SME Securitization and Basel II: Europe vs America
Book: Structured Finance and Collateralized Debt Obligations | Author: Janet M. Tavakoli | Publisher: John Wiley & Sons (2008) | ISBN: 978-0-470-44344-6
The second half of Chapter 12 covers small and medium-size enterprise securitization, the secured loan trust structure, and Basel II. These are three separate topics that connect through the same underlying theme: banks trying to reduce regulatory capital charges through securitization, and regulators trying (and mostly failing) to keep up.
SMEs in Europe
Small to medium-size enterprises are typically defined as companies with annual sales under EUR 250 million, though the definition varies by program.
SME CDOs are backed by claims against SMEs: loans, receivables, undrawn revolvers, undrawn guarantees. Most of these claims are unrated. Implied ratings tend to cluster from BB+ to single A. To figure out what ratings mean in the context of an SME portfolio, originators try to map their internal rating criteria to rating agency criteria. If the originator does not have a solid internal rating system with a track record, the rating agencies may reject this mapping. In that case, historical default and loss data on the SME portfolio becomes critical. If all else fails, a rating agency may apply its own corporate rating model.
The key challenge with SMEs is data quality and stability of portfolio characteristics. If you want to use historical data to make projections about future performance, you need confidence that the character of the SME loan portfolio is not changing. That stability matters especially for revolving portfolios. These challenges contribute to a longer lead time in bringing SME CDOs to market.
The first SME CDOs were funded. Now both funded and synthetic SME CDOs exist, with synthetics dominating. They can be static or revolving.
European SME Programs
Germany’s Kreditanstalt für Wiederaufbau (KfW) is one of the biggest names in European SME securitization. KfW is guaranteed by the Federal Republic of Germany, which gives all its obligations a 0% BIS risk weighting and a AAA rating.
KfW runs the PROMISE program (Program for Mittelstand-loan Securitization) for SME loans and the PROVIDE program for non-Pfandbriefe eligible residential mortgages. KfW can consolidate loans from multiple German bank originators to bring enough volume for a single transaction.
The structure works like this: German bank originators buy credit default protection from KfW on their SME loan portfolios. From the originating bank’s perspective, they are transferring 100% BIS risk-weighted exposure to a 0% risk-weighted counterparty. They get credit protection and free up regulatory capital for new lending.
KfW then transfers the risk in one of two ways. Method one: KfW issues certificates of indebtedness linked to the mezzanine tranches of the SME portfolio, placing them in an SPV. Since the certificates repay principal with AAA certainty if there are no defaults at the mezzanine level, the AAA tranche of the SPV notes can be AAA rated. Super senior risk transfers via a super senior swap, and first-loss risk transfers separately.
Method two: KfW transfers mezzanine risk to the PROMISE SPV using credit default swaps. Note proceeds are invested in KfW-issued medium-term notes, which are themselves AAA rated. Those MTNs serve as collateral for the CDSs.
Bank Austria Creditanstalt (BaCa), an Austrian subsidiary of HypoVereinsbank, became the first non-German user of the KfW PROMISE platform in December 2002. The deal was EUR 1 billion and securitized Austrian SMEs. The EUR 90.7 million AAA tranche sold at three-month Euribor plus 38 bps.
Spain has its own program through the Instituto de Credito Oficial (ICO). ICO does not act as intermediary but serves as paying agent. Spanish SME portfolios often include ICO-sponsored loans, which have higher recovery rates than unrestricted Spanish SME loans. The Kingdom of Spain guaranteed timely payment on senior notes and gave sliding-scale partial guarantees on junior notes until 2002, when it stopped guaranteeing lower tranches. Spain was upgraded to Aaa from Aa2 by Moody’s in 2001.
Deutsche Bank’s nonpromotional SME deals: CORE used a true sale structure and a static reference portfolio. CAST issued credit-linked notes linked to tranched SME risk, similar to the Glacier structure, meaning the CLNs carried Deutsche Bank’s own credit risk (AA- at time of writing) plus the tranched SME risk.
BBVA used a synthetic structure with a revolving SME reference portfolio.
A few reinsurers entered this market by seeking diversified pools of European loans and avoiding the actively traded investment-grade names. One French reinsurer guaranteed first-loss risk (30% of the deal) on diversified European loan pools with concentration limits. Since the reinsurer was single-A rated and covered 30% of the deal, the senior tranches could achieve investment-grade ratings on acceptable SME collateral. This type of structure was expected to become more popular as Basel II came into effect.
SMEs: US vs Europe
European loans are predominantly floating rate, tied to Euribor. The euro’s introduction in January 1999 created a broader market for euro-denominated SME securitization within the European Community.
In the US, the Small Business Administration (SBA) securitized SBA loans through Guaranteed Loan Pool Certificates (GLPCs). Pools of around 50 loans, roughly $5 million each, backed by the full faith and credit of the US government. The government guarantee gives these securities 0% BIS risk weighting and AAA ratings. The government guarantees timely payment of both principal and interest.
GLPCs can have maturities from 5 to 25 years. Coupons adjust monthly or quarterly. But the coupons are indexed to the US prime rate, not USD LIBOR. Prime is a US consumer lending rate. For European investors who fund on a LIBOR basis, the prime/LIBOR spread creates basis risk. The SBA International program addressed this by creating grantor trusts that issue AAA “A” trust certificates. The trust buys premium GLPCs with the note proceeds and also issues a subordinated tranche that absorbs prepayment risk up to the 95% confidence band based on historical CPR data.
Coupons can be hedged for basis risk or left unhedged. Unhedged investors accept that the prime/LIBOR spread can move either way.
A structural note for European investors: the grantor trust structure, unlike an SPC structure, causes European banks to look through to the underlying 0% risk-weighted collateral, attracting zero regulatory capital requirement. An SPC would attract 100% risk weighting. The grantor trust structure was a deliberate design choice to improve European investor capital treatment.
Secured Loan Trusts
The first CLOs to use synthetics were Chase Manhattan Bank’s secured loan trust (SLT) structures, with maturities around three years. These arbitrage CLOs sometimes used Chase balance-sheet loans supplemented by secondary market purchases. Underlying loans were mostly highly leveraged transactions (HLTs) with implied ratings of B to BBB.
The distinctive feature: secured loan trusts use total rate of return swaps (TRORS) to transfer first-loss risk to investors, and bifurcate that risk into two types.
One tranche is less leveraged and got a single-A rating in some structures, or BBB from Duff & Phelps in others. The investor in this tranche is the noteholder. The other tranche is more leveraged and is held by the certificate holder, who is treated as the equity investor. Both tranches share first-loss risk on a pari passu basis, but with different leverage: noteholder at 5:1, certificate holder at 8:1 in the example.
The investment-grade rating on the noteholder tranche was based on return of the initial cash investment plus a nominal coupon, not on the potentially high leveraged coupon. This matters because US insurance companies drove these transactions. Insurance companies cannot invest in high-yield loans under NAIC rules. If the noteholder tranche gets a single-A rating, it achieves NAIC 1 classification, the highest, with the lowest capital charge. The insurance company earns equity-level returns while enjoying investment-grade capital treatment.
Hedge funds were the certificate buyers. They asked for full portfolio disclosure and built their own stress test models. Ironically, hedge funds that kept their own balance sheets confidential demanded full transparency from the deal.
Hedge funds also tried to extract more leverage by receiving certificate cash flows in TRORS form, financed by a third bank. From a US tax perspective, swap income is often cleaner than equity income for reducing state tax and itemized deduction complications.
The bank sponsor receives a stream of floating-rate payments backed by risk-free collateral equal to about 30% of the trust. This stream would have investment-grade or even super senior risk characteristics. The bank books a TRORS in its trading book, with a back-to-back TRORS to a securitization conduit. The conduit pays the TRORS on the loans; the bank pays a funding cost; the bank charges a higher funding cost to the loan trust than what it pays to the conduit, earning annual income. Counterparty risk is managed by frequent mark-to-market of the TRORS.
If the HLT portfolio deteriorates rapidly, the TRORS behaves more like equity. Unwind triggers mitigate this: the entire structure unwinds when overcollateralization falls to the trigger threshold. For regulatory capital, the bank uses the mark-to-market value of assets plus a regulatory add-on factor for counterparty risk. Economic capital required for the long position is offset by the short position in the back-to-back TRORS. True sale plus low remaining capital charge makes it an efficient structure for balance sheet reduction.
Bank Regulatory Capital and Basel II
Basel I came out of the Bank for International Settlements in July 1988. It may have had an effective shelf life of about a picosecond, in Tavakoli’s framing. As soon as countries set their own implementations of the BIS guidelines, they interpreted them inconsistently. As soon as they did that, banks started looking for ways to arbitrage the inconsistencies.
New products like credit derivatives created further problems. Basel I was not designed for them, so they were essentially unregulated from a capital perspective. Basel I also had arbitrary distinctions unrelated to actual credit risk, like the 0% risk weight for all OECD sovereign risk regardless of actual creditworthiness.
Basel II was meant to fix all of this. The new rules were due in 2005 but special-interest groups disputed BIS decisions, causing delays. Revisions came in November 2005. The US agreed to adopt Basel II in November 2007, but full implementation by some global banks was not expected until 2015.
The minimum ratio of qualifying total capital to BIS risk-weighted assets remains 8% after December 31, 1992. But how much capital gets reserved against a specific asset depends on its BIS risk weighting, which varies by jurisdiction. That gives local regulators discretion, which brings inconsistency back in.
For credit derivatives in the bank book, they must meet hedge accounting criteria: the seniority and legal entity of the reference asset must match the loan held in the portfolio, and cross-default provisions must exist. If these criteria are met, the regulatory treatment treats the default swap as a guarantee, substituting the counterparty’s risk weight for the obligor’s.
Since 1998, many venues allowed credit derivatives to be booked in either the bank book or the trading book. Under market risk proposals, three components of risk matter:
Counterparty risk: The mark-to-market exposure for the credit derivative due to the counterparty potentially defaulting.
General market risk: Changes in the reference asset’s value from broad market movements (interest rates, FX, commodity prices, equity prices). Calculated using the bank’s internal VAR model.
Specific risk: Changes in the reference asset’s value from factors specific to the reference asset, like credit quality deterioration. Accounted for by BIS-specified risk weightings or standard add-on factors.
Banks with reproducible and consistent VAR models can use model-based approaches. The difference is dramatic. For a matched credit default swap position (bought and sold protection on the same asset):
Without a VAR model, regulatory capital is calculated as: Notional × Counterparty Risk Weight × 8%
For a non-OECD BBB-rated bank counterparty at 100% risk weight, that is 800 bps of regulatory capital charge. For an OECD AAA-rated bank at 20% weight, it is 160 bps. A fivefold difference just from the counterparty’s geography.
With a first-generation VAR model, the calculations change significantly. For the BBB-rated non-OECD counterparty, return on regulatory capital jumps 17 times with the model. For the AAA-rated bank counterparty, it jumps 8.6 times.
That model-based capital benefit drove banks to set up CDS trading desks, which then enabled synthetic CDO structures for clients who had no models or less sophisticated ones. It also allowed banks to trade CDSs in a regulatory capital efficient manner purely for creating synthetic arbitrage CDOs. Some bank arrangers with poor distribution capabilities entered the synthetic CDO business knowing their only advantage was helping clients reduce balance-sheet regulatory capital while parking the assets in their trading books at much lower capital cost.
Basel II: An Honest Assessment
Under Basel II, Total Capital = Capital Requirements for Credit Risk + Market Risk + Operational Risk.
For credit risk, banks can use the standardized approach, a foundation model internal ratings-based (IRB) approach, or an advanced model IRB approach. US banks cannot use the standardized approach.
Under Basel I, OECD sovereigns had 0% risk weight regardless of actual creditworthiness. Under Basel II, risk weights reflect external ratings. In theory this is better. The IMF, EC, and ECB continue at 0%. All others map to external ratings.
Basel II also tries to design a framework for ABS, CDOs, and other securitized products. The standardized approach gives the worst-case capital charge. No bank will adopt a model that produces a larger charge than necessary. Internal ratings-based approaches typically offer more relief than standardized approaches.
For ABS tranches with long-term ratings, Basel II introduces new concepts: granularity of pools and thickness of tranches. The risk weighting tables changed between the original proposal and the final version, generally going down, which Tavakoli describes as “an arbitrary exercise, since Basel II failed to account for much greater risks presented by misrated products and insufficiently vetted portfolios.”
The new rules encouraged genuine risk transfer and discouraged banks from bailing out failing deals. Banks could only provide credit enhancement at the outset of the transaction.
Tavakoli’s summary verdict: “Basel II is part of a series of failures by various financial regulators to capture the most fundamental risk in the banking system: lending money to someone who may not pay you back. If you do not properly measure that risk at the outset, no amount of securitization or credit derivatives technology will correct the problem for you.”
As securitizations began coming back onto bank balance sheets in 2007, BIS had no adequate framework for what leverage, synthetics, and securitizations could do to systemwide risk. Citigroup became a notable example, taking back tens of billions in off-balance-sheet SIV assets plus CDO tranches it had underwritten with full liquidity lines. The regulatory framework had not anticipated this.
Previous: Balance Sheet CLOs: Moving Risk Off the Books Next: Super Senior Sophistry: The AAA Disappearing Act