Many decisions made in the CDO structuring process are a function of the arb. If certain classes are more expensive than combinations of other classes, those classes more expensive are less apt to be created.
We make this point by looking at trade-offs inherent in deal structures.
We show that greater subordination and more overcollateralization, can, at times, result in greater extension risk. It is very important for investors to examine an entire deal structure in light of their portfolio objectives.
Rules of CDO Deal Structuring
Following are two rules of CDO structuring. Rule #1 is never leave money on the table! If (all things being equal) a deal structure can support
c21-ConsiderationsCreating Page 422 Monday, March 6, 2006 11:27 AM
80% AAA rated bonds, it is unlikely that any issuer will construct that deal with 78% AAA rated bonds. So if two tranches of different deals, both rated AAA by the same rating agency, are in the market simulta- neously, then it is likely that both contain the maximum amount of AAA bonds supportable by their structures. If one deal carries a higher percent of AAAs, then trade-offs were made elsewhere in the structure.
Rule #2 is optimize deal structure. It is survival of the fittest out there. Issuers try various deal structures, and come up with one or two that look the “best.” (Structurers’ trash cans generally overflow with printouts of trials failed because they were “nonoptimal.”) Optimal structure in CDOs is that for which each of the rated notes can receive a market-determined interest rate, with IRR maximized on the equity piece. If one dealer structures an equity return of 17% while another offers 18% off the same collateral at similar leverage then an investor will clearly run, not walk, to that higher return.
Interest Costs Drive Subordination
We now look at how the CDO arbitrage and current spread configura- tion dictates structure. Note that many investors (particularly at the AAA level) look at percent subordination as an indicator of protection.
While it is certainly one such bellwether, it should not be used as the be- all and end-all. In fact, Rules #1 and #2 above are so powerful that if two tranches are created at the same time with the same rating, there is unlikely to be any strict dominance of one over the other.
Exhibit 21.4 displays three different structures, using typical combi- nations of structured finance collateral (ABS, MBS, CMBS, and REITs).
The three structures (labeled Deals A, B, and C) are backed by exactly the same collateral. The cost of that collateral was assumed to be
$97.88, which includes the CMBS IO that is often included in these deals. The diversity score is 17, also very typical of mortgage deals, and the weighted average rating factor is 345, which corresponds to the BBB level. The WAC on the collateral is 8.30%, which again, includes the effects of the CMBS IO.
In Deal A, liabilities were tranched into notes rated AAA, A−, and BB and equity, proportioned 86.67%, 9.0%, 1.67%, and 2.67%, respec- tively. Note that this structure maximized the amount of AAA rated notes permitted (shown in the middle section of Exhibit 21.4). The bot- tom part of the exhibit shows that after paying a liability holder the spreads shown separately in Exhibit 21.4, the return-to-equity (assuming no defaults) is 17.09%.
Deal B has essentially the same structure. The only difference is that the A− and BB amounts are collapsed into a BBB class. Equity yield then
424 OTHER CDO TOPICS
expands to 17.24%, although that is not all that much different from Deal A’s 17.09%.
In Deal C we recarved the AAA and A− cash flows into notes rated AAA, AA, and A−, holding constant the amounts of BB and equity. This structure would be considered suboptimal, as return-to-equity drops to 15.69%, versus Deal A’s 17.09%. No matter how gifted a salesperson is, they would not be able to sell equity at this level when other deals are in the market with equity returns 150 bp higher.
Anyway, the reason the arb is much less attractive in Deal C is due to the spread configuration. The deals shifted 10% of the AAA rated bonds (paying LIBOR + 50) and 1.67% of the A− bonds (paying LIBOR + 150) into the AA bucket, which pays LIBOR + 90. That raised our interest costs 25.7 basis points on 11.67% of the deal. By the way, note that there is a relative value implication here: The AA rated bonds are quite attractively priced.
Debt Holders Do Not Look Exclusively at Subordination
Suppose an investor is considering buying AAA rated paper. The inves- tor is trying to decide between the bond in Deal A or that in Deal C.
Which should the investor want in his portfolio? At first glance, the one in Deal C looks “better” because of its 23.23% subordination, rather
EXHIBIT 21.4 Structuring Tradeoffs
Liability Spread
Deal
A B C
Principal face 300 300 300
Total cost 97.88 97.88 97.88
Rating factor 345 345 345
WAC 8.30% 8.30% 8.30%
AAA L + 50 86.67% 86.67% 76.67%
AA L + 90 — — 11.67%
A− L + 150 9.00% — 7.33%
BBB L + 250 — 10.83% —
BB T + 700 1.67% — 1.67%
Equity 2.67% 2.50% 2.67%
Min. AAA I/C required 122 122 144
Min. AAA O/C required 110 110 128
Equity yield 17.09% 17.24% 15.69%
c21-ConsiderationsCreating Page 424 Monday, March 6, 2006 11:27 AM
than the 13.67% subordination in Deal A. Furthermore, the minimum I/C and O/C levels are much higher for Deal C’s AAA rated bond.
But do not forget that according to Rule #1, you should “never leave money on the table.” Each of the deals has already maximized the amount of AAA bonds that can be created in that structure. Thus, rating agencies consider the bonds roughly equivalent.
Many believe that the AAA rated bond in Deal C could never be worse than Deal A. That is wrong. The application-of-cash waterfall typically pours collateral interest and principal cash flow into AA inter- est payments prior to paying any AAA principal. More precisely, rating agencies will not allow a AA rated class to defer interest payments.
However, if defaults are very high, cash flows in the deal will be lower.
Thus, Deal C’s AAA rated notes are more likely to extend than are Deal A’s. Intuitively, an additional 40 basis points of interest on 10% of the deal plus the entire interest payment on 1.67% of the deal are “ear- marked” to pay the AA rated noteholders their interest; the interest on this tranche must be paid before the AAA rated notes get principal back.
CONCLUSION
In this chapter we have seen that the CDO arbitrage is the key to CDO transactions. And the arbitrage calculations also shown in this chapter allow investors to gauge activity levels. Those activity levels are espe- cially crucial to equity buyers and to CBO managers.
Equity buyers should check the equity returns pitched by an invest- ment bank. If that return is materially different from what you figure on your hand calculator—investors should find out what assumptions are being made about the structure, and be very sure that they are comfort- able with those.
Additionally, these arbitrage calculations allow potential CBO man- agers to determine when they really want to press to get a deal done.
Marginal arbitrage may portend that it is better to sit tight and wait for better timing. Deal performance is certainly important to any deal man- ager. It impacts future deals, and a manager’s own pocket is directly impacted in that he or she typically retains a large chunk of the equity.
The CDO arbitrage is also a major determinant of deal structure.
We have seen how different spread configurations and different collat- eral can make for very different deal structures. That is, deals are gener- ally optimized to maximize returns to equity holders, while making sure to pay rated note holders their appropriate market levels. Yet with very dissimilar deal structures, investors are unable to figure relative value
426 OTHER CDO TOPICS
simply by looking exclusively at subordination in a deal, or exclusively at the amount of overcollateralization. That is because the benefits of higher subordination can be offset, depending on the waterfall rules.
The benefits of higher overcollaterization can be offset by lower interest coverage ratios.
The bottom line is that since deal structures have been optimized, there are always trade-offs. Investors need to be fully aware of those choices they implicitly make.
c21-ConsiderationsCreating Page 426 Monday, March 6, 2006 11:27 AM
CHAPTER 22
427
How to Evaluate a CDO and Manage a CDO Portfolio
n this chapter, we look at evaluating CDOs individually and as part of a portfolio. We make two points about buying CDO deals individually.
First, we make the case that one of the most important points to look for in a CDO purchase is the structural protections inherent in a CDO. There is a natural tension between the interest of debtholders and equity holders that the CDO structure tries to address. Buyers of CDO debt will want to look at both the incentive structures in a CDO, as well as how the man- ager has done on outstanding CDOs. Second, in picking managers, track record cannot be taken at face value. Common sense goes a long way.
We then move on to managing a CDO portfolio. Some portfolio man- agers have quite an extensive collection of CDOs, owning 100 to 200 dif- ferent CDOs. Yet they often tend to look at buying each additional CDO as if they were buying their first. They do not place the CDOs within a general portfolio framework. In the second half of the chapter, we make the case that investors should buy CDOs backed by different types of col- lateral. Asset class is a far more important determinant of returns than is choice of specific managers. Finally, we look at diversity on a portfolio basis. A low diversity CDO may add more diversity to a portfolio because it is backed by a different type of collateral than a high diversity CDO.