[This analysis was prepared by David Litvack and Brady Tournillon of Fitch Investors Service, L.P.]
As part of this study, Fitch Investors Service L.P. (Fitch) was asked to develop a model for evaluating the default risk for a federal credit program involving direct loans, loan guarantees, and standby lines of credit for surface transportation projects. This Appendix reviews the current approach for scoring transportation loans and proposes a new approach adapted from Fitch's model for evaluating the capital adequacy of private, for-profit bond insurance companies. Fitch also recommends specific scores or capital charges for projects that are rated differently.
The Alameda Corridor loan was scored using a yield premium approach to assess potential default cost. The loan was assumed to be made at an interest rate equal to the U.S. Treasury bond yield. The net present value of loan repayments on the project was then calculated, discounted at both the Treasury yield and the assumed market yield, based on the project's preliminary rating (105 basis points above the Treasury yield). The difference in net present values was deemed to represent the cost of the default risk. In our opinion, the yield spread approach is not a valid measure of the expected default risk. The yield spread takes into account other factors beside default risk, such as liquidity risk and call risk. Since liquidity risk and call risk are not relevant factors in the cost of the program to the federal government, the yield premium results in a cost estimate that is too high.
The fact that market yield spreads overestimate default risk is evidenced by the existence of a private, for-profit bond insurance industry that guarantees the principal and interest on municipal bonds, as well as asset-backed and mortgage-backed securities. Bond insurers guarantee municipal bonds that have mostly A and BBB underlying ratings; the insurance raises the bonds' public ratings to AAA. The premiums on these policies, which are usually paid by the issuer, amount to about half of the issuer's interest cost savings as a result of the higher credit rating.
The yield spread between the insured AAA and the uninsured bonds is sufficient to support premiums that cover the bond insurers' default risk plus an adequate profit margin. From January 1, 1992, through June 30, 1996, the bond insurers guaranteed a total of $544.4 billion in par. Premiums written over this period totaled $4.5 billion, or 0.8 percent of par insured. Losses from bond defaults over this period totaled $129.9 million, or 5.7 percent of premiums earned and 0.03 percent of average par outstanding (premiums for municipal bonds are typically written in the form of a one-time, up-front payment, but income is earned over the outstanding life of the bond). Bond insurers incur other costs besides losses, such as underwriting, surveillance, and administration; however, during the period from 1992 to mid-year 1996, bond insurers averaged 13.1 percent return on equity (income after expenses and taxes, divided by average shareholder's equity).
Bond insurers do not currently insure start-up toll roads because they tend to avoid stand-alone projects with construction risk and ramp-up risk. In our opinion, however, the shortcomings of the yield premium approach also apply to these risks as well. In other words, it should be possible for the federal government to provide credit for start-up toll road projects at a lower cost than the yield premium would imply.
The fact that the yield premium method overestimates the expected default cost means a different methodology is required. We propose a method derived from that used by the rating agencies to determine how much capital bond insurers need to allocate for their insurance policies. This will result in a more direct measure of the default risk on the projects.
Fitch rates the claims-paying ability of the bond insurers. A large part of this analysis focuses on the insurers' capital adequacy. To measure capital adequacy, Fitch uses a stress test model that subjects a bond insurer's portfolio to an economic downturn that produces an extraordinary level of bond defaults. For an insurer to receive a AAA claims-paying ability rating, it must be able to pay all projected claims through the peak years of the stress period and be left with sufficient resources to write new business when more stable economic conditions resume.
Claims during the stress period are forecast using capital charges that Fitch developed based on bond defaults experienced during the Great Depression of the 1930's. Fitch has adjusted the capital charges to reflect regulatory changes and the relative probability and severity of defaults for the types of insured risks in today's market. For example, current banking laws enacted after the Great Depression reduce the potential severity of another depression. However, in the 1930's all municipal bonds were backed by a general obligation pledge; most municipals today are revenue bonds which have potentially greater risk. For this reason, Fitch has developed different benchmark capital charges for various types of insured bonds. For example, transportation bonds on existing facilities are more risky and, therefore, have higher benchmark capital charges than tax-backed and water and sewer bonds. They are, however, less risky and have lower benchmark capital charges than private higher education and hospital bonds. These benchmark capital charges are then adjusted further based on Fitch's evaluation of the actual credit quality and diversity of the bonds within each sector of the individual insurer's portfolio.
Bond insurers do not currently insure start-up toll roads. Fitch developed capital charges for this category specifically for this report. Our methodology for developing these charges is described in the following paragraphs.
Based on historical evidence, while some start-up toll road projects experience late payment delinquencies in years one through five, and less frequently in years six through ten, almost all do get built, begin operations, and eventually pay off their debt, including interest on interest. Subordinate lenders to projects of investment-grade quality should get paid as well, although perhaps over a somewhat longer time frame than the senior bondholders. It is estimated that only about one percent of the loans rated BBB will not be recovered within a reasonable time frame, which for discussion purposes is defined as 30 years.
A project is rated below investment grade (lower than BBB-) if there is a foreseeable risk that it will not be successfully completed on time or generate sufficient revenues to fully pay creditors. Indeed, default rates are much higher for unrated and below investment-grade municipal bonds than they are for investment-grade bonds. Because start-up toll roads have only recently received ratings, there is little empirical data on default rates specifically for this sector. Based on the default experience in other sectors of the municipal market, Fitch estimates that a portfolio of loans on start-up toll road projects rated BB will experience a four percent loss rate and start-up toll road projects rated B an eight percent loss rate (net of recoveries).
Highly rated financial institutions not only require enough capital for an expected level of losses, but for a multiple of such losses. Fitch has concluded that for start-up infrastructure projects, a multiple of four to five times expected losses is needed to provide our highest credit standard of AAA. Multiplying the expected losses by five produces the capital charges that should be used on loans to start-up toll road projects; these charges (expressed as a percentage of original principal) are shown in Table A.1.
Table A.1 Capital Charges for Start-up Toll Roads
| Project Rating | Expected Loss (%) | Multiplier | Capital Charge AAA Scenario (%) |
|---|---|---|---|
| BBB | 1.0 | 5 | 5.0 |
| BBB- | 1.6 | 5 | 8.0 |
| BB+ | 2.6 | 5 | 13.0 |
| BB | 4.0 | 5 | 20.0 |
| BB- | 5.0 | 5 | 25.0 |
| B+ | 6.4 | 5 | 32.0 |
| B | 8.0 | 5 | 40.0 |
Fitch recognizes that, in many cases, the federal loan will be junior to the senior debt, but believes the same capital charges are applicable for subordinate, flexible payment debt. The flexibility in the federal credit program reduces the demands on a project to make timely payments; however, full repayment is still required. An important element in Fitch's capital charge calculation is that most loan defaults that occur during the initial ten-year period will be recovered. Fitch assumes that interest on delinquent loan payments is equal to the U.S. Treasury rate, so timing defaults will not affect the net present value cost of the loan credit program. The same analysis should hold true whether the federal credit takes the form of a direct loan, a guaranteed loan, or a contingent standby line of credit.
It should also be noted that the capital charge methodology for private, for-profit bond insurers applies to a large and diversified portfolio of loans. If an insurer were to guarantee loans to only a handful of projects, and one of these projects defaulted, then the overall cost could conceivably be higher than the weighted average capital charge. Fitch would require considerably more capital to assign a rating of AAA to a private company insuring only a small, non-diversified portfolio of loans. Considering the fact that the federal government has no liquidity constraints and these transportation loans would be only one piece of an existing diversified portfolio of approximately $1 trillion of federal government loans and guarantees in a wide range of industry sectors, this capital charge method is considered appropriate.
The capital charges Fitch recommends are consistent with AAA security. For an ongoing federal credit program that encompasses a portfolio of loans and guarantees, the likelihood of underestimating default cost is remote. In other words, the capital reserves should absorb all anticipated default risk, in essence representing a proxy for federal subsidy cost. This makes it a useful and conservative tool for budgeting purposes.
