by Hal Cochrane
This article presents research conducted to date on the development of models to forecast the economic impacts of earthquakes. Comments and questions should be directed to Hal Cochrane, Colorado State University, at (970) 491-6493; email: firstname.lastname@example.org.
The economic repercussions of earthquakes are often misunderstood and misrepresented. Much of what we think we know about post-disaster recovery is derived from case studies and computer simulations, both of which produce conclusions which should be viewed with a healthy dose of skepticism.
This brief summary begins by defining both real and financial indirect economic losses. Several important generalizations about regional economies are employed to explain why post-disaster recovery is a product of a unique combination of outside assistance and the preexisting state of the economy. The paper then assesses the indirect losses from a hypothetical large Midwestern earthquake, examines financial ripples and systemic losses, and evaluates claims about the vulnerability of the national insurance and financial markets.
Direct vs. Indirect Loss
Indirect loss is any loss other than direct loss. This seems tautological, however, it is important to begin with the obvious. Direct loss is a loss linked directly to an earthquake. It includes all damages, plus employment losses due directly to the closure of the damaged facility. Indirect losses/gains are anything else. Direct losses can produce dislocations in factories or commercial ventures (real ripple effects), banking, which causes uncertainty and a subsequent tightening of overall credit conditions (bank systemic risk), and insurance, thereby limiting coverage and triggering subsequent economic dislocations (insurance systemic risk).
Real ripple effects stem from supply shortages or sudden reductions in demand. The relative size of the ripple hinges on the geographic area of focus, and the time frame under scrutiny. Losses can be displaced both geographically and temporally. As will be shown in the following discussion, ripples must be negative for the nation, but can, with the receipt of outside assistance, prove positive for the impacted region.
Generalizations About Regional and National Indirect Losses
This simple presentation draws upon sound economic principles to portray the full range of possible outcomes in the post-disaster period. All losses, both direct and indirect, are scaled from 0 to 1, the scaling representing the percentage of the activity that is lost.
Figure 1 maps all combinations of direct and indirect loss for a region. The upper part of the diagram shows a ray from zero reflecting the income gains which occur as an unconstrained region rebuilds (100 percent financed by outside sources, i.e., insurance and federal aid). Economists will recognize this uppermost part of the envelope as simply the amount of reconstruction spending times the income multiplier for construction. At the extreme, sufficient alternate supplies and markets void any bottlenecks. This is the best the impacted region can hope for. As more of the burden of financing reconstruction is shifted to the region's victims, the net positive effect of spending diminishes, and the ray rotates toward zero indirect gain. At the extreme, when all costs are borne internally, the gains from rebuilding are offset by reduced spending later on as households pay off disaster-related debt. This type of analysis is based on traditional input-output techniques and is all too familiar to regional economists.
The lower half of the diagram is not as simple to understand, at least initially. It reflects an economy with bottlenecks; the closer to zero, the less constraining these bottlenecks are. As in the explanation about indirect gains, the extremes are readily interpreted. Line segment A-C traces the outer edge of the loss envelope. Point B, the uppermost (where upper means the most negative) level of indirect loss, results from a maximum shock to the smallest sector, when no means of mitigating supply and demand shortages present themselves. At this point, the economy implodes to the level of output dictated by the constraining sector. Indirect loss can be a multiple of direct loss at B. Such an event is most likely when critical lifelines, such as power and water are lost. One might think of B as the terrorist's target of choice; the smallest amount of direct damage produces the greatest amount of economic disruption.
Point C on the indirect loss frontier shows no indirect loss when direct loss is total (100 percent). If all is lost then no forward or backward linked losses are possible. Line segment D-B shows the influence of (1) an increased variance in the pattern of loss (zero variance at D and maximum at B); and (2) reduced flexibility in the region's ability to mitigate shortages.
The line segment D-C shows the effect of a uniform damage pattern, which causes the economy to shrink proportionately. This occurs when all sectors are damaged proportionately, in which case forward and backwardly linked losses disappear. In this case, the economy remains balanced regardless of the amount of damage observed.
From a national accounting stance, indirect losses can be measured by deriving regional indirect impacts, adjusted for the liability the federal government incurs in providing disaster relief, and for offsetting increases in outputs elsewhere.
The positive effects outside aid produces for the region are to some degree negated by the fiscal drag produced by debt financed federal relief. Figure 2 shows how the generalized loss envelope changes to reflect a national accounting stance. First, direct damage and subsequent indirect loss is transmitted to other regions via altered trading patterns (imports and exports). If the region produces critical materials for which ready substitutes are unavailable, then local indirect losses can spill over to affect the national economy. Hence, it is possible for indirect loss to exceed that sustained by the region alone. Second, the national economy is impacted in that external aid has to be financed. As a result, indirect gains disappear; the costs incurred outside the region negate local gains. So from the vantage point of the nation as a whole indirect loss must exist. Indirect gains are an illusion, a byproduct of focusing only on the stricken region.
New Madrid Earthquake and the Memphis Economy
The following simple example is provided as an illustration of how constraints amplify indirect losses. Suppose that the Memphis economy is fully constrained, and there are no means for circumventing forward or backward linked losses. No excess capacity exists within the disaster stricken area, and trade with other regions cannot provide the needed supplies or markets. Something like this might result from the loss of high voltage transmissions towers, or a vital natural gas pipeline. Assume that only the region's sources of power, its lifelines, are damaged (30 percent destroyed). In such a situation, the output of transportation services, the sector which includes pipeline services, declines by 30 percent.
Table 1 demonstrates a solution to the problem based on the NIBS HAZUS indirect loss model. Since no adjustments were permitted in this example, the direct loss of lifeline services causes all sector outputs to decline by 30 percent. Indirect loss in such a situation is several times greater than direct loss.
Table 1: Losses Due to the Effect of a Pipeline Shock
In this simple example of a highly constrained economy, the 30 percent shock to transportation, produces 4.12, 3.76, and .77 percent change in direct output, income and employment, respectively. Because of the constraints assumed, indirect income and output losses are approximately six times the direct loss. Indirect employment losses are 35 times the direct employment effects.
Wider Regional and National Economic Consequences
The wider regional effects are easily modeled by tracing sector disruptions to import and export changes. National impacts are derived by reducing government spending by the amount of aid provided to the disaster stricken area. Table 2 provides an illustrative computation based on a hypothetical $50 billion earthquake, where rebuilding is fully financed by outside sources. The results show that rebuilding stimulates not only the immediately impacted region but surrounding areas (through imports) as well. So, overall the wider region enjoys a $22.99 billion gain. However, the implied reconstruction subsidy leaves the federal government with a budget shortfall which is assumed to be resolved by cutting spending elsewhere. As a result, incomes around the nation fall by more than the region gains, resulting in an overall reduction in income of $2.57 billion.
Table 2: Distribution of Losses and Gains
|Impacted Areas||Income Loss (-)
or Gain (+)
|Other regions||8.67||Change in imports due to reconstruction|
|Impacted region||14.32||Net effect of bottleneck effects and rebuilding expenditures. All outside assistance assumed.|
|Regional net gains||22.99||The sum effect on the local impacted area and the surrounding region.|
|National net gains||-2.57||The net effect on the region less the negative effect on the national due to a reduction in government spending elsewhere.|
So, what may appear to be a windfall gain for the region proves to be a wash or, in this case, an overall loss for the nation.
Dynamics of Loss and Gain
The solutions just presented are static, that is, they reflect an adjustment process which is presumed to last a fixed period of time, typically a year. Such a portrayal, although useful for some events, may produce misleading results for others, particularly for sizable events such as the Kobe earthquake, where dislocations are prolonged and the burden of reconstuction falls primarily on the victim. In such instances, time plays an important role in determining the ultimate costs of an event. The severity of bottlenecks vary over the recovery period, and the financial burden indebtedness imposes on survivors lingers long past the event.
Much has been written about the fragility of the insurance industry and the potential for large systemic losses resulting from a sudden reduction in surplus. The industry's position regarding financial ripple effects is that a catastrophic earthquake will deplete the insurance industry's surplus. Surviving firms will be forced to ration coverage by: (1) restricting availability and (2) raising premiums. Without coverage, firms not suffering direct damage will be forced to shut down and substantial national indirect economic loss will result.
The evidence available thus far does not support the contention that surplus shocks produce general rationing and escalating premiums (aside from region specific adjustments motivated by factors other than supply considerations). Arguments that disasters, such as Andrew and Northridge, would drain industry reserves, forcing them to curtail other product lines, simply have not materialized. Premiums have not changed appreciably for property and casualty coverage in general (as opposed to earthquake and wind). It appears that supply restrictions are restricted to those risks which the industry feels are uninsurable. Their other lines do not appear to have suffered.
"Despite the massive losses caused by the California quake, rates for most lines of coverage remained flat throughout the year. Only some accounts with coastal properties or earthquake exposures experience significantly higher rates accompanied by shrinking capacity." (Business Insurance, December 26, 1994, p. 4).
It is highly likely that even if gaps in coverage were to emerge, they may be filled by alternate financial arrangements. Additional capacity offered by a maturing foreign insurance industry could help fill the void. So too, a more integrated and concentrated domestic financial sector is likely to weather surplus shocks more efficiently and smoothly than the current fragmented system. What is clear here is that predictions about the fallout from future earthquakes should not be grounded in outdated institutional arrangements.
From the little that can be gleaned from the literature, it is not at all clear that post-disaster financial repercussions will induce measurable real indirect loss. The insurance industry's vulnerability to the most recent string of catastrophes is debatable. The links between insurance and macroeconomic performance have not been credibly demonstrated. Lastly, financial institutions are in the throws of a revolution, one which is likely to cause us to rethink what we thought we already knew.
A differing set of pre-disaster conditions, along with a decentralized program of disaster relief without the involvement of the insurance industry, would have produced a dramatically different set of outcomes, where the region's economy might have deteriorated rather than turned sharply upward. The most important conclusion that can be drawn from this work on indirect loss is that it is difficult to generalize from any single event; real economic dislocations are a product of a complex set of forces. However, if forced to stand on a statement, it would be that the most important aspect of recovery is outside assistance. This is true for both the private and public sectors. Scenarios which others have painted regarding the national economic consequences of a string of catastrophic disasters should be viewed with a dose of skepticism. Financial markets are unlikely to collapse, regional bank failures will not produce contagion effects, and it seems unlikely that the economy will sustain sizable indirect losses if insurance has to be rationed. The word if is underscored in that recent events have not produced rationing. Lastly, trends in the financial sector to agglomerate and offer a full range of services is likely to invalidate much of what we thought we knew about the insurance industry and its response to natural disasters.
Boisvert, R., (1992), "Indirect Economic Consequences of a Catastrophic Earthquake," Chapter 7, Direct and Indirect Economic Losses from Lifeline Damage, Development Technologies Inc. under FEMA Contract EMW-90-3598.
Brookshire, D., and McKee, M., (1992), "Indirect Economic Consequences of a Catastrophic Earthquake," Chapter 8, Indirect Loss Measurement and Computable Equilibrium Model, Development Technologies Inc. under FEMA Contract EMW-90-3598.
Development Technologies, Inc., (1992), "Indirect Economic Consequences of a Catastrophic Earthquake," Development Technologies Inc. under FEMA Contract EMW-90-3598.
Doherty, N., Kleffner, A.E. and Kunreuther, H., (1991), "The Impact of a Catastrophic Earthquake on Insurance Markets," Philadelphia, Pennsylvania: University of Pennsylvania, Wharton School, Center for Risk and Decision Process.
Ellson, R.W., Milliman, J.W. and Roberts, R.B., (1984), "Measuring the Regional Economic Effects of Earthquakes and Earthquake Predictions," Journal of Regional Science, 24 (4):559-79.
Friesma, P., Caporaso, J., Goldstein, G., Lineberry, R., and McClearly, R., (1979), "Aftermath: Communities and Natural Disasters." Beverly Hills: Sage Publications.
National Research Council, (1992), "The Economic Consequences of a Catastrophic Earthquake," Washington, D. C.: National Academy Press.
Rose A., Benavides, J., Chang, S., Szczesniak, P. and Lim, D., (1995), "The Regional Economic Impact of an Earthquake: Direct and Indirect Effects of Electricity Lifeline Disruptions," 42nd North American Meeting of the Regional Science Association Meeting International, Cincinnati, November 10.
Roberts, R. Blaine, Milliman, Jerome W. and Ellson, Richard W., (1982), "Earthquakes and Earthquake Predictions: Simulating Their Economic Effects," Technical Report prepared for National Science Foundation under Grant PRF 80-19826.
Wright, James D. et al., (1979), "After the Clean-up: Long-Range Effects of Natural Disasters," Beverly Hills: Sage Publications.
Yezer, A. M., (1990), "Measuring the Effects of Catastrophic Earthquakes on Different Regions of the Country," Proceedings of a Forum on the Economic Consequences of a Catastrophic Earthquake, Washington, D. C.: National Academy Press.
Click here to return to the Table of Contents
| Home | NCEER |
Information Assistance | Publications |
Databases, Software |
| Events | New! | Links | Data Resources |