Moreno and Cardona, 2011).



These figures indicate that if decision-making were based on a realistic assessment of the social and economic costs and benefits, DRM should have a similar public policy importance as controlling inflation or resolving armed conflict. In other words, a larger share of the national budget should be allocated to reducing disaster risks. Making these costs visible is also a key step towards identifying the trade-offs in DRM investment.

These figures on disaster loss do not include the cost of indirect disaster impacts documented in the 2009 Global Assessment Report (de la Fuente and Dercon, 2008; UNISDR, 2009), for example increasing poverty and declines in human development. Moreover, disasters reduce the savings level in society and thus the amount of capital and product per person. As a result, recurrent disasters, even though they may be small-scale, affect per capita income rates in the long term (IDEA, 2005; Cavallo et al., 2010; Moreno and Cardona, 2011).



Notes

Indonesia, Maldives, Sri Lanka and Thailand.