Subprime mortgage lending: recognising its potential and managing its risks
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
256112 HFI September 06 25/10/2006 11:40 am Page 39
HOUSING MORTGAGE
SUBPRIME
& MORTGAGE
HOUSING TRANSACTION
LENDING IN CHINA
Subprime mortgage lending: recognising its
potential and managing its risks
By Edoardo Turano, Experian – Scorex’s Business Consultant,
Experian - Scorex
Risks and opportunities in subprime for example, by their credit history) and that and collecting the loans) compared to prime
mortgage lending do not qualify therefore for the prime loans. The difference [1] is generally in the
market. The availability of subprime order of 3 to 4.5 per cent over the base rate.
Subprime mortgage lending offers many products is not limited to mortgages but it Given the higher risk implied by such loans,
opportunities for growth and profitability. includes other typical retail products such usually subprime lenders try to cover the
While this market is already well developed as personal loans and credit cards. higher related costs through [1]:
in the US and the UK, in the rest of Europe • frequent use of early redemption charges;
there is still a low availability of non- Considering mortgage products, subprime • frequent requirement for Mortgage
standard mortgage offers. However, recent loans are offered to borrowers who Insurance;
research suggests there is a significant represent a higher level of risk with respect • higher commission fees (with differences
untapped demand for this type of product in to standard mortgage underwriting compared to prime loans up to 1% of the
Europe. Many lenders are therefore trying to guidelines. These loans are characterised credit facility).
enter this market by expanding their by interest rates (and fees) higher than the
product offering to reach a wider customer standard prime rate that is available in the An overview of the European subprime
base. market at a given point in time. The most mortgage market
common categories of applicants who fall
To be successful in this market though, into the subprime mortgage segment are: Recent research [2] indicates that there is a
lenders will need to address the issues of - Borrowers with a poor credit history significant demand in Europe for high risk
funding and risk management. Many (previous credit problems); mortgage loans. This has been valued as
lenders, in fact, underestimate the higher - Borrowers asking for high Loan to Value ranging from 400 billion euros to over 1,200
risks in subprime lending and its proper (LTV) mortgages; billion euros in the medium term. This
management, ending up with losses. This - Borrowers with a high Debt to Income forecast growth depends on future house
article presents an outline of a quantitative ratio; prices and on regulatory issues.
approach to risk evaluation and risk based - Borrowers who cannot document all of the
pricing as an effective tool to deal with underwriting information in their Currently, the availability of high risk
regulatory compliance and risk application. products in Europe is fairly limited,
management. This type of approach brings especially compared to the US and UK. For
many benefits when correctly implemented, Statistics show that loans to this type of example, in most European countries it is
although it does present some difficulties in borrower are characterised by high default very difficult for borrowers to succeed in
its practical implementation especially to rates (borrowers falling into arrears which obtaining a mortgage with a loan to value
entrants to the subprime mortgage market. cannot be recovered) and high loss rates higher than 80%. Other examples are loans
(lender’s loss after repossession compared to people with poor credit histories, self-
What are subprime mortgages? to original loan amount) and are therefore certified income or for older age applicants.
considered as posing significant risk. All of these products are generally not easily
The term subprime (also called non- available although there is quite a difference
conforming or non-standard) defines a Subprime loans are offered with higher than among countries with the UK being
specific lending market sector where standard interest rates, as the lender needs recognised as the most developed
borrowers are considered as posing a to cover higher credit losses and overhead European mortgage market. Estimates of
higher than standard credit risk (as revealed, costs (related to underwriting, servicing, the subprime market in the UK indicate a
HOUSING FINANCE INTERNATIONAL – September 2006 37256112 HFI September 06 25/10/2006 11:40 am Page 40
HOUSING MORTGAGE
SUBPRIME
& MORTGAGE
HOUSING TRANSACTION
LENDING IN CHINA
share in the order of 10% of total mortgage credit history [1]. In the UK, the most loss-on-default rates involved with subprime
market value [3]. In the US instead, the developed European mortgage market, the lending, as well as the higher overhead
market share of subprime loan originations corresponding percentage is much lower. costs. Moreover, they frequently lacked the
was approximately 20% of the total Furthermore, in the US there is wider management expertise, business planning
mortgage market in 2004, and subprime availability of historical data on subprime processes, and risk management processes
mortgages originations have been growing loans on which robust quantitative analysis necessary to manage these risks in a safe
at an average annual rate equal to 25% from can be based. Regarding analysis, and sound manner.”
1994 to 2003 [4]. European lenders have obviously learned
from the US business model in terms of risk Pricing is the key to a successful
There are however many factors that justify management. However, many lenders still subprime lending business
expectations for a likely expansion of the lack the necessary know-how and expertise
European subprime mortgage market. First to develop sophisticated pricing models It is evident then that the most crucial
of all, a general widespread growth in credit necessary to handle the risk in subprime element for a sub-prime lender is the
use (not limited to mortgages) is evident. mortgages. assessment and the pricing of risk. This
The demand for mortgage credit has been includes both a careful evaluation of the
increasing in particular among non- The high risk in subprime mortgages customer’s riskiness at application and
traditional borrowers. A typical example of effective methods of dealing with loans that
this class of borrower is immigrants, whose The increased risk in subprime lending start to fall into arrears. An objective
share of the European population is needs to be carefully assessed and assessment of risk at application allows the
becoming more and more substantial. managed. Clearly, a lender who decides to lender to reject applicants whose higher risk
Immigrants are generally considered as enter the subprime mortgage market has to cannot be offset by higher interest rates
subprime borrowers as they are be prepared to deal with higher delinquency (considering obvious practical and legal
characterised by limited credit histories, low rates than those relative to standard prime constraints) and that are profitable only for
incomes and are likely to apply for high loan mortgages. An analysis of a real portfolio lenders engaged in equity lending.
to value mortgages. Moreover, the growth in showed that for subprime loans serious Accepted applicants are then assigned an
self-employment or in flexible working delinquency rates [4] can be 20 times higher interest rate that reflect their risk.
arrangements (with less certain and regular than for prime loans. Apart from high
income streams) has also somehow delinquency rates, lenders should also Although all subprime lenders have a
modified the “traditional” consumer expect higher than average loss rates from methodology to adjust the price on the
landscape, pushing further the demand for their subprime portfolio. The combination of basis of the borrower’s risk, there are clearly
innovative mortgage products that do not high delinquency rates and high loss rates different levels of sophistication. Many
rely on standard underwriting practices. may result in risk that could be lenders, in fact, employ a fairly basic
unmanageable. This is why within the calculation, based on a coarse risk
Another driver of subprime loans growth subprime segment it is necessary to segmentation of their subprime customers,
worth highlighting is the increasing distinguish among different classes of denoted by letters such as A- (the least risky
popularity in Europe of mortgage insurance. customers in terms of risk. These range category), B, C, or D (the riskiest category).
This allows the lender to transfer part of the from lower risk customers (near prime), Borrowers within each category are
risk to the insurer. If the borrower defaults through medium risk, and to high risk generally charged the same interest rate
on his payments and the loan is customers. While the majority of lenders will with minor differences among borrowers in
subsequently written off, part of the lender’s only target the first two groups, the highest the same risk group due to the negotiating
loss is covered by the insurance risk group is generally the target of ability of the borrower or the mortgage
underwriter. With mortgage insurance, specialized lenders, prepared to cope with broker, and certain risk indicators such as
many lenders are willing to offer higher risk the greater risks implied by such loans or property type or ability to document
products. In Italy for example, mortgages who aim to profit through the repossession income. The risk segmentation, however, is
with a loan to value up to 100% are of the collateral (equity lending). often based on a limited set of risk drivers
becoming popular thanks to mortgage (eg LTV) that may not result in risk being
insurance, as generally regulatory The reality is that many lenders try to enter homogeneous among each class. This
requirements would not allow banks to offer the subprime market without the necessary causes lower risk customers within a class
housing loans with an LTV higher than 80%. requisites. As recognised by the US to be overcharged, effectively subsidising
Comptroller of the Currency [5]: “A number the higher risk customers within the same
Two important aspects in which European of institutions have incurred significant class who are being undercharged.
mortgage lenders are lagging behind their losses and other problems because of
US counterparts regard data capture and poorly structured subprime lending The segmentation may also be the result of
analysis. In the US, credit-scoring agencies programs. Generally, these institutions expert judgment that, although it is an
can access 80–90 per cent of people’s underestimated the higher default rates and important element in all business decisions,
38 HOUSING FINANCE INTERNATIONAL – September 2006256112 HFI September 06 25/10/2006 11:40 am Page 41
HOUSING MORTGAGE
SUBPRIME
& MORTGAGE
HOUSING TRANSACTION
LENDING IN CHINA
lacks the objectivity and consistency of the probability of a future event. Credit PD is the likelihood that a client will default
quantitative analysis. As for the correct scoring is an instrument widely used by on his repayments in a given timeframe. The
pricing for each level of perceived risk, this companies for the internal processes of EAD is the value of the bank’s exposure at
is a task that requires a significant effort in portfolio risk measurement and the time of the borrower’s default. This can
terms of data collection and analysis. Many management. It allows a lender to estimate be expressed either in absolute terms or in
lenders employing simplified pricing criteria the probability of future insolvency of a terms of a percentage of the loan exposure
find themselves incurring higher than person requesting credit or of an existing at origination. The LGD is the loss on a
expected losses. customer. The profile is calculated on the credit instrument after the borrower has
basis of the information available at the defaulted. It is therefore a percentage of the
The most advanced lenders rely on moment of the decision. Application scoring exposure at default that takes into account
automated underwriting systems and on estimates, at the moment of the request, the not only the amounts recovered but also the
quantitative analysis for price setting. level of risk associated with each associated direct and indirect costs. By
Automated underwriting systems evaluate application before it is effectively approved. building scoring models for each of these
applications using algorithms based on In practice, credit scoring results in the three parameters, a lender is able to have an
statistical analyses of historical data definition of a table listing the objective and accurate estimation of the
containing previous applicants’ characteristics that provide the most expected loss of a prospective client at the
characteristics and their performance in predictive information, together with the application stage. An example of the
terms of debt repayment. This is generally associated attributes and weightings. A estimation of the expected loss using
referred to as credit scoring. Credit scoring total score is obtained as the sum of the scoring models is given in the figure below.
techniques are recognised as being more points in each characteristic. The total score A given applicant would be associated to a
accurate and consistent than manual corresponds to an estimate of the different level of PD, LGD and EAD
underwriting in determining the riskiness of probability of default. (expressed as a percentage of the original
individual loan applications. An accurate exposure) using the output of three scoring
estimation of risk allows a more Even though scoring models have models. The product of the resulting values
sophisticated quantitative approach to traditionally been used only to assess the of these three parameters with the
pricing to be taken, which in turn allows a default probability of a client, the true risk of requested loan amount results in the
lender to define clear and objective goals a loan is given by its expected loss. Not estimate of the expected losses.
such as profit maximisation or volume every client with the same default
increase. probability has the same potential loss. Two
clients with the same default probability can
An objective risk evaluation of subprime react differently to collection procedures
borrowers and generate very different losses for the
lender. The product of three different
Scoring can be defined in general as a elements gives the expected loss:
statistical technique to predict, at a specific probability of default (PD), loss given default
point in ti.me with the available information, (LGD) and exposure at default (EAD). The
Figure 1 – Calculation of expected losses
Loan and applicant characteristics
PD scoring model EAD scoring model LGD scoring model
Quantile PD % Quantile EAD % Quantile EAD %
1 5 1 50 1 10
2 10 2 58 2 30
3 12 3 65 3 40
4 15 4 68 4 42
5 20 5 70 5 48
6 25 6 72 6 57
7 30 7 74 7 60
8 35 8 78 8 68
9 40 9 80 9 73
Loan amount = $100,000
Expected Loss = PD x EAD x LGD x Loan Amount = 0.12 x 0.58 x $100,000 = $3,967
HOUSING FINANCE INTERNATIONAL – September 2006 39256112 HFI September 06 25/10/2006 11:40 am Page 42
SUBPRIME MORTGAGE LENDING
There are three main drivers of risk for while retaining all the evidence available in Cost of funding 4.50%
mortgages: LTV (there is in fact significant the existing portfolio (for example regarding
evidence that default rates rise sharply the correlation between the variables). Operating costs 1.25%
when the LTV is higher than 80%), debt to
Expected loss allowance 2.50%
income ratios (DTI, measuring the ratio Another issue which affects the estimation
between instalment and monthly income) of risk for mortgages in general is that Desired return 2.00%
and credit bureau scores (accounting for collateral risk can have a greater impact on
experiences with other institutions for a loss than default risk. Credit risk estimation Final price 10.25%
more comprehensive assessment of the techniques are limited by the difficulty in
applicants’ creditworthiness). These forecasting changes in collateral values, Figure 2 – Cost model for interest
variables represent the key information used regardless of their accuracy in assessing the rate determination
by credit scoring models to assess the PD creditworthiness of individual borrowers.
and the LGD. Socio-demographic Experience shows that collateral risk is This cost model incorporates all the items
information (age, residential status, etc) of difficult to measure [6] and, even more so, it that should be taken into consideration in
the applicant is also used to increase the is difficult to forecast over the typical life of determining the price:
accuracy of the models. As for the EAD, the a mortgage.
main driver is the maturity of the loan. It is a) Cost of funds: the interest rate at which
very rare for borrowers to default soon after Some lenders, while recognising that credit the lender can borrow money;
they have received credit (unless fraud is scoring is a valuable tool, do not think that it b) Operating costs: rate increase to include
involved) and when they have nearly can be applied to subprime lending. This is costs to originate, service and terminate
completed the repayments. Estimating the because the risk in subprime mortgages is the loan;
time of default and knowing the repayment driven by many qualitative factors. c) Expected Loss Allowance: rate increase
amount enables the lender to estimate the Therefore it is argued that an individual to take into account the expected loss
exposure at the time of default. approach to underwriting is necessary. rate (PD x EAD x LGD) that can be
However qualitative evaluation criteria can estimated using scoring models as
New entrants to the subprime segment are always be combined to the results of the described above;
likely to have some difficulties in developing quantitative evaluation performed by the d) Desired return: takes into account the
scoring models. Scoring models allow scoring model to increase the quality of the target return on the investment and
accurate predictions when they are risk estimation. This approach is also taken should be such as to adequately
developed on the basis of historical data for the credit risk assessment of small remunerate the cost of the economic
that is representative of the characteristics businesses (SME) and corporate clients capital (for example, as from Basel II
of the target population. Therefore models where there are a lot of qualitative aspects advanced internal rating approach or
developed for prime borrowers should not (such as the quality of the management or from any value at risk model) employed
be used for risk assessment of subprime the business perspective of the market by the financial institution (also known as
borrowers. Institutions should instead build sector where the company operates) used the cost of risk).
models on data representative of the to integrate the quantitative risk evaluation.
targeted subprime borrowers. However new In a more comprehensive view of the
entrants to the subprime segments will not A quantitative approach to risk based approach, there are other items that can be
have this data available, given that pricing for subprime lending considered such as relationship and
mortgage defaults appear generally several competition adjustments. Relationship
years after origination. In this case it may be Risk based pricing (RBP) is the practice of adjustments take into account the
possible for them to rely on external data. It charging different interest rates depending relationship and value of existing customers
is also possible to infer the behaviour of on the risk of the loan. A basic approach to in terms, for example, of potential additional
risky customers from an existing portfolio of the determination of the interest rate is revenues to the lender in the form of a
prime customers by incorporating prior based on identifying costs and then adding purchase of additional products or the
assumptions regarding the relationship the required rate of return to determine introduction of additional clients. More
between risk and risk drivers. For example, pricing. An example of this approach, in its valuable customers can be offered a higher
a portfolio of prime loans may not have simplified version, is given in the following discount. Competition adjustments allow
customers with LTV > 80%. A possibility is figure. lenders to take into consideration existing
then to assume a relationship between market conditions with a price adjustment
default rates and LTV for LTV values higher that makes sure that the final price is
than 80%, using for example available competitive. Whilst the latter is quite usual,
evidence from the US market. This enables the former is more difficult to estimate given
the subprime lender to build a scoring the several difficulties that have to be faced
model for an extended range of LTV values when dealing with the estimation of future
40 HOUSING FINANCE INTERNATIONAL – September 2006256112 HFI September 06 25/10/2006 11:40 am Page 43
SUBPRIME MORTGAGE LENDING
cross- (or up-) selling opportunities or the Flat rate lenders could then face a significant pricing strategy. With no differentiation, the
evaluation of the value of new customers issue since, over a period of time, the profile interest rate is independent of the risk of the
potentially introduced by the applicant. of the portfolio could change with a lower applicant (denoted in the figure as AAA for
proportion of good quality customers and an the less risky, to the more risky CCC). By
Given that under a risk-based pricing altered risk profile, which can significantly charging a lower interest rate to low risk
framework the price is more sensitive to risk, change the financial exposure of the applicants it is possible to improve the
financial institutions adopting it can avoid organisation (as offered prices would be in volume of applications (maximisation of size)
the adverse selection effect. Adverse fact more attractive to risky borrowers). while still being profitable. A maximisation of
selection can affect lenders not applying risk adjusted returns results in a flat price for
risk-based pricing in a market where other Another benefit of a quantitative RBP the lower risk grades and an increase in
lenders are. Higher risk customers will be approach is that it allows for a differentiated price for the higher risk applicants. A
offered higher rates by lenders operating a set of strategies to be implemented. The combination of the two previous approaches
risk-based pricing strategy and will following figure illustrates three possible results in a full price discrimination with a
potentially migrate to those lenders offering strategies that can be adopted with a trade off between volume and profitability.
a flat rate as this will offer them better value. variable interest rate compared to a flat
Figure 3 – Implementation of different strategies for risk based pricing
Now: market taker, no differentation Maximisation of risk adjusted returns
2500 2500
2000 2000
1500 1500
1000 1000
500 500
0 0
AAA AA A BBB BB B CCC AAA AA A BBB BB B CCC
Y axis: basis points
X axis: credit class
Maximisation of size Full price discrimination
2500 2500
2000 2000
1500 1500
1000 1000
500 500
0 0
AAA AA A BBB BB B CCC AAA AA A BBB BB B CCC
The adoption of risk-based pricing brings Some limitations of risk-based pricing arise on interest rate, beyond which customers
benefits not only to lenders but also to from constraints set by regulators. One have to be declined.
consumers. Lower risk consumers are example is the need to advertise interest
rewarded for their good financial rates. In the UK, for example, lending Although the RBP practice is not without
performance and can benefit from organisations must give the advertised rate, criticism, it allows lenders to set prices in an
increased choice and competitive rates. For or a better rate, to 66% of applicants. This objective and transparent manner. This is
higher risk consumers, the gap between is simple when using a flat rate; however it very important in the subprime market
mainstream and alternative finance sources becomes much more difficult with risk- where there are many concerns regarding
can be significant. Organisations that based pricing, where there needs to be fair lending. An objective assessment of
previously would not have lent to these high accurate portfolio information to determine client risk allows reputable lenders to offer
risk consumers are now lending at an the advertised rate. Another example is fairly priced product to individuals who
appropriate price, so risk-based pricing caps on interest rates set by regulators. This have difficulties in obtaining finance on
brings more of the population into the means that if risk-based pricing is used standard terms and conditions. Serious
mainstream lending process. there may need to be a cut-off level based lenders can therefore differentiate
HOUSING FINANCE INTERNATIONAL – September 2006 41256112 HFI September 06 25/10/2006 11:40 am Page 44
SUBPRIME MORTGAGE LENDING
themselves from predatory lenders [1] who must assess all costs needed to service [2] Mortgage Insurance Trade Association
instead attempt to make a profit in the these assets and the expected losses that and Mercer Oliver and Wyman News
subprime segment by: will impact on profits. Some subprime Release (2005) Untapped mortgage
a) marketing loans explicitly at those in lenders, in fact, charge borrowers high up- demand tops 3500 billion across Europe
debt; front fees that allow them to originate a high
b) offering loans with limited or no enquiries volume of loans that include very high risk [3] Financial Times (2006). Fears over surge
about income; customers, and this can put purchasers at a in high risk mortgage
c) being interested in the value of the disadvantage. Furthermore, subprime
collateral rather than the borrower’s loans, especially those purchased from [4] Inside B&C Lending, Vol. 10, Issue 4,
credit- worthiness (‘equity lending’); outside the institution’s lending area, are February 14, 2005, p 1.
d) charging high brokers’ or other advance particularly at risk of fraud or
fees; misrepresentation (ie, the quality of the loan [5] OCC bulletin 99-10 (1999) Interagency
e) charging very high interest rates not may be lower than the loan documents Guidance on Subprime Lending
commensurate with the real risk of the indicate).
applicant; [6] Temkin, K., Johnson, J.E.H and Levy, D.
f) charging increasing interest when a loan is One issue not to be overlooked regards the (2002) Subprime markets, the role of GSEs
in arrears. impact of an economic downturn. Adverse and Risk Based Pricing. The Urban Institute
macro economic conditions have a
A key issue for the development of the particularly relevant impact on the less [7] Bradley, E., Kane, B. (2005) Europe's
subprime loan market is funding. While financially stable subprime borrowers. Whole Loan Sales Market Burgeoning As
banks can raise money from savers to offer However, mortgage products have Mortgage Credit Market Comes Of Age.
mortgage loans, smaller players, who may developed considerably in the last decade, Standard & Poor’s Structured Finance
have no branch network, cannot rely on without related risks being tested in times of
deposits to finance their lending. These economic stress. The need for stress testing [8] Cataldi, G. (2003) Profitability
institutions can raise funds for further required in the Basel II framework is Optimisation: The use of Risk Adjusted
lending by selling part of their loans (their therefore particularly relevant for subprime Performance Measures. Experian-Scorex
assets) to institutional investors. This lenders who should test the effect of White Paper
practice, called securitisation, is becoming economic downturns on the
more widely used for subprime lending. The creditworthiness of their portfolio. [9] Putt, J. (2006) A profit driven approach to
credit risks are passed from the original risk based pricing. Experian-Scorex White
lender to the institutional investor while the Conclusions Paper
originator retains the loan servicing rights.
An objective and consistent assessment of There is a growing demand in Europe for [10] White, A.M. (2004) Risk-Based
risk through quantitative analysis is higher risk mortgage products. This market Mortgage Pricing; Present and Future
therefore a key aspect of these offers many opportunities for profitability Research. Housing Policy Debate Vol 15
transactions. This is particularly critical for but it also involves considerable risk. Some Issue 3.
the whole loan sale market. In a whole loan prime and near prime lenders have already
sale, an originator will sell a pool of loans approached the market directly or through [11] Lax, H., Manti, M. , Raca, P. and Zorn,
directly to another institution, rather than other companies they control or with which P. (2004) Subprime Lending: An
through securitisation. The buyer pays an they have ad hoc agreements for the sale of Investigation of Economic Efficiency.
originator face value plus a premium for this subprime credit products. A quantitative Housing Policy Debate Vol 15 Issue 3.
pool. While in the US this type of fund approach to risk evaluation and more
raising is widespread, the European market tailored pricing strategies can provide the [12] Collins, M., Belsky, E., and Case K.E.,
has only recently reached a level of competitive edge that will enable the first (2004) Exploring the Welfare of Risk-Based
structuring sophistication and an movers to acquire a dominant position in Pricing in the Subprime Mortgage Market.
understanding of fundamentals of the credit these markets. BABC 04-8
risk, liquidity, and relative value pricing to
embrace the whole loan sales market [7]. A References [13] International Convergence of Capital
strong whole loan sales market is likely to Measurement and Capital Standards. A
be a key driver of growth for the European [1] Munro, M., Ford, J., Leishman, C. and Revised Framework (2004). Bank for
subprime mortgage market. Kofi Karley, N. (2005) Lending to higher risk International Settlements
borrowers: Sub-prime credit and
The evaluation of risk is a critical task for sustainable home ownership. Joseph [14] Crews Cuts, A. and Van Order, R.A.
institutions purchasing subprime loans Rowntree Foundation (2003) On the economics of subprime
securities from other lenders. Purchasers lending. Freddie Mac
42 HOUSING FINANCE INTERNATIONAL – September 2006You can also read
