While the major issuers of asset-backed securities are corporations, municipal governments use this form of financing rather than issuing municipal bonds and several European central governments use this form of financing. In the United States, the first type of asset-backed security (ABS) was the residential mortgage loan. We discussed the resulting securities, referred to as mortgage-backed securities, in the previous chapter. Securities backed by other types of assets (consumer and business loans and receivables) have been issued throughout the world. The largest sectors of the asset-backed securities market in the United States are securities backed by credit card receivables, auto loans, home equity loans, manufactured housing loans, student loans, Small Business Administration loans, corporate loans, and bonds (corporate, emerging market, and structured financial products). Since home equity loans and manufactured housing loans are backed by real estate property, the securities backed by them are referred to as real estate-backed asset-backed securities. Other asset-backed securities include securities backed by home improvement loans, health care receivables, agricultural equipment loans, equipment leases, music royalty receivables, movie royalty receivables, and municipal parking ticket receivables. Collectively, these products are called credit-sensitive structured products.

In this chapter, we will discuss the securitization process, the basic features of a securitization transaction, and the major asset types that have been securitized. In the last section of this chapter, we look at collateralized debt obligations. While this product has traditionally been classified as part of the ABS market, we will see how the structure of this product differs from that of a typical securitization.

There are two topics not covered in this chapter. The first is the valuation of an ABS. This topic is covered in Chapter 12. Second, the factors considered by rating agencies in rating an ABS transaction are not covered here but are covered in Chapter 15. In that chapter we also compare the factors considered by rating agencies in rating an asset-backed security and a corporate bond.

The credit department of QHT makes the decision as to whether or not to extend credit to a customer. That is, the credit department will request a credit loan application form be completed by a customer and based on criteria established by QHT will decide on whether to extend a loan. The criteria for extending credit are referred to as underwriting standards. Because QHT is extending the loan, it is referred to as the originator of the loan. Moreover, QHT may have a department that is responsible for servicing the loan. Servicing involves collecting payments from borrowers, notifying borrowers who may be delinquent, and, when necessary, recovering and disposing of the collateral (i.e., home theater equipment in our illustration) if the borrower does not make loan repayments by a specified time. While the servicer of the loans need not be the originator of the loans, in our illustration we are assuming that QHT will be the servicer.

Now let’s see how these loans can be used in a securitization. We will assume that QHT has $100 million of installment sales contracts. This amount is shown on QHT’s balance sheet as an asset. We will further assume that QHT wants to raise $100 million. Rather than issuing corporate bonds for $100 million, QHT’s treasurer decides to raise the funds via a securitization. To do so, QHT will set up a legal entity referred to as a special purpose vehicle (SPV). In our discussion of asset-backed securities we described the critical role of this legal entity; its role will become clearer in our illustration. In our illustration, the SPV that is set up is called Homeview Asset Trust (HAT). QHT will then sell to HAT $100 million of the loans. QHT will receive from HAT $100 million in cash, the amount it wanted to raise. But where does HAT get $100 million? It obtains those funds by selling securities that are backed by the $100 million of loans. These securities are the asset-backed securities we referred to earlier and we will discuss these further in Section II.C.

In the prospectus, HAT (the SPV) would be referred to as either the “issuer” or the “trust.” QHT, the seller of the collateral to HAT, would be referred to as the “seller.” The prospectus might then state: “The securities represent obligations of the issuer only and do not represent obligations of or interests in Quality Home Theaters Inc. or any of its affiliates.”

The transaction is diagramed in panel a of Exhibit 1. In panel b, the parties to the transaction are summarized.

The payments that are received from the collateral are distributed to pay servicing fees, other administrative fees, and principal and interest to the security holders. The legal documents in a securitization (prospectus or private placement memorandum) will set forth in considerable detail the priority and amount of payments to be made to the servicer, administrators, and the security holders of each bond class. The priority and amount of payments is commonly referred to as the “waterfall” because the flow of payments in a structure is depicted as a waterfall.

There is a good deal of legal documentation involved in a securitization transaction. The attorneys are responsible for preparing the legal documents. The first is the purchase agreement between the seller of the assets (QHT in our illustration) and the SPV (HAT in our illustration).¹³² The purchase agreement sets forth the representations and warranties that the seller is making about the assets. The second is one that sets forth how the cash flows are divided among the bond classes (i.e., the structure’s waterfall). Finally, the attorneys create the servicing agreement between the entity engaged to service the assets (in our illustration QHT retained the servicing of the loans) and the SPV.

An independent accounting firm will verify the accuracy of all numerical information placed in either the prospectus or private placement memorandum.¹³³ The result of this task results in a comfort letter for a securitization.

The trustee or trustee agent is the entity that safeguards the assets after they have been placed in the trust, receives the payments due to the bond holders, and provides periodic information to the bond holders. The information is provided in the form of remittance reports that may be issued monthly, quarterly or whenever agreed to by the terms of the prospectus or the private placement memorandum.

The underwriters and rating agencies perform the same function in a securitization as they do in a standard corporate bond offering. The rating agencies make an assessment of the collateral and the proposed structure to determine the amount of credit enhancement required to achieve a target credit rating for each bond class.

Finally, a securitization may have an entity that guarantees part of the obligations issued by the SPV. These entities are called guarantors and we will discuss their role in a securitization later.

A simple transaction can involve the sale of just one bond class with a par value of $100 million in our illustration. We will call this Bond Class A. Suppose HAT issues 100,000 certificates for Bond Class A with a par value of $1,000 per certificate. Then, each certificate holder would be entitled to 1/100,000 of the payment from the collateral after payment of fees and expenses. Each payment made by the borrowers (i.e., the buyers of the home theater equipment) consists of principal repayment and interest.

A structure can be more complicated. For example, there can be rules for distribution of principal and interest other than on a pro rata basis to different bond classes. As an example, suppose HAT issues Bond Classes A1, A2, A3, and A4 whose total par value is $100 million as follows:

As with a collateralized mortgage obligation (CMO) structure described in the previous chapter, there are different rules for the distribution of principal and interest to these four bond classes or tranches. A simple structure would be a sequential-pay one. As explained in the previous chapter, in a basic sequential-pay structure, each bond class receives periodic interest. However, the principal is repaid as follows: all principal received from the collateral is paid first to Bond Class A1 until it is fully paid off its $40 million par value. After Bond Class A1 is paid off, all principal received from the collateral is paid to Bond Class A2 until it is fully paid off. All principal payments from the collateral are then paid to Bond Class A3 until it is fully paid off and then all principal payments are made to Bond Class A4.

The reason for the creation of the structure just described, as explained in the previous chapter, is to redistribute the prepayment risk among different bond classes. Prepayment risk is the uncertainty about the cash flow due to prepayments. This risk can be decomposed into contraction risk (i.e., the undesired shortening in the average life of a security) or extension risk (i.e., the undesired lengthening in the average life of a security). The creation of these bond classes is referred to as prepayment tranching or time tranching.

Now let’s look at a more common structure in a transaction. As will be explained later, there are structures where there is more than one bond class and the bond classes differ as to how they will share any losses resulting from defaults of the borrowers. In such a structure, the bond classes are classified as senior bond classes and subordinate bond classes. This structure is called a senior-subordinate structure. Losses are realized by the subordinate bond classes before there are any losses realized by the senior bond classes. For example, suppose that HAT issued $90 million par value of Bond Class A, the senior bond class, and $10 million par value of Bond Class B, the subordinate bond class. So the structure is as follows:

The purpose of this structure is to redistribute the credit risk associated with the collateral. This is referred to as credit tranching. As explained later, the senior-subordinate structure is a form of credit enhancement for a transaction.

There is no reason why only one subordinate bond class is created. Suppose that HAT issued the following structure

Now we will add just one more twist to the structure. Often in larger transactions, the senior bond class will be carved into different bond classes in order to redistribute the prepayment risk. For example, HAT might issue the following structure:

As explained in the previous chapter, a bond class in a securitization is also referred to as a “tranche.” Consequently, throughout this chapter the terms “bond class” and “tranche” are used interchangeably.

In contrast to amortizing assets, non-amortizing assets require only minimum periodic payments with no scheduled principal repayment. If that payment is less than the interest on the outstanding loan balance, the shortfall is added to the outstanding loan balance. If the periodic payment is greater than the interest on the outstanding loan balance, then the difference is applied to the reduction of the outstanding loan balance. Since there is no schedule of principal payments (i.e., no amortization schedule) for a non-amortizing asset, the concept of a prepayment does not apply. A credit card receivable is an example of a non-amortizing asset.

The type of collateral—amortizing or non-amortizing—has an impact on the structure of the transaction. Typically, when amortizing assets are securitized, there is no change in the composition of the collateral over the life of the securities except for loans that have been removed due to defaults and full principal repayment due to prepayments or full amortization. For example, if at the time of issuance the collateral for an ABS consists of 3,000 four-year amortizing loans, then the same 3,000 loans will be in the collateral six months from now assuming no defaults and no prepayments. If, however, during the first six months, 200 of the loans prepay and 100 have defaulted, then the collateral at the end of six months will consist of 2,700 loans (3,000 - 200 - 100). Of course, the remaining principal of the 2,700 loans will decline because of scheduled principal repayments and any partial prepayments. All of the principal repayments from the collateral will be distributed to the security holders.

In contrast, for an ABS transaction backed by non-amortizing assets, the composition of the collateral changes. The funds available to pay the security holders are principal repayments and interest. The interest is distributed to the security holders. However, the principal repayments can be either (1) paid out to security holders or (2) reinvested by purchasing additional loans. What will happen to the principal repayments depends on the time since the transaction was originated. For a certain amount of time after issuance, all principal repayments are reinvested in additional loans. The period of time for which principal repayments are reinvested rather than paid out to the security holders is called the lockout period or revolving period. At the end of the lockout period, principal repayments are distributed to the security holders. The period when the principal repayments are not reinvested is called the principal amortization period. Notice that unlike the typical transaction that is backed by amortizing assets, the collateral backed by non-amortizing assets changes over time. A structure in which the principal repayments are reinvested in new loans is called a revolving structure.

While the receivables in a revolving structure may not be prepaid, all the bonds issued by the trust may be retired early if certain events occur. That is, during the lockout period, the trustee is required to use principal repayments to retire the securities rather than reinvest principal in new collateral if certain events occur. The most common trigger is the poor performance of the collateral. This provision that specifies the redirection of the principal repayments during the lockout period to retire the securities is referred to as the early amortization provision or rapid amortization provision.

Not all transactions that are revolving structures are backed by non-amortizing assets. There are some transactions in which the collateral consists of amortizing assets but during a lockout period, the principal repayments are reinvested in additional loans. For example, there are transactions in the European market in which the collateral consists of residential mortgage loans but during the lockout period principal repayments are used to acquire additional residential mortgage loans.

As with nonagency mortgage-backed securities, described in the previous chapter, modeling defaults for the collateral is critical in estimating the cash flows of an asset-backed security. Proceeds that are recovered in the event of a default of a loan prior to the scheduled principal repayment date of an amortizing asset represent a prepayment and are referred to as an involuntary prepayment. Projecting prepayments for amortizing assets requires an assumption about the default rate and the recovery rate. For a non-amortizing asset, while the concept of a prepayment does not exist, a projection of defaults is still necessary to project how much will be recovered and when.

The analysis of prepayments can be performed on a pool level or a loan level. In pool-level analysis it is assumed that all loans comprising the collateral are identical. For an amortizing asset, the amortization schedule is based on the gross weighted average coupon (GWAC) and weighted average maturity (WAM) for that single loan. We explained in the previous chapter what the WAC and WAM of a pool of mortgage loans is and illustrated how it is computed. In this chapter, we refer to the WAC as gross WAC. Pool-level analysis is appropriate where the underlying loans are homogeneous. Loan-level analysis involves amortizing each loan (or group of homogeneous loans).

The expected final maturity of an asset-backed security is the maturity date based on expected prepayments at the time of pricing of a deal. The legal final maturity can be two or more years after the expected final maturity. The average life, or weighted average life, was explained in the previous chapter.

Also explained in the previous chapter is a tranche’s principal window which refers to the time period over which the principal is expected to be paid to the bondholders. A principal window can be wide or narrow. When there is only one principal payment that is scheduled to be made to a bondholder, the bond is referred to as having a bullet maturity. Due to prepayments, an asset-backed security that is expected to have a bullet maturity may have an actual maturity that differs from that specified in the prospectus. Hence, asset-backed securities bonds that have an expected payment of only one principal are said to have a soft bullet.

There are two general types of credit enhancement structures: external and internal. We describe each type below.

As explained earlier, the creation of a senior-subordinate structure is done to provide credit tranching. More specifically, the senior-subordinate structure is a form of internal credit enhancement because the subordinate bond classes provide credit support for the senior bond classes. To understand why, the hypothetical HAT structure with one subordinate bond class that was described earlier is reproduced below:

Note that one subordinate bond class can provide credit enhancement for another subordinate bond class. To see this, consider the hypothetical HAT structure with two subordinate bond classes presented earlier:

The basic concern in the senior-subordinate structure is that while the subordinate bond classes provide a certain level of credit protection for the senior bond class at the closing of the deal, the level of protection changes over time due to prepayments. Faster prepayments can remove the desired credit protection. Thus, the objective after the deal closes is to distribute any prepayments such that the credit protection for the senior bond class does not deteriorate over time.

In real-estate related asset-backed securities, as well as nonagency mortgage-backed securities, the solution to the credit protection problem is a well developed mechanism called the shifting interest mechanism. Here is how it works. The percentage of the mortgage balance of the subordinate bond class to that of the mortgage balance for the entire deal is called the level of subordination or the subordinate interest. The higher the percentage, the greater the level of protection for the senior bond classes. The subordinate interest changes after the deal is closed due to prepayments. That is, the subordinate interest shifts (hence the term “shifting interest”). The purpose of a shifting interest mechanism is to allocate prepayments so that the subordinate interest is maintained at an acceptable level to protect the senior bond class. In effect, by paying down the senior bond class more quickly, the amount of subordination is maintained at the desired level.

The prospectus will provide the shifting interest percentage schedule for calculating the senior prepayment percentage (the percentage of prepayments paid to the senior bond class). For mortgage loans, a commonly used shifting interest percentage schedule is as follows:

The shifting interest percentage schedule given in the prospectus is the “base” schedule. The set of shifting interest percentages can change over time depending on the performance of the collateral. If the performance is such that the credit protection for the senior bond class has deteriorated because credit losses have reduced the subordinate bond classes, the base shifting interest percentages are overridden and a higher allocation of prepayments is made to the senior bond class.

Performance analysis of the collateral is undertaken by the trustee for determining whether or not to override the base schedule. The performance analysis is in terms of tests and if the collateral fails any of the tests, this will trigger an override of the base schedule.

It is important to understand that the presence of a shifting interest mechanism results in a trade-off between credit risk and contraction risk for the senior bond class. The shifting interest mechanism reduces the credit risk to the senior bond class. However, because the senior bond class receives a larger share of any prepayments, contraction risk increases.

Typically, for a corporate, federal agency, and municipal bond the trigger event for a call provision is that a specified amount of time has passed.¹³⁶ In the case of asset-backed securities, it is not simply the passage of time whereby the trustee is permitted to exercise any call option. There are trigger events for exercising the call option based on the amount of the issue outstanding.

There are two call provisions where the trigger that grants the trustee to call in the issue is based on a date being reached: (1) call on or after specified date and (2) auction call. A call on or after specified date operates just like a standard call provision for corporate, federal agency, and municipal securities: once a specified date is reached, the trustee has the option to call all the outstanding bonds. In an auction call, at a certain date a call will be exercised if an auction results in the outstanding collateral being sold at a price greater than its par value. The premium over par value received from the auctioned collateral is retained by the trustee and is eventually distributed to the seller of the assets.

Provisions that allow the trustee to call an issue or a tranche based on the par value outstanding are referred to as optional clean-up call provisions. Two examples are (1) percent of collateral call and (2) percent of bond call. In a percent of collateral call, the outstanding bonds can be called at par value if the outstanding collateral’s balance falls below a predetermined percent of the original collateral’s balance. This is the most common type of clean-up call provision for amortizing assets and the predetermined level is typically 10%. For example, suppose that the value for the collateral is $100 million. If there is a percent of collateral call provision with a trigger of 10%, then the trustee can call the entire issue if the value of the call is $10 million or less. In a percent of bond call, the outstanding bonds can be called at par value if the outstanding bond’s par value relative to the original par value of bonds issued falls below a specified amount.

There is a call option that combines two triggers based on the amount outstanding and date. In a latter of percent or date call, the outstanding bonds can be called if either (1) the collateral’s outstanding balance reaches a predetermined level before the specified call date or (2) the call date has been reached even if the collateral outstanding is above the predetermined level.

In addition to the above call provisions which permit the trustee to call the bonds, there may be an insurer call. Such a call permits the insurer to call the bonds if the collateral’s cumulative loss history reaches a predetermined level.

Home equity loans can be either closed end or open end. A closed-end HEL is structured the same way as a fully amortizing residential mortgage loan. That is, it has a fixed maturity and the payments are structured to fully amortize the loan by the maturity date. With an open-end HEL, the homeowner is given a credit line and can write checks or use a credit card for up to the amount of the credit line. The amount of the credit line depends on the amount of the equity the borrower has in the property. Because home equity loan securitizations are predominately closed-end HELs, our focus in this section is securities backed by them.

There are both fixed-rate and variable-rate closed-end HELs. Typically, variable-rate loans have a reference rate of 6-month LIBOR and have periodic caps and lifetime caps. (A periodic cap limits the change in the mortgage rate from the previous time the mortgage rate was reset; a lifetime cap sets a maximum that the mortgage rate can ever be for the loan.) The cash flow of a pool of closed-end HELs is comprised of interest, regularly scheduled principal repayments, and prepayments, just as with mortgage-backed securities. Thus, it is necessary to have a prepayment model and a default model to forecast cash flows. The prepayment speed is measured in terms of a conditional prepayment rate (CPR).

There are differences in the prepayment behavior for home equity loans and agency MBS. Wall Street firms involved in the underwriting and market making of securities backed by HELs have developed prepayment models for these deals. Several firms have found that the key difference between the prepayment behavior of HELs and agency residential mortgages is the important role played by the credit characteristics of the borrower.¹³⁷

Borrower characteristics and the amount of seasoning (i.e., how long the loans have been outstanding) must be kept in mind when trying to assess prepayments for a particular deal. In the prospectus of a HEL, a base case prepayment assumption is made. Rather than use the PSA prepayment benchmark as the base case prepayment speed, issuer’s now use a base case prepayment benchmark that is specific to that issuer. The benchmark prepayment speed in the prospectus is called the prospectus prepayment curve or PPC. As with the PSA benchmark, faster or slower prepayments speeds are a quoted as a multiple of the PPC. Having an issuer-specific prepayment benchmark is preferred to a generic benchmark such as the PSA benchmark. The drawback for this improved description of the prepayment characteristics of a pool of mortgage loans is that it makes comparing the prepayment characteristics and investment characteristics of the collateral between issuers and issues (newly issued and seasoned issues) difficult.

Since HEL deals are backed by both fixed-rate and variable-rate loans, a separate PPC is provided for each type of loan. For example, in the prospectus for the Contimortgage Home Equity Loan Trust 1998-2, the base case prepayment assumption for the fixed-rate collateral begins at 4% CPR in month 1 and increases 1.45455% CPR per month until month 12, at which time it is 20% CPR. Thus, the collateral is assumed to be seasoned in 12 months. The prepayment ramp begins in month 1 and ends in month 12. If an investor analyzed the deal based on 200% PPC, this means doubling the CPRs cited and using 12 months for when the collateral seasons. For the variable-rate collateral in the ContiMortgage deal, 100% PPC assumes the collateral is seasoned after 18 months with the CPR in month 1 being 4% and increasing 1.82353% CPR each month. From month 18 on, the CPR is 35%. Thus, the prepayment ramp starts at month 1 and ends at month 18. Notice that for this issuer, the variable-rate collateral is assumed to season slower than the fixed-rate collateral (18 versus 12 months), but has a faster CPR when the pool is seasoned (35% versus 20%).

Typically, home equity loan-backed securities are securitized by both closed-end fixed-rate and adjustable-rate (or variable-rate) HELs. The securities backed by the latter are called HEL floaters. The reference rate of the underlying loans typically is 6-month LIBOR. The cash flow of these loans is affected by periodic and lifetime caps on the loan rate.

Institutional investors that seek securities that better match their floating-rate funding costs are attracted to securities that offer a floating-rate coupon. To increase the attractiveness of home equity loan-backed securities to such investors, the securities typically have been created in which the reference rate is 1-month LIBOR. Because of (1) the mismatch between the reference rate on the underlying loans (6-month LIBOR) and that of the HEL floater and (2) the periodic and life caps of the underlying loans, there is a cap on the coupon rate for the HEL floater. Unlike a typical floater, which has a cap that is fixed throughout the security’s life, the effective periodic and lifetime cap of a HEL floater is variable. The effective cap, referred to as the available funds cap, will depend on the amount of funds generated by the net coupon on the principal, less any fees.

Let’s look at one issue, Advanta Mortgage Loan Trust 1995-2 issued in June 1995. At the offering, this issue had approximately $122 million closed-end HELs. There were 1,192 HELs consisting of 727 fixed-rate loans and 465 variable-rate loans. There were five classes (A-1, A-2, A-3, A-4, and A-5) and a residual. The five classes are summarized below:

The collateral is divided into group I and group II. The 727 fixed-rate loans are included in group I and support Classes A-1, A-2, A-3, and A-4 certificates. The 465 variable-rate loans are in group II and support Class A-5.

Tranches have been structured in home equity loan deals so as to give some senior tranches greater prepayment protection than other senior tranches. The two types of structures that do this are the non-accelerating senior tranche and the planned amortization class tranche.

An example of a HEL PAC tranche in a HEL-backed deal is tranche A-6 in ContiMortgage 1998-2. We described the PPC for this deal in Section III.A.1 above. There is a separate PAC collar for both the fixed-rate and adjustable-rate collateral. For the fixed-rate collateral the PAC collar is 125%-175% PPC; for the adjustable-rate collateral the PAC collar is 95%-130% PPC. The average life for tranche A-6 (a tranche backed by the fixed-rate collateral) is 5.1 years. As explained in Chapter 3, the effective collar for shorter tranches can be greater than the upper collar specified in the prospectus. The effective upper collar for tranche A-6 is actually 180% PPC (assuming that the adjustable-rate collateral pays at 100% PPC).¹⁴⁰

For shorter PACs, the effective upper collar is greater. For example, for tranche A-3 in the same deal, the initial PAC collar is 125% to 175% PPC with an average life of 2.02 years. However, the effective upper collar is 190% PPC (assuming the adjustable-rate collateral pays at 100% PPC).

The effective collar for PAC tranches changes over time based on actual prepayments and therefore based on when the support tranches depart from the initial PAC collar. For example, if for the next 36 months after the issuance of the ContiMortgage 1998-2 actual prepayments are a constant 150% PPC, then the effective collar would be 135% PPC to 210% PPC.¹⁴¹ That is, the lower and upper collar will increase. If the actual PPC is 200% PPC for the 10 months after issuance, the support bonds will be fully paid off and there will be no PAC collateral. In this situation the PAC is said to be a broken PAC.

Manufactured housing-backed securities are issued by Ginnie Mae and private entities. The former securities are guaranteed by the full faith and credit of the U.S. government. The manufactured home loans that are collateral for the securities issued and guaranteed by Ginnie Mae are loans guaranteed by the Federal Housing Administration (FHA) or Veterans Administration (VA).

Loans not backed by the FHA or VA are called conventional loans. Manufactured housing-backed securities that are backed by such loans are called conventional manufactured housing-backed securities. These securities are issued by private entities.

The typical loan for a manufactured home is 15 to 20 years. The loan repayment is structured to fully amortize the amount borrowed. Therefore, as with residential mortgage loans and HELs, the cash flow consists of net interest, regularly scheduled principal, and prepayments. However, prepayments are more stable for manufactured housing-backed securities because they are not sensitive to refinancing.

There are several reasons for this. First, the loan balances are typically small so that there is no significant dollar savings from refinancing. Second, the rate of depreciation of mobile homes may be such that in the earlier years depreciation is greater than the amount of the loan paid off. This makes it difficult to refinance the loan. Finally, typically borrowers are of lower credit quality and therefore find it difficult to obtain funds to refinance.

As with residential mortgage loans and HELs, prepayments on manufactured housing-backed securities are measured in terms of CPR and each issue contains a PPC.

The payment structure is the same as with nonagency mortgage-backed securities and home equity loan-backed securities.

The standard mortgage loan is a variable rate, fully amortizing loan. Typically, the term of the loan is 25 years. As in the U.S. mortgage market, borrowers seeking a loan with a high loan-to-value ratio are required to obtain mortgage insurance, called a “mortgage indemnity guarantee” (MIG).

The deals are more akin to the nonagency market since there is no guarantee by a federally related agency or a government sponsored enterprise as in the United States. Thus, there is credit enhancement as explained below.

Because the underlying mortgage loans are floating rate, the securities issued are floating rate (typically, LIBOR is the reference rate). The cash flow depends on the timing of the principal payments. The deals are typically set up as a sequential-pay structure. For example, consider the Granite Mortgage 00-2 transaction, a typical structure in the United Kingdom.¹⁴³ The mortgage pool consists of prime mortgages. There are four bond classes. The two class A tranches, Class A-1 and Class A-2, are rated AAA. One is a dollar denominated tranche and the other a pound sterling tranche. Class B is rated single A and tranche C is rated triple BBB. The sequence of principal payments is as follows: Class A-1 and Class A-2 are paid off on a pro rata basis, then Class B is paid off, and then Class C is paid off.

The issuer has the option to call the outstanding notes under the following circumstances:

Credit enhancement can consist of excess spread, reserve fund, and subordination. For the Granite Mortgage 00-2, there was subordination: Class B and C tranches for the two Class A tranche and Class C tranche for the Class B tranche. The reserve was fully funded at the time of issuance and the excess spread was used to build up the reserve fund. In addition, there is a “principal shortfall provision.” This provision requires that if the realized losses for a period are such that the excess reserve for that period is not sufficient to cover the losses, as excess spread becomes available in future periods they are used to cover these losses. Also there are performance triggers that under certain conditions will provide further credit protection to the senior bonds by modifying the payment of principal. When the underlying mortgage pool consists of nonconforming mortgage loans, additional protections are provided for investors.

Since prepayments will reduce the average life of the senior notes in a transaction, typical deals have provision that permit the purchase of substitute mortgages if the prepayment rate exceeds a certain rate. For example, in the Granite Mortgage 00-02 deal, this rate is 20% per annum.

Loan maturities are typically between 20 and 30 years. As in the United States, there is a wide range of mortgage designs with respect to interest rates. There are fixed-rate, variable-rate (both capped and uncapped), and rates tied to a benchmark.

There is mortgage insurance for loans to protect lenders, called “lenders mortgage insurance” (LMI). Loans typically have LMI covering 20% to 100% of the loan. The companies that provide this insurance are private corporations.¹⁴⁵ When mortgages loans that do not have LMI are securitized, typically the issuer will purchase insurance for those loans.

LMI is important for securitized transactions since it is the first layer of credit enhancement in a deal structure. The rating agencies recognize this in rating the tranches in a structure. The amount that a rating agency will count toward credit enhancement for LMI depends on the rating agency’s assessment of the mortgage insurance company.

When securitized, the tranches have a floating rate. There is an initial revolving period—which means that no principal payments are made to the tranche holders but instead reinvested in new collateral. As with the U.K. Granite Mortgage 00-02 deal, the issuer has the right to call the issue if there is an imposition of a withholding tax on note holders’ interest payments, after a certain date, or if the balance falls below a certain level (typically, 10%).

Australian mortgage-backed securities have tranches that are U.S. dollar denominated and some that are denominated in Euros.¹⁴⁶ These global deals typically have two or three AAA senior tranches and one AA or AA− junior tranche.

For credit enhancement, there is excess spread (which in most deals is typically small), subordination, and, as noted earlier, LMI. To illustrate this, consider the Interstar Millennium Series 2000-3E Trust—a typical Australian MBS transaction. There are two tranches: a senior tranche (Class A) that was rated AAA and a subordinated tranche (Class B) that was AA−. The protection afforded the senior tranche is the subordinated tranche, LMI (all the properties were covered up to 100% and were insured by all five major mortgage insurance companies), and the excess spread.

Historically, auto loan-backed securities have represented between 18% to 25% of the asset-backed securities market. The auto loan market is tiered based on the credit quality of the borrowers. “Prime auto loans” are of fundamentally high credit quality and originated by the financial subsidiaries of major auto manufacturers. The loans are of high credit quality for the following reasons. First, they are a secured form of lending. Second, they begin to repay principal immediately through amortization. Third, they are short-term in nature. Finally, for the most part, major issuers of auto loans have tended to follow reasonably prudent underwriting standards.

Unlike the sub-prime mortgage industry, there is less consistency on what actually constitutes various categories of prime and sub-prime auto loans. Moody’s assumes the prime market is composed of issuers typically having cumulative losses of less than 3%; near-prime issuers have cumulative losses of 3-7%; and sub-prime issuers have losses exceeding 7%.

The auto sector was a small part of the European asset-backed securities market in 2002, about 5% of total securitization. There are two reasons for this. First, there is lower per capita car ownership in Europe. Second, there is considerable variance of tax and regulations dealing with borrower privacy rules in Europe thereby making securitization difficult.¹⁴⁷ Auto deals have been done in Italy, the U.K., Germany, Portugal, and Belgium.

Prepayments due to repossession and subsequent resale are sensitive to the economic cycle. In recessionary economic periods, prepayments due to this factor increase. While refinancings may be a major reason for prepayments of mortgage loans, they are of minor importance for automobile loans. Moreover, the interest rates for the automobile loans underlying some deals are substantially below market rates since they are offered by manufacturers as part of a sales promotion.

There is a mathematical relationship between the SMM and the ABS measures. Letting M denote the number of months after loan origination, the SMM rate can be calculated from the ABS rate using the following formula: where the ABS and SMM rates are expressed in decimal form.

For example, if the ABS rate is 1.5% (i.e., 0.015) at month 14 after origination, then the SMM rate is 1.86%, as shown below:

The ABS rate can be calculated from the SMM rate using the following formula:

For example, if the SMM rate at month 9 after origination is 1.3%, then the ABS rate is:

Historically, when measured in terms of SMM rate, auto loans have experienced SMMs that increase as the loans season.

The student loans that have been most commonly securitized are those that are made under the Federal Family Education Loan Program (FFELP). Under this program, the government makes loans to students via private lenders. The decision by private lenders to extend a loan to a student is not based on the applicant’s ability to repay the loan. If a default of a loan occurs and the loan has been properly serviced, then the government will guarantee up to 98% of the principal plus accrued interest.¹⁴⁹

Loans that are not part of a government guarantee program are called alternative loans. These loans are basically consumer loans and the lender’s decision to extend an alternative loan will be based on the ability of the applicant to repay the loan. Alternative loans have been securitized.

Student loans are floating-rate loans, exclusively indexed to the 3-month Treasury bill rate. As a result, some issuers of SLABs issue securities whose coupon rate is indexed to the 3-month Treasury bill rate. However, a large percentage of SLABS issued are indexed to LIBOR floaters.¹⁵¹

Prepayments typically occur due to defaults or loan consolidation. Even if there is no loss of principal faced by the investor when defaults occur, the investor is still exposed to contraction risk. This is the risk that the investor must reinvest the proceeds at a lower spread and in the case of a bond purchased at a premium, the premium will be lost. Studies have shown student loan prepayments are insensitive to the level of interest rates. Consolidations of a loan occur when the student who has loans over several years combines them into a single loan. The proceeds from the consolidation are distributed to the original lender and, in turn, distributed to the bondholders.

The Small Business Secondary Market Improvement Act passed in 1984 permitted the pooling of SBA loans. When pooled, the underlying loans must have similar terms and features. The maturities typically used for pooling loans are 7, 10, 15, 20, and 25 years. Loans without caps are not pooled with loans that have caps.

Most variable-rate SBA loans make monthly payments consisting of interest and principal repayment. The amount of the monthly payment for an individual loan is determined as follows. Given the coupon formula of the prime rate plus the loan’s quoted margin, the interest rate is determined for each loan. Given the interest rate, a level payment amortization schedule is determined. It is this level payment that is paid for the next month until the coupon rate is reset.

The monthly cash flow that the investor in an SBA-backed security receives consists of

Prepayments for SBA-backed securities are measured in terms of CPR. Voluntary prepayments can be made by the borrower without any penalty. There are several factors contributing to the prepayment speed of a pool of SBA loans. A factor affecting prepayments is the maturity date of the loan. It has been found that the fastest speeds on SBA loans and pools occur for shorter maturities.¹⁵² The purpose of the loan also affects prepayments. There are loans for working capital purposes and loans to finance real estate construction or acquisition. It has been observed that SBA pools with maturities of 10 years or less made for working capital purposes tend to prepay at the fastest speed. In contrast, loans backed by real estate that are long maturities tend to prepay at a slow speed.

Interest to security holders is paid periodically (e.g, monthly, quarterly, or semiannually). The interest rate may be fixed or floating—roughly half of the securities are floaters. The floating rate is uncapped.

A credit card receivable-backed security is a nonamortizing security. For a specified period of time, the lockout period or revolving period, the principal payments made by credit card borrowers comprising the pool are retained by the trustee and reinvested in additional receivables to maintain the size of the pool. The lockout period can vary from 18 months to 10 years. So, during the lockout period, the cash flow that is paid out to security holders is based on finance charges collected and fees. After the lockout period, the principal is no longer reinvested but paid to investors, the principal-amortization period and the various types of structures are described next.

In a passthrough structure, the principal cash flows from the credit card accounts are paid to the security holders on a pro rata basis. In a controlled-amortization structure, a scheduled principal amount is established, similar to the principal window for a PAC bond. The scheduled principal amount is sufficiently low so that the obligation can be satisfied even under certain stress scenarios, where cash flow is decreased due to defaults or slower repayment by borrowers. The security holder is paid the lesser of the scheduled principal amount and the pro rata amount. In a bullet-payment structure, the security holder receives the entire amount in one distribution. Since there is no assurance that the entire amount can be paid in one lump sum, the procedure is for the trustee to place principal monthly into an account that generates sufficient interest to make periodic interest payments and accumulate the principal to be repaid. These deposits are made in the months shortly before the scheduled bullet payment. This type of structure is also often called a soft bullet because the maturity is technically not guaranteed, but is almost always satisfied. The time period over which the principal is accumulated is called the accumulation period.

We begin with the concept of the gross portfolio yield. This yield includes finance charges collected and fees. Charge-offs represent the accounts charged off as uncollectible. Net portfolio yield is equal to gross portfolio yield minus charge-offs. The net portfolio yield is important because it is from this yield that the bondholders will be paid. So, for example, if the average yield (WAC) that must be paid to the various tranches in the structure is 5% and the net portfolio yield for the month is only 4.5%, there is the risk that the bondholder obligations will not be satisfied.

Delinquencies are the percentages of receivables that are past due for a specified number of months, usually 30, 60, and 90 days. They are considered an indicator of potential future charge-offs.

The monthly payment rate (MPR) expresses the monthly payment (which includes finance charges, fees, and any principal repayment) of a credit card receivable portfolio as a percentage of credit card debt outstanding in the previous month. For example, suppose a $500 million credit card receivable portfolio in January realized $50 million of payments in February. The MPR would then be 10% ($50 million divided by $500 million).

There are two reasons why the MPR is important. First, if the MPR reaches an extremely low level, there is a chance that there will be extension risk with respect to the principal payments on the bonds. Second, if the MPR is very low, then there is a chance that there will not be sufficient cash flows to pay off principal. This is one of the events that could trigger early amortization of the principal (described below).

At issuance, portfolio yield, charge-offs, delinquency, and MPR information are provided in the prospectus. Information about portfolio performance is then available from Bloomberg, the rating agencies, and dealers.

Typically, early amortization allows for the rapid return of principal in the event that the 3-month average excess spread earned on the receivables falls to zero or less. When early amortization occurs, the credit card tranches are retired sequentially (i.e., first the AAA bond then the AA rated bond, etc.). This is accomplished by paying the principal payments made by the credit card borrowers to the investors instead of using them to purchase more receivables. The length of time until the return of principal is largely a function of the monthly payment rate. For example, suppose that a AAA tranche is 82% of the overall deal. If the monthly payment rate is 11% then the AAA tranche would return principal over a 7.5-month period (82%/11%). An 18% monthly payment rate would return principal over a 4.5-month period (82%/18%).

When the underlying pool of debt obligations are bond-type instruments (high-yield corporate, structured financial products, and emerging market bonds), a CDO is referred to as a collateralized bond obligation (CBO). When the underlying pool of debt obligations are bank loans, a CDO is referred to as a collateralized loan obligation (CLO).

The funds to purchase the underlying assets (i.e., the bonds and loans) are obtained from the issuance of debt obligations (i.e., tranches) and include one or more senior tranches, one or more mezzanine tranches, and a subordinate/equity tranche. There will be a rating sought for all but the subordinate/equity tranche. For the senior tranches, at least an A rating is typically sought. For the mezzanine tranches, a rating of BBB but no less than B is sought. As explained below, since the subordinate/equity tranche receives the residual cash flow, no rating is sought for this tranche.

The ability of the asset manager to make the interest payments to the tranches and payoff the tranches as they mature depends on the performance of the underlying assets. The proceeds to meet the obligations to the CDO tranches (interest and principal repayment) can come from (1) coupon interest payments of the underlying assets, (2) maturing assets in the underlying pool, and (3) sale of assets in the underlying pool.

In a typical structure, one or more of the tranches is a floating-rate security. With the exception of deals backed by bank loans which pay a floating rate, the asset manager invests in fixed-rate bonds. Now that presents a problem—paying tranche investors a floating rate and investing in assets with a fixed rate. To deal with this problem, the asset manager uses derivative instruments to be able to convert fixed-rate payments from the assets into floating-rate payments. In particular, interest rate swaps are used. This derivative instrument allows a market participant to swap fixed-rate payments for floating-rate payments or vice versa. Because of the mismatch between the nature of the cash flows of the debt obligations in which the asset manager invests and the floating-rate liability of any of the tranches, the asset manager must use an interest rate swap. A rating agency will require the use of swaps to eliminate this mismatch.

The first breakdown in the CDO family is between cash CDOs and synthetic CDOs. A cash CDO is backed by a pool of cash market debt instruments. We described the range of debt obligations earlier. These were the original types of CDOs issued. A synthetic CDO is a CDO where the investor has the economic exposure to a pool of debt instrument but this exposure is realized via a credit derivative instrument rather than the purchase of the cash market instruments. We will discuss the basic elements of a synthetic CDO later.

Both a cash CDO and a synthetic CDO are further divided based on the motivation of the sponsor. The motivation leads to balance sheet and arbitrage CDOs. As explained below, in a balance sheet CDO, the motivation of the sponsor is to remove assets from its balance sheet. In an arbitrage CDO, the motivation of the sponsor is to capture a spread between the return that it is possible to realize on the collateral backing the CDO and the cost of borrowing funds to purchase the collateral (i.e., the interest rate paid on the obligations issued).

Cash CDOs that are arbitrage transactions are further divided in cash flow and market value CDOs depending on the primary source of the proceeds from the underlying asset used to satisfy the obligation to the tranches. In a cash flow CDO, the primary source is the interest and maturing principal from the underlying assets. In a market value CDO, the proceeds to meet the obligations depends heavily on the total return generated from the portfolio. While cash CDOs that are balance sheet motivated transactions can also be cash flow or market value CDOs, only cash flow CDOs have been issued.

Now we can put this all together. Let’s look at the interest coming into the CDO:

The interest to be paid out to the senior and mezzanine tranches and to the swap counterparty include:

Netting the interest payments coming in and going out we have:

Since 70 bp times $80 million is $560,000, the net interest remaining is $3,140,000 (= $3,700,000 − $560,000). From this amount any fees (including the asset management fee) must be paid. The balance is then the amount available to pay the subordinate/equity tranche. Suppose that these fees are $634,000. Then the cash flow available to the subordinate/equity tranche for the year is $2.5 million. Since the tranche has a par value of $10 million and is assumed to be sold at par, this means that the annual return is 25%.

Obviously, some simplifying assumptions have been made. For example, it is assumed that there are no defaults. It is assumed that all of the issues purchased by the asset manager are noncallable and therefore the coupon rate would not decline because issues are called. Moreover, as explained below, after some period the asset manager must begin repaying principal to the senior and mezzanine tranches. Consequently, the interest rate swap must be structured to take this into account since the entire amount of the senior tranche is not outstanding for the life of the collateral. Despite the simplifying assumptions, the illustration does demonstrate the basic economics of an arbitrage transaction, the need for the use of an interest rate swap, and how the subordinate/equity tranche will realize a return.

The credit risk associated with the reference asset is divided into two sections: (1) senior section and (2) junior section. In a typical synthetic CDO structure, the senior section is about 90% and the junior section is about 10%. (We’ll see what we mean by 90% and 10% shortly.) The losses that are realized from the reference are first realized by the junior section up to a notional amount and then after that full loss is realized, the senior section begins realizing losses.

For example, let’s suppose that the reference asset is a high-yield corporate bond index. An amount of credit risk exposure in terms of market value must be determined. Suppose that it is $500 million. The $500 million is referred to as the notional amount. Suppose further that the credit risk associated with the $500 million credit exposure is divided into a $450 million senior section and $50 million junior section. The $450 million is the notional amount for the senior section and the $50 million is the notional amount for the junior section. The first $50 million of loss to the reference asset due to a credit event (explained later) is absorbed by the junior section. Only after the junior section absorbs the first $50 million in loss will the senior section realize any loss.

You may wonder why we refer to senior and junior “sections” rather than senior and junior “note holders.” The reason is that in a synthetic CDO structure, no debt obligations are issued to fund the senior section. However, for the junior section, debt obligations are issued. In our illustration, $50 million of junior notes are issued. They are issued in the same way as in a cash CDO structure. That is, there is typically several tranches of junior notes issued by the special purpose vehicle (SPV). There will be the most senior tranche of the junior notes and there will be the subordinate/equity tranche.

The proceeds received from the issuance of the junior notes are then invested by the asset manager. However, the investments are restricted to high quality debt instruments. This includes government securities, federal agency debentures, and corporate, mortgage-backed, and asset-backed securities rated triple A.

Now we introduce the key to a synthetic CDO—a credit derivative instrument. An interest rate derivative is used by an investor to protect against interest rate risk. (We actually illustrated the use of one type of interest rate derivative, an interest rate swap, earlier when we demonstrated the economics of an arbitrage CDO transaction.) A credit derivative, as the name indicates, is used to protect against credit risk. The type of credit derivative used in a synthetic CDO is a credit default swap. Here we discuss the essential elements of a credit default swap, just enough to understand its role in a synthetic CDO.

A credit default swap is conceptually similar to an insurance policy. There is a “protection buyer” who purchases protection against credit risk on the reference asset. In a synthetic CDO, the insurance buyer is the asset manager. The protection buyer (the asset manager in a synthetic CDO) pays a periodic fee (like an insurance premium) and receives, in return, payment from the protection seller in the event of a “credit event” affecting any asset included in the reference asset. Who is the seller of the protection seller? It is the SPV on behalf of the junior note holders.

As with an interest rate swap, a credit default swap has a notional amount. The notional amount will be equal to the senior section, $450 million in our example.

Let’s clarify this by continuing with our earlier illustration and look at the return to the junior note holder in the structure. The junior note holders are getting payments that come from two sources:

Effectively, the junior note holders are receiving the return on a portfolio of high-quality assets subsidized by the insurance premium (i.e., the payment from the credit default swap). However, this ignores the obligation of the junior note holders with respect to the credit default swap. If there is a credit event (discussed below) that requires the junior note holders to make a payment to the protection buyer, then this reduces the return to the junior note holders. As noted earlier, the effect on a particular tranche of the junior section depends on its priority. That is, the subordinate/equity tranche is affected first before the most senior tranche and the other tranches superior to the subordinate/equity tranche is affected.

So what becomes critical for the junior note holders’ return is when it must make a payment. In credit derivatives, a payoff by the protection seller occurs when there is a credit event. Credit events are defined in the credit derivative documentation. On a debt instrument a credit event generally includes: bankruptcy, failure to pay when due, cross default/cross acceleration, repudiation, and restructuring. This credit event applies to any of the assets within the reference asset. For example, if a high-yield corporate bond index is the reference asset and Company X is in the index, a credit event with respect to Company X results in a payment to the protection buyer. If a designated portfolio of bank loans to corporations is the reference asset and Corporation Y’s loan is included, then a credit event with respect to Corporation Y results in a payment to the protection buyer.

How much must be paid by the protection seller (the junior tranches in our illustration) to the protection buyer (the asset manager)? Should a credit event occur, there is an intent that the protection buyer be made whole: The protection buyer should be paid the difference between par and the “fair value” of the securities. How this is determined is set forth in the credit derivative agreement.

What is the motivation for the creation of synthetic CDOs? There exist two types: synthetic balance sheet CDOs and synthetic arbitrage CDOs. In the case of a synthetic balance sheet CDO, by embedding a credit default swap within a CDO structure, a bank can shed the credit risk of a portfolio of bank loans without having to notify any borrowers that they are selling the loans to another party, a requirement in some countries. No consent is needed from borrowers to transfer the credit risk of the loans, as is effectively done in a credit default swap. This is the reason synthetic balance sheet CDOs were initially set up to accommodate European bank balance sheet deals.

For a synthetic arbitrage CDO, there are several economic advantages of using a synthetic CDO structure rather than a cash CDO structure. First, it is not necessary to obtain funding for the senior section, thus making it easier to do a CDO transaction.¹⁵⁶ Second, the ramp-up period is shorter than for a cash CDO structure since only the high-quality assets need be assembled, not all of the assets contained in the reference asset. Finally, there are opportunities in the market to be able to effectively acquire the assets included in the reference asset via a credit default swap at a cheaper cost than buying the assets directly.¹⁵⁷ It because of these three advantages that issuance of synthetic CDO structures has increased dramatically since 2001 and is expected to continue to increase relative to cash CDO structures.