Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

?

Proposed U.S. Reverse Mortgage Securitization Ratings


Methodology

Morningstar Credit Ratings Introduction


February 2019 The methodology described herein applies to Morningstar Credit Ratings, LLC ratings of U.S. reverse
Version: 1.0
mortgage securitizations. It pertains to both new issuance and ratings surveillance. A rating Morningstar
_____________________________ assigns based on this methodology will provide market participants with a benchmark that they can use
Contents to gauge the relative risk of a reverse mortgage securitization making ultimate principal payment on or
1 Introduction prior to a rated final distribution or maturity date.
2 Rating Methodology
2 Analysis
5 Operational Review Reverse mortgage loans are typically taken out by borrowers 62-years-old or older who wish to borrow
6 Legal Review against the equity in their houses. No mortgage payments are made until the house is sold after the
6 Surveillance borrower's death or when the borrower moves out. As such, the future value of the property backing a
7 Performing Loans
reverse mortgage loan is the sole source for repayment. These loans typically take the form of lump sum
17 Non-Performing Loans
_____________________________ payments, serial disbursements, lines of credit, or some combination thereof. Some reverse mortgages
are negative amortization loans because the borrowers do not make periodic interest payments. Instead,
the accrued interest amounts are added to the loan balance and repaid at maturity.

Reverse mortgages have been part of the U.S mortgage market for the past five decades. However,
securitization had not been used in this sector until 1999, when the first reverse mortgage securitization
deal (SASCO 1999-RM1) was issued comprising private-label collateral. Since then, the transaction
structure has evolved as the industry has developed, with the first transaction containing government-
insured collateral being issued in a Mortgagee Equity Conversion Trust in 2006. This Home Equity
Conversion Mortgage, or HECM, collateral is insured by the Federal Housing Administration, which is
part of the U.S. Department of Housing and Urban Development, or HUD.

The U.S. reverse mortgage market thus consists of two loan groups: a) government-guaranteed HECM
loans, which are typically underwritten to Ginnie Mae guidelines and insured by HUD, and b) private-
label loans, which are not. The loan universe can be further partitioned into performing and non-
performing loans. Morningstar uses two different models, one for performing loans, and another for non-
performing loans.
Page 2 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Section 1 – Rating methodology


In Morningstar’s view, reverse mortgage securitizations have unique risk attributes, which include:

a) Maturity risk: The maturity date underlying a reverse mortgage is not fixed, and a loan could
be repaid by an event such as borrower death or a voluntary prepayment.
b) Default risk: Default can be triggered by events such as non-occupancy as a primary residence
or a borrower's failure to pay property taxes, make insurance payments, or maintain the property up to a
certain standard, and so on.
c) Housing price risk: The net proceeds of a property may be less than the final loan amount.
d) Liquidity risk: Cash flow from assets may not cover scheduled payments.
e) Interest-rate basis risk: This could arise from a mismatch between the rates on assets and
liabilities of the transaction.
f) Operation risk: Operational issues related to transaction parties.

Morningstar’s reverse mortgage securitization methodology tailors its analytical approach to address
each of these risks. Our rating process typically starts with processing loan-level data tape through
proprietary collateral analysis models. The model projects total cash flows (interest and principal) from a
pool of underlying reverse mortgages through a loan-level Monte Carlo simulation. The rating process
may also include reviews of originators, servicers, and third-party due diligence providers.

The information gathered in these reviews may be incorporated into the model to help assess the
relative risk of the specified pool of mortgages being analyzed. In addition, Morningstar may apply
qualitative adjustments to model results based on additional information reviewed as part of the rating
process. Finally, Morningstar may perform additional stress testing to incorporate the impact of
performance triggers, amortizing credit enhancement, and tail-end default risk, if necessary.

Morningstar’s U.S. RMBS General Ratings Methodology describes other factors considered in ratings of
reverse mortgages that are not included in this document. For details on the RMBS General Ratings
Methodology, please visit www.morningstarcreditratings.com. In particular, the general methodology
augments the Legal Review, Operational Review, and Surveillance sections, below. The Analysis section
highlights differences between the reverse mortgage approach and the approach for traditional forward
mortgages, which are the focus of the general methodology.

Section 2 – Analysis

Maturity Risk
One of the bigger uncertainties in reverse mortgage transaction is the maturity date of the underlying
loans. A non-defaulting loan is typically terminated by either the death of the borrower (mortality event)
or prepayment (mobility or morbidity event). An example of mobility would be the borrower selling or
transferring the home, perhaps moving to a state with a better climate. Morbidity is a moving out of the
property sometimes to an assisted living/dependent care facility.
Page 3 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

The Morningstar Reverse Mortgage Model addresses death probability by utilizing the life expectancy
tables from the Social Security Administration and other population characteristics. Mobility and
morbidity prepayment profiles are similarly data-driven. Stress levels typically vary by letter rating, for
example at higher rating levels Morningstar may modify maturity risk profiles to stress cash flows.

Morningstar may adjust assumptions to reflect developments that may have a material impact on the
maturity date of the reverse mortgage. Morningstar may also apply additional maturity stress to verify
transaction performance. For example, transactions with high advance rates are typically more
vulnerable to high prepayment rates, and Morningstar may add additional stress tests in response to
these risks.

Housing Price Risk


The Morningstar Reverse Mortgage Model incorporates housing price index data from 1976 to the
present for stressed scenarios forecasts. The B rating environment is Morningstar’s base-case scenario.

The most stressful national economic environment that occurred since the inception of the housing price
index used by Morningstar was the housing crisis of 2007-12, which Morningstar believes to be an A
stress environment at the national level. For the A stress scenario, Morningstar generally applies house
price decline assumptions that are at least as stressful as the worst regional house price decline
experienced by that specific region since the inception of the related regional home price index.

The worst historical regional HPI decline is measured by the steepest peak-to-trough HPI percentage
decline experienced during any housing cycle in the available data set. However, Morningstar reverts to
the national HPI decline when the regional level decline is not as severe. Further, Morningstar may
adjust property value due to other factors, such as usage of broker price opinions, or BPO, or missing full
appraisal reports.

Because a security rated AAA should survive an extremely stressful environment, we apply an economic
scenario equivalent to a catastrophic event when running a AAA rating scenario. To simulate a
catastrophic event, we apply a house price decline that is more severe than the A house price decline.

For example, at the time of this publication and subject to change, the AAA house price decline
assumption is 1.3x the A decline. HPI forecast assumptions are regional, so deal-level HPI assumptions
may vary based on the geographic composition of the pool.

To arrive at forecast performance results for ratings between the above-listed anchor scenarios (AAA, A,
and B), we interpolate the forecast performance results for rating notches between the anchors. We
extrapolate the forecast performance results for rating scenarios below B.
Page 4 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Default Risk
Reverse mortgages will enter default for various reasons, such as if the borrower fails to pay property
taxes and insurance or does not maintain the property to certain standards or was not occupying the
property as a primary residence. If applicable, failure to pay home owner association fees is also a
default. Morningstar addresses default risk through its Monte Carlo based simulation models. (see
Sections 6 and 7 for details). To further stress performance projections, our analysis will typically
incorporate additional variables, such as liquidation timelines and foreclosure expenses.

Liquidity/Cash Flow Timing Risk


The cash flow from reverse mortgages can be highly volatile due to the uncertainty related to the loan
maturity date. In addition, reverse mortgage loans typically capitalize interest and do not produce
regular cash flows from interest, instead typically repaying principal and accrued interest in a single
lump sum at maturity.

The resulting cash flows can be sporadic, and available cash may not be enough to cover any scheduled
interest payment on the bonds. This could cause disruption in interest payment stream for bond
investors. Morningstar will typically review the deal structure for mitigation or remediation mechanisms,
such as availability of the reserve fund or servicer advance coverage and will typically adjust model
inputs to reflect the above characteristics.

Interest-Rate Risk
The interest-rate environment can vary from security to security based on several factors, including
collateral interest-rate type (fixed versus adjustable rate), security interest-rate type, the security’s
position in the waterfall, and exposure to interest-rate derivatives. Therefore, Morningstar generally
tests three interest-rate paths at the various rating levels in both the credit and cash flow models.

Morningstar's economic outlook for each index typically includes a forward curve, as well as two
stressed interest-rate paths for each index: a high path and a low path. These paths are updated
periodically. All three interest-rate paths (middle, high, and low) will generally be tested independently,
with potentially varying degrees of stress per rating level, in the credit model and the cash flow model.

Draw Risk
Some reverse mortgage loans offer the borrower a line of credit from which the cash can be drawn at
the borrowers' discretion. These draws require funding and are typically capitalized to the balance (UPB)
of the loan. This is discussed in more detail in Section 6.

Risk Factors - Reverse versus Forward Mortgages


Certain risk factors are more germane to reverse mortgages versus traditional forward mortgages,
whereas other risk factors are common, such as interest-rate risk. Reverse mortgages do not require the
borrower to make monthly payments, hence events triggering default for reverse mortgages differ from
those for forward mortgages.
Page 5 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Mortality and morbidity are important factors in reverse mortgages, and these are highly influenced by
borrower age, gender (women typically have higher life expectancies), and the presence of co-borrowers
(loans extend for the life of the last living borrower). Pools of reverse mortgages can include many well-
maintained, high value homes, that may exhibit different HPI trajectories than average, and Morningstar
may make value adjustments to reflect this.

Section 3 – Operational Review

Operational Risk Assessments


The Morningstar reverse mortgage rating process may involve reviews of the originator/aggregator and
the servicer. Operational Risk Assessment (ORA) rankings may provide Morningstar additional comfort
with the origination practices and underwriting guidelines of the originators, and the servicing
capabilities of the loan servicers in proposed transactions. In Morningstar’s view, how well an
originator/aggregator or a servicer performs its duties can have an impact on the loan's performance.
The ORA rankings are forward-looking assessments of the originators and servicers regarding how well
these entities mitigate operational risk inherent in their business processes. Based on the rankings and
findings from the third-party due diligence review, Morningstar may adjust the expected losses.

If an originator/aggregator or servicer that is not ranked by Morningstar is part of a securitization to be


rated by Morningstar, Morningstar may perform a less comprehensive review to determine whether the
originator or servicer is acceptable. If the aggregator/originator or servicer is not reviewed or ranked,
Morningstar may make conservative assumptions when assessing the expected losses to address the
lack of detailed information.

Originator Ranking
Operational reviews for originators may be performed where Morningstar believes there will be material
exposure to the securitization. As previously mentioned, in the first level of review, which may consist of
a questionnaire and, in some cases, an on-site visit, Morningstar assesses the relative risk of the
policies, procedures, and quality-control operations of the company against its peers. The second
optional level of review, a comprehensive ORA ranking, provides the reviewed company with a ranking
among its peers.

Morningstar will rank a mortgage originator on a scale of 1 to 4: A ranking of 1 indicates the originator
exceeds prudent standards, 2 indicates the originator demonstrates proficiency in standards, 3 means
that the originator demonstrates adequacy in standards, and a ranking of 4 indicates that originator fails
to meet one or more requisite standards. The ranking is an assessment of the originator’s operational
risks and the effectiveness of its control mechanisms for mitigating that risk. Morningstar believes that
the ability of the originator to mitigate these risks in its loan production process affects the performance
of the loans.

In the comprehensive ranking, Morningstar’s ORA group will conduct an on-site visit to an originator to
understand underwriting policies, sales and marketing practices, appraiser selection and property
Page 6 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

valuations, closing and funding procedures, infrastructure setup, risk-management practices, legal and
regulatory compliance, corporate governance, and management experience. Special attention will be
focused on fraud, repurchases, representations and warranty performance, and mortgage-insurance
experience. If the originator is a wholesaler, Morningstar will focus on broker and correspondent
selection and oversight and the process for managing that business line.

Servicer Ranking
In some cases, Morningstar may provide a similar review of a mortgage servicer and rank the servicer on
the 1-to-4 scale. The Morningstar servicer ranking is a forward-looking operational risk assessment of
the mortgage servicer’s capabilities. The Morningstar ORA group will conduct on-site evaluations with a
focus on management tenure and business strategy, audit and quality assurance methodologies, loan
administration and customer-relationship management, collection strategies, loss-mitigation
effectiveness and foreclosure-caseload management, real estate owned marketing timelines, loss
accounting and related financial strategies, technology architecture and organizational infrastructure,
legal and regulatory controls, and vendor oversight and certification. For additional details, please refer
to Morningstar’s publication on “Operational Risk Assessments Methodology and Process for Residential
and Consumer-Finance Servicers and Vendors,” available at www.morningstarcreditratings.com.

Section 4 – Legal Review


Morningstar, along with its legal counsel, will generally review the transaction documents, including
formation documents and operative documents to determine if a transaction is exposed to noncredit
risks.

Morningstar will typically review the prospectus and the pooling and servicing agreement to make sure
the documentation and the modeling of the transaction structure agree with one another. A key
consideration is whether the legal issuer is reasonably protected from the bankruptcy or insolvency of
the sellers to the trust. As part of this review, Morningstar will typically review legal opinions as to
whether the transfer of collateral to the securitization trust constitutes a true sale.

Section 5 – Surveillance
After Morningstar initially rates a reverse mortgage transaction, it will review and periodically update
the letter ratings if contracted to do so. Morningstar will analyze the performance of the securitizations
periodically, generally in accordance with the methodology articulated herein, and in accordance with
any changes to the methodology made during the life of the securitization. The surveillance may include
reviewing transaction performance relative to expected performance, the likelihood of transaction
triggers being tripped, the pool factor, the bond factor (the ratio of the current value of a metric to
starting value), and changes to the expected interest-rate environment. Additionally, information
obtained from ORA reviews can affect Morningstar’s opinion on the reverse mortgage securitization.

Surveillance is typically like a new issue credit rating insofar as we will review available loan data and
current performance, including general market conditions. Similar to new issue, surveillance can run our
models if this is deemed appropriate. In certain circumstances, such as highly seasoned or factored-
Page 7 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

down deals, other evaluation methods may be more appropriate than re-applying the standard model.
In particular, Morningstar will review loan data and apply appropriate modifications. For instance, if it is
known that a mortality event has occurred on a loan, it may be deemed appropriate to reflect this in our
model.

Morningstar may be requested to provide a no-downgrade letter addressed to the securitization trust.
Morningstar may waive, deny, or approve such request in its sole discretion. A no-downgrade letter is
confirmation from Morningstar that a proposed change or amendment to the transaction, in and of itself
and solely as of the date requested, will not result in a qualification, withdrawal, or downgrade on any
of the current ratings of the rated deal securities by Morningstar.

The letter shall not constitute any consent, approval, agreement, or advice or affirmation and is based
solely on the proposed transaction documents reflecting the change or amendment and provided to
Morningstar at the time of request. In addition, any no-downgrade letter and analysis related thereto by
Morningstar does not address (i) whether the proposed change or amendment is permitted, consistent,
or otherwise approved under the rated deal documents; (ii) any benefits or effect of the proposed
change or amendment on the security holders or parties to any rated deal documents or any such
parties’ interests; or (iii) any factors not considered or enumerated pursuant to the no-downgrade letter
and the ratings letters issued in connection with the rated deal.

Any ratings of the rated deal remain subject to and qualified by the initial ratings letter. At Morningstar’s
sole discretion, a legal review of the proposed change or amendment may be performed in conjunction
with Morningstar’s analysis of the proposed transaction.

Section 6 – Performing Loans


Morningstar uses two different models, one for performing loans and one for non-performing loans. This
section describes the model for performing loans.

Model Cash Flows


A reverse mortgage loan accrues interest that is added to the principal balance until the loan is repaid or
defaults. Cash flows from these loans are unlike regular forward mortgages, which typically pay monthly
principal and interest. No cash is received until a final single lump sum, which contains all cash flows
accrued.

Generally, there are three types of reverse mortgages: monthly payment, line of credit, and bulk
payment. In a monthly payment reverse mortgage, the borrower receives a monthly payment from the
lender, which is added to the principal balance along with the accrued interest. In a line of credit reverse
mortgage, the borrower can draw up to the maximum amount of the line of credit, and interest is only
accrued on the drawn balance. In a bulk payment reverse mortgage, the borrower receives a one-time
payment from the lender and interest is accrued on that balance from day one.
Page 8 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Three results can occur with a reverse mortgage: death, default, and prepayment. Morningstar’s
Reverse Mortgage Model uses a Monte Carlo simulation to generate these results and averages the
cash flows from each simulation for each mortgage to calculate the available funds at the various rating
levels. Cash flows to the liabilities when a loan experiences one of the three types of events for each
borrower: death, default, or prepayment.

Event: Mortality
If a borrower dies, the heirs have the option of selling the house and repaying the mortgage. If they
choose not to do this, such as if the house value is less than the mortgage balance, the heirs can give up
the house to the originator (i.e. deed-in-lieu). If the heirs decide not to do these things, the servicer will
generally move to foreclose on the property. The time for foreclosure for reverse mortgages may be
shorter than for forward mortgages, the because there is no one to contest it, no requirement to attempt
to modify the loan, or other actions that are part of servicing defaulted forward mortgages.

The modeled probability of death in the B rating scenario is determined using the life expectancy tables
from the Social Security Administration, and these values may be stressed for higher rating levels. If the
borrower dies with ample equity in the home, the model will typically assume the property will be sold
and the mortgage repaid. If the borrower has insufficient equity in the home, the model will default the
loan and calculate a loss severity that is generally the same as used for forward mortgages.

Morningstar determines various parameters influencing the path a loan may take, for example,
maximum loan-to-value ratios, or LTV, and the duration of various steps, such as the timeline lags for
foreclosure and real estate owned properties (typically longer for higher letter ratings). The REO timeline
is separate because the loan no longer accrues interest when the home is in REO. If it meets certain
requirements such as LTV, the model may dictate selling the property instead of it being foreclosed on.
The number of months it takes to sell a property after the borrower’s death would also be modeled.

As mentioned above, death probabilities are calculated from Social Security Administration actuarial
tables (refreshed), and loan-level inputs, such as age and gender. A probability of a borrower dying is
computed for every borrower and at every timestep and rating level. These probabilities are influenced
by inputs such as actuarial data, lifetime expectancy stress percentages, a borrower’s gender and age,
the presence of a co-borrower, and other factors.
Page 9 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Morningstar typically regards the actuarial data to inform the B rating probability of death. We can
modify life expectancy by altering death probability for a given age. To produce more conservative
estimates of credit risk for higher rating levels we would typically lower death probability, say by around
3% for A ratings and around 5% for AAA.

Source: Social Security Administration

Event: Default
Reverse mortgages can default for several reasons, such as a) failure to occupy the property, b) failure to
pay taxes and insurance, or T&I, c) failure to maintain the property, or d) failure to pay home owner
association fees. Reverse mortgages must be secured by a lien on a primary residence only. A borrower
who moves or occupies a different primary residence must repay the mortgage, or the servicer will
foreclose. The borrower must also continue to pay T&I on the property.

Prior to April 2015, no assessment was made on whether borrowers could afford the T&I on their homes,
so this type of default happened more frequently, especially in HECM loans. Since April 2015, HUD has
required that lenders assess borrowers' financial positions and determine if they can pay their T&I and
other expenses out of their incomes, which are generally fixed.
Page 10 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

The charts below show examples of default curves. In these examples, the monthly default probability
(left-side chart) is tuned to produce a cumulative default curve (percentage of loans that have defaulted)
with the desired shape, rise rate and peak. This example shows cumulative defaults of 20% for the base
A scenario, and 1.3x this for the AAA, or 26%. These cumulative default levels were chosen based on
historical market data. Note that a loan's individual monthly default probability is typically scaled with
LTV. So, for example, loans with >80% LTV could be roughly double these probabilities.

Source: Morningstar

If the borrower cannot afford the T&I, the originator will calculate the expected lifespan of the borrower
and hold back the total T&I for that amount. Each year, when the T&I payments are due, the servicer will
pay them from these held-back funds and add the payment amount to the balance of the loan. This has
significantly reduced the percentage of T&I defaults that have been experienced. Morningstar
determines inputs influencing the default curve, as well as multipliers in the model that increase the
probability of default in response to variables such as LTV.

We apply different REO and foreclosure timelines per rating. Base default curves will be adjusted for
each borrower at each timestep using the current LTV of the borrower at that timestep, with higher LTV
loans showing higher default probabilities. We use our standard mortgage data to calculate foreclosure
costs. These vary by state and span the following data: annual property tax (percentage, annually),
annual insurance premium (percentage, annually), annual repairs (dollar amount, annually), annual
maintenance (dollar amount, annually), legal expenses (dollar amount, one-time fee), miscellaneous
expenses (dollar amount, one-time fee), broker rebate (percentage, one-time fee), and closing costs
(percentage, one-time fee).

Event: Prepayment
The borrowers can repay the mortgages at any time, just like a forward mortgage. The borrowers repay
either because they refinance their reverse mortgage with another reverse mortgage or because they
move. As borrowers age, the prepayment level increases because many older borrowers move into
assisted living facilities. The model includes a prepayment ramp, and the input is the maximum
prepayment rate, the levels of which Morningstar sets on a case-by-case basis. The model increases the
likelihood of prepayment for older borrowers and reduces it for younger borrowers.
Page 11 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

The following charts show examples of base prepayment curves. The curves are set for the average age
borrower (71-75) and scaled lower or higher for younger or older borrowers respectively. The 60-65-year-
old age range would typically be around 70% of base, whereas 86-90 could be 3x and above 91 can be
4x. The base curves will scale with new data, this illustration shows monthly prepay probability of
around 0.25% for A, which is equivalent to an annualized figure of 3.0% CPR. The graphics show curves
for illustrative purposes only. Actual stresses used may vary.

Source: Morningstar

Prepayment probabilities are modeled using a standard curve describing single-month mortality (SMM),
which is translated into a conditional prepayment rate (CPR) metric that is an annualized expression of
SMM. The SMM, or monthly prepayment rate, is shown in the left-side graphic above. If a particular
month's data show that 1% of collateral has prepaid, then SMM is 1%, which annualizes to just under
12%, CPR = 1 - (100% - 1%) ^ 12 = 11.36%.

The model assumes a base curve, which will be same for every borrower. Then this base curve is
adjusted for every borrower using loan-level data. Probabilities from adjusted prepayment curves will be
used in Monte Carlo computations. In each rating scenario, a maximum CPR, or annualized prepayment
rate, is assumed at a certain level. Time to reach that level of CPR is provided as a "time to level off"
parameter. The base CPR curve is calculated as illustrated in the right-side graphic above.

Once base repayment curves are established, they will be adjusted for each borrower using multipliers.
The base prepayment curve will be adjusted for each borrower in each timestep depending on the
borrower’s age, using age-based multipliers. If a borrower prepays a reverse mortgage, he or she might
not be able to pay back the entire balance due. Prepay recovery is a percentage of the positive
difference between balance due and current property value that a borrower will prepay.

Draws and Payments


Many reverse mortgage loans are fully drawn, which means the borrower receives the full lump sum
principal limit at closing but is not eligible to receive any further cash. Other loans permit borrowers to
withdraw less than the principal limit at closing, and they then are eligible to receive cash in future,
either as draws on a line of credit, or LOC, or as scheduled payments.
Page 12 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

The timing and sum of and draws are at the borrower's discretion, so in one extreme the borrower may
never have a draw. In other cases, the borrower can draw all of the eligible remaining line of credit.
Payments are different from draws insofar as payments are made to the borrower on a schedule that is
known up-front. Certain loans feature a combination of scheduled payments and a line of credit for
elective draws. The base (B rating) settings for these are informed by New View Advisors’ Reverse
Mortgage Draw Index, which publishes monthly draw rates for seasoned and unseasoned loans. This
index tracks draw rates for monthly adjustable HECM mortgage-backed securities. Most recent data
show unseasoned loans are drawing at roughly 2.5%-3% per month, and seasoned loans are slightly
slower at roughly 1.3%-1.5%.

Morningstar models draw events driven by a) probability of a draw event and b) size of draw, the
combination of which would be set to produce draw rates similar to the above, which could be modeled
as a high draw probability combined with lower draw amounts, or vice versa--less frequent but larger
draws. For example, if the goal was to produce a 2.5% monthly draw in a B rating scenario, then one
setup could be 5% monthly probability of draw coupled with size of draw, also probability-based,
centered around the 50% of available LOC.

Loans are partitioned into two sets: new loans, and seasoned loans. The boundary between seasoned
and new varies by rating letter stress scenario but could be set, for example, at around 48-72 months.
Using example probability numbers similar to that above, in the B rating scenario, say the monthly draw
rates are chosen to be 2.5% and 1.5% for new and seasoned respectively, with a 50% of LOC average
size of draw, and a transition from new to seasoned at a loan age of 72 months. The following example
illustrates the framework. A new loan would have a 5% chance of a draw in month 1, and month 2, and
so on, until a draw happens. A 68-month-old loan in the same scenario would have 5%, 5%,5%, 5%, 5%,
3%, 3%, and so on, with the step change to 3% happening in period 72, the transition month shown in
the graphic below.

If a draw happens in a particular month, say month 10 in the above example, the probability of the next
draw in month 10 is zero, and probability will ramp linearly back up over the next several months (over 6
months for example) to the appropriate rate. So, the probability of a draw in month 11 is 1/6 x 5% and is
2/6 x 5% in month 12. If a draw happens ahead of the seasoning step change, say in month 70, the
probability ramp will reflect a step change, so months 71, 72, and 73 will show 1/6 x 5%, 2/6 x 3%, 3/6 x
3% respectively. Again, once a draw event occurs, the random number simulator chooses the
percentage of LOC that is drawn. These mechanics are illustrated in the "Single-Loan, Ten-Iteration
Illustrative Model Run" section below.

Home Price Valuation


Home price and LTV are large drivers of cash flow and loan default probability. In order to calculate LTV
for every borrower at each timestep, we first need to establish the adjusted appraisal value at the tape
run date. Using historical HPI data we adjust appraisal value found in the tape from the loan origination
date to the present using the following formula. Furthermore, we may apply a haircut to these values to
reflect data quality, for example if the method is a BPO versus a full appraisal.
Page 13 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

After this, for every timestep, a borrower’s LTV is computed as loan balance over property value, with
the following considerations: a) loan balance includes accrued interest and b) Morningstar may adjust
the given original property value. The figure below shows example home valuation trajectories, where
the A rating scenario is intended to offer a similar peak-to-trough trajectory as was seen in the credit
crisis. The AAA rating scenario is typically around 30% more arduous than A.

Different ratings follow different paths depending on location, property type, and other factors. Data are
indexed to a base value of 100 on Jan. 2000. A property's value on any particular date is calculated
using index ratios versus a given reference value and date.

Sources: CoreLogic, Morningstar

Illustrative Analysis from a Rated Deal


Although the cash flow from each loan is a discrete single event and single cash flow, our model runs
numerous Monte Carlo simulations, and this typically produces a smoother set of resulting cash flows
than may be the case in any single simulation. The figure below shows an example set of AAA cash
flows from the three payment modes smoothed over many scenarios. For reference, these are actual
runs taken from a public deal rated in Nov. 2018, CFMT 2018-RM2, the presale for which is available at
www.morningstarcreditratings.com.
Page 14 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Source: Morningstar

This deal had highly-seasoned collateral, originated pre-crisis, so these curves look different from what
may be expected from newly originated loans for which prepayment rates would build from zero with
loan age. These highly-seasoned loans are in the flat part of the CPR curves shown in the prepayment
section above. The black line tracks cash flows from mortality, where the lack of cash flows in the first
few periods illustrates the lag between mortality and ultimate cash flow described in the Mortality
section above.

Source: Morningstar
Page 15 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

The charts above show trajectories versus time for different rating scenarios from the same illustrative
deal. The deal structure had 6 sequential notes. Three of the charts show data pertaining to the senior,
Class A, notes. The bottom right chart is for the sum of all 6 notes. The right-side scale is HPI (upper left
chart) and loan count (bottom right chart).

Single-Loan, Ten-Iteration Illustrative Model Run


The following charts show a AAA result for an artificially simple simulation: a single loan and ten Monte
Carlo simulations. The current UPB is $400,000, the index is six-month Libor with a 3.5% margin, and a
semi-annual reset. The home is a single-family with a value of $454,000. The sole borrower is an 80-year-
old male. The loan has a $10,000 line of credit (LOC) and $30,000 in payments scheduled for 60 monthly
installments of $500.

The charts illustrate 10 cash flows, one from prepayment, two from mortality events, and seven defaults.
Note that a prepayment event that has a resulting loss is reclassified as a default. Loss severities (not
shown) associated with each default range from 33%-72%. In the chart bottom left, multiple elective
draws are shown, as well as payments. The payments vector (black line in lower left-side chart) drops in
concert with loan count but terminates in month 60.

The black line in the lower right-side chart shows the UPB growing as interest and advances accrue to
the balance and dropping with cash flows as each of the scenarios pay off. The dashed red line in the
lower right-side chart shows loan count, which decrements in 10% steps as the single loan pays off in
each of the 10 scenarios.

Source: Morningstar
Page 16 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Illustrative Levels Data


The following tables use a hypothetical pool to express an example set of outputs from a model run in
order to illustrate the relative influence of various factors. Prepayments that fail to repay UPB in full are
included in the conditional default rate (CDR) data and Loss Severity data.

Events by count
Rating Default (%) Prepay (%) Death (%) TOTAL (%)

AAA 25 51 24 100
A 17 58 25 100
B 7 67 26 100

Source: Morningstar

Cash flow from event


Rating Default (%) Prepay (%) Death (%) TOTAL (%)

AAA 18 62 20 100
A 14 63 23 100
B 5 68 27 100

Source: Morningstar

Simple average of vectors


Rating Default CDR (%) Loss Severity (%) Prepay CDR (%)

AAA 13 42 13
A 13 35 14
B 10 24 15

Source: Morningstar
Page 17 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Section 7 – Non-Performing Loans


Morningstar uses two different models, one for performing loans, and a second for non-performing
loans. This section describes the model for non-performing loans.

State Transition Model


Non-performing loans that serve as collateral for securitizations are typically either: 1) in default, 2)
called due and payable, 3) undergoing foreclosure process, or 4) real estate owned. In the diagram
below, we illustrate the loan state transition model.

Per the diagram notation, loans typically enter a transaction in state 1, 2, 3, or 4, and transition down the
line after a certain lag period. Loans in status 1) through 4) can exit the typical state transition flow by
being either cured (a) or bought (c). Cured means the defaulting situation is resolved (for example, the
borrower failed to mail in occupancy statement documents but is later found to in fact occupy the
property. If cured, the loan is again performing. Bought typically means the loan is paid off in full, by a
borrower or an heir. Bought loans are paid and enter 7) paid status. If a loan is not bought, it will move
Page 18 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

through Real Estate Owned (REO) to liquidation and claim. Loans with guarantees may be eligible to
make an insurance claim from HUD, which can take one of two forms, a sales-based claim, or SBC, or an
appraisal-based claim, or ABC. The process is detailed below.

State 1: In Default
Although no payments are due on a reverse mortgage, the loan can still default, as discussed above, by
several means such as non-payment of T&I, non-occupancy, failure to maintain the property, or death.
Once an event of default occurs, the loan is usually called due and payable the following month. Our
Monte Carlo model transitions loans from the default event to due and payable, but time scales may be
extended for higher ratings or other considerations.

State 2: Called Due and Payable


Once the loan is called due and payable, or D&P, the UPB and D&P date become reference points for
claim calculations (see the Liquidation and Claim section below for more details). A loan typically takes
60 days to transition from this state to the start of the foreclosure process, although time frames can
again be extended for higher ratings and other considerations.

State 3: Foreclosure Process


Generally, within six months of being called due, the loan will reach its "1st legal" date (unless this is
extended by HUD), which is when legal proceedings commence. From here the time to proceed can vary
greatly depending on state. Non-judicial states can take as little as two months to advance, whereas
judicial states can take many years if contested. The courts are often severely backed up, and even
scheduling the hearing can take a while.

Ultimately a foreclosure sale, or auction, will take place. Loans can also be sold in a pre-foreclosure short
sale. This is necessarily at a loss, but the loan then becomes eligible to qualify for a sales-based claim, or
SBC, (see details below). The foreclosure auction has a required minimum bid amount, which is the
minimum of a recent appraisal and the debt UPB. If the auction bid amount is met, typically either a
third-party could buy the loan or else it goes to REO status. The model will simulate all these variables,
including using our home price valuation model. If the model determines purchase of the loan at
auction, it projects cash recovery after applying typical fees and costs.

The charts on the following page illustrate the state transition framework for advancement from
foreclosure to resolution (either REO or sale to third party), and from REO to claim. These profiles can
vary greatly for different ratings levels and loan-level considerations such as the state of jurisdiction. For
example, curves for loans from judicial states, which typically have longer lag times, are slower to build
than non-judicial states.

The left-side chart below shows the cumulative percentage of loans that advance out of the foreclosure
status versus time in months since the loan entered this status.
Page 19 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

State 4: Real Estate Owned


If a loan enters REO then a new appraisal is typically performed. In many such cases the new owner has
little incentive to put any money into the property, such as for improvements. As described above, the
model will apply an S-curve on state transition, which will vary by rating and loan-level considerations.
If the loan is not sold quickly, it is likely to fail to qualify for an SBC, instead it will ultimately likely be an
ABC, which results in significantly higher loss severities, as described in the Claim section below.

The right-side chart below shows the cumulative percentage of loans that advance out of the REO status
versus time in months since the loan entered this status.

Source: Morningstar

Action: (a) Cured


At any stage until the loan goes REO, the loan can cure and become reperforming. For example, when a
reverse mortgage borrower may have overlooked paying T&I, repaying the expense can remedy the
situation. The Monte Carlo model uses a probabilistic approach to this event, testing every loan every
month for a potential cure. Loan-level characteristics can affect the probability of curing. For instance, if
the UPB is greater than the home value, curing is less likely. If the loan is modeled to cure, our model
will transition the loan to the Performing state, or alternatively, if the issuer has pledged to buy such
loans out of the pool, the loans can be transitioned instead to Bought (see below) and then to Paid.

Action: (c) Bought


From any state upstream of REO, a loan can be bought outright out of the pool. The model tests this
transition in a Monte Carlo simulation for each loan each month. Loan-level factors such as ratio of UPB
to value affect the probability of being bought. The loan is more likely to be bought if there is more
residual value in the property, but in many cases the size of the loan will be larger than the value of the
underlying property.

State 0: Performing
Loans mostly enter this state through curing from some defaulted state. Depending on the deal
specifics, these loans can also be transitioned straight to the Paid status, for example if the deal requires
these re-performers to be bought out of the pool. If a loan is modeled to be reperforming, our model
applies tests to the loan each month for re-defaulting or for being bought.
Page 20 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

States 5+6: Liquidation and Claim


Once a loan is liquidated, the servicer can make a claim with HUD. The mechanics of this are perhaps
best conveyed by a worked example (see below). A loan that qualifies for an SBC is preferable because
more costs can be recouped than if an ABC is filed. Correspondingly, ABC outcomes typically report
higher loss severities.

The marketable title date is the reference for an REO sale, which will determine if it goes ABC or SBC. If
the sale happens within 6 months, then the loan may be eligible for a sales-based claim. The
qualification criteria for SBCs are somewhat involved, and, also require the servicer to meet all filing
deadlines. For example, if the D&P date is on or after July 1, 2015, the rules are as follows:
1) If an eviction was completed, the marketable title date is the later of the following dates: foreclosure
sale date, redemption date, and vacancy date.
2) If an eviction was NOT completed, the marketable title date is the later of the following dates: the
foreclosure sale date and the redemption date.

In the case of an SBC, claims can be made on advanced mortgage insurance payment, or MIP, servicing
advances, and certain foreclosure costs and brokerage fees. One-third of foreclosure costs and legal fees
are not HUD reimbursable. Broker costs are claimable in SBC but not in ABC.

From the date the loan is declared due and payable, interest is claimable from HUD without regard to
maximum claim amount, or MCA, which is a limit for each loan set at origination as a function of
borrower age, home value, interest rate, and so on. But this can receive a haircut in cases of filing errors
or failure to comply with HUD guidelines. Claimable interest accrues at the debenture rate--not the loan
rate--which is a level HUD sets at the time of loan origination. In the worked example below, the
debenture rate is shown as 4% versus 5% for the loan itself.

Status Transition Output


The chart below shows examples of loan-level state-transition trajectories activity for a single specific
Monte Carlo simulation at a given rating level. Other simulations of the same scenario will produce
different paths. Status numbers correspond to the State Transition Flowchart diagram on page 17 above.
The colors in this chart serve to illustrate transition between early statuses (red), intermediate statuses
(pink and white) and final resolution statuses (green).

Loan #1 for example was in Due and Payable status at the cut-off date and is modeled--in this one single
example simulation--to enter foreclosure in period 3, and REO in period 15. Loan #16 is in REO, and an
ABC claim is filed during period 3, and the claim is paid three months later, resulting in cash to the deal
in period 6.
Page 21 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Source: Morningstar
Page 22 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Cash Flow – SBC Worked Example

Source: Morningstar
Page 23 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Transaction Output Illustration


The charts below show an example model output, 1560 loans, and 600 Monte Carlo universes.

Source: Morningstar
Page 24 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Source: Morningstar
Page 25 of 25 Proposed U.S. Reverse Mortgage Securitization Rating Methodology | February 2019 Pro

Morningstar Credit Ratings, LLC

For More Information


+1 800 299-1665
ratingagency@morningstar.com

?
4 World Trade Center
150 Greenwich Street, 48th Floor
New York, NY 10007 USA

Copyright © 2019 by Morningstar Credit Ratings, LLC (“Morningstar”). All rights reserved. Reproduction or transmission in whole
or in part is prohibited except by permission from Morningstar. Morningstar is a credit rating agency registered with the Securities
and Exchange Commission as a nationally recognized statistical rating organization (“NRSRO”). Under its NRSRO registration, MCR
issues credit ratings on financial institutions (e.g., banks), corporate issuers and asset-backed securities. While MCR issues credit
ratings on insurance companies, those ratings are not issued under its NRSRO registration.

The information, credit ratings and other opinions (the “Content”) contained herein have been obtained or derived from sources
Morningstar believed to be reliable. Morningstar is not an auditor and, it does not and cannot in every instance independently
verify or validate information used in preparation of this report, credit ratings or any other opinions contained herein. THE
CONTENT IS PROVIDED “AS IS” AND NOT SUBJECT TO ANY GUARANTIES OR ANY WARRANTIES, EXPRESS OR IMPLIED,
INCLUDING WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE OR USE. The Content herein is
provided for information purposes only. The Content, including credit ratings and other opinions issued by Morningstar, are and
must be solely construed as statements of opinion and not statements of fact as to credit worthiness or recommendations to
purchase, hold, or sell any securities or make any other investment decisions. Morningstar may receive compensation for assigning
credit ratings or engaging in other analytical activities from issuers of the securities or third parties participating in marketing the
securities. Morningstar credit ratings and related analysis should not be considered without an understanding and review of
our methodologies, disclaimers, disclosures and other important information found at www.morningstarcreditratings.com.
Morningstar shall not be responsible for any damages or other losses of any kind resulting from, or related to, the use of the
Content contained herein.

Your use of this report is further governed by Morningstar’s Terms of Use located at
https://ratingagency.morningstar.com/MCR/about-us/terms-of-use.

To reprint, translate, or use the data or information other than as provided herein, contact Vanessa Sussman (+1 646 560-4541) or
by email to: vanessa.sussman@morningstar.com.

You might also like