The ENHANCE project participated in the 2016 Understanding Risk Forum, held in Venice, Italy, from 16 to 20 May 2016.
Below is a summary of the ENHANCE session at the Forum. The summary was originally published in the UR2016 proceedings.
Lorenzo Carrera, World Bank Group
Jaroslav Mysiak, Fondazione Eni Enrico Mattei; Euro-Mediterranean Centre on Climate Change
Elco Koks, Institute for Environmental Studies, VU University Amsterdam
Accounting for and Predicting the Economic Impacts of Natural Hazards
Natural hazard risks, including weather- and climate-related extreme events, are able to undo sizable development and poverty reduction efforts, upset financial and economic stability and growth, and devastate communities and individual lives. The 2015 Global Assessment Report (UNISDR 2015a) valued the global annual average losses from natural hazards as topping $300 billion, more than any previous estimate. But even this value does not account for the whole magnitude of tangible and intangible damage and losses.
The Sendai Framework for Disaster Risk Reduction (UNISDR 2015b) has made substantial reduction of disaster losses a top priority of international efforts. To assess the progress toward this end, the global disaster risk reduction (DRR) community will have to fill the gaps in the loss data records and substantially improve the practice of damage and loss assessment. Moreover, the DRR community will need to link up with the climate change community in order to value the economic impacts of climate change and the costs of extreme weather and climate events. There is ample scope for the two groups to learn from one another—and to advance knowledge beneficial to both.
The Sendai Framework represents a commitment to a transformative change in how natural and human-made risks are dealt with (van der Vegt et al. 2015; Wahlström 2015). Because disaster accountancy was largely neglected in the past, it is not an easy task, or sometimes even possible, to portray the spatial and temporal patterns of disaster damage and losses with reasonable precision. For years, the United Nations Office for Disaster Risk Reduction and the international community have worked to fill in the knowledge gaps and to promote a culture of evidence- and knowledge-based DRR. But as we try to compensate for past negligence, we should not waste the opportunity to collect information and knowledge on the full economic costs of disasters, including their ripple and spillover effects on the increasingly interconnected economies.
Understanding the Potential Impacts of Natural Hazards
There are multiple approaches to estimating the distributions of natural hazard economic risks. Statistical approaches look at the past records of loss data, and estimate risk from historical loss data using extreme value theory. A fundamental challenge is how to model the rare phenomena that lie outside the range of any available observation, and cannot be accounted for in extreme value theory methods.
Catastrophe models – computer-based representations that estimate the potential damage of disasters (Grossi and Kunreuther 2005) – are able to perform extreme value analysis. This is usually done by overlaying the properties or assets at risk (the exposure module, such as classification based on a land cover data set) on the potential sources of natural hazards (hazard module) in a specific geographical area. A vulnerability module estimates the damage (e.g., of a hurricane) that occurs based on a function of the hazard intensity (e.g., wind speed), the environmental conditions (e.g., the region’s terrain), and the exposed value characteristics (e.g., the structural types).
Because of their outputs – the potential damage to the stock of assets – catastrophe models are mainly used in the insurance industry. Over the last three decades (since the late 1980s), catastrophe models have been quite effective in contributing to the shift from reactive catastrophe reinsurance pricing to technically informed pricing. This shift has led to a more resilient catastrophe reinsurance industry: one based in scientific and technical knowledge, and more affordable to buyers. In turn, this change has allowed the development of new financial instruments for disaster risk management and climate change adaptation, including capital market products and international risk pooling. Although in the past catastrophe models have focused on a limited set of perils, such as hurricanes, earthquakes, and extreme precipitation, more and more applications are being developed for other perils, such as drought, terrorism, and pandemics, and for areas of the world that have been neglected in the past.
Assessing Wider Economic Impacts
On the other hand, the estimation of wider economic impacts of extreme weather and climate events has been less exploited by disaster risk management practitioners than damage estimates. Typically, models such as input-output (IO), computable general equilibrium (CGE), social accounting matrix (SAM), and econometric models are able to provide the impacts of extreme weather and climate events on the economic flows—for example on the production of economic sectors and the regional or national gross domestic product (GDP). Although these models have advanced over time, their effective applicability to real-world cases has been constrained by a number of factors, including their intrinsic level of uncertainty (arising from the number of assumptions) and the difficulty in modeling the complex dynamics of a system in the aftermath of a disaster.
Standard IO models are relatively simple, static, and linear models imitating the interrelationships between economic branches within a national or regional accounting system. CGE models are nonlinear models of circular flows of goods and services between agents, where representative households and firms choose their demand and supply following constrained optimization problems, taking prices as given. Prices are determined by market equilibrium conditions, allowing substitution effects and more realistic behavioral content and working of both factor and product markets compared to IO models (Rose 2004). Econometric models, based on time-series data, have the advantages of being statistically rigorous and possessing forecasting capabilities, but they can provide only estimates of the total impacts. Thus they are often unsuitable for a detailed analysis of the specific losses of a disaster.
Models of this type may also be used to inform policy making in some areas of disaster risk management, such as flood risk management (Koks and Thissen 2014) and water resources management under scarcity conditions (Distefano and Kelly 2016).
The assessment of the impacts of climate change on human welfare are generally performed by using integrated assessment models (IAMs), such as GCAM (GCAM 2012). IAMs are mathematical computer models that integrate both social and economic components with biogeochemical cycles to assess the resultant effect of greenhouse gas emissions. Economic losses are mainly determined by a damage function that relates temperature and precipitation variations to the economic effects across different hypothetical futures, also called Shared Socioeconomic Pathways (O’Neill et al. 2015). Compared to disaster risk models, IAMs are generally used to assess the effects of slow-onset changes, such as temperature increase, with few experiences on catastrophic risk (Bosello, De Cian, and Ferranna 2015; Pindyck and Wang 2013).
Old Issues and New Challenges: Modeling under Changing Conditions
Many studies have already highlighted the increase in value of global annual average losses from natural hazards (see for example Munich Re 2014). Evidence shows that, in the future, the increasing exposure and vulnerability of assets, economic activities, and population to natural hazards will continue contributing to the increase of global losses. Moreover, climate change will further exacerbate this trend. Unfortunately, the capacity of economic and financial risk models to reproduce systems’ dynamics is still limited. High levels of uncertainty, particularly under changing conditions, still characterize the outcomes of the models. The recent critique of integrated assessment models (Pindyck 2013; Stern 2013)— echoed by the Intergovernmental Panel on Climate Change Fifth Assessment Report (IPCC 2014)— voices a growing frustration with contemporary models reckoned too simple and arbitrary. Contemporary economic risk analysis and assessment practices could face comparable critiques. It remains the case that disaster risk assessments are rarely dynamic exercises and cannot represent the overall and interrelated systems’ reaction to and recovery from a disaster, the multifaceted conditions of fast-growing economies, or changes in land use and the environment.
Over the lasts decades, catastrophe models have made substantial improvements in their capacity to assess the physical damage of extreme weather and climate events. This success has contributed to the development of a number of financial instruments targeting disaster risk. Despite this accomplishment, our understanding of the full economic cost of disasters is still limited. Economic risk models can help to fill this gap, but if real-world policies and investments are to be based on them, they need to be more robust and reliable. Moreover, there is a need for tools that are affordable, credible, transparent, and open access, as well as tailored to specific perils and scales of analysis (regional to municipal). Recent experiences of model coupling have demonstrated the capacity of economic risk models to provide outputs at local scale. For example, Carrera et al. (2015) coupled spatial analysis and regionally calibrated CGE models to assess climate change effects on flood risk at regional level in Italy. For the same country, Pérez-Blanco et al. (2016) coupled a revealed preference model calibrated at local level and a regional CGE model to inform water resources management policies under drought conditions. Other models (CGE) have also been developed at municipal scale, for example to assess flood risk in São Paulo, Brazil (Haddad and Teixeira 2015). Recent efforts to add realistic behavioral features, through evolutionary methods such as agent-based modeling (Safarzyńska, Brouwer, and Hofkes 2013), network analysis, and supply chain principles (Rose et al. 2016), hold promise for improving models’ capacities to capture systems’ response in the aftermath of a disaster.
Better evidence of economic losses and ex post analysis of disasters’ effects are needed to improve models’ robustness, through calibration and verification. This need becomes more and more pressing under global change conditions. The ongoing efforts of the Directorate General for European Civil Protection and Humanitarian Aid Operations (ECHO) to promote a standardized European disaster loss database are steps in the right direction. With better disaster loss data, more improvements will surely come in the near future.
Bridging the gap between the disaster risk management and climate change adaptation communities will be key to improving our ability to assess and estimate the economic effects of disasters and climate change—with a specific focus on the scope and scale of analysis and consideration of the complexity of dynamic and interconnected systems.
Download the full UR2016 Proceedings.
Laura Bonzanigo, Climate Change Group, World Bank Group
Lorenzo Carrera, World Bank Group
Grant Cavanaugh, Nephila Capital Ltd.
Tiziano Distefano, Polytechnic of Turin
Jaroslav Mysiak, Fondazione Eni Enrico Mattei; EuroMediterranean Centre on Climate Change
C. Dionisio Pérez-Blanco, Fondazione Eni Enrico Mattei; Euro-Mediterranean Centre on Climate Change
David Simmons, Willis Re Analytics