Mr Cooper Group Inc (COOP)

[Petewamu]

Should Financial Institutions Use Covered Bond Financing? by
Karan Bhanot and Carl Larsson1
Abstract
Recent bond offerings by US Banks are backed by collateral that is ring-fenced but held on the balance sheet of the bank (also called “covered bonds”). Using a theoretical model we show that contingent liquidity provided by covered bonds benefits equity holders especially when their private assessment about the quality of collateral is high, and there is limited access to bank deposit growth. The cost of covered bond financing is borne to an extent by the taxpayers depending on the nature of collateral. Regulatory uncertainty coupled with higher leverage makes covered bond financing more costly for precisely those institutions that could benefit from contingent liquidity. We relate the agency cost implications of covered bond financing to hand-collected quarterly cover pool statistics of a recent US covered bond issue by Washington Mutual.
JEL classification: G2, G21, G28
1 Professor of Finance and Graduate Student respectively, Department of Finance, College of Business Administration, University of Texas at San Antonio, One UTSA Circle, San Antonio, TX 78249. Emails [e-mail] and [e-mail]. Phone number for both authors: 210-458-7429, Fax for both authors: 210-458-6320. Initial work on this paper was completed when the first author was a Visiting Professor at the Stern School, New York University. Corresponding author- [e-mail].
1

Should Financial Institutions Use Covered Bond Financing?
Abstract
Recent bond offerings by US Banks are backed by collateral that is ring-fenced but held on the balance sheet of the bank (also called “covered bonds”). Using a theoretical model we show that contingent liquidity provided by covered bonds benefits equity holders especially when their private assessment about the quality of collateral is high, and there is limited access to bank deposit growth. The cost of covered bond financing is borne to an extent by the taxpayers depending on the nature of collateral. Regulatory uncertainty coupled with higher leverage makes covered bond financing more costly for precisely those institutions that could benefit from contingent liquidity. We relate the agency cost implications of covered bond financing to hand-collected quarterly cover pool statistics of a recent US covered bond issue by Washington Mutual.
2

Should Financial Institutions Use Covered Bond Financing?
1. Introduction
Covered bonds are debt obligations of banks that are secured by a pool of assets, usually consisting of residential mortgages. The cover pool is ring-fenced, i.e., the assets serve as collateral for the covered bonds only. Also, the cover pool assets remain on the balance sheet of the issuer (Section II provides an overview of the covered bond market). While covered bonds have a long history in Europe (Bernanke 2008), recent issues by Washington Mutual and Bank of America constitute the first two issues of covered bond offerings in the US. A Federal Deposit Insurance Corporation (FDIC) policy statement notes2:
“Insured depository institutions are showing increasing interest in issuing covered bonds. Proponents argue that covered bonds provide new and additional sources of liquidity and diversity to an institution's funding base.”
After these bonds were first introduced in the United States in 2006, the Federal Deposit Insurance Corporation (hereafter FDIC) issued a Covered Bond Policy Statement on actions it would take in handling the covered bonds of a failed financial institution. However, the FDIC gives little indication about how the collateral of covered bonds would be treated in the case of issuer bankruptcy. Ideally, covered bond holders should hold a priority claim on the cover assets. Uncertainty surrounding the legal ownership of the cover assets during bankruptcy, however, can decrease recovery rates expected by investors, and thus increase the cost of financing via covered bonds. This paper asks the following questions not previously addressed in the literature:
• When is the liquidity provided by covered bonds optimal relative to other sources of liquidity? How does regulatory uncertainty about the status of covered bonds impact this choice?
2See FDIC policy statement at: http://www.fdic.gov/news/news/press/2008/pr08060a.html.
3

• What are the systemic implications of covered bond financing and the overall costs imposed by the banking system on the FDIC?
We answer the questions posed using a framework in the spirit of Flannery (2010). We start with a two period model wherein a bank may encounter a profitable project in the interim. These assets can be financed by covered bonds, unsecured debt or securitization via a Special Purpose Entity (SPE). We address the costs and benefits of each mode of financing from the perspective of the firm’s equity holders.
Our model shows that covered bond financing is costly for existing equity holders relative to unsecured debt when there is full information because equity holders implicitly insure the covered bond holders. However, the upside potential of the asset payoffs that accrue to equity holders may exceed these costs in some instances. When assets are financed via unsecured debt, the cash flows from existing and new assets are pooled and cannot be disentangled. The comingling of new and existing assets changes their risk profile and alters the probability of insolvency. Equity values reflect the impact of the change in the asset risk profile and the costs of insolvency plus the gains on the upside. Hence the incremental benefit of covered bonds over unsecured debt accrues only if the tax benefits coupled with the comingling of assets and their impact on the probability of bankruptcy exceeds the corresponding payoffs for covered bonds.
Covered bond structures have certain similarities with securitization in that investors have a claim to a ring-fenced, bankruptcy remote pool of assets that is segregated from the parent bank. However, covered bonds are distinct from securitization insofar that investors have recourse to both the cover pool and the parent bank. Also, the cover pools assets can be switched in the interim. The key benefit of an SPE is that it can be structured so that there is no event that may trigger default. In the absence of bankruptcy costs and with full information, we show that payoffs to Equity holders for SPE financing and covered bond financing are similar.
Our analysis focuses primarily on US covered bonds wherein potential covered bond issuers include US depository institutions that hold deposits insured by the FDIC. We examine the change in expected cost borne by the FDIC when the capital structure consists of covered
4

bonds. When project financing opportunities are correlated, many banks may use covered bonds at the same time thereby increasing the amount of covered bond financing in the system. In the aggregate, this may increase the overall perceived systemic risk under certain conditions. Counteracting this rise in systemic risk is that weaker institutions are naturally discouraged from the use of covered bond financing because of regulatory uncertainty.
An interesting feature of covered bonds is that the cover pools are actively managed by the issuer. This allows nonperforming collateral to be replaced on a periodic basis. In the absence of a vigilant asset monitor, however, it also makes the cover pool susceptible to ex-post manipulation. In particular when the private assessment of asset quality is low, the issuer has an incentive to move collateral to the cover pool and switch it with better quality assets. We motivate a theoretical analysis of the potential for these incentives with hand-collected quarterly cover pool statistics of a recent US covered bond issue.
Our paper contributes to the literature on the moral hazard problem between banks and providers of deposit insurance (e.g., Merton (1977), Kareken and Wallace (1978), and Boyd, Chang, Smith (1998)). Covered bonds allow issuing institutions to keep loans on balance sheet in contrast with the traditional securitization model whereby credit risk is transferred to end investors. Thus, the structure of covered bonds solves issues examined in the literature that allow financial intermediaries to dominate direct investment, including monitoring (Diamond 1984), collateral and monitoring (Rajan and Winton (1995)) information reliability (Leland and Pyle (1977), Campbell and Kracaw (1980)), and banks as commitment mechanisms (Calomiris and Kahn (1991)). The covered bond structure provides an alternative solution that mitigates the moral hazard problem of loan sales and securitization identified in Pennacchi (1988) and in some cases. The paper also related to the stream of research on securitization (Wynant (1980), John and John (1991), Gorton and Souleles (2007), Gorton and Metrick (2012)).
The article is organized as follows. Section II has the institutional background. Section III outlines the basic model and our benchmark case and provides a model of covered bond financing with full information. Section IV analyzes the optimality of covered bonds in the
5

presence of regulatory uncertainty. Section V discusses other sources of financing, including unsecured debt and securitization. Section VI analyzes the impact of covered bonds issuance for the FDIC. Section VII discusses the role of private information and agency issues related to the Washington Mutual experience on covered bond issuance. Section VIII concludes.
II. Institutional Background
Covered bonds are debt obligations of banks secured by a ring-fenced, bankruptcy remote pool of assets, the cover pool. The cover pool is typically overcollateralized, remains on the balance sheet of the issuer, and is dynamically managed such that any nonperforming loans must be periodically (e.g., on a monthly or quarterly basis) replaced with eligible collateral. As long as no default occurs, the interested and principal payments made on covered bonds originate from the general cash flows of the issuer. In the case of issuer bankruptcy, covered bond investors retain a dual-recourse claim which consists of (1) a priority claim on the ring-fenced cover pool, and (2) an unsecured claim on the general insolvency estate of the bank. See Figure 1 for a stylized diagram of key covered bond characteristics.
Covered bonds have a long history that can be traced back to 18th Century Prussia. Following the destruction of the Seven Years’ War (1756 to 1763), the Prussian nobility faced a credit shortage that threatened to hamper rebuilding efforts (Verband Deutscher Pfandbriefbanken (2005)). As a solution, Frederick the Great of Prussia introduced the first covered bonds, or in German, Pfandbriefe, literally, “letters of pledge” (Quirk (2010)). Today, roughly 20% of European Union private mortgages outstanding are financed by covered bonds, and total European covered bond principal outstanding exceeds 2.5 trillion Euros (ECBC (2011)).
More recently, Washington Mutual and Bank of America became the first US-based financial institutions to issue covered bonds in 2006 and 2007, respectively. In 2008, the Federal Deposit Insurance Corporation (FDIC) released a covered bond policy statement and the
6

US Treasury released a best-practices guide to support the fledgling market and potentially foster additional covered bond issuance by US financial institutions.
Despite the efforts of US regulators to establish a set of rules and best-practices for covered bonds, the US does not currently have statutes in place needed to codify the legal status of the on-balance sheet, ring-fenced cover pool in the case of issuer bankruptcy. As a result, US firms have taken to issuing what are typically referred to as structured covered bonds. As an example of a structured covered bond, we present the Washington Mutual Covered Bond Program structure in Appendix A. Washington Mutual issues Mortgage Bonds secured by an on-balance sheet pool of residential mortgages. These Mortgage Bonds are held by a statutory trust, and are then used as security for covered bonds issued to investors. In the case of issuer insolvency, the Mortgage Bonds are accelerated and proceeds to the trust are invested in a guaranteed investment contract that provides the cash flows needed to continue making coupon and principal payments on the covered bonds
Of course, the cost of funding via covered bonds is tied directly to the perceived value of the underlying cover assets during times of financial distress. Importantly, covered bonds are predominantly backed by pools of high-quality assets. These typically consist of high-quality residential mortgages, though commercial mortgages, shipping loans, and public sector debt may sometimes be used. In legislative covered bond countries, the quality of the cover assets can be mandated. For example, legislation can set minimum requirements on loan-to-value ratios, geographical distribution, and rating levels with which all cover assets must comply (Packer, Stever, Upper (2007)). In the US, the FDIC covered bond policy statement and US Treasury best practices document provide guidance for the type and quality of cover assets.
Investors in covered bonds depend on an appointed asset monitor to monitor and periodically assess the loans and other eligible assets within the cover pool. The asset coverage tests involved checking that an adjusted measure of the aggregate loan amount remains above the aggregate principal amount of the outstanding bonds and ensuring that all cover assets meet minimum underwriting requirements related to loan-to-value ratios, FICO scores, and
7

delinquency rates, among others. Failure to meet minimum requirements can trigger technical default as well as an increase in required credit enhancements. Some monitoring responsibilities also fall on the ratings agencies, such as Standard and Poor’s and Fitch, whose expected loss rates on underlying mortgages affected the minimum acceptable standards of the asset coverage tests.
Cover pools are dynamically managed by (and at the expense of) the issuing institution; that is, banks are required to bolster the cover pool with additional eligible mortgages should it fail to meet any of the required asset coverage tests. The result of this feature of covered bonds is that it allows covered bond issuers to “put skin in the game”. Covered bond issuers bear the full brunt of the pool’s expected losses (default rate x loss given default). This stands in sharp contrast to the “originate to distribute” aspect of traditional securitization structures in which much of the credit risk of the underlying pool can be transferred to outside investors.
The demand for covered bond products is evidenced by the issuance of around $30 billion of US dollar denominated covered bonds in 2010 (Marlatt, (2011)). In 2013, the Royal Bank of Canada issued the first covered bonds registered with the US Securities and Exchange Commission (SEC).
III. Model – The Benefit of Covered Bonds
A. Existing assets and liabilities of a bank and the investment opportunity
Figure 2 depicts the capital structure of a bank and the evolution of bank assets between t=0
and t=2. The assets are financed by insured deposits ( D0 ) and equity ( E0 ) at time t=0.3 The value of unlevered assets at t=0 is denoted G0 . We first exclude unsecured debt from the
financing mix. This provides considerable reduction in complexity without changing the main results.
3 These are FDIC insured deposits.
8

The risk-free rate is set to zero and all market participants are risk neutral. The value of secured deposits is D0 . These deposits are returned to the depositor at t=2. Because secured
depositors are backed by deposit insurance, they are made whole in case of insolvency. Therefore, the face value of secured deposits equals the t=0 deposit: FD = D0 . In other words,
depositors do not require a premium for bankruptcy risk.
Assume that the bank encounters an investment with probability q at t=1 that requires a cash
outflow, N1 . The unlevered assets at t=1 conditional on the financing are G0 + N1 . We assume that the time 2 asset values are uncertain and the return is uniformly distributed between
a high and low value: rG ?[rH ,rL ] so that the corresponding asset values are G2 ?[GH ,GL ] where GL < FD < GH . Therefore with risk free rates at zero, the expected present value of
time 2 asset values equals the current value of the asset: G0 = GH + GL . Similarly, the new 2
asset values change in consonance with the existing bank assets rN ?[krH ,krL ] where k is a constant so that N2 ? [NH , NL ] and N1 = N H + N L .
2
Case when bank does not invest (q=0)
First consider the case where there is no organic deposit growth, and the bank does not
invest at t=1. This is equivalent to the case where q=0. The end-of-period unlevered asset value, G2 , determines the payoffs to the firm’s depositors and equity holders. At time t=2 the
bank survives with probability pno such that asset values increase so that depositors can be repaid in full. Insolvency occurs at t=2 with probability (1- pno ), when G2 < FD . The
probability of these events depends on leverage: p no (G0 , FD ) = GH - FD . As in standard GH -GL
9

structural models of default, the probability of survival is decreasing in the extent of leverage: p''no (FD )< 0 . For simplicity, we assume that equity holders do not receive any payouts in
bankruptcy. Thus, the equity value Eno at time t= 0 is given by the payoffs were the asset 0
values to increase:
no??GH -D0 ??
E0 =?(G2-D0)f(G)dG=p?? 2 ?? (1)
no
G2 >D0 ?? ??
where pno = GH - FD . Insured deposits do not include a tax benefit since the risk free rate is
GH -GL
set to zero, and there are no insolvency costs borne by the depositors in the presence of the FDIC
guarantee.
Investing with covered bonds
Given the benchmark case we are now in a position to analyze the implications of covered bond financing of the new assets. Assume the bank realizes a new investment opportunity at
t=1 with value N1 and finances it with covered bonds. Denote the proceeds from selling covered bonds as C1 = N1, and the face value of covered bonds owed at the end of the period as
FC . The collateral for the new assets is a cover pool composed of the new assets. In addition
to the new assets the cover pool also includes a proportion, ? <1, of existing bank assets (the credit enhancement in covered bonds). Suppose that the extent of over-collateralization, ? , is large enough to pay off the covered bonds even if there is a capital loss in the final asset values:
?GL +N2 =N1 =FC (2)
Figure 2 shows the outcomes for covered bond financing. When the bank survives and the cover pool asset values are higher than the face value of covered bonds at maturity, the excess collateral is returned to the bank. However in bankruptcy, excess collateral accrues to the
10

depositors, and equity holders do not receive any inflow. The new equity value at time t=0 is given by (suppressing the dependence of the probability of each state):
( "G-D%+ ( "G-D% "N-F%+ E0cb=(1-q)*pno$ H 0'-+q*pcb$ H 0'+pcb$ H C'- (3)
) # 2 &, ) # 2 & # 2 &, !##"##$ !#####"#####$
payoff with no new investment payoff with new investment
where the superscript cb denotes covered bond financing and the probability of insolvency is pcb = pcb ((1-?)G0 ,FD )< pno . Equation (3) has two parts. The first component is the payoff
to equity holders when there is no new project investment opportunity, which occurs with probability (1-q). The second component is the payoff were the firm to sell covered bonds, with probability q. In this case, equity holders receive the residual value of the original assets, plus the return of excess cover assets after retiring the covered bonds at face value. The latter payoff is represented in the second term of the second bracket. Comparing equation (3) with our benchmark case in equation (1) gives:
Proposition 1: Equity values with covered bond financing are higher than the case when the bank does not invest in new assets when:
cb no cb no ??GH -D0 ?? cb??NH -FC ??
E0 -E0 =(p -p )???? 2 ????+qp ???? 2 ????>0
$!!!#!!!!" $!!#!!"
Proof:
. (4) Follows directly from equations (1) and (3) where pcb = pcb ((1-?)G0 ,FD )< pno .
impact of change in upside of new asets prob. of solvency
The key point made in Proposition 1 is that, in the absence of deposit growth, covered bonds offer contingent liquidity. The net benefit of this contingent liquidity is tied to the solvency of the bank (its probability of default) and the extent of upside of the new assets. Thus, there are two main factors driving the choice of covered bond financing in this case. First, covered bonds require an upfront over-collateralization where assets are transferred to the cover pool. This can reduce the value of the equity claim because the probability of insolvency increases
11

with the amount of covered bond financing (i.e., pcb = pcb ((1-?)G0 ,FD )< pno ). It follows
that if regulators were to set the over-collateralization level to be high enough, equity holders may not find it beneficial to issue covered bonds. On the other hand, covered bonds allow equity holders to gain from the upside potential of new assets, while the downside risk is borne by the FDIC. Under the current assumptions, equity holders can therefore use covered bonds to garner the benefits of volatile investments with large upside (and downside) potential, at the expense of taxpayers. Hence, in the full information case, the benefit to equity holders of choosing covered bond financing is driven by the fact that any bankruptcy costs are borne by the deposit insurance entity.
Figure 3 charts a numerical example of the t=2 payoffs to the depositors, equity holders, and covered bond holders of a hypothetical firm. Again, the FDIC cannot use the ring-fenced cover pool of an insolvent firm to pay back depositors. Thus, covered bond holders remain whole even in cases when the FDIC must tap the deposit insurance fund to pay back depositors.
IV. TheImpactofRegulatoryUncertainty
As we noted earlier, the cover pool is ring-fenced. Although the assets remain on the balance sheet of the issuer, they serve as collateral exclusively for the covered bond holders. From the investor’s perspective, covered bonds offer dual-recourse on both the underlying cover pool, and the issuing entity in the case of default. After these bonds were first introduced in the United States in 2006, the Federal Deposit Insurance Corporation (hereafter FDIC) issued a Covered Bond Policy Statement on actions it would take in handling the covered bonds of a failed financial institution. However the FDIC gives no indication about whether cover pool assets would remain segregated from other bank assets in the event of bankruptcy. The dual recourse structure and regulatory status of covered bonds issued in the US is therefore unclear.4
4 Conversations with the covered bond team at Blackrock, one of the largest fixed income portfolio management companies, reveals that these issues are at the fore when investments in covered bonds are evaluated.
12

When handling the covered bonds of a failed financial institution, the FDIC can continue to perform on the covered bond transaction under its terms, pay-off the covered bonds in cash up to the value of the pledged collateral and allow liquidation of the pledged collateral to pay-off the covered bonds. Alternately the FDIC can choose to not honor the cover pool agreement and use cover pool assets to pay depositors and other creditors. Suppose the probability of honoring
the cover pool is phon. Then the face value of covered bonds needs to be adjusted upwards to
reflect the possibility that the bondholders would not be repaid in certain cases. This is economically equivalent to a higher over-collateralization level (where the superscript ru indicates regulatory uncertainty:
ph?ruGL +NL =N0 =FC (6)
Thus the probability of insolvency with regulatory uncertainty is higher because more collateral
isrequiredtomakethebondsrisk-freepcb,ru =pcb,ru((1-?ru)G0,FD)<pcb. Inotherwords,the parent bank requires a larger equity infusion to make the bonds more palatable to investors.
Proposition 2: With regulatory uncertainty:
a. Investors in covered bonds optimally require larger overcollateralization.
b. Given the level of collateralization, regulatory uncertainty increases the cost of
covered bond financing.
Proof: Follows from the preceding discussion.
Note that weaker financial institutions are less likely able to afford higher overcollateralization requirements imposed by regulatory uncertainty. Hence it is all the more likely that these institutions that require contingent liquidity do not perceive this as a viable source of liquidity.
V. Covered Bonds vs. Unsecured Debt and Securitization
Our analysis above compares the benefits of covered bond financing when there is insufficient deposit growth to the case where the bank passes up the opportunity. How does covered bond
13

financing compare to the other options available to potential issuers? In this section, we compare covered bond financing to financing via unsecured debt and securitization.
A. Unsecured debt
Suppose the bank finances the new project assets N1 with unsecured debt. Again, the
underlying premise is that organic secured deposit growth takes time while the market for unsecured debt is deep and cash can be raised at short notice. The sequence of events is similar to Figure 2.
Secured depositors have the first claim on the firm’s assets in the case of insolvency, and remaining cash is distributed to unsecured debt holders. The face value of the newly issued unsecured debt is scaled to finance the new assets. The unsecured debt required is such that the present value of the new face value of total unsecured debt equals the new issuance (assuming no recovery on junior debt in case of insolvency),
N=pbsF , (7) 1U
where FU is the face value of unsecured debt, the superscript bs denotes balance sheet financing and the probability of solvency is pbs = p(G + N , F , F ). The new bank value
00UD
and equity value at time 0 are given by (suppressing the dependence of probability on the asset
values and face value of deposits):
Vbs =G +t(qpbsF ) 00U
(8)
(9)
tax benefits
?? ??G-D????
Ebs =(1-q) pno?? H 0 ?? +qpbs
??G+N-F-D
??
$!#!"
H U 0 +tF
0 ??????2??????????2 U????
H
$!!!!#!!!!"
payoff with no new investment
Equation (9) shows that with full information equity holders benefit from balance sheet financing when the tax benefits outweigh the change in the probability of insolvency.
Proposition 3: In the absence of deposit growth unsecured debt financing is preferred to covered bond financing when:
14

cb"GH +NH -D0 -FC % bs"GH +NH -D0 -FU % p $# 2 '&<p $# 2 +tFU'&.
Proof: Follows from equations (3) and (9). The payoff to equity holders conditional on covered bond financing equals the bank asset payout minus the value of debt claims. t represents the incremental tax benefits relative to covered bond financing per dollar of face value.
B. Securitization via Special Purpose Entities
Securitization via the transfer of assets to a special purpose vehicle is another potential choice for the bank. The special purpose entity (SPE) is a legal entity sponsored by the bank for the specific purpose of raising financing for new assets. The key benefit of a SPE that it can be structured so that there is no event that may trigger default. The avoidance of bankruptcy costs is a key benefit reflected in the financing costs Gorton and Metrick (2012). However, there are no tax benefits because the off balance sheet interest expenses are not tax deductible. In addition, there are set up costs that we abstract from in the discussion below.
Consider again the financing of assets with value N 0 via an SPE vs. using covered bonds. The collateral for these new assets is similar to the cover pool except that there is no
over-collateralization. Instead, the bank provides equity. Denote the proceeds from selling SPE debt as S 0 and the face value owed at the end of the period as FSPE . The face value of
the SPE debt is FSPE = NL , and the equity stake of the bank equals the balance amount. In the
case that new asset values decline, the SPE equity investors receive nothing. In this setting, when the bank maintains an equity stake, the new bank value and equity value at time t=0 are given by the original capital structure of the bank plus the equity payoff in the up state. Thus, SPE payoffs are qualitatively identical to those of covered bonds when the new asset values exceed the face value of SPE debt. In the absence of bankruptcy costs and informational asymmetries, both avenues provide similar cash flows. However, were we to include deadweight costs of bankruptcy, SPE financing would be preferred by equity holders.
15

VI. Cost to the FDIC
A key issue in covered bond financing is whether such a structure is beneficial in the context of US financial markets. The tradeoff facing US regulators stems from the fact that bank deposits are insured. Over-collateralization of the cover pool results in a lower amount of collateral for the insured depositors. The potential downside risk of equity and covered bond holders is borne by the FDIC and taxpayers.
Using the full information model, it is possible to estimate expected losses to the FDIC insurance fund under covered bond financing. Suppose there are a number of banks in the financial system indexed i=1,2,...,B. Each realizes investment opportunities with probability
qi and has a recovery rate given insolvency of ai ? [0,1] .
Remark 1: The direct incremental cost to financing via covered bonds for the FDIC relative to
Proof:
Follows from equation (3).
the case where the opportunity is passed equals:
C
=
q(p -p )(D -aG )+q(p -p )(a?G ) ???ii i 0,i L,i ii i iL,i?? i=1 ?? $!!!!#!!!!" $!!!#!!!"??
???? FDIC B?? no cb no cb ??
costs due to change in prob. cost due to lower
?? of insolvency recovery ??
????
(10)
The expected cost to the FDIC deposit insurance fund is something that must be weighed against other potential costs and benefits of allowing U.S depository institutions to issue covered bonds backed by a ring-fenced, bankruptcy remote cover pool. This issue has made its way into policy discussions and has helped shape proposed covered bond bills in both the US Senate and House of Representatives. For example, in a hearing on July 22nd, 2011 on the House bill (H.R. 940), sponsored by Scott Garrett (R-NJ), Congressman Barney Frank proposed an amendment that would allow the FDIC, when handling a firm under receivership, the power to repudiate an issuer’s covered bonds by paying the principal and accrued interest back to investors early, ahead of schedule. This amendment would allow the FDIC greater flexibility and reduce expected shortfalls to depositors, but would also introduce additional prepayment and reinvestment risk to
16

covered bondholders not typically observed in most European covered bonds. The amendment was eventually defeated 28 to 26, but the debate is on-going.
VII. AgencyIssuesandtheManagementoftheCoverPool
A. Private Information and cover bond financing
Securitization markets are an important source of financing for both financial and non-financial corporations. However, certain securitization markets collapsed in 2008 when issuance volumes declined significantly.5 An oft-cited explanation for this collapse lies in investors’ concerns over asymmetric information around underlying asset valuations. This section will examine covered bond financing in the context of information asymmetries.
Consider the case where equity holders of the bank have private information about the quality of the new assets. We think of private information in terms of knowledge about the volatility of the asset values. Thus, when the market perceives volatility to be lower than its
true, privately known value, the projected payoffs for each state are: N pvt > N mkt and HH
N pvt < N mkt , where pvt and mkt denote the private and market projections, respectively. The LL
inequalities are reversed when private information indicates a lower volatility than the market perception.
Proposition 4: When private information regarding the collateral indicates that there is more upside potential and downside risk, the benefit of covered bond financing to existing equity holders is higher because of the lower overcollateralization required by the market:
??Npvt -F ?? Ecb,pvt -Ecb,mkt =q(pcb,mkt -pcb,pvt)???? H C ????
00 ??2??.
Proof: Using the condition for computing the overcollateralization: ?GL + NL = N1 , it follows
that ? pvt > ? mkt when N pvt < N mkt . Hence the market requires lower insurance via LL
5 See www.sifma.org
17

overcollateralization. Because the probability of survival is decreasing in leverage, pcb, pvt < pcb,mkt , using equation (4) gives the desired result.
Proposition 4 also implies that when faced with a choice between Securitization and Cover Bond Financing, it is possible that cover pool financing is not viable because equity holder costs from the increased probability of financial distress outweigh any upside benefits. In such instances the bankruptcy remote SPEs are a viable outcome.
B. Management of the Cover Pool
A special feature of the cover pool is that the bank may switch cover assets from time to time after the issuance of covered bonds. The cover pool administrator maintains oversight but may not have all information that is private to the bank about asset values. Suppose a bank receives private information about the quality of its own assets and the likelihood of its own bankruptcy. The equity holders may have an incentive to move assets with lower private valuations into the cover pool and take out assets that have potentially higher values. The incentive to switch is large when its private assessment of its own bankruptcy is larger than the market view.
This section examines the Washington Mutual Covered Bond Program cover pool composition prior to Washington Mutual’s seizure by the Office of Thrift Supervision (OTS) and subsequent receivership under the FDIC. We will analyze a hand collected, quarterly time series of Washington Mutual Covered Bond Program investor reports with the goal of motivating the above conjecture that equity holders may have an incentive to switch cover assets ex-post.
For each quarter between Q1 2007 and Q4 2012, we have the following statistics: cover pool mortgage type composition, overcollateralization measures, delinquency rates, documentation distributions, weighted-average FICO score, and weighted-average LTV ratio. Summary statistics for the entire period are reported in Table 1, and charts for each item are presented in Figure 4.
18

Despite Washington Mutual’s fall into receivership, none of the key statistics broke the minimum required levels during our sample period. For example, the mortgage bond excess credit chart in Figure 4 shows the percentage of excess credit, or overcollateralization, increases from an average of about 20% to around 50% in 2008. This sudden increase in overcollateralization was trigged by a downgrade of Washington Mutual’s general senior unsecured credit rating during the summer of 2008. Further, delinquency rates are kept below the minimum required level throughout the period, and we observe in Figure 4 that delinquent loans appear to either be cured or replaced with current loans by the next reporting period for the majority of the sample.
That key ratios are met does not, however, mean that the underlying composition of the mortgages remained unchanged during the sample period. The charts in Figure 4 clearly show changes in average underwriting statistics. There is a slight, but noticeable deterioration in average FICO score and loan-to-value ratio throughout the sample period. The composition of the mortgage pool by loan type (e.g., interest only, payment option, etc.) also changes through time. Table 1 measures averages for a number of underwriting statistics on the Washington Mutual cover pool both before and after Washington Mutual’s acquisition by JP Morgan Chase in September, 2008. Difference of means tests reveal statistically significant changes in the majority of the categories, including the proportion of interest only loans, the weighted average loan-to-value of the pool, the weighted average FICO score of the pool, the proportion of loans that are 30-59 days past due, and the proportion of loans that are originated with low versus full documentation.
Without loan-level cover pool data, it is not possible to conclude the extent to which any potential deterioration in the quality of the cover pool is due to issues of risk-shifting or moral hazard versus simply remnants of a market-wide drop in loan quality. However, we can conclude that the cover pool quality clearly does change through time, within the constraints of the bond’s covenants. Any potential manipulation of the cover pool in times of issuer distress is counterbalanced by high underwriting standards as well as objective triggers that increase the
19

minimum overcollateralization level in response to issuer credit rating downgrades. Simple covenants such as these, coupled with periodic surveillance by a third party asset monitor, can certainly be effective in mitigating the ability of the covered bond issuer to potentially change the composition of the cover pool assets to the detriment of the covered bondholders.
Given the ability of issuers to actively manage the cover pool, it is important to stress the importance of high underwriting standards coupled with a vigilant asset monitor. In related work, Snowden (2010) examines a type of mortgage-backed bond that gained popularity among US mortgage companies in the late 19th Century. The bonds were similar in structure to present-day covered bonds, but did not have verifiable underwriting statistics or an active asset monitor. He argues the mortgage bond issuer’s reputational capital and willingness to bear the credit risk of the pool was not enough to keep underwriting standards from deteriorating in the absence of regulatory oversight of the bonds and their underlying collateral. Our empirical results corroborate this viewpoint.
VIII. Discussion and Conclusions
We investigate the impact of covered bond issuance on the various claimants of a bank. We show that a bank’s private assessment about the quality of collateral determines whether covered bond financing is optimal relative to financing via off-balance sheet special purpose entities or unsecured debt. A bank has an incentive to use a cover pool in times of liquidity constraints and when its private information about the quality of assets differs from the consensus view of the market. Specifically, covered bonds allow the bank to minimize financing costs relative to alternative funding choices when expected losses to the cover pool are below a certain threshold. When there is limited access to bank deposit growth, the cover pool offers an avenue for immediate liquidity.
The public policy debate on covered bonds ought to focus on their proven track record of providing a liquid market for privately financing mortgages, while also considering the potential social costs. One such tradeoff demonstrated in our model is between the funding costs of the
20

bank and the potential increase in cost to the FDIC insurance fund. This paper provides a starting point for these, along with other, important issues that are currently being fleshed out by policy makers and potential covered bond issuers.
21

Appendix:
A. Sample of a covered bond contract
The Washington Mutual Covered Bond Program structure is presented below. Washington Mutual issues Mortgage Bonds secured by an on-balance sheet pool of residential mortgages. These Mortgage Bonds are held by a statutory trust, and are then used as security for covered bonds issued to investors. In the case of issuer insolvency, the Mortgage Bonds are accelerated and proceeds to the trust are invested in a guaranteed investment contract that provides the cash flows needed to continue making coupon and principal payments on the covered bonds
Because the covered bonds are denominated in Euros, the SPV entered into a fixed-for-floating, USD-for-Euro foreign exchange swap agreement to change floating rate, USD denominated receipts from the underlying mortgages into fixed rate, Euro denominated payments to investors.
An asset monitor is appointed to periodically assess the quality and overcollateralization of the cover pool.
22

Bibliography
Ben, S. Bernanke. (2008). The future of mortgage finance in the United States: a speech at the UC Berkeley/UCLA Symposium: The Mortgage Meltdown, the Economy, and Public Policy, Berkeley, California, October 31, 2008: Board of Governors of the Federal Reserve System (US).
Besanko, David, & Thakor, Anjan V. (1987). Collateral and rationing: sorting equilibria in monopolistic and competitive credit markets. International Economic Review, 28(3), 671-689.
Bester, Helmut. (1985). Screening vs. rationing in credit markets with imperfect information. The American Economic Review, 75(4), 850-855.
Boyd, John H., Chang, Chun, & Smith, Bruce D. (1998). Moral Hazard under Commercial and Universal Banking. Journal of Money, Credit and Banking, 30(3), 426-468. doi: 10.2307/2601249
Calomiris, Charles W, & Kahn, Charles M. (1991). The role of demandable debt in structuring optimal banking arrangements. The American Economic Review, 497-513.
Campbell, Tim S, & Kracaw, William A. (1980). Information production, market signalling, and the theory of financial intermediation. The Journal of Finance, 35(4), 863-882.
Chan, Yuk-Shee, & Thakor, Anjan V. (1987). Collateral and competitive equilibria with moral hazard and private information. The Journal of Finance, 42(2), 345-363.
Diamond, Douglas W. (1984). Financial Intermediation and Delegated Monitoring. The Review of Economic Studies, 51(3), 393-414. doi: 10.2307/2297430
European Covered Bond Council. (2011). European Covered Bond Fact Book. Brussels, Belgium. Flannery, M. (2010). "Stabilizing large financial institutions with contingent capital certificates."
CAREFIN Research Paper (04).
Gorton, Gary, & Metrick, Andrew. (2012). Securitization. National Bureau of Economic Research
Working Paper Series\, No. 18611\.
Gorton, Gary B, & Souleles, Nicholas S. (2007). Special purpose vehicles and securitization The Risks
of Financial Institutions (pp. 549-602): University of Chicago Press.
John, Teresa A, & John, Kose. (1991). Optimality of project financing: Theory and empirical
implications in finance and accounting. Review of Quantitative Finance and Accounting, 1(1),
51-74.
Kareken, John H, & Wallace, Neil. (1978). Deposit insurance and bank regulation: A partial-equilibrium
exposition. Journal of Business, 413-438.
Leland, Hayne E, & Pyle, David H. (1977). Informational asymmetries, financial structure, and financial
intermediation. The Journal of Finance, 32(2), 371-387.
Marlatt, J.R. (2011, May 17). Covered Bonds: 2011 -- The House. Morrison and Foerster.
Merton, Robert C. (1977). An analytic derivation of the cost of deposit insurance and loan guarantees an
application of modern option pricing theory. Journal of Banking & Finance, 1(1), 3-11.
Packer, Frank; Stever, Ryan; Upper, Christian. (2007). The covered bond market. BIS Quarterly Review,
September, 2007.
23

Pennacchi, G. G. (1988). "Loan Sales and the Cost of Bank Capital." The Journal of Finance 43(2): 375-396.
Quirk, Patrick. (2010). Cover Me: The Economy Is on Fire (The German Pfandbrief). German Law Journal, 11(1323-1346).
Rajan, Raghuram, & Winton, Andrew. (1995). Covenants and collateral as incentives to monitor. The Journal of Finance, 50(4), 1113-1146.
Verband Deutscher Pfandbriefbanken (Association of German Pfandbrief Banks). (2005). History of the Mortgage Bank Act. Retrieved July 2012, from http://www.pfandbrief.de/cms/_internet.nsf/tindex/en_116.htm
Washington Mutual Covered Bond Program. (September 22, 2006). Prospectus for €20 billion Covered Bond Program.
Wynant, Larry. (1980). Essential elements of project financing. Harvard Business Review, 58(3), 165-173.
24

Figure 1: Covered bond structure
This figure depicts a stylized covered bond structure that highlights the dual-recourse claim of investors and the important features of the over-collateralized, dynamically managed cover pool.
25

Figure 2: Bank with Covered Bond Financing
This figure depicts the initial balance sheet structure of a bank. The bank is financed by insured deposits ( D0 ) with face value FD and equity ( E0 ). The value of original, unlevered assets of the bank is denoted by G, and the value of the new investment project is denoted by N. The probability of realizing a new investment opportunity is q and the probabilities of insolvency with and without covered bond financing are pcb and pcb , respectively.
q G0
pcb
G0 +N1
1- pcb
FD ≤(1-?)G2 ≤(1-?)GH (1-?)G2 <FD
FD ≤G2 ≤GH G2 <FD
1-q
G0
pno
1-pno
26

Figure 3: Payoffs with Covered Bond Financing
This figure presents a payoff diagram for equity, depositors, and covered bond holders as a function of the firm’s asset value at t=2. The original asset value is G0 =100 and the value of
the new investment at t=1 is N1 = 10 . We set k=1 so that the end-of-period new project value, N2 , changes in consonance with G2 , and G2 +N2 ?[GH +NH =80,GL +NL =120]. The face
value and initial amount of deposits are set at FD = D0 = 90 . The overcollateralization of the cover pool is set to 1.5 times the face value of the covered bonds, such that ?=1.5and
?GL + N2 = N1 = FC =10 . Equity holders receive the residual amount after depositors and covered bond holders have been paid in full.
27

Figure 4: Cover Pool Statistics
This figure presents Washington Mutual Covered Bond Program cover pool statistics hand collected on a quarterly basis between first quarter, 2007, and fourth quarter, 2012, through time. We chart the cover pool delinquency distribution (the red line shows proportion of loans that are 30-59 days past due), the weighted average FICO score, the weighted-average loan-to-value ratio, the mortgage bond excess credit provided by overcollateralized loan pool, and the composition of the cover pool by interest only versus payment option products (the uncharted balance consists of hybrid ARMs with various fixed interest period lengths).
0.04 0.035 0.03 0.025 0.02 0.015 0.01 0.005 0
Cover Pool Delinquency Distribution
Date
60 days or more
30 to 59 days
760 755 750 745 740 735 730 725 720
Non-Zero Weighted FICO Score
Date
0.68 0.66 0.64 0.62
0.6 0.58 0.56 0.54
Weighted-Average Loan-to-Value Ratio
Date
60% 50% 40% 30% 20% 10%
0%
Mortgage Bond Excess Credit
Date
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1
0
Cover Pool Mortgage Composition
Interest Only Products (Proportion of Total)
Payment Option Products (Proportion of Total)
Date
28
Proportion of Cover Pool Weighted Average LTV Ratio % of Cover Pool
Excess Credit % Weighted Average FICO Score

Table 1: Washington Mutual Cover Pool Statistics
This table summarizes key cover pool statistics hand collected on a quarterly basis from the Washington Mutual Covered Bond Program investor reports provided by JP Morgan Chase between 2007Q1 and 2012Q4. We report statistics for the full sample, as well as for the samples before and after Washington Mutual's bankruptcy, along with differences in means and associated t-statistics (*** signifies p-values < 0.01, ** p-values < 0.05, and * p-values < 0.10).
Full Sample (n = 24) 2007Q1-2008Q3 (n = 7) 2008Q4-2012Q4 (n = 17) After - Before
Variable Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Mean t-statistic
Excess Credit Support
Millions of USD
Outstanding mortgage bonds
Total loan balance
Percentage of outstanding mortgage bonds
Excess credit support
Product Type Distribution
Proportion of total cover pool
Interest only products Payment-option products
Loan-to-Value and FICO Score
Weighted average LTV Weighted average FICO
Delinquency Rates
Proportion of total cover pool
Current
30-59 days past due 60+ days past due
Documentation Distribution
Proportion of total cover pool
Full
Low Streamline/Unknown/None
7,036.72 1,190.51 10,036.11 1,768.73
0.43 0.12
0.58 0.11 0.24 0.08
0.63 0.02 743.50 8.50
0.99 0.01 0.01 0.01 0.00 0.00
0.36 0.07 0.56 0.07 0.08 0.01
7,397.80 1,022.58 9,327.47 1,430.42
0.26 0.11
0.47 0.09 0.24 0.12
0.60 0.01 753.29 3.15
0.99 0.00 0.01 0.00 0.00 0.00
0.42 0.04 0.50 0.04 0.07 0.01
6,888.04 1,250.89 (509.76) -0.95 10,327.91 1,849.22 1,000.44 1.28
0.50 0.01 0.24 *** 9.14
0.63 0.06 0.17 *** 5.09 0.25 0.07 0.01 0.17
0.64 0.01 0.04 *** 5.72 739.47 6.40 -13.82 *** -5.39
0.98 0.01 -0.01 * -2.01 0.02 0.01 0.01 * 1.86 0.00 0.00 0.00 0.87
0.34 0.07 0.58 0.07 0.08 0.01
-0.08 *** -2.83 0.07 ** 2.52 0.01 ** 2.29
29