A growing consensus in the scientiﬁc community holds that climate change could be worsening certain natural disasters. The Intergovernmental Panel on Climate Change (IPCC) released a special report in early 2012, which notes that climate change could be altering the frequency, intensity, spatial extent, duration, and/or timing of many weather-related extreme events. Even non-experts are perceiving a trend toward more or worse extreme events: a 2012 poll of US residents found that, by a margin of 2:1, people believe that the weather is getting worse, and a large majority believe that climate change contributed to the severity of several recent natural disasters.
This paper reviews what we know about the economic impacts of natural disasters to inform both the estimation of potential climate damages using integrated assessment models and the potential extent of climate adaptation to extreme events. The paper limits focus to empirical estimates of the economic costs of natural disasters and ﬁndings on the determinants of economic damages and fatalities. The paper then also provides an overview of the handful of empirical papers to date on the likely extent of adaptation in response to changes in extreme events. Given the focus on informing climate scholarship and policy, the paper looks speciﬁcally at hydro meteorological (or weather related) disasters and not geophysical disasters, since conﬁdence in the impact of climate change on hydro meteorological events is greater.
The review is focused on the empirical literature; it does not cover the theoretical literature on the economic impacts of disasters or simulation- and modeling-based studies. The focus of this review is also limited to economic impacts. While natural disasters can have profound social and political impacts, those are not covered here.
Finally, as a further limit to the scope, this review is largely focused on literature published within the past couple of decades, a period during which new data sets and improved understanding of disaster losses has emerged. Recent working papers are included, in addition to peer-reviewed studies. Estimating the full range of economic costs from natural disasters is difﬁcult—both conceptually and practically. Complete and systematic data on disaster impacts are lacking, and most data sets are underestimates of all losses. Best estimates for the average annual cost of natural disasters worldwide between 2000 and 2012 ranges between $94 billion and a little over $130 billion. The work reviewed here suggests negative consequences of disasters, although communities tend to have a lot of resilience, recovering in the short to medium-term from all but the most devastating events. The worst disasters, or multiple disasters close in time, can have very long-term, negative economic consequences. Natural disasters generate many transfers and can have substantial distributional consequences, with some groups suffering devastating losses and others coming out ahead, even if overall impacts are close to neutral. Consequences are less severe in higher-income countries, countries with better institutions, and those with a higher level of education. Risk reduction options are available, but predicting increases in adoption in response to climate change is difﬁcult. The occurrence of a disaster has been shown in some cases to increase investments in reducing risks. In addition, some evidence suggests that areas more prone to hazards invest more in reducing their impacts, providing some limited insight on potential future adaptation. Recent research is attempting to move beyond correlations, particularly by addressing the external of many disaster measures, and more work on this is needed.
The theoretically correct measure of economic impacts from a natural disaster is the change in welfare that occurred as a result of the event. Welfare can be evaluated ex post, as the compensation required to avoid loss, or ex ante, which accounts for uncertainty. Although thinking in terms of hypothetical welfare measures can be instructive, a complete welfare analysis is usually quite difﬁcult empirically and would require making a number of assumptions and simpliﬁcations in the analysis. If society were risk-neutral, ex ante welfare could be evaluated with the expected economic loss. Scholars interested in empirical estimates (as opposed to modeling results, which can be useful in estimating welfare calculations) have attempted to measure observable disaster damages and follow-on economic impacts as a rough approximation of the net economic costs of a disaster.
Various lists and typologies of disaster impacts have been created. Most scholars of disasters have broadly classiﬁed disaster impacts into direct and indirect impacts. Direct impacts refer to the physical destruction from a disaster, and indirect impacts (some authors prefer the term
higher-order impacts) are considered to be the follow-on consequences of that destruction. Note that although it is convenient to speak in the shorthand of losses, costs, or damages from a disaster, in practice, this review like the work it summarizes focuses on the net impact of disasters. It investigates the question of whether and when disasters can have a positive economic impact.
In theory, it should be possible to sum up all direct and indirect losses to generate a measure of the total economic costs of a disaster. Two overarching complications arise when trying to measure the full economic costs in each of the categories. First, it is necessary to be very clear about the spatial and temporal scale being examined because different boundaries for analysis can generate different results.
For example, consider the economic costs of a disaster from the point of view of a homeowner who lost her home. Some direct losses, such as the home, are reimbursed by insurance or aid from government or other groups, and some losses are borne fully by the victims. If the individual receives aid, the economic cost of the disaster to that person will be the value of the lost home minus the amount of the aid. From the perspective of society, however, the aid is just a transfer from one taxpayer to another and thus should not be added or subtracted from the damage.
Temporal boundaries can also matter. As an example, it has been shown that construction sectors can experience a boom right after a disaster as people rebuild. A couple of years afterward, however, they may face a lull because people undertake upgrades during the post-disaster reconstruction that they would have otherwise deferred.
Looking only one year post-disaster may suggest a beneﬁt to the construction sector, but looking over three years might diminish this beneﬁt. And although the construction sector may get a beneﬁt, had the disaster not occurred, the funds spent on rebuilding would have been spent elsewhere in the economy, with a higher utility to the homeowner, thus, post-disaster spending should not simply be counted as a beneﬁt of the disaster.
The economic costs that ﬁrst come to mind when thinking about natural disasters are damages to buildings and contents. Though seemingly straightforward to measure, getting the precise economic costs of this impact is not theoretically trivial. Consider a house that is completely destroyed. The economic loss could be measured as either the market value of the house right before the disaster hit or the replacement cost to rebuild it. The most appropriate measure is the market value at the time of disaster impact (or, for other assets, depreciated value). The replacement cost could be higher or lower for several reasons. Post disaster, some materials may be in short supply and more expensive substitutes used or higher prices charged, for example, or labor may be in short supply and thus wages higher, driving the cost of rebuilding above what it would have been before the disaster. This is often referred to as demand surge. Although these higher costs are a loss to the homeowner, they are a gain to the suppliers and builders. On the ﬂip side, if business interruption is severe and more laborers are looking for temporary work, rebuilding costs could be lower. This again would be a savings to the homeowner that, from the perspective of society as a whole, would offset the loss to the worker.
The homeowner could also receive insurance payouts if insured. This would again lessen the negative wealth shock to the homeowner. Homeowners who suffer capital losses but have insurance are paying for the loss ex ante through the insurance premiums instead of export. The insurance is a mechanism to smooth the loss over time. Insurance payments are often used as a proxy for economic costs, as they should theoretically be closely correlated with the lost value of the homes and structures—at least in areas with high take-up rates of insurance. Further, insurance companies usually keep extremely good records and so can be an excellent data source, but they are not synonymous with total direct costs.
In addition to the cost of the lost home, other direct losses to the homeowner include the time lost to the rebuilding effort, emotional trauma or stress, and loss of nonmarket items of value, such as baby photographs or family keepsakes. These losses are rarely included in disaster damage estimates and obtaining estimates of many would require non-market valuation studies.
Destruction to the buildings, contents, inventory, and other capital of ﬁrms can be similarly analyzed. For destroyed capital, the correct measure of economic loss is the depreciated value of the lost asset. If production is lost from a delay in replacing damaged capital, then this lost production from downtime should also be counted as an economic loss. It is possible the replaced capital could be more productive than the capital destroyed if there has been technological change. This productivity bump will offset some of the economic loss but would presumably also be paid for, as the new capital would cost more than the depreciated value of the lost asset. If the ﬁrm receives disaster aid such that the upgrade is, in a sense, free to the ﬁrm, then it could, in theory, be better off post-disaster. Again, though, from the point of view of society, the aid is simply a transfer.
Infrastructure damage is another category of direct loss from a natural disaster. Again, the depreciated value is the correct measure of economic loss for reasons already discussed. Delays in repair and rebuilding can trigger indirect costs, discussed next, through an interruption in use or service.
Especially in the developing world, loss of life and injury from disasters can be large, and these are direct costs of a disaster. An enormous debate centers on how to value loss of life and injury, and I will not summarize that here, except to note that a value-of-a-statistical life (VSL) estimate based on disaster risk explicitly would be the best measure. To my knowledge, very few, if any, VSL estimates have looked explicitly at natural disaster risk, although one comparative stated preference study ﬁnds that willingness to pay (WTP) to reduce mortality risk is greater for terrorism than for natural disasters and that reducing the mortality risk from natural disasters is valued about the same as that from trafﬁc accidents, even though the latter is a much higher risk. Injury and illness can be measured in quality-adjusted life-years or similar measures.
Disasters can be viewed as a negative capital shock to a region. This has follow-on economic consequences in addition to the value of the lost assets. First, economic losses are not exclusive to ﬁrms or households that sustain direct physical damage. If electricity or water is lost, for instance, it can cause business interruption. Similarly, the loss of such services could lead to a decline in the quality of life for households, and thus a utility loss, and could also lead to the need for costly measures to compensate, although this is rarely discussed in the literature.
Such compensating actions could involve longer travel times due to a road outage or the purchase of battery-powered lighting in response to a loss of electricity, for instance. These are indirect damages to include in estimates of total costs.
Some of the literature has focused on possible multiplier effects post-disaster. Consumer demand post-disaster may be higher for some sectors—such as construction—and lower for others as consumers forgo some expenditures to use their funds for rebuilding.
These types of expenditure changes could have economic multiplier effects within the community (positive or negative). A similar story can be told for business interruption. This could decrease demand for inputs and reduce production, having negative ripple effects in the supply chain. Aid and insurance could mute these impacts if such funds allow for a faster resumption of normal business activity. From the perspective of the whole economy, however, multiplier effects may well be zero, with positive and negative impacts canceling out. For instance, if a ﬁrm fails to produce an output, its customer may simply purchase the good elsewhere. As another example, tourists may avoid a hurricane-stricken coast, but instead of not traveling, they may just frequent another area. Clearly the distributional impacts of a disaster could be quite large and could have signiﬁcant consequences for individuals, ﬁrms, or communities.
If a government does make changes to taxation or resource allocation post-disaster, this could have indirect economic effects as spending in other areas is reduced or taxes increased on certain groups. For hard hit countries, particularly small or poor countries, this is a distinct possibility and would need to be evaluated. Countries can also receive international assistance (which, again, would be a transfer from a global perspective). Case study evidence suggests that donors do not necessarily provide additional aid after a disaster, but simply reallocate aid budgets.
Future Research need
This review has suggested several remaining gaps in the empirical literature that warrant further research. First, as previously discussed, more work is needed that explicitly addresses externality concerns. One approach is the search for possible instruments. Another, and which has been pursued by several papers recently, particularly with hurricanes, is to use physical measures of a disaster, such as wind speed. It has suggested the creation of an index of disaster intensity, but notes that collecting data from primary sources to create such an index for multiple hazards and countries would be a signiﬁcant undertaking. There are some gaps in the literature that may be difﬁcult to ﬁll due to limited data. For instance, little empirical work has assessed the impact of multiple disasters occurring fairly close in time or the cumulative impact of many small events. These questions are hard to tackle with the EM-DAT data and thus may require taking a single-country and single-hazard focus. In addition, few studies have empirically estimated indirect damages from disasters. This is an area in need of much more investigation. Similarly, very little work has evaluated nonmarket impacts of disasters. Finally, more empirical work on the economic impacts of shifts in post-disaster spending, altered risk perceptions, demographic shifts, or political changes, would be intriguing to pursue. Without comprehensive data sets, however, all such work will most likely have to be in the form of disaster-speciﬁc studies and then general ﬁndings drawn by looking across many empirical case analyses.
The empirical work on adaptation to potential changes in extreme events is quite small. More studies like those proﬁled which compare current risk reduction investments for different levels of risk, could help inform the extent of adaptation that is possible. In addition, there is a dearth of studies investigating the extent to which there is an adaptation deﬁcit—that is, are we not even currently adapted to today’s climate everywhere, let alone future climate? More work on the costs and beneﬁts of different adaptation strategies—especially beyond one-off, household-level investments, but including larger community-level changes—would also be a helpful contribution to this emerging literature.
Finally, this review has limited itself to empirical studies of the economic impacts of weather-related disaster events. Parallel reviews of modeling studies, engineering estimates, case studies, and impacts of disasters on socio-political and health outcomes would be useful complements to this work.
Author: Ali Hassan Shabbir
MSc (Hons) Agricultural Economics
Institute of Agricultural and Resource Economics,
University of Agricultural Faisalabad, Pakistan.