Bank Risk Management and Collateralized Debt Obligations

Part I: The case against securitization: CDOs an exception to the rule?

Part II: Copula exercises

Collateralized debt obligations -CDOs- are well known instruments in the financial industry however as being one of the principal triggers of the global financial crisis -GFC- causing more than $1.3trillions losses (IMF, 2013). They originate through a process called securitization, involving the pooling, “slicing” and repacking of debt-instruments into innovative products. Although the whole sector of securitized products came under fire in the aftermaths of the crisis, the process provides important benefits for both banks and investors. On the bank’s side the transfer of securities and thus risks to a special vehicle purposes -SVP-, it improves balance sheet management by decreasing capital requirements for regulatory purposes, as well as improving liquidity from selling to a wide range of investors and can constitute a source of long-term funding and finally allow to transform illiquid assets into securities (OECD, 2011). On the other hand, investors can choose investments according to their risk profile while enjoying diversification effects as different types of loans, with different investment grades and different maturities are brought together (2011). The purpose of this paper is to establish the reasons why benefits turned into doom at the wake of the GFC, section 1 will examine structure of popular CDOs and point their role in the GFC, section 2 examine CDOs structural complexity supported by evidences while section 3 focus on remedies.

Section 1. Role in the GFC

CDOs have been around since 1987 however it is only for the period 2000-07 that issuance boomed as denoted in figure 1 (IMF, 2013). Indeed, low interest rates -1%- set by the FED back in 2001 together with the 1995 community reinvestment act allowed individuals with no prior or poor credit history to access funds for home-buying purposes and therefore encouraged subprime growth. Their issuance grew to reach $600bn from $100bn for the period 2000-06 and accounted for 20.6% of the total mortgages by 2006 (2013).

Fig. 1 Global issuance by type of securitized products

Source: IMF (2013)

Those mortgages loans were then securitized into assets backed securities -ABS- and divided into tranches (see fig. 1.2); senior, mezzanine and equity. The latter, called the leveraged or unsubordinated, carried the higher return but was the first to absorb losses in case of default followed by mezzanine and so on. The most senior, the thicker and safest received the highest rating AAA (see appendix 1 for ratings details) and then was sold to investors while unbought, riskier mezzanine tranches could be re-securitized through financial engineering into AAA CDOs senior tranches (Fligstein and al., 2010).  CDOs are thus by definition claims on ABS cashflows, which themselves are claims on mortgages. Squared and cubic CDOs were produced in a similar fashion. The higher yield on senior tranches despite similarly rated US treasury securities implied more risk thus those ratings were in fact inflated, a direct consequence of “ratings shopping”. Credit rating agencies -CRAs-, institutions responsible to assess creditworthiness of borrowers derived colossal revenues from structured finance products thus created an incentive towards upward bias and contributing to increase systemic risk. Moody’s 2006 total revenue from securitization accounted for $881bn, more than its combined profits in 2001 (IMF, 2013). The reason the GFC has been so hard to predict is because elements from the along the securitization chain -including inflated ratings- all had participated in increasing systemic risk without regulators, investors and issuers being fully aware of the situation. As adjustable interest rates on mortgages reset higher to 5.25% in 2006 (Fligstein and al., 2010) prime borrowers started to default on their obligations and the large supply of houses in the market caused prices to decline and huge losses started to emerge for banks, which were left of illiquid assets. The collapse of the CDOs market damaged other ABS reputation as denoted in figure 1.3. CDOs lost 72.5% of their par value in spite of their small account in the securitized market (ABDI, 2010). The next section will tackle CDOs complexity while demonstrating that securitization itself is not necessarily bad but rather dependent on the type of product.

Fig 1.2 ABS and CDOs structure

Source: IMF (2012)

Fig 1.3. Structured finance: mtm losses

Source: ADBI (2010)

Section II. Empirical

CDOs structure is complex but when that same structure is continuously enhanced to meet certain credit standards, risk spreading all level becomes uncontrollable. At some point, simulations determined whether a tranche could withstand a certain level of default thus could earn a higher rating (ABDI, 2010). Overcollateralizing -several collaterals for a single security- and CDS -insurance against default- became popular credit enhancement methods. The average investor cannot associate risk and returns anymore caused by asymmetric information thus making adequate risk valuation and management challenging. CDOs and ABS being sisters but not twins due to differences in structural complexity require different rating models -in fact each structured product should have its own model- and especially react differently to change in any macroeconomic factor. It is now widely known that deteriorating lending criteria were not accounted for while default correlation where largely underestimated, thus leading to increased overall systemic risk. CDOs exhibit higher sensitivity to credit risk as re-securitization of ABS mezzanine tranches implies lower assets quality, thus even the most senior tranches can be considered unsafe. Furthermore, as the process is repeated, each tranche gets smaller and smaller thus exhibit increased sensitivity to default risk. ADBI research evaluated a base scenario -BS- with alternate views of both higher probabilities of default -4.5% against 3%- and default correlation – 0.5 against 0.1-

Comes out the impact of higher default probabilities is the largest for mezzanine ABS, with 28.2% more sensitivity than in BS, which explains the large increase in the value-at-risk +79.7% and expected shortfall +36.7% of senior CDOs. Mezzanine and junior CDOs remain unchanged as they are 100% “at risk” by definition. Higher default correlations lead to a small positive effect for RMBS equity but in generally there is also increase sensitivity for the senior tranches reflected in ES at +36.7%.  This shows that senior tranches are more sensitive to change in systemic risk thus explaining the why AAA CDOs’ tranches lost between 4 to 7 notches when downgrades happened back in 2009 (IMF, 2012).

Section III. Remedies and conclusion

Reestablishing trust about securitization on the long run is essential instead of shying away from the process. On the short run increased product transparency and simpler structures to facilitate risk valuation should be efficient measures from investors’ point of view. The former can be implemented through information collection (assets’ riskiness and performance), avoiding re-securitization and shortening securitization chain (OECD, 2011).  On the other hand, limitations can arise, firstly collecting information and produce reliable measurements can be challenging when products are new and secondly in a booming economy, investors’ confidence is high thus incentive towards financial innovation increases and is reflected by more complex products with complex structures (ADBI, 2010). Changes in policies/regulations have been applied for the purpose of long-term horizon, firstly, the 5% minimum risk retention rule in effect since 2011 in Europe has been established, in order to increase originators “skin in the game” to disincentive excess risk taking and enforce due diligence (IMF, 2013). There is a lack of established consensus on which tranches retention scheme is the most effective with regard to equity -most conventional method-, mezzanine or vertical slice retention since it is highly dependent on both tranche thickness and the economic cycle (BIS, 2009). Furthermore, the amount to be retained -despite the rule- varies with each transaction and market segment. If it is too low, screening incentives could be misaligned while high retention will lead to higher securitization costs. To solve both those issues, information disclosure regarding the amount retained at issuance and over the security’s lifetime as well as whether tranche exposure have been hedged allow for flexibility while also helping regulatory bodies to help track changes the market (BIS, 2009). The same year, the ECB increase the haircut applicable to ABS from 12% to 16% however this could revive the use of overcollateralization to reach desired threshold (OECD, 2011).   The simplified supervisory formula approach -SSF- suggest that ratings could be generated internally by banks to avoid overreliance on CRAs ratings but conflict of interest between the bank’s role as regulator and profit-seeker could arise (2011). Finally, Basel III stricter capital requirements for securitized products provide additional cushion in case of losses but limit securitization positive spillovers (2013). The policies mentioned above are therefore to be manipulated with careful considerations. An alternative to ABS products is covered bonds securitization, which are very popular in Germany whereby loans are backed by a pool of high-quality assets and if default arise, the loan is replaced with a performing loan though such situation never happened since their creation (Fligstein and al., 2010).  In the united states, the bond will require to go through an SVP instead of remaining on the balance sheet and this represent a potential risk it is untested.

Appendix 1. Credit Ratings

Source: IMF (2013)

Bibliography

• Asian Development Bank Institute, (2010), Securitized products, Financial Regulation and Systemic Risk, [online] Available at: https://asianbondsonline.adb.org/publications/adb/2010/financial_regulation_systemic_risk.pdf [Accessed 27th April 2019]
• BIS, (2009) quarterly review. The future of securitization: how to align incentives. [online] Available at: https://www.bis.org/publ/qtrpdf/r_qt0909e.pdf [Accessed 27th April 2019]
• Fligstein, N., and Goldstein, A., (2010), ‘The anatomy of the mortgage securitization’, Emerald Group Publishing Limited, pp.29 – 70, [online] Available at: https://www.emeraldinsight.com/doi/pdfplus/10.1108/S0733-558X(2010)000030A006 [Accessed 27th April 2019]
• IMF, 2013. Securitisation : lesson learned and the road ahead. [online] Available at: https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Securitization-Lessons-Learned-and-the-Road-Ahead-41153 [Accessed 27th April 2019]
• OECD journal: financial markets. Outlook for securitization market. Vol 2011, (1). [online] Available at:  http://www.oecd.org/finance/financial-markets/48620405.pdf [Accessed 27th April 2019]