5

Insuring Properties Behind Levees

Insurance is a method of prefunding (usually over time) the adverse consequences of a loss so as to have available funds to pay damages when they arise. It is a legal contract that transfers some of the financial consequences of an uncertain low-probability but high-consequence peril to another party (called an insurer) in return for periodic relatively low payments (premiums). If an event covered by the insurance contract occurs, and if coverage is triggered by conditions detailed in the contract, then the insurer will indemnify the insurance purchaser up to the limits of the insurance policy in accordance with the terms of the contract (e.g., the deductible, the policy limits, proof of loss, etc.). Thus, the value of insurance arises at the time of a covered loss, and can thus be viewed as a hedging instrument against the financial consequences of a loss.

In insurance nomenclature, the causes of loss is called a peril, and a condition or situation that increases either the likelihood or severity of the peril occurring is called a hazard. Risk, on the other hand, is the term used to designate the loss consequence of the realization of the uncertain peril, and it may be financial or nonfinancial in nature (e.g., reputational loss). Thus for example, fire is a peril, storing oily rags next to a gas water heater is a hazard for the peril of fire, and the loss of property and the financial losses due to fire-related property destruction is the risk. Concerning the peril of flooding, building unelevated property below the base flood elevation in a floodplain is a hazard, and the financial consequences of partial or total destruction of property due to flooding is the risk that the property owner faces. Risk is an adverse consequence of uncertainty concerning perils, and without uncertainty there can be no risk.

Not all risk can be insured in the private market (i.e., by private insurers) but when available, insurance can be part of an effective risk management strategy. Other parts of the risk management process include risk identification, risk assessment, risk mitigation or control, and risk communication. Insurance is not a risk mitigation technique, however, because the legal insurance contract does not change the physical probability or the severity of a realized peril, but rather it just allocates the cost to another party.1

From the perspective of the purchaser, insurance is a viable risk transfer technique when the probability of the event occurring, p, is relatively small and the consequence or loss, L, is relatively large (so as to be difficult to financially assume all at once by the purchaser without insurance), but such that the expected value in any given

_______________

1 Insurance that is truly risk-based priced can motivate policyholders to mitigate risks to lower premium costs, but it is not that insurance itself that mitigates the risk, but rather the affirmative actions of the policyholder to lower frequency or severity.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 63
5 Insuring Properties Behind Levees Insurance is a method of prefunding (usually over time) the adverse consequences of a loss so as to have avail- able funds to pay damages when they arise. It is a legal contract that transfers some of the financial consequences of an uncertain low-probability but high-consequence peril to another party (called an insurer) in return for peri- odic relatively low payments (premiums). If an event covered by the insurance contract occurs, and if coverage is triggered by conditions detailed in the contract, then the insurer will indemnify the insurance purchaser up to the limits of the insurance policy in accordance with the terms of the contract (e.g., the deductible, the policy limits, proof of loss, etc.). Thus, the value of insurance arises at the time of a covered loss, and can thus be viewed as a hedging instrument against the financial consequences of a loss. In insurance nomenclature, the causes of loss is called a peril, and a condition or situation that increases either the likelihood or severity of the peril occurring is called a hazard. Risk, on the other hand, is the term used to designate the loss consequence of the realization of the uncertain peril, and it may be financial or nonfinancial in nature (e.g., reputational loss). Thus for example, fire is a peril, storing oily rags next to a gas water heater is a hazard for the peril of fire, and the loss of property and the financial losses due to fire-related property destruc- tion is the risk. Concerning the peril of flooding, building unelevated property below the base flood elevation in a floodplain is a hazard, and the financial consequences of partial or total destruction of property due to flooding is the risk that the property owner faces. Risk is an adverse consequence of uncertainty concerning perils, and without uncertainty there can be no risk. Not all risk can be insured in the private market (i.e., by private insurers) but when available, insurance can be part of an effective risk management strategy. Other parts of the risk management process include risk iden- tification, risk assessment, risk mitigation or control, and risk communication. Insurance is not a risk mitigation technique, however, because the legal insurance contract does not change the physical probability or the severity of a realized peril, but rather it just allocates the cost to another party. 1 From the perspective of the purchaser, insurance is a viable risk transfer technique when the probability of the event occurring, p, is relatively small and the consequence or loss, L, is relatively large (so as to be difficult to financially assume all at once by the purchaser without insurance), but such that the expected value in any given 1  Insurance that is truly risk-based priced can motivate policyholders to mitigate risks to lower premium costs, but it is not that insurance itself that mitigates the risk, but rather the affirmative actions of the policyholder to lower frequency or severity. 63

OCR for page 63
64 LEVEES AND THE NATIONAL FLOOD INSURANCE PROGRAM year p × L is affordably small as a basic for an annual insurance premium calculation. 2 This expected value is called the “pure premium” in insurance and is subsequently “loaded” or modified to reflect insurance underwriting costs, claim adjusting and settlement costs, and returns to the providers of capital to the insurance organization. An additional loading is added to reflect the level of uncertainty in the estimation of the pure premium. Once loaded, the result is the actual premium charged to the purchaser of insurance. As part of the National Flood Insurance Program (NFIP), the insurance component is one pillar to address catastrophic loss potential from floods (other pillars include motivating building and land-use restrictions in vul- nerable areas and construction of dams, levees, and other structures). NFIP insurance is available to all property owners in communities that participate in the program. Coverage for flood damages extends up to $250,000 dol- lars for residential structures and $100,000 dollars for residential contents and $500,000 for business structures and $500,000 dollars for business contents.3 Those that require greater amounts of coverage than available from the NFIP have access to additional flood insurance through private industry, e.g., their property insurance carrier. BASICS OF INSURANCE PRICING Pricing is done in essentially the same manner in all areas of insurance with some differences in models and assumptions corresponding to the characteristics of the different types of insurance. Property and casualty insur- ance (also known as hazard insurance or general insurance in some areas) generally has a shorter contract dura- tion, usually exists in a more highly regulated environment than life insurance, and has more complicated claims structure (with more correlated losses, the potential for multiple claims within the policy period, etc.). These characteristics have led financial/actuarial pricing formulas specifically tailored to the property or liability arena, which are somewhat distinct from the models used for pricing life and health insurance. A brief introduction to various aspects of property insurance pricing including manual rate setting and individual risk rating is presented. (More detailed examination of pricing of general property insurance is given by Ai and Brockett [2008] and by the Casualty Actuarial Society [1990].) An (oversimplified) essential representation of pricing of insurance policies is that prices (or charged rates) should equal the expected present value of the future losses on the contract augmented by a load for expenses, plus an insurer profit load, plus a load to compensate the insurer for risk bearing (see equation (5-1)). Hence, the price for insuring a risk exposure is given by N P = E[∑ LiV (Ti )](1 + k1 )((1 + k2 )(1 + k3 ), (5-1) i =1 where P is the price, E is the expectation operator representing the taking of the statistical expectation of the random variables involved in the brackets, and where k1 = the load for expenses (marketing underwriting, administration, taxes, claims adjusting, commissions, and general operating expenses); 2  For insurance to be a viable alternative for an insurer to offer, there needs to be additional conditions that mitigate the financial hazard being assumed by the insurer so that the risk-assuming insurer itself does not go insolvent and can adequately manage the financial exposure they take on. These ideal conditions for a risk to be insurable include that (1) there be a large number of independent homogeneous expo- sure units to be insured (so the law of large numbers and the central limit theorem from statistics is available to the insurer for pooling and spreading risk among clients); (2) losses that occur are accidental; (3) a catastrophe cannot occur that affects a large number of exposure units simultaneously (this high correlation between losses thus defeating the independence of the exposure units and mitigating the insurers’ use of the law of large numbers and the central limit theorem); (4) losses must be definite and measurable (so they can be priced); and (5) the prob- ability distribution for losses should be determinable (so prices can be set). In addition to the above characteristics of a risk to be insurable, there are behavioral conditions that the customer can create which will make the insurer reluctant to offer insurance. One of these is adverse selection, which can occur when the individuals who have a higher likelihood of experiencing losses are disproportionately drawn to buying insurance defeating the risk pooling needed by insurers (e.g., if only high-flood-risk property owners seek flood insurance). Finally moral hazard also needs to be considered by insurers, where moral hazard is the term used to describe the behavioral property that can arise wherein the people who are insured will take fewer precautions than those who are not insured simply because they have insurance and are not fuly responsible for their losses. This escalates costs to insurers, and if severe moral hazard or adverse selection is anticipated, then insurance will not be offered (Baranoff et al., 2009). 3  See floodsmart.gov.

OCR for page 63
INSURING PROPERTIES BEHIND LEVEES 65 k2 = profit load (included so as to allow the capital suppliers to obtain a “fair” or competitive rate of return on capital); k3 = load for bearing uncertainty in future cash flows; Ti = (possibly random) time until payment of the ith claim during the policy coverage period; Li = (possibly random) dollar severity realized in the ith claim, also called the severity (or size of loss) of the loss exposure; N = (possibly random) number of claims incurred during the policy period, also called the frequency; V(Ti) = the discounted present value of $1 payable at time Ti (a priori this being a random stochastic process or a random variable). T Usually V(Ti) is taken to have the form V(Ti) = v i where v = 1/(1 + r) is the discount factor based on an interest rate r. Stochastic models taken for N often include Poisson (for rare events) and binomial distributions. Statistical distributions chosen for Li generally depend on the particular hazard situation or peril involved. Indeed, the determination of appropriate statistical distributions and the dependence structure for N and Li is a crucial part of property insurance pricing. (See Klugman et al. [2008] for explicit loss distributions models used in insurance pricing.) Once calculated, these theoretical prices may be modified for political, social, or market competition reasons. For example, political or regulatory considerations may force the company to deviate from the theoretically derived prices and constrain or suppress rates (Witt and Hogan, 1993). Such regulatory intervention (and social influences) upon the mathematically derived pricing formulas occurs more in property insurance lines wherein purchase is mandated by contractual obligations (e.g., homeowners’ insurance on mortgaged homes, comprehensive colli- sion insurance on financed automobiles, etc.) and whose purchase is deemed .mandatory for social reasons (e.g., automobile insurance, workers’ compensation, flood insurance4) (Ai and Brockett, 2008). Most state regulations impose the conditions that rates (or premiums) cannot be inadequate (to ensure solvency), excessive (to prevent excessive profits), or of an unfairly discriminatory nature (for social equity) (Ai and Brockett, 2008). Pricing Property Insurance Using “Manual Rates” The notion of “manual rating” is setting the rate (or price) based on a basic rating table (or tables) that produces a price per unit for different classes of risks. From such tables the user can obtain a numerical value (say, dollars to charge per $1,000 of coverage) that is then applied to obtain a price for any designated risk within the class of risks covered by the table. Such manual rates are used generally to price homogeneous groups of exposure risks and are not tailored to the idiosyncratic characteristics of any specific individual. For such manually rated policies, the issue is how to determine the table entries from which the appropriate rates can be read. Manual rate encompasses two basic methods: the loss ratio method and the pure premium method (Casualty Actuarial Society, 1990). The pure premium method bases the rate on the expected fundamental loss exposure probability distribution itself. Mathematically, E[ L ] + F R= , (5-2) 1− V − Q In equation (5-2), R denotes the rate per unit of exposure, E[L] denotes the expected loss per unit of exposure (called the pure premium or actuarial value), F denotes the per-exposure unit fixed expense, V is a variable expense factor, and Q is a risk contingency factor that also incorporates anticipated profit. Essentially, this is analogous to equation (5-1) in that it says the rate should be sufficient to pay expected losses after allowing for required profits and fixed and variable expenses (Ai and Brockett, 2008). The expected loss per unit of exposure is calculated as the discounted present value of the loss(es) per exposure 4  According to the Federal Emergency Management Agency revised edition of the Mandatory Purchase of Flood Insurance Guidelines (FEMA, 2007), “The mandatory purchase law directed the Federal agency lender regulators and Government-Sponsored Enterprises (GSEs) to develop and adopt regulations requiring lenders subject to their jurisdiction not to make, increase, renew, or extend any loan on applicable property unless flood insurance is purchased and maintained to protect that property securing loans in high flood risk areas.”

OCR for page 63
66 LEVEES AND THE NATIONAL FLOOD INSURANCE PROGRAM N unit, E[∑ LiV (Ti )]. After calculating this value for each possible rating class, a table (or set of tables) of values i =1 is created such that every possible exposure unit is uniquely found within a table. The coverage rate for a new exposure is ascertained by first determining the exposure’s class, and then reading the multiplicative factor for this class from the corresponding table. Ultimately, by multiplying the number of exposure units encompassed by the risk exposure to be priced times this factor, a premium is determined. The difficulty in applying this method is appropriately stochastically modeling the frequency, N, and severity, Li, variables and their dependencies. Many books on statistical analysis have been devoted entirely to this topic (cf. Klugman et al., 2008). The second method, called the loss ratio method, is really a premium adjustment method as opposed to a premium-setting method. It assumes a current rate (R0) and determines a rate change for the next year by mul- tiplying by an adjustment factor (A) to obtain a new rate R = AR0. The adjustment factor (A) is the ratio of the experienced loss ratio (Y) to a prespecified target loss ratio (T). Mathematically, R = AR0, where R0 is the current rate, A = Y/T, and Y = L/[(R0)(e)], where L denotes the current period experienced losses and e is earned exposure for the experience period. The target T = (1 − V − Q)/(1 + M), where V denotes the premium-related expense factor, Q is the factor for contingencies and profit, and M is the ratio of nonpremium-related expenses to losses. The loss ratio method (by construction) is not applicable when a current rate is unavailable (such as for new lines of business). Still, it allows the insurer to obtain new rates for the next period by updating the previous period’s rates. Brockett and Witt (1982) show that “when current prices are set by a regression methodology based on past losses, and the loss ratio method is used to adjust prices, then an autoregressive series is obtained, thus partially explaining insurance pricing cycles”. Although the loss ratio method and the pure premium method are constructed from different perspectives, one can show that the methods yield the same prices given the same data (cf. Casualty Actuarial Society, 1990). Individual Risk Rating To obtain prices for individual exposure units, one can make modifications to the manual rates to accommodate for individual characteristics. A basic approach is to use a so-called credibility formula, 5 where individual loss experience is incorporated with an estimation of expected losses for the risk class derived from some other source to obtain a blended or averaged rate concatenating individual loss experience and exogenous derived expected losses (Ai and Brockett, 2008). Two such rating schemes are prospective rating and retrospective experience rating (Casualty Actuarial Society, 1990), where the rate for a future period is determined using a weighted combination of expected losses (from some other source such as industry data) an individual experienced losses. In addition to the above methods, a somewhat different method for individual risk rating is called schedule rating (Ai and Brockett, 2008). It adjusts the manual rates to reflect individual risk characteristics that may affect future losses. Unlike the above individual risk rating methods, however, it does not necessarily use individual loss experience to make this adjustment, but instead creates a factor to apply to the manual rate to account for char- acteristics known to affect the likelihood of losses (e.g., a certain building standard, such as height of first floor, may result in a multiplicative factor less than one being applied to the manual table rates for flood insurance) (Ai and Brockett, 2008). CURRENT RATE-SETTING PRACTICES WITHIN THE NFIP FEMA’s National Flood Insurance Program (NFIP) has approximately 5.6 million policies insuring over a trillion dollars in assets. Currently, 21,881 communities participate in the NFIP.6 Broken down by structure type, this includes 5  The concept of credibility theory, which dates back to Mowbray (1914), is said to be “the casualty actuaries’ most important and enduring contribution to casualty actuarial science” (Rodermund, 1990). 6  As of June 2012.

OCR for page 63
INSURING PROPERTIES BEHIND LEVEES 67 • 3,839,254 single-family policies, • 281,508 nonresidential (i.e., commercial) policies, • 341,607 multifamily (non-condominium) policies, • 1,123,427 units (in 78,821 contracts or buildings) Residential Condominium Building Association policies. The above numbers include all NFIP policies regardless of flood zone designation and 10,361 properties that have been designated as severe repetitive loss properties7 (Andy Neal, FEMA, personal communication, July 19, 2012). The purpose of this section is to explain the current process by which the NFIP sets rates and discuss the similarities and differences related to traditional private insurance processes for setting risk-based rates. Next, comments on the actuarial and fiscal soundness of the NFIP on the basis of the risk premium methodology used are presented. Finally, improvements and advancements that can be made to better serve the goals of the NFIP and the pricing of flood insurance are discussed. Full-risk Class Rates FEMA evaluates the flood risk of policy holders and determines rates for the “full-risk” class through a bal- ance of elements including the extent and type of flood hazard, the base elevation of the insured structure and the structure type,8 the contents location (first floor, second flood, etc.), and whether or not the community participates in the Community Rating System9 that provides discounts for communities that actively manage their flood risk (Box 5-1). At the core of FEMA’s risk-based rate setting is the calculation of the expected loss for a property. FEMA calculates community losses using average annualized loss data (Box 5-2). In theory, this is calculated for each property by conditioning upon the floodwater level and the probability of that level of water inundation occurring during the year. Thus, they assess the frequency or probability that the floodwaters will reach an elevation of i (which they denote by probability of elevation or PELVi). This number is then multiplied by the “loss severity” that would occur at the structure if the floodwaters reached level i (i.e., damage based on water depth in a given structure), which they denote by damage by elevation (DELVi). This is then summed over all possible water eleva- tion levels to arrive at an expected loss.10  Max  Expected Loss =  ∑ (PELVi × DELVi )  (5-3)  i = Min  This is consistent with a standard actuarial method for calculating the expected loss due to flooding similar to the pure premium method described in equation (5-2) earlier. In practice, properties having similar risk-related covariates (flood risk elevation, construction type, zone, etc.) are grouped together, and a single rate is given to all properties in this class nationwide. This is essentially similar to risk-based rate setting by private insurers in property insurance and automobile insurance where entities having the same set of underwriting classification characteristics are grouped together and given the same rate. Risk pooling in this regard is also used by private insurers to spread the risk.11 7  Severe repetitive loss properties, as established in Section 1361A of the National Flood Insurance Act, as amended, 42 U.S.C. § 4102a) is defined as a residential property that is covered under an NFIP insurance policy and has at least four claim payments and for which two of these payments in total have exceeded the market value of the property. See http://www.fema.gov/severe-repetitive-loss-program. 8  Condominium, single or multi-family home, residential or commercial use, number of floors, with or without a basement, ventilated crawlspace, etc. 9  The discounts given under the Community Rating System are modeled back in to the expected aggregate premium calculation for the NFIP by adjusting all premiums upward so that the aggregate premium collected is sufficient to cover expected losses accounting for CSR discounts for the CSR discounts. Thus, CRS discounts constitute a relative discount which is adjusted for in the aggregation and is not an uncompensated for diminution of income as occurs with other types of non-full-risk class policies. (Andy Neal, FEMA, personal communica- tion, August 22, 2012). 10  See Hayes and Neal (2011, Appendix). 11  Although risk pooling is used in private insurance, there have been questions about the appropriateness of the breadth (nationwide) of NFIP’s grouping of all properties within the same zone classification for the purpose of obtaining a single rate applicable to all properties in the

OCR for page 63
68 LEVEES AND THE NATIONAL FLOOD INSURANCE PROGRAM BOX 5-1 FEMA’s Community Rating System FEMA’s Community Rating System (CRS) is a voluntary incentive program that encourages flood- plain management activities exceeding the minimum NFIP requirements. The CRS has 1,211 participating communities among the approximately 21,000 communities participating in the NFIP. However, policies held within CRS communities represent 68 percent of the policies in force (Bill Blanton, FEMA, personal communication, July 10, 2012). Typically, communities receive between a 5 and 15 percent discount for participating in the program with a maximum discount of approximately 45 percent (FEMA, n.d.a; Bill Blanton, personal communication, July 10, 2012). Discounts are calculated based on a credit point system for actions that help save lives and property in the event of a flood—and that exceed the minimum NFIP requirements. There are four catego- ries of activities that are credited: (1) public information programs that advise the public about the hazard, flood insurance, and strategies for reducing the hazard; (2) mapping and regulation programs that increase protection to development that postdates existing maps; (3) flood damage reduction programs, such as relocating or retrofitting vulnerable structures); and (4) flood preparedness programs, such as flood warning and dam and levee safety programs (FEMA, n.d.a). For additional insight regarding the CRS, see Chapter 7. BOX 5-2 Average Annualized Flood Losses Average annualized losses (AALs) due to flooding represent the estimated value of flood losses in any given year in a selected region. Analysis of such data provides information on changes in losses by geographic area that are useful in identifying loss trends and geographic anomalies in loss patterns. The data are also useful to leaders in establishing policies and priorities and developing flood risk management strategies. FEMA’s Hazus-Multi Hazard (Hazus-MH) offers analysts the opportunity to quantify annualized flood losses in a given community over multiple flood recurrence intervals. The Hazus-MH program, which can calculate damages, casualties, and economic losses, is based on national data from U.S. government agencies. It may also use, when available, higher resolution data available within the communities. For flood AALs, it calculates losses under each of five flood recurrence intervals (10, 25, 1.0, 0.5, and 0.2 an- nual chance floods or 10, 50, 100, 200, and 500 years, respectively) and then multiplies the loss values by their respective annual frequencies of occurrence, then summing the values for an annual average. This approach is focused on the recurrence intervals within the bounds of FEMA FIRMs’ upper delineation, the 0.2 percent flood. Additional calculations using recurrence intervals for extreme events (>0.2 percent) would provide a more robust portrayal of the potential AALs but would likely increase the costs of the AAL process. FEMA began the calculation AALs in 2009. Pilot data on flood AALs have been released for initial review within the floodplain management community but release of the entire report is awaiting final FEMA review. The use of Hazus-MH provides a relatively straightforward and simple but time-consuming method of making such calculations. Because the quality and resolution of the data used are subject to considerable variation across the country, the resultant AAL figures should be taken as general estimates and not high- resolution information. Such an approach is appropriate for the management uses envisioned by FEMA.

OCR for page 63
INSURING PROPERTIES BEHIND LEVEES 69 The determination of the components in the above expected-value calculation is as follows: The event fre- quency (probability) or PELV is obtained by reference to a collection of PELV probability curves created to describe the probability of various water depths relative to the 1 percent flood depth, that is, at this property the probability of a flood during the year reaching the height +2 (2 feet above the base flood level of the one percent annual chance flood) is a certain point on the PELV curve.12 The PELV curves provide probabilities up to the 0.2 percent annual chance event (flood recurrence interval of 1 in 500 years). For all flood events greater than the 0.2 percent level, the program assumes a “catastrophic” water depth level by doubling the 0.2 percent depth. Because of the low likelihood of this “catastrophic” water depth level and because of the generally small marginal increase in the damage assumption between the 0.2 percent water depth and the damage at the assumed catastrophic depth, this approximation does not affect the rate significantly. Currently, “residual risk” resulting from the likelihood of overtopping a levee, levee failure, human development resulting in other topographic changes affecting inundation probability, and so on are not mapped or considered in the PELV determinations (FEMA, 2006). The second component in the above expected-value calculation, the damage estimate as a function of the water depth elevation, or DELV, is based on historical damage data obtained by FEMA associated with different water depths in the given zone, and varies with structure content and location. When the NFIP’s historical damage data are sufficiently credible13 based on the number of claims and the variability of those claims, the NFIP data are used to develop the statistical damage estimate. When historical NFIP data are absent, U.S. Army Corps of Engineering (USACE) data are used. When some historical NFIP data are available but are is not fully credible (in an actuarial sense), the NFIP program uses a blend of NFIP data and USACE data to determine the DELV variable. This is consistent with standard property and casualty actuarial practice in private insurance. 14 Once the expected loss is determined, the actual premiums or rates for the NFIP are determined by “loading” the expected losses in a manner similar to equation (5-3). Specifically, the charged rate for the group charged at the full-risk costs is determined by the formula LADJ × DED × UINS Rate =  ∑ i = Min (PELVi × DELVi )  × (5-4) Max   EXLOSS where LADJ is the loss adjustment expense loading, similar to the factor k1 in equation (5-1), and DED is a factor that adjusts the DELV variable to compensate for the amount of the deductible chosen (so that the loss actually paid reflects the deductible amount) (Hayes and Neal, 2011). The variable UINS represents an adjustment to account for the underinsurance amount because not all properties can (or do) insure to their full value. The UINS factor is determined by the NFIP using a review of past insurance claims data. Incurred losses are a nonlinear function, with most losses being smaller and relatively fewer losses being much larger (i.e., the loss distribution is skewed); hence the variable UINS adjusts the loss for the percentage coverage to account for this nonlinearity (Hayes and Neal, 2011). This is similar to traditional homeowners insurance. Currently the value used for LADJ is 1.05 and zone nationwide, without regard to state or topography. For example, the comparison of losses to premiums over time differs dramatically across states even within similar zones. This suggests that some properties in one state are cross-subsidizing other properties in other states because they are charged the same rate but have much different loss probabilities or risks (GAO, 2008). In a competitive insurance market with many participants, such cross-subsidization would be arbitraged away because an insurer who can identify those policyholders being overcharged relative to their risk (the subsidizers) would recruit such policyholders by offering lower premiums until perceived discrepancies disappear. Because the NFIP is essentially a monopolistic insurance provider, this cross-subsidization may continue to exist, and innovation to identify better pricing methods and cross-subsidization opportunities may not be pursued with the same vigor as in the private competitive market. 12  The curve for a given flood zone (A01-A30) is based on the difference between the 1 percent flood elevation and the 10 percent flood elevation with each policyholder being assigned a PELV curve. The original PELV curves are based on studies conducted at the time the program was established. FEMA is currently collecting data to reevaluate the original PELV curves (Andy Neal, FEMA, personal communica- tion, November 7, 2011). 13  In actuarial practice “creditability theory” delineates when there is sufficient historical data to give a confident assessment of losses for estimation purposes, and how to augment the estimate of loss using auxiliary datasets when there are not sufficient historical data. See Casualty Actuarial Society (1990). 14  The accuracy of the loss data used for determining the values of DELV has been questioned since many entries in the NFIP data set obtained from claims history have loss values but are missing data on elevation at the site. In these cases NFIP uses a zero elevation in their calculation, which will bias the losses experienced at zero elevation, and distort the DELV curve values. See GAO (2008) for details and refer- ences to further studies validating this criticism.

OCR for page 63
70 LEVEES AND THE NATIONAL FLOOD INSURANCE PROGRAM the value used for DED is 0.98 (Andy Neal, FEMA, personal communication August 22, 2012). Finally, the factor EXLOSS represents the expected loss ratio and contingency loading 15 and adjusts the rate to accommodate com- missions, acquisition costs, and other costs such that the rate times the expected loss ratio is sufficient to cover the expected loss accounting for the loss adjustment expenses as well as the idiosyncratic choices by the purchaser of the deductible and underinsurance amount. Again, this formula is actuarially appropriate given the constraints and attributes of the program (absence of bankruptcy risk, ability to borrow from the government, etc.). Discounted Rates By statute, Congress mandated that the NFIP make flood insurance available to certain properties at less than their full-risk rates, that is, at discounted rates below those specified by equation (5-2). This occurs. for example. if (1) a structure was built before the flood insurance maps were available, that is, it was a “pre-FIRM” (Flood Insurance Rate Map) structure or built before December 31, 1974; (2) a structure was built in a V zone before 1981 and before maps that consider wave height were adopted in setting flood insurance rates; (3) a structure is in an AR or A99 zone for levees in the course of reconstruction or construction (so their current actuarial risk does not correspond to their current risk charge); or (4) the policyholder participates in a group flood insurance policy. Discounts for structures that were built pre-FIRM or where the NFIP does not have any elevation informa- tion are by far the largest group—with 1,082,201 policies or a little less than 20 percent of policyholders. Only 7,508 policies are discounted because of entering the program prior to including wave height in rate setting, that is, pre-1981 V zones. Discounted rates are applied to approximately 24,907 policies for property owners behind levees in construction or deaccredited levees but where the community is showing a good-faith effort to construct levees or repair the levees (AR/A99). Pre-FIRM Policies The largest group of discounted policies consists of those that predate the existence of the flood insurance maps. By legislative directive, pre-FIRM policies were given a discount in order to encourage community participation in the NFIP, to help maintain property values for homeowners who might not have known of their flood risk at the time of purchase, and to encourage NFIP participation among those with a higher risk of floods (Michel-Kerjan, 2010). Currently, of the 5.6 million NFIP policies, there are 1,082,201 or approximately 19.3 percent of the poli- cies that are pre-FIRM discounted policies (Andy Neal, FEMA, personal communication, July 19, 2012). These can be further decomposed as • 736,066 single-family policies (150,674 of those are nonprincipal residences), • 82,387 nonresidential policies (commercial) policies, • 93,669 multifamily (non-condominium) policies, and • 170,079 Residential Condominium Building Association policies (corresponding to 11,516 contracts or condominium buildings). By statute, highly discounted premiums have been made available for pre-FIRM buildings in the Special Flood Hazard Area (SFHA). Not all structures that were built in SFHAs prior to the FIRM are discounted, how- ever, because some structures meet more stringent building codes and qualify for better rates on a full-risk basis because of elevation above the base flood level than they would receive taking the discount off the higher SFHA rate. For the discounted older buildings, the average full-risk premium is about five times greater than the average full-risk premium for compliant buildings. With the discounted premium level, the noncompliant NFIP discounted buildings only pay between 40 percent and 45 percent of what they should pay would they be charged using the actuarially determined full-risk premium for being in the SFHA. Even so, their discounted premiums are still much 15  Currently, the contingency loading incorporated into the EXLOSS factor includes a 10 percent additional load in nonvelocity zones and 20 percent in velocity zones (cf. Hayes and Neal, 2011).

OCR for page 63
INSURING PROPERTIES BEHIND LEVEES 71 higher than those that would be paid by actuarially rated policyholders for buildings constructed in compliance with the building codes. The status of the pricing subsidies given to certain policyholders will change in the future. The Biggert-Waters Flood Insurance Reform Act of 2012 included provisions that were intended to help the fiscal soundness of the NFIP by eliminating certain pre-FIRM discounts and increasing the ability of the NFIP to raise rates to achieve fiscal soundness (Nechamen and Inderfurth, 2012). Effective July 1, 2012, pre-FIRM discounts are being removed from certain types of properties, with permissible increases in their flood insurance rates of up to 25 percent per year until the actuarial rate level is achieved for the property. Discounted policies losing their discount include nonprimary residences (e.g., second homes), severe repetitive loss properties, 16 all business properties, homes that have had substantial damage or improvements (of over 30 percent of the market value of the property) after the implementation of the Act, and several other classes of properties (Nechamen and Inderfurth, 2012). Group Flood Insurance Policies A group flood insurance policy is a discounted policy offered after a presidentially declared flood event for property owners who qualify for federal individual assistance. Qualifications include low-income property owners who do not qualify for a Small Business Administration disaster loan. On a one-time basis, owners are provided by FEMA with a 3-year policy with a coverage amount equal to FEMA’s Individual and Households Assistance Program maximum (currently $31,400). The policy premium is set according to 44 CFR §61.17 at $600 per year, which is typically subtracted from the insurance adjustment payment received in the event of a loss (Andy Neal, FEMA, personal communication, January 15, 2013). Approximately 42,000 group flood policies were issued in the wake of Hurricane Katrina. These policies will eventually become nondiscounted and will be charged actuarial rates, and so they constitute a short-term imbalance, but new group flood insurance policies will be created as new floods occur (continuing to affect the fiscal soundness of the NFIP). Grandfathered Rates Grandfathered policy rates are applied to certain policies when revisions to FEMA’s FIRMs result in increased premiums. Unlike the discounts described above, grandfathering is achieved through cross-subsidies with other policyholders. Grandfathered rates are available to structures that were built in compliance with a particular map or to structures that purchased a policy prior to the revised map’s effective date, and whos purchasers have maintained continuous coverage since the map change. Grandfathering can apply to both the flood zone changes, resulting in zone grandfathering, and the one percent flood elevation changes, resulting in elevation grandfathering. The most common form of zone grandfathering occurs when a policyholder was once outside the SFHA and paying a lower rate, but is now included in the SFHA because of to remapping. In this case, non-SFHA to SFHA zone grandfathered policies do not pay the lower preferred risk policy (PRP) non-SFHA rate. Instead they pay an average rate, called the X zone standard rate that is sufficient for the spread of risk for all non-SFHA to SFHA zone grandfathered structures. Effective October 1, 2010, FEMA introduced a 2-year PRP extension, allowing structures that were recently mapped into the SFHA to use PRP rates for a limited, 2-year period before moving to the higher X zone standard rate. This was achieved through a cross-subsidy by increasing the premium for all PRP policyholders. Approximately 90,000 PRP 2-year extension policies have been issued. There are approxi- mately 1.6 million non-SFHA PRP policyholders (Andy Neal, FEMA, personal communication, August 9, 2012). Zone grandfathering can also occur when a policy moves from a lower risk zone to a higher risk zone, for example, from an AE zone to a VE zone. In this case, the grandfathered structure pays the lower AE zone rate. Elevation grandfathering applies when new maps increase the elevation of mapped 1 percent flood without chang- 16  TheBiggert-Waters Act defines severe repetitive loss properties as those that have “incurred flood-related damage (i) for which 4 or more separate claims payments have been made under flood insurance coverage under this title, with the amount of each claim exceeding $5,000, and with the cumulative amount of such claims payments exceeding $20,000; or (ii) for which at least 2 separate claims payments have been made under such coverage, with the cumulative amount of such claims exceeding the value of the insured structure.”

OCR for page 63
72 LEVEES AND THE NATIONAL FLOOD INSURANCE PROGRAM ing the zone. For example, a property that was 3 feet above the 1 percent flood elevation according to the previous FIRM and is now only 1 foot above the 1 percent flood elevation according to the revised FIRM would be eligible to use the +3-feet rate. FEMA is currently engaged in a study to quantify the number of zone and elevation grand- fathered policies (Andy Neal, FEMA, personal communication, August 9, 2012). The use of grandfathered rates does allow the grandfathered properties pay a lower rate than they would pay if the property were truly risk rated, and hence grandfathering implies that not all properties having the same NFIP premium are actually of the same risk level. This is a move away from risk-based pricing. Because properties receiving a higher risk classification under the Map Modernization program continue to receive insurance at the lower, pre-Map Modernization rate in spite of their now-recognized higher risk, the effect is similar to subsidiza- tion in that some properties are not charged their risk-based price, but rather the price that would have occurred if they had remained in their previous risk classification. Moreover, the grandfathered status continues indefinitely, even upon sale of the property (GAO, 2008). Expected Annual Income from Premiums from the NFIP To ensure an adequate aggregate expected annual income from premiums for the NFIP as a whole, the pre- miums are adjusted for the discounted risk classes. First, the aggregate expected full-risk premium anticipated to be received using the rating formula for the nondiscounted classes is calculated, the calculated base rates for the subsidized classes is added in, and then this value is subtracted from a calculated target “average historical loss year” value. The resulting difference is then used to determine an upward adjustment of rates among the discounted classes so as to accomplish a balance between the total anticipated income received and the targeted average historical loss year cost set by the NFIP. Note that by this process, some properties will be paying more than their actuarial rate. There is a problem with this method, however. As noted by the Government Accountability Office (GAO) the“average historical loss year value is calculated using loss data over a limited loss experience time frame coupled with extreme loss values, such as the 2005 loss year (Box 5-3), which are heavily discounted when input into the historical loss year calculation (GAO, 2008).17 Thus, this historical loss year average value may balance the books in an historically typical year, but this value is neither a reasonable estimate of either actual historical average yearly losses (which now do include catastrophic loss years such as 2005), nor a reasonable long-term estimate of the actual expected future loss because it diminishes the impact of real or anticipated catastrophic loss events that have hap- pened in the past and can be expected to happen again in the future.18 The method used to calculate the average historical loss year ensures that in the long run there will be inadequate premiums collected to cover costs in significant flood years.19 Even with the mandate in the Biggert-Waters Act that catastrophic loss years be fully incorporated into the NFIP calculation of the “historical loss year average,” there is still a potential long-run shortage becausee larger but less frequent catastrophic floods (500-year floods, 1,000-year floods, etc.) may not have been recorded in the flood record at a particular location (although their likelihood might be modeled), yet these events might occur in the future. Hence, the historical average does not reflect the real expected long-term average loss. NFIP’s Approach to Insurance Compared with Private Insurer Approaches Insurance companies started to appear in the United States in the mid-1700s but did not expand rapidly, and encountered many bankruptcies at first. Early insurers were small, localized in geographic coverage, and usually 17  For example, since 2007 the catastrophic losses from the 2005 year were given only a 1 percent weight in the historical loss averaging process in “an attempt to reflect the events of 2005 without allowing them to overwhelm the pre-Katrina experience of the Program” (Hayes and Neal, 2011). 18  The full-risk class and the risk classes outside the 100-year flood level have premiums calculated using formula (5-2) and already incor- porate the possibility of catastrophic risk. 19  FEMA itself recognized the inadequacy of using the mandated “historical average loss year” goal in premium setting even prior to the dramatic 2005 losses. From the NFIP’s 2004 Annual Rate Review, “[t]he underwriting experience period has, to date, included 7 heavy-loss years. Despite these heavy-loss years, the absence of extremely rare but very catastrophic loss years leads to the conclusion that the historical average is less than what can be expected over the long term” Hayes and Sabade (2004).

OCR for page 63
INSURING PROPERTIES BEHIND LEVEES 73 BOX 5-3 The 2005 Loss Year The record-breaking hurricane season of 2005 included five Category 4 and 5 hurricanes—Dennis, Emily, Katrina, Rita, and Wilma—that battered the Gulf Coast of the United States along with neighboring countries. In addition, the Los Angeles River experienced flooding in January 2005, and there was flooding in New England and the Mid-Atlantic during October 2005 caused by excessive rainfall due to a subtropical depression, Tropical Storm Tammy, and another tropical disturbance. The NFIP experienced an unprec- edented amount of paid losses, 212,235 claims, totaling over $17 billion (Figure 5-1). FIGURE 5-1 Loss dollars paid by the NFIP per calendar year. SOURCE: http://www.fema.gov/policy-claim-statistics-flood-insurance/policy-claim-statistics- flood-insurance/policy-claim-13/loss. specialized in a single type of insurance (e.g., fire insurance). Thus, initially, private localized insurers were reluc- tant to include flood coverage because of potential highly correlated losses and catastrophic total loss payment possibility (floods would simultaneously affect many of their customers, undermining the law of large numbers and central limit theorem for advantageous risk pooling) and because of the lack of ability to calculate appropriate actuarial rates that were protective of the insurer and at the same time affordable to the policyholder. Additionally, moral hazard and adverse selection considerations would imply that in a voluntary market the higher risk properties would disproportionately seek insurance and those that had insurance would be less motivated to take precautionary measures if they had insurance coverage. Nevertheless, in the late 1920s, there were still several dozen insurers willing to sell flood insurance, perhaps in part because USACE had declared in 1926 that improvements had been made to the levee system sufficient that they could now “prevent the destructive effects of floods” (Daniel, 1977; see also Moss, 1999; King, 2005). The problems with private companies writing flood insurance policies were brought home dramatically by the great Mississippi River flood of 1927 and other riverine flooding in 1928, so that all of these several dozen insurers

OCR for page 63
86 LEVEES AND THE NATIONAL FLOOD INSURANCE PROGRAM TABLE 5-2 Number of new policies and duration from 2001 through 2009. Market Penetration and Compliance Under the NFIP, flood insurance is available to both those subject to the MPR and to any property owner living within the NFIP community, whether in or out of the SFHA or in front of or behind a levee. From a flood risk management perspective, the NFIP encourages all who live in areas subject to any level of flooding to purchase insurance so as to provide loss coverage to them if their covered property is damaged from a flood event and to reduce the cost to the federal government for their recovery. Therefore, FEMA is interested in knowing what percentage of those living in flood-prone areas have insurance, that is the market penetration: 35 Number of buildings in SFHA eligible for insurance purchase that have policies a Market Penetration = Number of buildings in SFHA eligible for insurance purchase i Compliance with the NFIP MPR is a subset of the market penetration: 35  “Insurance may be written on any building eligible for coverage with two or more outside rigid walls and a fully secured roof that is af- fixed to a permanent site. Buildings must resist flotation, collapse, and lateral movement. The structure must be located in a community that participates in the NFIP . . . . Buildings entirely over water or principally below ground, gas and liquid storage tanks, animals, birds, fish, aircraft, wharves, piers, bulkheads, growing crops, shrubbery, land, livestock, roads, machinery or equipment in the open, and most motor vehicles are not insurable through the NFIP” (FEMA, 2011).

OCR for page 63
INSURING PROPERTIES BEHIND LEVEES 87 FIGURE 5-4 National Flood Insurance Program total number of policies in force, by state, as of September 30, 2012. SOURCE: http://www.fema.gov/policy-claim-statistics-flood-insurance/policy-claim-statistics-flood-insurance/policy- claim-13-2, accessed January 11, 2012. Number of buildings in SFHA eligible for insurance with federal mortgages that have flood insurance Compliance = Number of buildings in SFHA eligible for insurance with federal mortgages g e In a scenario with full lender application of the regulation, compliance with the MPR would be approximately 100 percent. There have been several analyses over the last two decades of the market penetration of the NFIP. According to Dixon et al. (2006), a 1999 report by PricewaterhouseCoopers estimated that 28 percent of residential structures in the SFHA carried flood insurance. In 2006, RAND estimated that 49 percent of single-family homes carried flood insurance (Dixon et al., 2006). Preliminary results from a J. Walter Thompson (JWT) analysis currently under way estimate that nationwide only approximately 30.6 percent of identified structures in the SFHA are insured.36 Because each of the three analyses used a different set of the total structures in the SFHA, it is difficult to compare the results, although the ongoing JWT analysis appears to provide the highest degree of resolution. Market penetration of flood insurance policies across the country varies with, for example, geography (coastal versus inland communities) and frequency of major flooding events in an area (Figure 5-4). The JWT analysis indicates that South Carolina had the highest penetration rate in the SFHA at 46.69 percent. Other coastal and flood-prone states have high penetration in the SFHA rates, such as Louisiana (43.69 percent), which is consis- tent with its history of high penetration particularly in the New Orleans area (Box 5-4, Louisiana), Florida (37.75 percent), New Jersey (37.94 percent), and California (33.49 percent). The development of statistics for compliance with the MPR is confounded by several factors. As indicated, some penetration studies focused on single-family homes whereas others included all residential structures or 36  A market penetration and mortgage compliance analysis, by state, in preparation for FEMA by J. Walter Thompson (JWT) advertising company. The committee used the version from January 2013, which is heretofore referred to as the “JWT analysis” in the report.

OCR for page 63
88 LEVEES AND THE NATIONAL FLOOD INSURANCE PROGRAM BOX 5-5 New York’s Market Penetration Experience Although detailed analyses of market penetration have not been conducted in areas impacted by Hurricane Sandy, preliminary figures indicate that approximately 62,000 of the owners of the 116,000 resi- dential structures damaged by Sandy in New York City held NFIP policies. This equates to approximately a 54 percent market penetration rate. In the devastated Breezy Point area of Queens, New York, owners of 1,394 of the 2,150 structures damaged submitted claims, approximately a 64 percent market penetration rate (Jeffrey Woodward, FEMA, personal communication, February 27, 2013). addresses, parcels, etc.37 Many structure owners do not carry mortgages and others do not insure with federally regulated lenders. Both of these groups are exempt from the MPR and, combining the results of several studies, represent somewhere between 15 and 40 percent of the residential structure ownership in the SFHA. Tobin and Calfee (2005) reported that 77 percent or more of the debt in the SFHA was associated with lenders subject to the MPR, whereas Dixon et al. (2006) assumed that 85 percent of the single-family homes in the SFHA were subject to the MPR. Dixon et al. (2006) also assumed that policies sold by private insurers, in addition to those sold under the NFIP, represented an additional 7 percent in policies, and that 38 percent of those not required to buy polices did purchase them. By applying these assumptions to the probable number of single-family homes with mortgages, by region, Dixon et al. identified compliance rates ranging from 43 percent in the Midwest to 88 percent in the West with a national average of 78 percent. The ongoing JWT analysis indicates, based on a sample of structures in the SFHA, that national compliance may be in the 50-60 percent range, with several states having compliance rates lower than 20 percent and other, high-risk states with compliance higher than 80 percent (Louisiana) (Box 5-5). According to Dixon et al. (2006), a 1999 PricewaterhouseCoopers study of purchase of flood insurance, assuming that 10 percent of homes compliant in 1 year are not compliant in the following year, resulted in MPR compliance rates between 55 to 60 percent. However, more recent retention data by Michel-Kerjan et al. (2012), noted above, found that the drop rate was considerably higher and would thus drive down the compliance rates. Review of all these studies indicates that estimates of both market penetration in the SFHA and compliance with the MPR have varied considerably and are very sensitive to the assumptions made in the study process. National market penetration most likely lies somewhere between 30 and 50 percent, with the most recent JWT analysis, which is seemingly the most definitive study, indicating that national market penetration is the lower value. National MPR compliance has been estimated to be between 50 and 78 percent. The JWT study, again which appears to be the more definitive study because of a higher degree of resolution, presents the values at the lower end of this range. For those who live in areas subject to flood risk both in front of (SFHA) and behind levees, insurance pro- vides an effective and efficient means of transferring financial risk and thus reducing the exposure of individuals and businesses to catastrophic consequences and the need to call on the government for assistance. In 1973, the Congress found that for the NFIP to maintain fiscal integrity and protect federal interests, it needed to mandate insurance for those in the SFHA. However, in spite of four decades of effort, the program has yet to achieve high levels of compliance. Given the emphasis that has been placed on market penetration and compliance with the MPR in FEMA’s risk communication activities, the efficacy of the mandatory purchase requirement is called into question. The current policy of mandatory flood insurance purchase appears to be ineffective in achieving widespread purchase of NFIP flood insurance policies. 37  Dixon et al. (2006) found that residential structures make up 87percent of parcels in the SFHAs examined, and 57 percent of the total were single-family homes.

OCR for page 63
INSURING PROPERTIES BEHIND LEVEES 89 Multiyear policies tied to the property rather than the individual, thus avoiding policy cancellation, have been proposed as one solution (NRC, 2012). FEMA views the fault to rest with the enforcement by those agencies charged with lender oversight as problematic (GAO, 2002). Discussions with personnel in the trade indicate that budget and manpower realities limit the number of oversight inspections. In addition, in the committee’s judgment, enforcement does not appear to have high priority within the lender community because it is seen as a tangential aspect of mortgage oversight. However, lenders assert that they are increasing their success (GAO, 2002). Retention statistics would also indicate lack of progress, in either oversight or enforcement. The rate of compliance with the mandatory purchase requirement indicates challenges with lender enforcement and federal oversight of this lending. On the other hand, statistics indicate that where residents are able to identify their own risk, some voluntarily acquire insurance. The fact that 41 percent of new policies between 2001 and 2009 were outside the SFHA, and were not mandated, gives support to this view (Table 5-2). Mandatory Purchase Behind Levees: The Current Approach Verses a Modern Risk Analysis Beginning in the early 2000s, FEMA took steps to modernize FIRMs via “Map Mod.” During Map Mod, FEMA discovered that because FIRM maps carried forward certification from previous years, many levees previously thought to be able to withstand the one percent annual chance flood were not able to withstand this event. Thus, increased attention was placed on ensuring that the accreditation status for all levee systems in the NFIP was up to date and accurate through Procedure Memorandum 34 and the Provisionally Accredited Levee (PAL)38 designation (Chapter 2). The process of ensuring that all one percent annual chance levees are properly constructed, certified, and accredited continues today, as affected communities are faced with the possibility of movement into the SFHA and the MPR. Often this is the catalyst for controversy as communities and levee owner/operators repair levees to the one percent annual chance standard to avoid MPR (Box 5-6). Thus, the one percent annual chance flood has become a de facto design standard that promotes the idea that if you live behind a levee you are “safe”—an unintended consequence of the MPR. During Hurricane Sandy, coastal V zone flows along open shorelines resulted in significant losses. Also, many residents and insurers learned that many areas outside the SFHA were subject to catastrophic loss. It is important to realize that overtopping of levees can produce flood flows behind levees whose velocity can match the coastal V zone flows on open shorelines, and results in damages similar to those associated with V zone flows. Mandatory purchase requirements have led many property owners to perceive that if they are not mandated to have insurance they are not susceptible to damage from floods. The committee examined the extension of mandatory purchase to include areas behind levees and took into consideration several factors. In the current NFIP approach to analyzing risk behind levees, areas behind accred- ited levees are considered to be at a low to moderate risk and insurance rates are priced accordingly. However, this evaluation is an incomplete reflection of flood risk (Chapter 3). Thus, extending the mandatory purchase requirement behind accredited levees would be extending the federal government’s exposure without a complete understanding of the risk that exists and the possible consequences. Furthermore, without the benefit of information provided in the modern risk-based analysis, those living behind levees will, like those in the SFHA, find it difficult to understand the reasoning behind the mandatory purchase requirement and the premiums associated with this requirement; a difficulty that is promoted by the idea that when you live behind an accredited levee, you are “safe.” The post-Katrina rise in insurance purchases outside of the SFHA may, if it continues, indicate that public awareness of flood threats is increasing and that given credible risk information, owners may voluntarily move to such purchases, thereby avoiding the necessity to extend the administrative challenge of the mandatory purchase requirement. Experience since Hurricane Katrina indicates that where residents are able to identify their own risk, some voluntarily acquire insurance (Figure 5-3). Furthermore, 41 percent of new policies between 2001 and 2009 were outside the SFHA and thus were not mandated. This gives support to the view that credible risk information 38  PAL is applied to a previously accredited levee on an effective FIRM for which FEMA is awaiting data or information that will show compliance with NFIP regulations.

OCR for page 63
90 LEVEES AND THE NATIONAL FLOOD INSURANCE PROGRAM BOX 5-6 Challenges in East St. Louis The American Bottom floodplain is home to approximately 156,000 people and businesses that employ 56,000 workers in the St. Louis area. The five levee systems (Figure 5-5) protecting the region were built in the 1940s and 1950s and, at the time of construction, were designed to provide protection against a 500-year flood. The levees were last upgraded in the 1960s. Since the flood of 1993, underseepage and sand boils have been recognized as a problem for these levees (FEMA, n.d.b). In 2004 and 2005, FEMA began revising and updating FIRMs that include levees in and around the St. Louis area. In 2007 and during the revision process, FEMA notified the impacted communities that five of the levee systems in the eastern suburbs (Metro East St. Louis) could no longer be depicted as accred- ited on revised flood maps without demonstrated compliance with the design requirements set forth in 44 CFR §65.10 (Chain of Rocks Levee, Fish Lake Drainage and Levee District, Metro East Sanitary District [MESD], Prairie du Pont Levee and Sanitary District, and Wood River Drainage and Levee District). If no longer accredited, the area was to be included by FEMA as in the SFHA (i.e., mapped as without a levee) and the mandatory purchase requirement would apply to the eastern suburbs. Inclusion in the SFHA and the MPR can have a significant impact on housing cost, economic de- velopment, and insurance rates. The East-West Gateway Council of Governments estimated mandatory insurance costs paid by residents and businesses in the American Bottom floodplain as a result of imple- mentation of the MPR to be approximately $50 million dollars per year (East-West Gateway Council of Governments, 2010). Thus, the community faced the question of how best to restore the levee systems to a level that provided adequate protection and FEMA accreditation as well as how to finance this restoration. This was even more challenging given the high poverty rate in the East St. Louis area—an estimated 41 percent of persons in East St. Louis were below the poverty level from 2006 through 2010 (U.S. Census Bureau, n.d.). In 2009 a local sales tax was passed, to collect the estimated $160 million dollars to repair the levees. As a result of legal action taken by local sponsors, FEMA withdrew issuance of a preliminary FIRM and is awaiting information from the sponsors as to what actions they propose to take to remediate identified deficiencies. The sponsors, using the $160 million dollars raised for the project, have proposed remediating the deficiencies to the point where the levee system can successfully pass the one percent chance flood and meet FEMA’s requirements for accreditation.  However, because the levee was originally built by US- ACE and then transferred to the local sponsors as a 500-year system, the sponsors must obtain approval from USACE to carry out any local work on the levee.1 USACE has indicated that the levee deficiencies should be remediated in a federal–local cost-shared project to meet the standards for a 500-year levee and is taking steps to obtain the federal funding. If the project were to proceed as a local project, the spon- sors must assure USACE that any work done to meet the one percent annual chance flood accreditation requirement will not interfere with the ability to eventually repair the levee system to the 500-year level. If federal funding can be obtained, USACE would move ahead in a process that would carry out the required work for a 500-year levee and, as part of this work, obtain accreditation from FEMA for one percent annual chance flood protection (Colonel Christopher Hall, USACE, personal communication, February 7, 2013) 1  33 U.S.C. §408 (Section 408) provides that any proposed modification to an existing USACE project must obtain permission from the Secretary of the Army by demonstrating that such proposed alteration or permanent use and oc- cupation of the federal flood control project is “not injurious to the public interest and will not impair the usefulness of such work.”

OCR for page 63
INSURING PROPERTIES BEHIND LEVEES 91 BOX 5-6  continued FIGURE 5-5 Map of the Metro East Levee System. SOURCE: USACE (2010).

OCR for page 63
92 LEVEES AND THE NATIONAL FLOOD INSURANCE PROGRAM will prompt voluntary policy purchase, although retention rates also raise questions about property owner long- term commitment (Table 5-2). To date, relying on federal supervisory agencies to oversee and lenders to require purchase has only margin- ally worked in the SFHA where the flood hazard is relatively obvious. There is little to suggest that most property owners in areas behind levees, if mandated to purchase insurance, would move to obtain it. Given the number of people behind levees, this would create an even greater challenge for lenders and oversight agencies. In summation, instituting this requirement would be imprudent where there is (1) limited and conflicting evidence to support the effectiveness of the enforcement mechanism, (2) a communication impediment to over- come that is associated with the without-levee approach (when you live behind a levee you are “safe”), and (3) an incomplete understanding of risk behind a levee.. At this time, there is no sound reason to institute mandatory purchase of flood insurance in areas behind accredited levees. This conclusion is not to be interpreted as the committee not supporting the purchase of flood insurance behind levees and throughout the SFHA—quite the opposite. Other paths to achieve increased insurance purchase are critical: for example, effective communication of the risk behind levees; strong community support for voluntary acquisition of insurance behind levees; meaningful engagement with insurance companies and agents, lenders, and real estate professionals; and a modern risk-based flood risk analysis generating more accurate pricing and flood risk information. Regardless of location (behind a levee or not behind a levee), because of the catastrophic nature of flooding, it is in a property owner’s and, when applicable, lender’s best interest to have the financial security provided by flood insurance coverage in the SFHA, assuming the cost of coverage is not an issue. As noted in Chapter 3 and above, a modern risk-based analysis will lead to a better informed, more differentiated assessment of properties’ susceptibility to flood damage. If insurance rates are priced accordingly, this would provide for more granular policy pricing that better reflects the actual risk to the property. This includes areas behind nonaccredited levees. For example, an area with a 50-year levee will be given some credit for risk reduction and the insurance rates will be priced accordingly. A perception exists that if the NFIP simply sold more policies the program’s fiscal issues (e.g., debt to the U.S. Treasury resulting from the 2005 loss year) would be solved. 39 However, simply adding new policies doesn’t automatically improve the program’s fiscal soundness. Adding new polices could lower the program’s overall risk if the policies increase the program’s diversification of risk acquired through those policies. However, adding new policies very well may increase the NFIP’s overall risk through concentration of policies in, for example, high-risk areas such as the SFHA. Furthermore, adding new policies would also increase the program’s exposure to risk accumulation, or the additive risk of multiple events occurring at the same time. If NFIP policies continue to be concentrated in the same high-risk areas around the country, the diversification of risk and the fiscal soundness of the program will remain relatively unchanged. Upon implementation of a modern risk analysis, insurance rates will more accurately reflect flood risk behind a levee. If the insurance rate goes up, it indicates that the property is at more peril than previously understood to be; if the insurance rate goes down, it indicates that the property is in less peril than previously understood to be. Thus, FEMA’s moving to a modern risk-based analysis and pricing premiums accordingly will have an impact on NFIP policy purchase. More accurate pricing and more information created by a modern risk analysis has the potential to prompt additional policy purchases. Property owners would be more favorably inclined to buy flood insurance if individual risk is well known and insurance rates are priced to match the probability of flooding and the financial impact of flooding events. This includes areas behind levees, both accredited and not accredited, in the SFHA at large, and also non-SFHA areas. It is important to note that the Biggert-Waters Act directs full, risk- 39  Asnoted by Baranoff (2009), the assumption that a large number of independent exposure units in a risk pool leads (via the law of large numbers and the central limit theorem) to an increased ability of the insurer to handle more risk at greater profitability. This is because the contingency loading for risk bearing added to the expected loss can decrease as the number of exposure units increase by the central limit theorem. This common perception, which is the mathematical underpinning of private insurance, can lead one to believe that a larger number of policies in the NFIP would lead to a similar result. Major differences, however, are that the NFIP premiums are not set in a fiscally sound manner (because of discounts, subsidies, no contingency reserve, etc.), and so, increasing the scale of an already cash-losing operation will not improve its position. Moreover, if the increase in policies is due to higher risk properties entering the NFIP (termed adverse selection, see Baranoff [2009]), then more policies can actually worsen the fiscal soundness of the NFIP.

OCR for page 63
INSURING PROPERTIES BEHIND LEVEES 93 based premiums and the elimination of certain discounts, also promoting a change to the NFIP that will impact policy purchase. Lenders may choose to protect their portfolio regardless of the role of the NFIP (Buckley, 2010). It is very likely that if a lender has more information through a risk-based analysis, the lender is more aware that requiring flood insurance purchase is in line with their fiduciary responsibility. Regardless of the MPR, lend- ers currently can require insurance regardless of any regulatory responsibility as a function of their normal underwriting process. Furthermore, the recent NFIP reauthorization includes a provision requiring that lenders provide a disclosure of the availability of flood insurance to all purchasers at loan closing or settlement 40—real estate agents might realize that disclosure of flood risk and promoting the purchase of a flood insurance policy is in their best interest. The reasons why residents in flood-prone areas do not buy flood insurance from the NFIP have been studied both theoretically and empirically. Both behavioral and economic reasons for nonpurchase include the avail- ability of aid from FEMA and other sources (i.e., charity hazard), underestimation of the risk probability, lack of knowledge of availability of the product and its price, the feeling that the coverage is not extensive enough, and several other rationales including experience with previous floods that did little damage, leading them to feel safe for the future (Kunreuther 1984, 1996; King, 2009). Effective communication of the risk behind levees is critical, for example, to overcome the impression that when you live behind an accredited levee you are “safe,” an impres- sion that is likely linked to policy purchase. Nuances in risk communication confound the premise that greater risk information will prompt additional policy purchases. For example, despite the positive association between risk perception and protective actions, awareness does not necessarily lead to actions to mitigate risk because of reasons judged to be more important, such as personal experience and evaluation of protective actions themselves. (See Chapter 7 for an expanded discussion.) An increase in the purchase of policies, particularly in non-high-risk areas, will diversify the program’s risk and ultimately generate a more fiscally sound NFIP. Thus, the necessity of the MPR in the SFHA may change, along with the decision to waive the MPR behind an accredited levee. A modern risk-based analysis has the potential to impact the purchase of flood insurance, diversify the NFIP’s exposure to flood risk, and generate a fiscally sounder program. Once the risk-based approach has been put in place and matures, FEMA should review and study the necessity of the mandatory purchase requirement, behind levees and throughout the SFHA. REFERENCES Ai, J., and P. L. Brockett. 2008. Insurance pricing/nonlife. Pp. 922-933 in Encyclopedia of Quantitative Risk Assessment and Analysis, E. Melnick and B. Everitt, eds. Chichester, UK: John Wiley and Sons, Ltd. American Academy of Actuaries, Actuarial Soundness Task Force. 2012. A Public Policy Special Report: Actuarial Sound- ness. Available online at http://www.actuary.org/files/publications/Actuarial%20Soundness%20Special%20Report%20 FINAL%205%2010%2012.pdf. Accessed September 21, 2012. ASCE (American Society of Civil Engineers). 2007. The New Orleans Hurricane Protection System: What Went Wrong and Why. Washington, DC: ASCE. Baranoff, E. G, P. Brockett, and Y. Kahane. 2009. Risk Management for the Enterprise and Individuals. Available online at http://www.flatworldknowledge.com/. Accessed February 27, 2013. Brockett, P. L., and R. C. Witt. 1982. The underwriting risk and return paradox revisited. Journal of Risk and Insurance 49: 621-627. Browne, M. J., and M. Halek. 2010. Managing flood risk: The National Flood Insurance Program and alternatives. Chapter 5 in Public Insurance and Private Markets, Jeffrey R. Brown, ed. Washington, DC: American Enterprise Institute for Public Policy Research. Browne, M. J., and R. Hoyt. 2000. The demand for flood insurance: Empirical evidence. Journal of Risk and Uncertainty 20: 291-306. 40  Notice of flood insurance availability is required under the Real Estate Settlement Procedures Act (RESPA); see Section 100222 of the Biggert-Waters Flood Insurance Reform and Modernization Act.

OCR for page 63
94 LEVEES AND THE NATIONAL FLOOD INSURANCE PROGRAM Buckley, M. 2010. The NFIP Outside the Box. Background Reading for the Gilbert F. White National Flood Policy Forum, 2010 Assembly, Managing Risk to Humans and to Floodplain Resources. Available online at http://www.asfpmfounda- tion.org/pdf_ppt/2010_GFW_Forum_Background_Reading.pdf. Accessed February 21, 2013. Casualty Actuarial Society. 1990. Foundations of Casualty Actuarial Science. New York: Casualty Actuarial Society. Czajkowski, J., H. Kunreuther, and E. Michel-Kerjan. 2012. A Methodological Approach for Pricing Flood Insurance and Evaluating Loss Reduction Measures: Application to Texas. Available online at http://opim.wharton.upenn.edu/risk/ library/WhartonRiskCenter_TexasFloodInsurancePricingStudy.pdf. Accessed April 4, 2012. Daniel, P. 1977. Deep’n as It Come: The 1927 Mississippi River Flood. New York: Oxford University Press. Dixon, L., N. Clancy, S. A. Seabury, and A. Overton. 2006. The National Flood Insurance Program’s Market Penetration Rate: Estimates and Policy Implications. Washington, DC: American Institutes for Research. East-West Gateway Council of Governments. 2010. Additional Flood Insurance Costs Resulting from the Federal Emergency Management Agency Flood Insurance Rate Map Modernization Program in the American Bottom. Available online at http:// www.floodpreventiondistrict.org/wpcontent/uploads/2011/03/InsuranceCostsReport.pdf. Accessed February 21, 2013. FEMA (Federal Emergency Management Agency). 2006. Designing for Flood Levels Above the BFE. Available online at http://www.fema.gov/library/file?type=publishedFile&file=fema549_apndx_e_ra8.pdf&fileid=143da3a0-0316-11dc- a1f1-000bdba87d5b. Accessed September 5, 2012. FEMA. 2007. Mandatory Purchase of Flood Insurance Guidelines. Washington, DC: FEMA. FEMA. 2011. National Flood Insurance Program: Answers to Questions About the NFIP. Available online at http://www.fema. gov/library/viewRecord.do?fromSearch=fromsearch&id=1404. Accessed February 15, 2013. FEMA. n.d.a. National Flood Insurance Program Community Rating System. A Local Official’s Guide to Saving Lives, Pre- venting Property Damage, and Reducing the Cost of Flood Insurance. FEMA 573. Available online at: http://www.fema. gov/library/viewRecord.do?id=3655. Accessed December 7, 2012. FEMA. n.d.b. Mapping the Risk, Flood Map Modernization, Metro East—Frequently Asked Questions. Available online at https://www.fema.gov/pdf/about/regions/regionv/faq_east_stlouis.pdf. Accessed February 21, 2013. GAO (Government Accountability Office). 2002. Flood Insurance: Extent of Noncompliance with Purchase Requirements Is Unknown. Washington, DC: GAO. GAO. 2008. Flood Insurance: FEMA’s Rate-Setting Process Warrants Attention. Available online at http://www.gao.gov/prod- ucts/GAO-09-12. Accessed August 27, 2012. Hayes, T. L., and D. A. Neal. 2011. National Flood Insurance Program Actuarial Rate Review in Support of Recommended October 1, 2011 Rate and Rule Changes. Available online at http://www.fema.gov/library/viewRecord.do?id=4853. Accessed June 12, 2012. Hayes, T. L., and S. S. Sabade. 2004. National Flood Insurance Program Actuarial Rate Review. Available online at http://www. fema.gov/library/viewRecord.do?id=2377. Accessed May 28, 2012. King, R. 2005. Federal Flood Insurance: The Repetitive Loss Problem. Congressional Research Service Report RL32972. Available online at http://www.fas.org/sgp/crs/misc/RL32972.pdf. Accessed February 26, 2013. King, R. O. 2009. National Flood Insurance Program: Background, Challenges, and Financial Status. Congressional Research Service 7-5700, R40650. Available online at https://opencrs.com/document/R40650/. Klugman, S. A., H. H. Panjer, and G. E. Willmot. 2008. Loss Models: From Data to Decisions, 3rd Rev. Ed. Chichester, UK: John Wiley and Sons, Ltd. Kunreuther, H. 1984. Causes of Underinsurance Against Natural Disasters. Geneva Papers on Risk and Insurance 31: 206-220. Kunreuther, H. 1996. Mitigating Disaster Losses Through Insurance. Journal of Risk and Uncertainty 12: 171-187. Mathewson, S. B. 2011. Actuarial Rates in the Context of National Flood Insurance Program (NFIP) Hazard Mapping. Avail- able online at http://www.actuary.org/files/publications/NCOIL_Simons_NFIP_Mapping_20110717_final.pdf. Accessed September 5, 2012. Michel-Kerjan, E. O. 2010. Catastrophe economics: The National Flood Insurance Program. Journal of Economic Perspectives 24: 165-186. Michel-Kerjan, E., S. L. de Forges, and H. Kunreuther. 2012. Policy tenure under the U.S. National Flood Insurance Program (NFIP). Risk Analysis 32: 644-658. Moss, D. A. 1999. Courting Disaster? The transformation of federal disaster policy since 1803, Chapter 8 in The Financing of Catastrophe Risk, Kenneth Froot, ed. Chicago: University of Chicago Press. Mowbray, A. H. 1914. How extensive a payroll exposure is necessary to give a dependable pure premium. Proceedings of the Casualty Actuarial Society 1: 24-30.

OCR for page 63
INSURING PROPERTIES BEHIND LEVEES 95 Nechamen, B., and M. Inderfurth. 2012. Summary of Contents: Biggert-Waters Flood Insurance Reform Act of 2012, H.R. Rep. No. 112-4348, Title III (Pages 521-576). Available online at http://www.floods.org/ace-files/documentlibrary/2012_ NFIP_Reform/2012_NFIP_Reform_Act_ASFPM_Summary_of_Contents.pdf. Accessed February 7, 2013. NRC (National Research Council). 2012. Dam and Levee Safety and Community Resilience: A Vision for Future Practice. Washington, DC: The National Academies Press. Rodermund, M. 1990. Introduction in Foundations of Casualty Actuarial Science. New York: Casualty Actuarial Society. Tobin, R. J., and C. Calfee. 2005. The National Flood Insurance Program’s Mandatory Purchase Requirement: Policies, Pro- cesses, and Stakeholders. Washington, DC: American Institutes for Research. U.S. Congress, House. 1966. A Unified National Program for Managing Flood Losses. 89 th Cong., 2d Sess. H. Doc. 465. USACE (U.S. Army Corps of Engineers). 2010. Metro East Levee System Maps. Available online at http://www.mvs.usace. army.mil/me_levees/maps.html. Accessed November 14, 2012. U.S. Census Bureau. n.d. State and County Quick Facts, East St. Louis, Illinois. Available online at http://quickfacts.census. gov/qfd/states/17/1722255.html. Accessed February 27, 2013. Witt, R. C., and A. M. B. Hogan. 1993. Economic, legal and social factors influencing insurance costs and prices. Journal of Insurance Regulation 11(3):314.

OCR for page 63