By Kelly Klima and Ismael Arciniegas Rueda
“The magnitude, frequency, location, duration, speed of onset, and other characteristics of many hazards can now be predicted and forecasted with more accuracy and further in advance than before,” observed geographer Ian Burton. The knowledge of people and infrastructure has likewise improved, allowing better understanding of damages and subsequent cost estimations. These improvements may lead to higher accuracy and precision, helping to speed disbursements to pay for repairing or replacing infrastructure. Yet there are other factors to consider that influence the accuracy of cost estimates before, during, and after a disaster. For example, what happens when there is money available from Federal Emergency Management Agency? When many different groups need that money, how should those limited funds be distributed?
Cost Estimation Uncertainties
Estimating the cost of disasters is a complex process. Consider, for example, a school superintendent seeking to renovate a school in a non-disaster setting. The superintendent might solicit cost estimates from contractors, who would provide estimates based on their knowledge.
Let’s say the school’s renovation is necessary because of flood damage, though. There are now other factors to consider, such as how quickly the repairs are needed (e.g., is school currently in session), if contractors will be too busy to take on the project, if government assistance is available to fund repairs, or if the infrastructure needs to be upgraded to meet new codes and standards,
Next, consider how disaster recovery is complicated even further during this new era of COVID-19, when extra precautions must be taken. For example, hotels (where workers might stay) may have to clean more often. Or, some workers might not be willing to stay in a hotel, thus driving down the supply of labor and increasing costs. In the field, more personal protective equipment, sanitizing agents, and other items would be required to protect against COVID-19, and supply shortages of these items could increase cost. Furthermore, local governments are under unusual stresses due to lack of tax revenue, which may delay payments.
Uncertainties also exist within the realm of repairs related to electric utilities. A utility might be faced with the tradeoff of salvaging material and equipment from one place to reuse in another as a temporary repair. In addition, some items are rare, difficult to transport, and require special expertise to install. For example, extremely high voltage transformers can weigh 500 tons or more, requiring special cars, helicopters, or rail to transport. Given all of these and other types of uncertainty, it is inevitable that even with excellent knowledge of damages, cost estimators still lack sufficient data to estimate costs.
Using Data to Improve Cost Estimate Accuracy
How can we estimate costs quickly and reliably, thus improving the flow of dollars back into the community? The answer is data, and the basis of a reliable cost estimation is collecting and cleaning the data to develop and accurate and accessible database.
Some data are available via proprietary databases. For example, when actual costs are not available, FEMA analysts typically use a product called RSMeans, which includes cost estimation books for everything from nuts and bolts to wind turbines. Other databases, such as Arizona State University’s Spatial Hazards Events and Losses Database for the United States (SHELDUS), provide information about specific types of damages from natural hazards dating from 1960 to present.
Other organizations are making some disaster-related data publicly available, from both government and private industry, which could be used to improve estimates. For example, the U.S. Bureau of Labor Statistics has databases that provide labor-related data to help estimate costs. Facebook has created Disaster Maps, which “share real-time information with response teams, helping them determine things like whether communities have access to power and cellular networks, if they have evacuated, and what services and supplies they need most.” From an electricity perspective, the National Oceanic and Atmospheric Administration’s Nighttime Lights makes it possible to determine changes in ground lighting, which can serve as a proxy for electricity usage before and after an event.
New Trends and Possibilities
As the amount of data increases, the potential value of analytics such as linear regressions also increases. This is especially true for new data-intensive methods such as artificial intelligence or machine learning, which can help identify patterns from previous disasters to increase cost estimation accuracy
Disaster managers have enough challenges to address during a disaster without adding the estimation of costs to repair or replace infrastructure to the list. A great deal of relevant data, stemming from years of federal cost estimating for the military, air force, and domestic programs, are available. An ever-increasing quantity of data are being generated, together with the development of technologies and cyberinfrastructure to process, publish, and share data. Cost estimation should embrace these two trends as an opportunity for improving knowledge on disaster cost estimation in an equitable fashion.
Acknowledgement: This work was supported by the RAND Corporation’s Homeland Security Operational Analysis Center (HSOAC) within the Recovery Cost Analysis Program.
Kelly Klima serves as associate program director for RAND’s Acquisition and Development Program for the Homeland Security Operational Analysis Center. Klima has more than ten years of experience in quantitative and qualitative decision analysis for risk reduction. Her research supports community resilience for extreme heat and flooding and has been applied in locations such as New York City and the City of Pittsburgh. She holds a certified floodplain manager designation from the Association of State Floodplain Managers and a certified cost estimator/analyst certification from the International Cost Estimating and Analysis Association.