Congressional Budget Office
Director’s Blog

In 2005, the National Flood Insurance Program (NFIP), administered by the Federal Emergency Management Agency (FEMA), experienced an unprecedented volume of claims resulting from hurricanes Katrina, Rita, and Wilma. Total payments on those claims were greater than the total for all of the program’s previous years combined and led the NFIP to borrow about $17 billion from the Treasury. The 2005 losses highlighted a number of factual and policy questions—discussed in a CBO paper released today—about the NFIP’s financial health, including the actuarial soundness of the premium rates charged on policies that are not explicitly subsidized and the cost of paying claims for properties that have suffered multiple flood losses.

As of July 31, 2009, the NFIP had 5.6 million policies, with a total insured value of $1.2 trillion and total premiums of $3.1 billion. For most of those policies, about 80 percent, FEMA charges “full-risk” premium rates, which it considers to be actuarially sound (that is, sufficient to cover the expected value of flood claims and administrative costs). About 20 percent of the premium rates are explicitly subsidized under current law; those explicit subsidies give the NFIP a built-in actuarial deficit of about $1.3 billion per year, by CBO’s estimate. Those policies mainly cover older structures in areas at high risk of flooding.

Are FEMA’s full-risk premium rates actuarially sound?

Historically, the NFIP’s full-risk premiums have been too low to cover the flood claims and administrative costs of the policies insured at those rates. Between 1978 and 2004, premiums for such policies totaled $10.2 billion in nominal dollars, while claims and expenses totaled $10.7 billion. The result was a loss of $0.5 billion over that period on policies with full-risk rates. Taking into account the large losses of 2005, income was about half of costs over the 1978-2006 period—total premiums were $12.6 billion while claims and expenses were $24.2 billion.

Previous experience does not necessarily imply that current premium rates are too low, however, because rates have risen over time and because the frequency of future catastrophic years like 2005 is highly uncertain. In part because of that uncertainty, CBO does not have enough information about the current distribution of flood risks to calculate whether the present full-risk premiums are actuarially adequate.

Further, analyzing the methods that FEMA uses to set the full-risk rates also does not provide definitive answers. Some aspects of those methods tend to contribute to an actuarial surplus—the primary one being the additional 10 percent that FEMA includes in the rates in high-risk areas (20 percent in high-risk coastal areas) as a safety margin for uncertainty. Other aspects of the agency’s rate-setting methods tend to contribute to an actuarial deficit. FEMA is not reviewing its flood maps every five years as required by law, and some older maps do not reflect significant changes in local conditions, such as coastal erosion, which can increase the probability of flooding. In addition, evidence suggests that climate change has increased the risk of flooding from rivers and perhaps also from coastal storms, making FEMA’s models of flood frequencies out of date. Those issues may warrant attention regardless of the overall adequacy of the program’s full-risk rates.

To what extent are the NFIP’s losses attributable to properties that have experienced multiple floods?

Currently insured repetitive-loss properties (RLPs)—defined by FEMA as those that have been the subject of at least two flood-claims payments of more than $1,000 apiece in any 10-year period—account for 2 percent of current policies and 3 percent of current premiums but about 12 percent of total claims since 1978. Including formerly insured RLPs, such properties accounted for almost one-quarter of claims payments since 1978. About 23,000 RLPs nationwide have been the subject of at least four claims payments while insured, and 10,000 of those have prompted six or more payments. FEMA’s approach to reducing the cost of repetitive-loss properties focuses more on measures to mitigate the worst flood risks—such as elevating, relocating, flood-proofing, or demolishing properties—than on charging higher premiums for flood insurance. More than half of the policies covering RLPs in high-risk areas have subsidized rates.

This paper was prepared by Perry Beider of CBO’s Microeconomic Studies Division.

The public and private sectors in the United States together spend over $500 billion a year on infrastructure projects, including highways and airports, water and energy utilities, dams, waste disposal sites and other environmental facilities, schools, and hospitals. The federal government makes a significant contribution to that investment through its direct expenditures and the subsidies it provides indirectly through the tax system, which are sometimes referred to as tax expenditures. Today CBO and the Joint Committee on Taxation (JCT) released a study on the importance of tax preferences, the types of tax-preferred bonds used in financing infrastructure, and the economic efficiency of such bonds.

That study concludes that the amount that the federal government forgoes through tax-exempt bond financing is greater than the associated reduction in borrowing costs for state and local governments. Some analysts have estimated the magnitude of that differential and conclude that several billion dollars each year may simply accrue to bondholders in higher income-tax brackets without providing any cost savings to borrowers.

The Importance of Tax Preferences in Financing Infrastructure

Most federal tax expenditures for infrastructure are the result of tax preferences granted for bonds that state and local governments issue to finance capital spending on infrastructure. Those tax preferences reduce borrowing costs. The amount of tax-preferred debt issued to finance new infrastructure projects undertaken by the public and private sectors totaled $1.7 trillion from 1991 to 2007. About three-quarters of those bond proceeds, or roughly $1.3 trillion, was for capital spending on infrastructure by states and localities, and the remainder was used to fund private capital investment for projects that serve a public purpose, such as schools and hospitals. That $1.3 trillion amounted to over one-half of the $2.3 trillion in capital spending on infrastructure by state and local governments (that is, net of federal grants and loan subsidies).

Tax preferences for debt are attractive to states and localities because they generally allow those governments to exercise broad discretion over the types of projects they finance and the amount of debt they issue. But unlike direct expenditures, tax expenditures—including tax preferences for state and local bonds—are not subject to the annual appropriation process that determines federal outlays for infrastructure and other discretionary programs. As a result, the cost of tax subsidies for infrastructure is not readily apparent, making the design of cost-effective tax preferences all the more important. For fiscal years 2008 to 2012, federal revenues forgone through the tax-exemot bond financing of infrastructure—both for new investments and for the financing of existing debt—are estimated to exceed $26 billion annually.

The Types of Tax-Preferred Bonds and Their Characteristics

The Internal Revenue Code provides for three forms of tax-preferred state and local bonds:

• Tax-exempt bonds pay interest to the bondholder that is not subject to federal income tax.  They are the most well established type of tax-preferred debt (tax exemption dates to the beginning of the federal income tax in 1913) and are issued to finance either the general functions of state and local governments or selected projects undertaken by the private sector. Tax-exempt bonds reduce the issuer’s borrowing costs because purchasers of such debt are willing to accept a lower rate of interest than that of taxable debt of comparable risk and maturity.
• Tax-credit bonds, by contrast, generally provide a credit against the bondholder’s overall federal income tax liability. They are much more recent in origin, and the outstanding amount of tax-credit bonds currently is minuscule in comparison with that of tax-exempt bonds.
• Direct-pay tax-credit bonds, in effect, require the federal government to make cash payments to the issuer of the bond in an amount equal to a portion of each of the interest payments the issuer makes to the bondholder. Such bonds, which were created by the American Recovery and Reinvestment Act of 2009 (ARRA, Public Law 111- 5), in part because the direct payment to the issuer represents a “deeper” subsidy to the issuer than the provision of a tax credit represents to the bondholder.

Increasing the Economic Efficiency of Tax-Preferred Bond Financing

Replacing tax-exempt interest with tax credits could, in principle, increase the efficiency of financing infrastructure with tax-preferred debt. Tax-credit bonds transfer to issuers all of the federal revenues forgone through the tax preference; in addition, the amount of the tax credit can be varied across types of infrastructure projects, thus bringing the federal revenue loss in line with the benefits expected from the investment.

Nevertheless, tax-credit bond programs have not been particularly well received by the market for a number of reasons, including the limited size and temporary nature of tax-credit bond programs and the absence of rules for stripping and selling credits. That situation is likely to change, however, as a result of the ARRA, which greatly expanded the size and range of tax-credit bond programs. As those new programs are implemented, it will be possible to gauge more accurately the practical advantages and disadvantages of tax-credit bonds.

This study was prepared by Nathan Musick of CBO’s Microeconomic Studies Division and the staff of JCT.