Fitch Rates COLT 2016-2 Mortgage Loan Trust
--$130,180,000 class A-1 certificates 'Asf'; Outlook Stable;
--$130,180,000 notional class A-1X certificates 'Asf'; Outlook Stable;
--$130,180,000 exchangeable class A-3 certificates 'Asf'; Outlook Stable;
--$59,666,000 class A-2 certificates 'BBBsf'; Outlook Stable;
--$59,666,000 notional class A-2X certificates 'BBBsf'; Outlook Stable;
--$59,666,000 exchangeable class A-4 certificates 'BBBsf'; Outlook Stable;
--$8,787,000 class M-1 certificates 'BBsf'; Outlook Stable;
--$8,787,000 notional class M-1X certificates 'BBsf'; Outlook Stable;
--$8,787,000 exchangeable class M-1E certificates 'BBsf'; Outlook Stable.
Fitch will not be rating the following certificates:
--$18,333,529 class M-2 certificates.
This is the second Fitch-rated RMBS transaction issued post-crisis that consists primarily of newly originated, non-prime mortgage loans. The most notable difference between COLT 2016-2 and COLT 2016-1 (which closed June 2016) is that not all of the loans in 2016-2 were originated by Caliber Home Loans, Inc. (Caliber). Roughly 15% of the COLT 2016-2 pool was originated by Sterling Bank and Trust, FSB (Sterling). Whereas the credit quality of the 85% of the pool originated by Caliber in 2016-2 is consistent with the credit quality of the loans in 2016-1, the loans originated by Sterling have a different borrower credit profile, with higher credit scores, lower loan-to-values and, notably, the use of bank statements to document the borrower's income rather than traditional income documentation. Despite projected loss penalties to reflect the weaker income documentation, Fitch projects meaningfully lower loan losses on the Sterling loans than for Caliber loans due to the relative strength of the remaining loan attributes.
In addition, Fitch made one change to its loss modelling approach for 2016-2 related to its Ability to Repay (ATR) claim probability. For 2016-1, Fitch doubled its standard ATR claim probability for all non-qualified (non-QM) and Higher Priced-QM (HPQM) loans in the pool. For 2016-2, Fitch did not double its standard ATR claim probability for non-QM and HPQM borrowers with Appendix Q income documentation, credit scores above 700 and household income above $100,000. Consequently, for 2016-2 only roughly 20% of the pool received double the standard ATR claim adjustment, while the remaining non-QM and HPQM borrowers received the standard adjustment. Fitch believes this adjustment more appropriately reflects the risk of ATR claims in the pool.
The combination of the higher credit quality loans from Sterling and a reduced ATR claim probability on a portion of the pool resulted in lower pool loss expectations for 2016-2 relative to 2016-1.
The transaction is collateralized with 53% non-QM mortgages as defined by the ATR rule while 41% is designated as HPQM and the remainder either meets the criteria for Safe Harbor QM or ATR does not apply. Due to the limited non-prime performance of the asset manager, Hudson Americas L. P. (Hudson), and originators, Fitch capped the highest possible initial rating at 'Asf'.
The certificates are supported by a pool of 501 mortgage loans with credit scores (702) similar to legacy Alt-A collateral. However, unlike legacy originations, many of the loans were underwritten to comprehensive Appendix Q documentation standards and 100% due diligence was performed confirming adherence to the guidelines. The weighted average loan-to-value ratio is roughly 76% and many of the borrowers have significant liquid reserves. The transaction also benefits from an alignment of interest, as LSRMF Acquisitions I, LLC (LSRMF) or a majority-owned affiliate, will be retaining a horizontal interest in the transaction equal to not less than 5% of the aggregate fair market value of all the certificates in the transaction.
Fitch applied a default penalty to 47% of the pool to account for borrowers with a mortgage derogatory as recent as two years prior to obtaining the new mortgage, and increased its non-QM loss severity penalty on lower credit quality loans to account for a potentially greater number of challenges to the ATR Rule. Fitch also increased default expectations by 258 basis points at the 'Asf' rating category to reflect variances from a full representation and warranty (R&W) framework.
Initial credit enhancement (CE) for the class A-1 certificates of 40.00% is substantially above Fitch's 'Asf' rating stress loss of 16.75%. The additional initial CE is primarily driven by the pro rata principal distribution between the A-1 and A-2 certificates, which will result in a significant reduction of the class A-1 subordination over time through principal payments to the A-2. The certificate sizing also reflects the allocation of collateral principal to pay only principal on the certificates and collateral interest to pay only certificate interest. Both of these features resulted in higher initial subordination to ensure that principal and ultimate interest (with interest accrued on deferred amounts) are paid in full by maturity under each class's respective rating stress scenario.
KEY RATING DRIVERS
New Asset Class (concern): Due to the limited non-prime performance of the asset manager, Hudson Americas L. P. (Hudson), and Caliber (as originator), Fitch capped the highest possible initial rating at 'Asf'. As Caliber and Hudson further develop a track record and more non-prime performance is established while upholding the same controls, Fitch will consider a higher rating.
Non-Prime Credit Quality (concern): The credit scores for the Caliber loans average 701, which resemble legacy Alt-A collateral, while the 760 average score for the Sterling loans more closely resembles recent prime quality loans. The pool was analyzed using Fitch's Alt-A model with positive adjustments made to account for the improved operational quality for recent originations, due diligence review, and presence of liquid reserves. Negative adjustments were made to reflect the inclusion of borrowers (47%) with recent credit events, increased risk of ATR challenges and loans with TILA RESPA Integrated Disclosure (TRID) exceptions.
Bank Statement Loans Included (concern): While Sterling has an established track record in originating both agency and non-agency mortgage loans, 65 of the non-QM loans included in this pool were underwritten to its Advantage Program where a 30-day (1-month) bank statement was used to verify income, which is not consistent with Appendix Q documentation standards. While employment and assets are fully verified, the limited income verification resulted in application of a probability of default (PD) penalty of approximately 1.4x for the Sterling loans and an increased probability of ATR claims.
Appendix Q Compliant (positive): Of the 435 loans contributed by Caliber, roughly 97% or 423 loans were underwritten to the comprehensive Appendix Q documentation standards defined by ATR. While a due diligence review identified roughly 2.8% of the Caliber loans as having minor variations to Appendix Q, Fitch views those differences as immaterial and all loans as having full income documentation.
Operational and Data Quality (positive): Fitch reviewed Caliber's, Sterling's and Hudson's origination and acquisition platforms and found them to have sound underwriting and operational control environments, reflecting industry improvements following the financial crisis that are expected to reduce risk related to misrepresentation and data quality. All loans in the mortgage pool were reviewed by a third party due diligence firm, and the results indicated strong underwriting and property valuation controls.
Alignment of Interests (positive): The transaction benefits from an alignment of interests between the issuer and investors. LSRMF, as sponsor and securitizer, or an affiliate will be retaining a horizontal interest in the transaction equal to not less than 5% of the aggregate fair market value of all the certificates in the transaction. As part of its focus on investing in residential mortgage credit, as of the closing date, LSRMF will retain the class M-2 certificates, which represent 8.45% of the transaction. Lastly, for the 435 Caliber-originated loans, the representations and warranties are provided by Caliber, which is owned by LSRMF affiliates, and therefore aligns the interest of the investors with those of LSRMF to maintain high-quality origination standards and sound performance, as Caliber will be obligated to repurchase loans due to rep breaches.
Modified Sequential Payment Structure (mixed): The structure distributes collected principal pro rata among the class A notes while shutting out the subordinate bonds from principal until both class A notes have been reduced to zero. To the extent that either the cumulative loss trigger event or the CE trigger event occurs in a given period, principal will be distributed sequentially to the class A-1 and A-2 bonds until they are reduced to zero.
R&W Framework (concern): Caliber and Sterling, as originators, will be providing loan level reps and warranties to the trust. While the reps for this transaction are substantively consistent with those listed in Fitch's published criteria and provide a solid alignment of interest, Fitch added 258 bps to the projected defaults at the 'Asf' rating category to reflect the non-investment-grade counterparty risk of the providers and the lack of an automatic review of defaulted loans. The lack of an automatic review is mitigated by the ability of holders of 25% of the total outstanding aggregate class balance to initiate a review.
Servicing and Master Servicer (positive): Servicing will be performed on 85% of the loans by Caliber and on 15% of the loans by Sterling. Fitch rates Caliber 'RPS2-'/Negative Outlook, due to its fast-growing portfolio and regulatory scrutiny, and reviewed Sterling to be acceptable. Wells Fargo Bank, N. A. (Wells Fargo; 'RMS1'/Stable Outlook, will act as master servicer and securities administrator. Advances required but not paid by Caliber and Sterling will be paid by Wells Fargo.
Fitch's analysis incorporates a sensitivity analysis to demonstrate how the ratings would react to steeper market value declines (MVDs) than assumed at the MSA level. The implied rating sensitivities are only an indication of some of the potential outcomes and do not consider other risk factors that the transaction may become exposed to or may be considered in the surveillance of the transaction. Two sets of sensitivity analyses were conducted at the state and national levels to assess the effect of higher MVDs for the subject pool.
This defined stress sensitivity analysis demonstrates how the ratings would react to steeper MVDs at the national level. The analysis assumes MVDs of 10%, 20%, and 30%, in addition to the model projected 8.3%. The analysis indicates that there is some potential rating migration with higher MVDs, compared with the model projection.
Fitch also conducted sensitivities to determine the stresses to MVDs that would reduce a rating by one full category, to non-investment grade, and to 'CCCsf'.
USE OF THIRD-PARTY DUE DILIGENCE PURSUANT TO SEC RULE 17G-10
Fitch was provided with Form ABS Due Diligence-15E (Form 15E) as prepared by AMC Diligence, LLC (AMC). The third-party due diligence described in Form 15E focused on three areas: a compliance review; a credit review; and a valuation review; and was conducted on 100% of the loans in the pool. Fitch considered this information in its analysis and believes the overall results of the review generally reflected strong underwriting controls. Fitch made the following adjustment(s) to its analysis: A total of 10 loans were identified as having material exceptions which are potentially at risk for statutory damages and were subject to an increase in Fitch's loss severity (LS) of $15,500 to account for the possible maximum statutory damage awarded to a borrower ($4,000); borrower legal costs associated with the TRID violation ($10,000); and the trust's incremental legal costs associated with the error ($1,500). Fitch received certifications indicating that the loan-level due diligence was conducted in accordance with its published standards for reviewing loans and in accordance with the independence standards outlined in its criteria.
REPRESENTATIONS, WARRANTIES AND ENFORCEMENT MECHANISMS
A description of the transaction's representations, warranties and enforcement mechanisms (RW&Es) that are disclosed in the offering document and which relate to the underlying asset pool is available by accessing the appendix referenced under "Related Research" below. The appendix also contains a comparison of these RW&Es to those Fitch considers typical for the asset class as detailed in the Special Report titled "Representations, Warranties and Enforcement Mechanisms in Global Structured Finance Transactions," dated May 2016.
A variation was made to Fitch's 'U. S. RMBS Loan Loss Model Criteria' in regard to treatment of loans with prior credit events. Historical data suggests that borrowers with similar credit scores as those in the pool are nearly 20% more likely to default on a future mortgage, as compared to all outstanding borrowers, if they had a prior mortgage-related credit event. This adjustment was applied to the roughly 47% of the pool that had a prior mortgage-related credit event, resulting in approximately a 10% increase to the pool's probability of default at each rating category.
Due to the structural features of the transaction, Fitch analyzed the collateral with customized versions of two of its standard models. Fitch's Alt-A Loan Loss Model was altered to include three additional inputs: due diligence percentage, operational quality and liquid reserves. These variables were not common in legacy Alt-A loans and were excluded in the derivation of Fitch's Alt-A model. Given the improvement in today's underwriting over legacy standards, these aspects were taken into consideration and a net credit was applied to the pool. The second customized model was based off of Fitch's Cash Flow Assumptions workbook. The customized version was created to allow for the consideration of delinquent loans at issuance.