The following paper was presented at the WEAO 2020 Collection Systems Committee Fall Webinar on October 28, 2020. The presentation made is included after the paper.
National Guideline Development for Benefit-Cost Analysis of Storm Drainage Infrastructure
Robert J. Muir, M.A.Sc., P.Eng., Dillon Consulting Limited, Fabian Papa*, M.A.Sc., M.B.A., P.Eng., FP&P HydraTek Inc.
*FP&P HydraTek Inc., 216 Chrislea
Road, Suite 204
Vaughan, Ontario L4L 8S5
Application of Benefit-Cost Analysis in Water Resources
2. An alternative method is to determine the damage caused by three or four recent floods whose hydrologic frequency can be determined and a smooth damage frequency curve plotted through these points; however, for most floodplains, changes in land use with calendar time prevent direct usage of a damage-frequency relationship from historical damages; and
3. The third method entails hydrologically determining various flood elevations for specific flood frequencies and deducing synthetically the damages that would occur given these flood events. This analysis provides a synthetic damage-frequency curve from which one can estimate average annual damages for a given study area.”
i) the loss estimate for sewer back-up in bottom-up assessments was often overestimated considering historical losses;
ii) bottom-up loss estimates using synthetic depth-damage curves were not calibrated to actual incurred losses from insurance industry data or disaster relief funding;
iii) the quantitative assessment of climate change impacts was not assessed (with the exception of sea level rise impacts);
iv) some losses were double-counted (both top-down and bottom-up);
v) the integration of event damages to derive annualized losses was not consistently calculated;
vi) there was inconsistent consideration of overall losses beyond insured losses;
vii) operation and maintenance costs were not evaluated in most assessments;
viii) present value discounting of costs and benefits was not considered in most assessments;
ix) effects of growth on future damages and potential benefits was not considered in any assessments; and
x) financing costs were not considered in any assessments.
In addition, avoided losses were assumed to be equal to total losses which is a reasonable approximation for large riverine mitigation projects but less reasonable for urban flood mitigation projects where infrastructure upgrades are not likely to be 100% effective given other causes. Later sections present losses based on the statistical analysis of recorded losses that may be used to derive top-down estimates used to assess adaptation funding needs at a national, regional or local scale, and be used to calibrate bottom-up estimates derived with synthetic depth-damage curves.
Estimation of Potential Benefits and Adaptation Funding (Total Cost)
Estimation of Adaptation Costs
Potential Adaptation Benefits (Avoided Damages) Based on Reported Losses
FIGURE 1. MUNICH RE INSURED AND OVERALL LOSSES IN CANADA 1980 – 2017 (Munich Re, 2018)
TABLE 1. MUNICH RE CANADIAN INSURED HYDROLOGICAL AND METEOROLOGICAL LOSSES, 2017 CAD – GUMBEL EXTREME VALUE DISTRIBUTION AND EXPECTED ANNUAL DAMAGE (EAD)
TABLE 2. MUNICH RE CANADIAN OVERALL HYDROLOGICAL AND METEOROLOGICAL LOSSES, 2017 CAD – GUMBEL EXTREME VALUE DISTRIBUTION AND EXPECTED ANNUAL DAMAGE (EAD)
The regional distribution of losses including those associated with particular flood event types can be assessed using the Catastrophe Indices and Quantification Inc. (CatIQ) loss database. CatIQ delivers detailed analytical and meteorological information on Canadian natural and man-made catastrophes on a subscription basis to serve the needs of the insurance / reinsurance industries, public sector and other stakeholders. National EAD for urban flood events was derived based on insured loss and loss expenses obtained through CatIQ. This event type categorization is unique from CatIQ’s broad Flood and Water peril classifications and is intended to represent those events with a significant proportion of sewer back-up/water damage within reported personal property damages. It is noted that events may be characterized by multiple peril types, including Hail, Windstorm, Winterstorm, and Fire; some water damages may occur primarily as a result of other perils not related to extreme rainfall conditions (e.g., due to power interruption disrupting sump pump operation, or wind damage to rooftops allowing water entry).
Based on the foregoing, some Flood and Water peril events may therefore be characterized by very limited water damage, and would represent damages that cannot be mitigated through storm infrastructure upgrades intended to address extreme rainfall conditions. Those events are excluded for the purpose of urban flood event analysis, and are identified as those with a minimum of 30% sewer back-up/water in overall personal property damages. As sewer back-up/water losses have been discretized only since 2013 in the CatIQ database, the time series of losses represents only part of the overall CatIQ dataset that begins in 2008, specifically 2013-2018. After filtering the data as discussed above, the derived damage-frequency values and EAD are summarized in Table 3.
TABLE 3. CATIQ CANADIAN URBAN FLOOD LOSSES, CAD – GUMBEL EXTREME VALUE DISTRIBUTION AND EXPECTED ANNUAL DAMAGE (EAD)
Sewer back-up/water losses represent a component of urban
flood event losses. On average these losses represent 70% of total event losses
for the urban flood events. The
component of total losses beyond sewer back-up/water include other personal
property, personal non-property, commercial and automobile line of business losses.
Table 4 summarizes derived damage-frequency values and EAD in Canada and in
provinces where CatIQ data is available.
The average sewer back-up loss is shown to vary across provinces and may
be used to calibrate losses derived from bottom-up damage assessments that
apply NRCan/PSC Method #3. For example,
where EAD is calculated by evaluating event losses for a range of return period
events in Method #3, the average EAD across many properties within a study area
or region should be comparable to the Event Average Sewer Back-up Loss in Table 4.
TABLE 4. CATIQ CANADIAN SEWER BACK-UP/WATER LOSSES, CAD – GUMBEL EXTREME VALUE DISTRIBUTION AND EXPECTED ANNUAL DAMAGE (EAD)
a. EADMROL = Overall Losses $1.347B ($2017 CDN)
b. EADMRIL = Insured Losses $0.695B ($2017 CDN)
ii) CatIQ Urban Flood Events (2013-2018)
a. EADCIUF = Loss and Loss Expenses $0.821B (CDN)
iii) CatIQ Sewer Back-Up/Water (2013-2018)
a. EADCISB = Loss and Loss Expenses $0.376B (CDN)
Assessment of Regional and Local Flood Damages
TABLE 5. REGIONAL ECONOMIC, DEMOGRAPHIC, HOUSING AND INFRASTRUCTURE INDICATORS
A case study in the Results section illustrates how the above provincial losses may be scaled to a municipality or project area to support the calibration or verification of bottom-up damage estimates based on NRCan/PSC’s Method #3.
Potential Adaptation Benefits (Avoided Damages) Based on Synthetic Depth-Damage Curves
The Need for Calibration of Depth-Damage Curves
Where more than a relative assessment of damages is required for prioritization, calibration of depth-damage curves is recommended such that cumulative losses broadly reflect recorded insured losses and estimated overall losses. Research in jurisdictions worldwide has shown that damages estimated using synthetic depth-damage curves may not represent actual reported losses without calibration. An evaluation of loss methods in Australia noted “most of the synthetic methodologies prepared for Australia are not calibrated with empirical loss data or express the magnitude of damage in absolute monetary values” (Nafari, 2018). Nafari evaluated two depth-damage curve methods (Geoscience Australia (GA) Depth-Damage Function and FEMA/USACE Depth-Damage Function (USACE)) and has shown that observed losses in Australia may be overestimated by a factor of almost 100%, considering a February 2012 flood event. A calibrated depth-damage function, called FLFArs, predicted damages within the 95% confidence interval of the reported losses. A case study in Denmark (Olsen et. al, 2015) concluded that insurance data can be used to calibrate inundation modelling even though estimating individual damages is a challenge:
“…it is shown that with the present data we can establish clear relationships between occurrences of claims and hazard maps on a basis of integrated hazard simulation and vulnerability assessment. This suggests that insurance data can be valuable for calibrating inundation modelling in terms of frequency and location of flooding, even when acknowledging that it is difficult to accurately identify the flooded properties, in particular for the low hazard category. The estimation of damage costs for individual claims remains a challenge in this study. Our results suggest that the main variation in per claim costs can, perhaps, be better described using socioeconomic variables in models of the value at risk in different households, rather than simple rainfall event variables.”
The following case studies illustrate i) the application of top-down Method #2 damage estimates at a national and municipal scale to inform national and local adaptation program funding, and ii) the calibration of bottom-up Method #3 damages for an urban infrastructure upgrade project using historical provincial sewer back-up/water damage data.
National Flood Adaptation Funding
National investments in disaster mitigation should be sufficient to address expected damages through cost-effective projects. This case study evaluates Canada-wide adaptation funding requirements considering historical expected damage values, estimated benefit/damage ratios that consider the cost effectiveness of mitigation projects, and estimated benefit-cost ratios considering recent DMAF project funding applications and international experience.
Damages to be addressed were presented in the Methodology section, including EADMROL and EADMRIL based on Munich RE Hydrological and Meteorological Events data and EADCIUF based on CatIQ Urban Flood Events data. While the expected benefit-cost ratios vary from project-to-project in an adaptation program, the overall ratio can be estimated considering a review of international and national programs. The ECONADPAT program reported a benefit-cost ratio of 4.1:1 for “hard flood control” (i.e., traditional grey infrastructure) based on a wide survey of international programs (Kuik et al., 2016). A survey of Canadian DMAF projects that required a minimum ratio of 2:1 revealed ratios from 5.5:1 to 17:1. The FCM and IBC suggested that a ratio of 6:1 could be achieved in a national adaptation program. For this case study, the potential benefit-cost ratio is assumed to be 6:1, while the actual benefit-cost is 4:1, considering that mitigation measures may be only partially effective.
Table 6 summarizes damages, benefits and adaptation budget values based on various EAD values and the assumed benefit-cost ratio. The service life of adaptation infrastructure over which benefits are achieved is assumed to be 100 years. Total adaptation funding and annual funding over a 20-year investment period is shown.
TABLE 6. NATIONAL DAMAGES AND ADAPTATION FUNDING
Infrastructure Canada has identified 59 eligible DMAF-eligible project costs (Infrastructure Canada, 2020) totaling $4.0B, with program spending over 10 years, or $0.4B/year on average. Based on expected damages and benefits, annual funding of $0.87-1.7B, about two to four times DMAF annual project investments, could be justified over twenty years based on expected benefits of hard flood control projects.
City-Wide and Project-Scale Flood Control Program Funding
Municipalities in Canada have pursued system-wide evaluation of risks and investment in adaptation upgrades to address recurring flood damages. The City of Markham’s Flood Control Program represents a system-wide flood risk reduction strategy that focuses on storm sewer upgrades, but that also includes culvert upgrades and floodplain reclamation / on-line storage (land purchase and use reassignment). The program cost is estimated to be up to $368M (City of Markham, 2019).
City-wide benefits from infrastructure upgrades can be estimated by scaling national damages to Ontario using the Table 5 scaling percentage (38%), and to Markham based on the ratio of Ontario and Markham populations of 2.45% (Muir and Papa, 2019) as follows:
i) EADMROL-Ontario = EADMROL × 0.38 = $512 M
ii) EADMROL-Markham = EADMROL × 0.0254 = $12.5 M
Benefits in the Markham study and project implementation areas can be scaled from city-wide values based on many approaches depending on available data, including insurance risk ratings obtained through the city’s insurer, hydrodynamic computer model surcharge risks (Xu and Muir, 2018), or based on city-records of reported flooding. The fractions of the West Thornhill study area and the Phase 1 project area flood reports to city-wide reports are 39% and 6.1%, respectively.
TABLE 7. CITY-WIDE, STUDY AREA AND PROJECT-LEVEL DAMAGES AND EFFECTIVE BENEFIT-COST RATIO FOR SEWER UPGRADES
TABLE 8. MUNICIPAL SEWER BACK-UP DAMAGE ESTIMATES
Applying the estimated average loss of $43,000 results in an EAD of $15,700,400 which increases the numerator in the project benefit-cost ratio by 230%. Of course, the benefits (numerator) should be tempered somewhat as the effectiveness of the investment is likely to be less than 100% efficient such that all basement flooding risk is completely eliminated, but rather that the risk profile is changed for the better. As the municipality also considered estimated losses of up to $100,000, reported losses and potential benefits are considerably lower than estimated values. While average event values are $18,500, more extreme events have been shown to have a higher average loss compared to smaller, more frequent events.
Reported flood damage losses have been analyzed and can support the estimation of overall adaptation funding requirements (i.e., mitigation infrastructure investment) across Canada. Approved DMAF disaster mitigation projects as of January 2020 have an estimated total cost of $4.0B. To mitigate national expected annual damages over the next 100 years, adaptation funding for hard flood control measures of between $17B and $34B is warranted based on an expected benefit/cost of 4:1 for such projects (ignoring any discounting effects). Current DMAF funding of $2B over a 10 year period of 2018-2028 could therefore be expanded to support more extensive adaptation efforts provided that eligible projects with suitable benefit-cost ratios are available. With an estimated value of wastewater and stormwater infrastructure in Canada of $368B, an investment of $17-34B represents 5-9% of current asset value. As most assets were installed over a long period covering the past 70 years, upgrades to infrastructure can be expected to occur over decades. The case study above demonstrates how Method #2 EAD estimates can be effectively applied, leveraging reported historical flood damages, both insured and overall to guide decision making on adaptation funding.
Investment in adaptation infrastructure includes large regional works including diversions and dams and also extensive municipal infrastructure upgrades. The level of funding for city-wide flood control programs can be reviewed considering expected benefits and the effective benefit-cost ratio at various scales including city-wide, high risk study areas, and individual project areas. The City of Markham Flood Control Program has been shown to have a potential benefit-cost ratio of up to 3.4:1, with higher risk areas exhibiting even higher ratios. The case study above demonstrates a top-down approach leveraging Method #2 EAD estimates that are scaled down to provincial, municipal, study and project area levels using suitable proxies. Such scaling is readily and reliably calculated using economic, demographic, development and infrastructure statistics, as well as records data characterizing localized flood risks (e.g., municipal flood reports or hydraulic system performance results).
The average cost of a flood incident is an important statistic in BCA of adaptation projects. Analyses that consider property-scale damages have often used high-level estimates of costs that do not consider local records that have previously not been available in sufficient detail or duration. Analysis of CatIQ datasets have provided provincial-scale average damage estimates that may be used to calibrate bottom-up Method #3 analysis damage estimates. In Ontario, the average CatIQ sewer back-up claim per event is approximately $18,500. This value provides a lower damage estimate than higher values that have been regularly applied in DMAF funding applications across Canada. To promote greater consistency in the estimation of flood damages and potential benefits, reported losses should be considered. This would support a more reliable prioritization of projects based on relative benefits and would support more reliable budget screening based on absolute benefit-cost ratios that can be achieved.
City of Markham (2018) 2018 First Quarter Investment Performance Review - Markham
City of Markham (2019) Flood Control Program and Stormwater Fee Update.
EPCOR (2019) Stormwater Integrated Resource Plan – Capital and Operational Plan Alternatives (May 2019 Utility Committee report) https://www.epcor.com/products-services/drainage/flood-mitigation/Documents/EPCOR_SIRP_May2019_Report.pdf
Formetta, G. (2019) Empirical evidence of declining global vulnerability to climate-related hazards, Global Environmental Change, Vol. 57
IBI Group (2015) Provincial Flood Damage Assessment Study
IBI Group and Golder Associates (2017) Report, Flood Mitigation Options Assessment.
Infrastructure Canada (2018) Disaster Mitigation and Adaptation Fund - Applicant’s Guide, Strengthening the Resilience of Canadian Communities,
Insurance Bureau of Canada and Federation of Canadian Municipalities (2019) Investing in Canada’s Future: The Cost of Climate Adaptation
Insurance Bureau of Canada (2019) New report shows urgent need for climate adaptation investment
Intact Centre on Climate Adaptation (2019)Weathering the Storm: Developing a Canadian Standard for Flood-Resilient Existing Communities.
Kuik, O. et al. (2016) Assessing the economic case for adaptation to extreme events at different scales
Klotzbach, P.J., Bowen, S.G., Pielke Jr., R., and Bell, M. (2018) Continental U.S. Hurricane Landfall Frequency and Associated Damage Observations and Future Risks, American Meteorological Society, Articles July, 2018,
National Resources Canada (NRCan), and
Public Safety Canada (PSC) (2017). Canadian Guidelines and Database of Flood
Vulnerability Functions, Draft.
Office of the Parliamentary Budget Officer (2016) Estimate of the Average Annual Cost for Disaster Financial Assistance Arrangements due to Weather Events
Olesen, L., Löwe, R., and Arnbjerg-Nielsen, K (2017) The Flood Damage Assessment, Literature review and recommended procedure
Olsen, A. S., Zhou, Q., Linde, J.J, Arnbjerg-Nielsen K (2015) Comparing Methods of Calculating Expected Annual Damage in Urban Pluvial Flood Risk Assessments
Muir, R. and Papa, F. (2019) An Economic Analysis of Green v. Grey Infrastructure, WEAO 2019 Technical Conference
Nafari, R. H. (2018) Flood Damage Assessment in Urban Areas
Sakshi, S. (2019) Risk and Return on Investment Tool (RROIT)
Toronto and Region Conservation Authority (2019) Toronto Flood Risk Ranking
Xu, L and Muir, R. (2018) Wastewater Collection System Performance Under Climate Change – Safety Factors and Stress Tests for Flood Risk Mitigation, WEAO 2018 Technical Conference.