When Congress authorized the National Flood Insurance Program (NFIP) in 1968, it intended for the program to encourage community initiatives in flood risk management, charge insurance premiums for new construction that was based on the flood risk at the property, and encourage the purchase of flood insurance by owners of flood-prone properties (that is, seek a high takeup rate), in part by offering affordable premiums. The NFIP has been reauthorized many times since 1968, most recently with the Biggert-Waters Flood Insurance Reform Act of 2012 (BW 2012). In this most recent reauthorization, Congress placed a particular emphasis on setting flood insurance premiums following actuarial pricing principles, which was motivated by a desire to ensure that future revenues were adequate to pay claims and administrative expenses (NRC, 2015a).
BW 2012 would have increased premiums for policyholders who had previously been paying less than NFIP risk-based premiums and possibly would increase premiums for all policyholders. Subsequently, congressional concern for the effect of that legislation on the affordability of flood insurance led to the Homeowner Flood Insurance Affordability Act (HFIAA 2014), modifying some provisions of BW 2012. HFIAA 2014 (Section 9) further emphasized that the Federal Emergency Management Agency (FEMA) report to Congress with a plan for an “affordability framework” following its submission of the affordability study that was originally required by Section 100236 of BW 2012.
The legislative language that called for an affordability study directed FEMA to seek advice from the National Academy of Sciences (NAS). In
response, NAS convened the Committee on the Affordability of National Flood Insurance Program Premiums that was directed by its task statement to prepare two reports. The first report (Report 1) was released in March 2015. It included chapters on the history of the NFIP pricing practices, the demand for flood insurance, considerations for design of an assistance program for persons who might be cost burdened by rising premiums, and policy options that could make premiums less expensive for all policyholders (NRC, 2015a). The summary of Report 1 is included as Appendix A.
This report (Report 2) proposes an analytical approach FEMA might use to evaluate affordability policy options such as those described in Report 1. In preparing Report 2, the committee’s work was informed by lessons learned from a proof-of-concept pilot study completed in North Carolina specifically for this committee’s work. The state of North Carolina has extensive data on floodplain properties and it has extensive experience in conducting analyses, using models, of flood risk management options. That proof-of-concept report1 undertaken by the North Carolina Floodplain Mapping Program (NCFMP) is an independently written, companion document to the committee’s report (see Box 1-1, Chapter 1, for the complete statement of task for Report 2).
Chapter 2 of this report describes model development for evaluating affordability policy options and their application to the NFIP. The analytical requirements to evaluate options led the committee to specifically consider microsimulation techniques to support a structured approach to assess NFIP policy options. How such analyses may be conducted is illustrated with examples of model output from the North Carolina proof-of-concept report. Chapter 3 discusses the data available to the NFIP and from other sources for conducting such analyses. Further, it describes ways to fill data gaps.
FEMA is directed to propose an affordability framework to Congress 18 months after completing the affordability study.2 The affordability framework was to include actions that advanced the original goals for the NFIP, which were to ensure reasonable insurance premiums for all, base all premiums on risk, secure widespread participation, and earn premium and fee income that covers claims and expenses. Ideally, FEMA would formulate affordability policy alternatives for consideration, conduct an evaluation of the alternatives, and propose a preferred affordability strategy. Policy analysis capacity and necessary data, however, currently are not available to complete a comprehensive analysis of affordability policy options. None-
1 The NCFMP (2015) report is publicly available at http://dels.nas.edu/resources/static-assets/wstb/miscellaneous/wstb-cp.pdf.
2 Section 100236 of BW 2012 states “methods to aid individuals to afford risk-based premiums under the NFIP through targeted assistance rather than generally subsidized rates, including means-tested vouchers.”
theless, FEMA can complete some limited analyses in the near term as the agency builds its analytical modeling capacity and database through time. Chapter 4 suggests such initial analyses.
In the process of preparing its second report, the committee had the opportunity to reflect on Report 1 findings and, as a result, develop additional findings for FEMA to consider as it prepares its affordability framework for Congress. Chapter 4 includes these additional findings. This report summary includes select findings from Chapters 2, 3, and 4. These are findings the committee believed were of the highest immediate priority. A complete list of all report findings are presented at the end of each chapter. All the committee findings are shown in bold text and reflect chapter number and sequence of findings in the respective chapter.
BUILDING MODELING CAPABILITY
A structured analytical process will provide answers to policy questions. For example, if BW 2012 reforms are in effect, will premiums exceed the ability of owners to pay for insurance for flood-prone properties? And what are the policy alternatives that can make premiums affordable for those who have limited ability to pay?
The need to answer such questions directs attention to development of models and data for estimating the effects of BW 2012 (the baseline condition) and then estimating how affordability policy options alter those effects of BW 2012. Microsimulation techniques often are well suited to making these estimates. Microsimulation models have two essential elements: (1) a microdatabase and (2) a computer program. To answer the policy questions implied in Section 100236 of BW 2012 the database would include information about each NFIP policyholder (or a sample of such policyholders), about their property characteristics, and about their policy. Ideally, the database also would include information about property owners and their properties located in flood-prone areas that do not purchase an NFIP policy, but might in the future. To estimate future flood damage to specific properties the database would require information that characterizes the likelihood of floods of different magnitudes and property-specific flood loss estimates based on the first-floor elevation. With these data available, the model’s computer program would simulate NFIP premium-setting practices and estimate premiums paid under both the “baseline condition” policy of BW 2012 and for any alternative affordability policy options that reduce flood insurance premiums (Report 1, Chapter 7).
If there are options that provide financial assistance to property owners who would be unable to pay the premium (Report 1, Chapter 6) then the program would have to simulate the assistance program’s rules that determine whether the property owner is eligible for assistance and estimate how
much assistance the property owner might receive. Other effects of interest might include, but would not be limited to, expected future insurance claims (especially if the assistance includes mitigation) and whether the policyholder would no longer purchase a policy at a higher premium (Report 1, Chapter 4). Making calculations for individual property owners and aggregating the results identifies effects on NFIP net revenues, on federal budget expenditures, and on the flood insurance takeup rate across all property owners and subgroups of interest (for example, low-income households).
Microsimulation is an attractive modeling approach because it focuses first on the individual policyholder and property owner. It can also aggregate results across policyholders to produce national estimates and estimates for categories of interest (i.e., income groups or geographic areas). This focus on the policyholder could address the concerns of those who will find premiums unaffordable, who might receive aid, and who might choose to purchase flood insurance. Microsimulation models, however, are necessarily complex to reflect the complexities of government programs and individual circumstances. As a result, their construction requires substantial time and resources. There are professionally recognized practices FEMA can rely upon if FEMA develops a microsimulation model. Among these accepted practices are building self-contained modules that can be readily added to or removed from a more comprehensive model. This in effect builds capacity incrementally as new and better data become available. The pace at which the modeling capacity grows will be determined by the resources available, access to appropriate expertise, and the support of agency leadership.
Finding 2.1: FEMA’s capability to evaluate affordability policy options is very limited but can be substantially advanced by embracing a microsimulation modeling approach and building the model incrementally through time. This would begin with conceptual microsimulation model design and the writing of computational algorithms for the self-contained modules as necessary data are identified and data gaps filled.
INFORMATION FOR MICROSIMULATION MODEL IMPLEMENTATION
The microsimulation approach requires the construction of one or more microlevel databases. Data records for each property could include data on variables that characterize property features, socioeconomic characteristics of the property owner and occupant (if different from the owner), and the NFIP policy (if there is a policy in force on that property). The committee reviewed the policy data records in the October 2013 NFIP policy database. The database does include some of the necessary data, but there are incomplete records. A particularly important gap in the data
was the absence of structure first-floor elevation data that are necessary for estimating the damage to the structure from floods of different magnitudes. While some of those data are now being collected for properties inside the special flood hazard area (SFHA), such data are not available and are not being collected for properties outside the SFHA. Also, even if all of the data in the NFIP database were complete and accurate, the database cannot be used to simulate affordability assistance programs that are means tested because the database does not contain income, wealth, or housing cost data. Furthermore, the NFIP database does not contain information for nonpolicyholders located in flood-prone areas and cannot be used to analyze whether an alternative policy option that would reduce premiums or provide assistance might promote takeup among such households.
Some microsimulation model analyses will be for flood claims and perhaps needed to simulate the effects on premiums from new alternative policy options. Making these estimates will depend on information about the likelihood that different floods reach different stages in different areas of the floodplain and the claims resulting from these floods. Only some flood insurance rate maps (FIRMs) include such needed information. Other data sources might be used to replace or supplement the data provided through the FIRMs such as the risk and damage assessment computer software tools from the U.S. Army Corps of Engineers and the tools found in FEMAs Hazus model.
Finally, response functions that can be used to simulate the behavioral response of property owners to changes in policy will be needed to improve the accuracy of estimates. Two examples of response functions include how takeup rates will vary with premiums charged and how premium levels and offers of assistance might affect the demand for flood mitigation. The professional literature provides only a limited basis for developing such response functions. Sensitivity analyses could be used to assess the uncertainty in modeling behavioral responses.
The committee reviewed options for filling data gaps from existing sources for use in microsimulation modeling. These included securing data on socioeconomic characteristics of property owners from the decennial census of population and the continuing American Community Survey (ACS) or administrative records from the Internal Revenue Service (IRS) or the Social Security Administration (SSA). Decennial census and ACS data are of limited usefulness, and although administrative records from IRS or SSA could provide useful data, it might be difficult (due to time constraints) to obtain access to and begin using such data in the near term. Property characteristics might be obtained from commercial enterprises that now collect data at the individual property level and perform their own analyses of home prices from tax assessor records. Some of these sources hold promise for securing necessary data, but none offered readily accessible data.
Certain data gaps could also be filled with data from a new sample survey conducted on behalf of FEMA specifically for the NFIP. Depending on the interviewing mode (personal interviews could be desirable for a FEMA survey because of the ability to capture information by observation of the property) and the extent of follow-up needed to bring response rates up to acceptable levels, however, a survey could be very costly. The necessary number of completed survey cases will be a function of the extent of disaggregation of microsimulation model results that is desired (greater disaggregation requires a larger sample for precision) and the budget the agency can allocate. It is unlikely that new survey data could be obtained in the near term.
Finding 3.3. Information available from the NFIP policy database and from FIRMs are missing data critical to a comprehensive analysis of affordability policy options. Numerous other sources of information, including new survey data collection, could be used to conduct microsimulation policy option analyses. Although the data for a national affordability study initially will be limited, numerous opportunities for database improvement for answering NFIP policy questions can be secured as budget resources permit.
FEMA’s current modeling capability and the data available cannot support building a microsimulation model for a comprehensive affordability analysis. Nonetheless, there are analyses FEMA can undertake in the near term. For example, some assistance program design questions require nonquantitative analysis. For example, who might administer an assistance program? Answers to such a question will affect the formulation of alternative policy options. Also, based on a conceptual-level argument and an understanding of each possible alternative, some might be initially removed from consideration (maybe to be reintroduced at a later date). For example, the discussion in Report 1 on disaster savings accounts, tax credits and deductions, and capping the NFIP responsibility to pay claims in high-loss years might be put aside if the alternative policy option is expected to have little applicability to low-income property owners.
After this kind of initial screening, some policy alternatives will remain candidates for inclusion in an affordability framework and could be subject to quantitative analysis. With this need in mind, FEMA could begin the conceptual development of modules that can answer some of the important policy questions. As one example, a question that can be answered now is how much premiums would increase if policyholders who were paying pre-FIRM subsidized (PFS) rates had to pay NFIP risk-based rates. This is the
baseline estimation needed to begin an analysis of affordability policy options. Answering this question requires a module to replicate the premium-setting practices of the NFIP to include rating tables, coverage selection, zone, and property characteristics including first-floor elevation. FEMA could use the data now being collected on first-floor elevations to impute first-floor elevations on PFS structures for which elevation data are not yet available. FEMA might be provided with the necessary resources and study schedule flexibility that will allow for a data development process to fill in critical, but missing, socioeconomic data on policyholders and property owners in specific geographic areas in North Carolina. This would allow FEMA to build on the proof-of-concept study to provide an evaluation of a number of affordability policy options, recognizing that the North Carolina results would be state specific.
Finding 4.1. Some decision-relevant analyses can be completed with currently available analytical tools and data, or with limited investments in methods and database development. In the process of doing such analyses, FEMA also will make progress toward building analytical capacity to conduct more comprehensive policy analyses in the future.
COMMITTEE REFLECTIONS AFTER REPORT 1
The committee was responsible for preparing two reports. The task for the first report was to describe concepts of affordability, assistance program design decisions, and policy options that may reduce the cost of premiums for those who were cost burdened by premium increases called for by BW 2012. To address its task, Report 1 was organized by chapters on pricing (Chapters 2 and 3), insurance demand (Chapter 4), location of affordability issues (Chapter 5), defining cost burden as ability to pay and assistance program design decisions (Chapter 6), and affordability policy options (Chapter 7). As a result of its work preparing Report 2, the committee developed additional findings regarding NFIP pricing after BW 2012, defining cost burden and ability to pay, and linking mitigation and premium assistance. In the process of preparing Report 2, the committee had the opportunity to reflect on its first report. The select new findings that the committee wishes to highlight are discussed below and other new findings can be found in Chapter 4.
NFIP PREMIUMS AFTER BW 2012 AND HFIAA 2014
Chapters 2 and 3 in Report 1 discussed NFIP rates and rate setting and the changes called for by BW 2012 and HFIAA 2014. With specific reference to rates and BW 2012, Congress instructed the NFIP to move toward
flood insurance premiums that better reflected the full risks of flooding at a given location, following actuarial pricing principles. In Report 2, the committee developed two additional findings regarding the effectiveness of BW 2012 in promoting actuarial-based pricing for the NFIP.
Grandfathered polices are allowed to maintain a lower flood insurance premium if a new FIRM moves the property into a higher flood-risk zone or identifies a new base flood elevation (BFE). BW 2012 eliminated grandfathering, but it was reinstated in HFIAA 2014. Report 1 found that the NFIP sought to compensate for the forgone revenues from grandfathering through a cross subsidy from other policyholders. This cross-subsidy violates the actuarial principle that each policyholder pays rates commensurate with its flood risk. Specifically, for policies that are grandfathered, premiums will be too low, and for those who pay the cross subsidy, premiums will be too high. In the future, and in the context of climate change, land development, and improved flood mapping, some properties will be mapped into SFHAs when they are not currently located or will have higher estimated BFEs. The owners of those properties will have the opportunity to pay grandfathered rates under HFIAA 2014, and the NFIP practice of increasing rates for all policyholders to account for revenue loss from grandfathering may continue.
Finding 4.2. HFIAA 2014’s reinstatement of grandfathering, which will perpetuate cross-subsidies in the NFIP, will result in the program increasingly violating actuarial pricing principles if flood risks increase in the future.
The NFIP divides the floodplain into the SFHA and the area beyond the SFHA. Within the SFHA, PFS rates, even after HFIAA 2014, will be phased out and replaced with NFIP risk-based rates. This means that about 20 percent of all policyholders will pay a rate that is more compatible with actuarial pricing principles. As noted, grandfathering was to be eliminated, but HFIAA 2014 reinstated the practice with the result that rates for those properties will be NFIP risk based, but perhaps rated in the wrong flood zone. The number of grandfathered polices is not known. Also, policyholders who live in communities that take actions to reduce flood risk can earn points in the Community Rating System (CRS). Policyholders participating in the CRS receive discounts on the NFIP risk-based premium; some community actions that earn CRS points may reduce expected losses and warrant the premium reductions, but others may not. Outside the SFHA, the NFIP does not require an elevation certificate for properties and also offers preferred risk policies3 (PRPs) for properties that have a favorable loss
3 Preferred risk policies can provide flood insurance coverage for both buildings and contents that are located in moderate to low flood-risk areas.
history. Neither premium is set using a rating table that considers first-floor elevation in relation to a BFE and, as such, cannot be an NFIP risk-based premium rate. BW 2012 does not direct FEMA to review and modify PRP and X zones, which are zones of moderate to low risk of flooding rates to make them risk based.
Finding 4.3. Full implementation of BW 2012 will not result in NFIP risk-based rates for properties located outside the SFHA.
ABILITY TO PAY
BW 2012, Section 100236 states that FEMA
. . . shall enter into a contract under which the National Academy of Sciences, in consultation with the Comptroller General of the United States, shall conduct and submit to the Administrator an economic analysis of the costs and benefits to the Federal Government of a flood insurance program with full risk-based premiums, combined with means-tested Federal assistance to aid individuals who cannot afford coverage, through an insurance voucher program.
Even though the language in the committee’s statement of task does not mirror the language in Section 100236, the committee did review Section 100236 to gain insights into the important questions being asked. For example, the phrase “cannot afford” can be understood as exceeding an individual’s ability to pay an NFIP risk-based premium. This focus on ability to pay requires FEMA to define when such premiums impose a cost burden on an individual. Report 1 discussed three possible measures of cost burden, two of which were related to an individual’s income. Specifically, Report 1 discussed an income approach and a housing cost as a percent-of-income approach to identify those who would be cost burdened by NFIP rate increases, as well as an approach suggested in HFIAA 2014 that identified premiums exceeding 1 percent of coverage as burdensome.
During the preparation of Report 2 the committee continued to discuss different definitions of cost burden and ability to pay as it considered the data needed to build a microsimulation model module to estimate who would be eligible for assistance under various affordability policy options. In having that discussion, the committee developed new findings relevant to the topic, including the measure of cost burden suggested by HFIAA 2014—premium as a percentage of flood insurance coverage. The commit-
tee found that a property owner’s income or wealth characteristics cannot be incorporated into the above-cited cost burden measure. For example, households with income of $500,000 and $50,000, but with the same coverage and premiums, would be considered equally cost burdened and, if a policy provided assistance to eliminate the entire cost burden, would receive similar amounts of assistance.
Finding 4.6. The use of premium as a percent of insurance coverage does not, by itself, satisfy the congressional directive to FEMA to consider providing “targeted assistance to flood insurance policyholders based on their financial ability.”4 Therefore, if ability to pay is the congressional concern, then FEMA will still need to develop a measure of cost burden based on policyholder income or wealth or both.
The committee’s review of the capped premium approach to defining cost burden and its assessment of policy analysis data needs and gaps led the committee to consider the premium as a percentage of the assessed value of the insured property as an alternative measure of cost burden. Property value, which is a substantial component of total wealth for many households, is used as a proxy for wealth. Wealth, in turn, would be employed as a metric for defining ability to pay for flood insurance. Adding this cost burden measure means that the committee considered four different approaches for defining when NFIP premiums become unaffordable. Each of these approaches has both advantages and disadvantages.
Finding 4.7. For the purpose of implementing an assistance program, policy makers will need to decide whether they want to define cost burden with reference to income, housing costs in relation to income, premium paid in relation to property value, or some other measure. This decision can be informed by technical analysis of the alternatives, but the final selection is a policy judgment.
Loss of Property Value and Household Wealth
Some property owners prior to BW 2012 were eligible to pay PFS rates. Eliminating those rates will increase premiums and, in turn, lower property value. The committee was aware that reduction in property values was a frequently expressed concern of property owners and communities following passage of BW 2012. An argument made by some was that, in many cases, the property value was a substantial component of a policyholder’s total wealth. Therefore, a premium increase that diminished property value
4 HFIAA, Sec. 9(b)(2).
would, in turn, have a negative impact on wealth. HFIAA 2014 was the congressional response to these and other expressed concerns, but it too will result in the eventual elimination of PFS rates and reduced property prices.
The committee considered the possibility of analytically identifying the effects on property values of losing PFS rates. However, isolating effects of premium increase from other determinants of market price will be difficult, even if the best data for making such a calculation were available. The committee also considered policy options to mitigate these effects if they could be identified. One option to mitigate these effects would be to cap rates at a level less than NFIP risk-based rates for all PFS rates and also allow that cap on the premium to transfer with the property to all future owners. This would be without regard to the future owners’ ability to pay. This option would be contrary to the goals of BW 2012 to have property owners pay NFIP risk-based rates and would result in lost revenues to the NFIP, unless offsetting increases in revenues were provided by the federal treasury or by cross subsidy. Another option would be for FEMA to offer financial compensation for property value loss when homeowners sell their house.
Finding 4.8. The negative effect on property values from allowing PFS rates to rise to NFIP risk-based rates is a market-driven reality but would be analytically difficult to isolate from other determinants of property price. A policy decision to compensate for some amount of property value loss may require public expenditure.
LINKING MITIGATION WITH PREMIUM ASSISTANCE
Report 1 described how assistance might be offered for making flood insurance premium payments, for paying for some or all of mitigation that can lead to reduced premiums, or for a combination of both. One way to link flood insurance premium assistance with flood mitigation is through providing an annual assistance payment that the property owner could use to cover the premium and implement mitigation through a long-term loan. If the property owner is expected to decide how to use an assistance payment, then owners should be provided with information on mitigation measures and the contributions of adopting such measures to premium reductions. This type of information is not currently available to homeowners and would require a new outreach and communication effort, especially if the NFIP offers premium reductions for mitigation actions in addition to elevating one’s home (Report 1, Chapter 7).
An alternative approach would be for the NFIP to make the calculation to determine the most cost-effective flood mitigation measure for the policyholder and, if assistance was offered, the NFIP would require that cost-effective mitigation be implemented. The argument for this mandatory
use of assistance is that people lack the information to make a financial assessment of the value of mitigation, so a calculation would be made on a policyholder’s behalf.
Finding 4.10. Linking mitigation with premium assistance can lead to property owners having a cost-effective combination of mitigation and insurance coverage. Identifying that combination, however, requires complex calculations and the roles and responsibilities of FEMA in assisting with that calculation need to be assessed and, potentially, enhanced.
A FINAL REFLECTION
Floodplains and coastal areas across the United States will continue to be inhabited, especially in places where ready access to water is essential to the economic activities of people and their communities. These geographic areas will sustain occasional damages from future riverine floods and coastal storms. The costs of these losses will be borne in three possible ways, or in some combination. One is that individual NFIP policyholders will bear location cost in the form of insurance premiums paid and damages falling within policy deductible amounts. The second is that the federal taxpayer might bear floodplain location costs if the federal treasury develops a premium assistance program, makes up for NFIP premium revenue shortfalls, pays for pre-flood mitigation, or makes post-flood disaster assistance payments to individual households. Third, property owners and other floodplain or coastal zone inhabitants will bear costs for the losses that are uninsured or otherwise uncompensated.
An original intent of the NFIP was to replace disaster aid payment with flood insurance purchase to the maximum extent possible, shifting more of the cost of floodplain location onto those persons who occupy such places (Report 1, Chapter 2). If this goal is to be pursued, then requests for premium assistance or pre-flood mitigation grants and loans may increase due to future possible premium increases and from changes in flood risk, stemming from changes in climate and changes in watershed runoff due to development. As an NFIP affordability framework is developed, FEMA and Congress will confront the central question, “Who will bear the costs of floodplain occupancy in the future?” With specific reference to the goal of “affordable premiums,” that question will be answered in recognition of the available governmental budget for premium or mitigation assistance and the adherence to the actuarial principle of minimizing cross-subsidies within the NFIP.