In the absence of significant new non-agency mortgage-backed security (MBS) origination, banks -- in particular, European financial institutions -- have looked to covered bond (CB) issuance as a means to raise capital and improve liquidity. The practice has become more prevalent in recent years following the global financial crisis.
Covered bonds are securities backed by a collateral pool, or “cover pool,” typically consisting of mortgages. A covered bond has a dual recourse feature that provides the bondholder, in case of issuer default, a priority claim to the assets in the covered pool and senior claim to the issuer itself, ranking pari passu (i.e., on equal footing) with senior unsecured creditors.
________________________________________
The U.S. covered bond market has seen a revival since 2010 when foreign issuers opportunistically entered the market. Since then, approximately $74 billion has been issued as of January 2012, according to data compiled by Barclays, and more foreign issuers seek to tap the U.S. investor base. But the U.S. issuer covered bond market itself is almost non-existent, with just a few legacy covered bonds outstanding that were issued during the financial crisis.
In November 2011 the U.S. Covered Bond Act was introduced in the U.S. Senate by Senator Kay Hagen (D-NC) and Bob Corker (R-TN), co-sponsored by Chuck Schumer (D-NY). The bill expands on the bill introduced in the House in April 2011 but some of the outstanding issues remain unresolved, such as the FDIC’s concern about covered bondholders’ seniority over depositors, the eligible collateral appropriate for covered pools and extent of overcollateralization permitted (amassing more collateral than is necessary to obtain financing is sometimes used to improve credit ratings). If the bill finds bipartisan support in both the House and Senate, it may pass later this year. And if so, then under the current 4% bank liability limit introduced by the FDIC in 2008, the domestic covered bond market could reach $500 billion, according to a Deutsche Bank research report dated 10 January 2012. Combined with ongoing foreign issuance, we believe the U.S. covered bond market has viability.
Covered bonds versus non-agency MBS
To evaluate the risks of investing in covered bonds (CBs), we believe a bottom-up approach is required, since a covered bond risk premium consists of multiple components such as liquidity, cover pool assets cash flow valuation, seniority to unsecured holders, cross-currency basis and sovereign risk. We therefore advocate a methodology for relative valuation between covered bonds and non-agency MBS. Our proposed approach shows that increased transparency on pool collateralization can facilitate investment decisions by allowing for more informative relative value decisions across a broader set of asset classes that includes structured products and non-agency MBS in particular.
Traditional credit investors deciding whether to invest in CBs typically investigate movement in the CB’s historical basis versus nominal senior unsecured debt. Investors should also consider details on pool (over)collateralization and the possibility of recourse to the sponsor bank (issuer) in case of residual covered claims above and beyond the cover pool. In case of the latter, recourse could be junior or on a pari-passu basis with the senior unsecured debtholders. This means that investment decisions between CBs and senior unsecured debt can result from relative value views on the difference between corporate recovery rates and recovery on CB pool collateral. In practice, we have found that pool recovery estimates are based on historical performance of similar mortgage types without adjusting to forward-looking economic scenarios. Forward-looking scenario analysis is made more complicated, in part, by the revolving nature of the collateral pool, and the lack of transparency on the underlying loan characteristics.
When considering investment decisions between non-agency MBS and CBs, we believe nominal spread analysis becomes insufficient, since investors must now consider recovery and default risk under various economic conditions.
To calculate relative value or break-even analysis between CBs and MBS, we apply a factor-based approach. The attractiveness of this approach is that it provides a means to quantify default probabilities across a range of outcomes instead of analyst-defined ad hoc assumptions. For a one-factor model in non-agency MBS, arguably the single most important variable is home price appreciation (HPA). A borrower's credit risk, as measured by a range of scores and levels of financial documentation, is an important underwriting criteria. However, as ample empirical evidence over the past decade has shown, the mortgagor's default incentive remains directly tied to the amount of negative home equity. By extension, the CB issuer's default probability is also impacted by home price depreciation via (i) direct primary impacts on bank portfolio retained loans and mortgage-backed securities, and through (ii) secondary effects as housing is strongly correlated to consumer and corporate credit held on banks’ balance sheets.
On the flip side to benefits, the fact remains that models are by definition simplified versions of complex financial mechanisms. As such, they are susceptible to biases on sensitivities that possibly could lead to an incorrect relative value view. Investment analysts should therefore be aware of the possibility of systematic deviation in statistical decisions and apply the appropriate adjustments.
PIMCO’s proprietary loan-level prepay and default models can be applied to non-agency MBS in conjunction with cash-flow structuring tools to model bond level valuations across a range of HPA scenarios. Similarly, this model can be used to value collateral backing a covered bond. Without knowing the exact details on every single loan, however, general assumptions need to be imposed using the limited information that is available about vintage, product type and originated collateral performance. For issuer default, corporate default models may look to interest-coverage or capitalization ratios to estimate the amount of earnings a company generates to cover debt payments or the amount of total assets relative to total debt. Correlating these factors to home price appreciation would be challenging.
Instead, we investigate historical credit default swap (CDS) spreads as a means to quantify default risk relative to national home price appreciation. Using data from Bloomberg and Moody’s on the nation’s largest lenders since early 2000, we arrived at the statistical results with derived relationship displayed in Figure 1. Note that the shape of the surface implies larger near-term HPA shocks could have a bigger impact on cumulative default rates, which tend to be magnified over time due to the rapid deterioration of housing prices after the 2008-09 crisis and its subsequent macroeconomic impact. Through this analysis, we now have a robust set of factor-based models that allow for relative value analysis between CBs and non-agency MBS.
Figure 2 shows loss-adjusted yields (reflecting the possibility of default and recovery), as calculated by PIMCO under a range of HPA scenarios for a representative CB. This CB is backed by a collateral pool of pay-option adjustable-rate mortgage (ARM) loans originated in 2006 (pay-option ARMs allow borrowers to choose from a number of different payment options each month). The chart also shows the loss-adjusted yield for a representative senior prime pass-through (SR FLT PT) security backed by floating rate collateral originated in the same year. According to PIMCO’s proprietary mortgage prepay and default models, the CB nominal yield without credit default possibility is 3.31%. The yield, however, declines to 3.12% (a reduction of 19bps) when calculating returns on a loss-adjusted basis by incorporating (CDS implied) default likelihood and estimated recovery from the CB collateral pool using PIMCO proprietary mortgage default models. The SR FLT PT, in comparison, returns significantly higher yields in base and modest stress scenarios – yet returns negative yields in the case of severe economic (HPA) shock, as shown on the left side of the chart. Investors in residential mortgage-backed securities (RMBS) typically demand a higher return over other securities due to embedded optionality (of mortgage prepayment and default) that result in cash flow uncertainty and which do not exist -- or exist to a much lesser extent -- in other securities such as the CB.
The break-even point, or macroeconomic economic shock in which hypothetical RMBS and CB returns are equal, in this example is -27% 2YR U.S. HPA. This implies that investors who can tolerate some risk of losing a significant portion of principal may have better relative value investing in RMBS versus CB. If one’s view is one of dire economic outlook, our model implies that the higher returning asset would be the CB. Different assumptions will lead to different results.
This relative value analysis would potentially be more accurate if the exact loan level details on the CB pool were known instead of estimating collateral performance from general pool level characteristics.
Conclusion: A better relative value framework
The U.S. covered bond market today is mainly foreign influenced, i.e. risk premiums on covered bonds from foreign issuers have embedded currency risk (“basis”), aside from liquidity risk due to the infant stage of the market. The credit risk however is not fully known, neither is it appropriately priced, in our view, versus adjacent asset classes such as senior unsecured securities and residential mortgage backed securities. We believe our proposed relative value methodology has merit since most covered pools consist of residential mortgage loans. Albeit the U.S. Covered Bond Act is still likely awhile away, covered bond issuance in the U.S. is increasing as more foreign issuers enter the market, evident by last year’s launch of the first Australian covered bonds. Our methodology could also be applied to covered bonds of foreign issuers, to better understand credit risk they are carrying. Above all, it can provide us with an effective relative value framework to compare covered bonds with different but related asset classes.