CA2295569A1 - Method for mortgage and closed end loan portfolio management - Google Patents

Abstract

A method for mortgage and closed end loan portfolio management in the form of an analytic tool designed to improve analysis of past and future performance of loan portfolios. In accordance with one aspect thereof, the invention aggregates loan units into loan vintages, wherein the loans in each vintage originate within a predetermined time interval of one another. The invention compares different vintages to one another in a manner such that the ages of the loans in the different vintages are comparable to one another. An early warning component of the system predicts delinquency rates expected for a portfolio of loans during a forward looking time window. A matrix link component of the invention combines the loan vintage analysis with the early warning component of the invention and predicts the default rate of the loan portfolios at a selected future point in time. The results of the analysis are graphically depicted and/or automatically fedback to provide "yes" or "no"
decisions regarding investments in various loan portfolios.

Description

METHOD FOR MORTGAGE AND
CLOSED END LOAN PORTFOLIO MANAGEMENT
BACKGROUND OF THE INVENTION
The present invention relates to banking and, more particularly, to a loan performance analytic tool designed to improve analysis of past and future performance of loan portfolios.
Financial institutions such as banks own large portfolios of mortgage and other closed end loan instruments. Further, there is a constant influx of applications for new loans and mortgages and, moreover, existing loans are treated by banks as commodities or products which they trade among themselves. Banks underwrite loans and/or purchase loan portfolios of other banks or sell portions of their own loan portfolios. In doing so, banks customarily continually assess and reassess the quality of various loan portfolios, which quality depends on the interest rates earned on those loans, the customer payment history on the loans and other criteria.
As regards newly originated loans, the process - begins with loan applicants submitting applications to financial institutions which then triggers an SUBSTITUTE SHEET (RULE 26) investigation by ei'-~.her the bank and/or related service organizations which check the cred_t history of the applicant before the loan is approved. Typically, the decision to grant or not grant a loan implicates various credit screens that examine such factors as the loan ", value ratio (LT~7) or the par ~icuiar appl;catien cr the debt to income ratio (D/y) of the apDl_cart and ether historical facis, ahi ch she? lochs on t'~e commercia-worthiness of the given loan cransac~ior.. Onca a loan _s granted, it becomes oar of 3n a~~r=mentioned vase portfolio of loans ahi ch a g-_ven financial ins ti ration owns and/or services. The "cuali w" ef the particular loan heavily depends on the interest fees earned by the financial institution on eac:~ loan and on the per f or_nance of the loan which is dependent on the timeliness of the payments by the loan applicant and/or on loan prepayment.
Loan portfolios represent to banks t~ro secarate and distinct lines of business or sources of income. One business line or source of income flows from the ownership of the loans and the earning of interest fees thereon. The second line of business involves the servicing of the loan, for example, the keeping of records, collection of periodic payments, enforcement in the form of loan foreclosures, etc. Banks can earn fees on servicing of loans which they ei~~.her own outright or which they service on behalf of other financial institutions. This is because it is traditional in the banking industry to attribute to each loan a basic cost of servicing which is included in the interest fees charged to the cus tourer . _T f a bank is able to car~~ ou or per f orn these servicing tasks at a cost s truczsre which is below the or~.ginall~r attr=bused servicing cosy, the bank is able to r ea l ize a profi t fr our i 's loan servicing business.
14 It is nct uncommon for large °inancial insti~utions to immediately tur:~ around and sell to other investors portions of the loans ''hat they have booked, to spread the credi t risks and in order to diver sify the types of loan instruments that they are holding. The same is true of the services end cf the business with respect to which decisions are constantly made as to whether retain or sell the servicing components of various groups of loans.
The loans that are retained far servicing are assigned to a subsidiary of the financial institution which is a purely service organization that has developed the methodology and procedures far servicing loans. A
portion of the loan portfolio can be sold to third party loan servicing bureaus. It is common for banks Which sell loans to retain ownership of the servicing rights to WO 99/03b52 PCT/US98/13195 earn the fee income thereon. _Tn addition, many financial institutions may decide to purchase servicing rights from other financial institutions.
In any case, ban3c managers are responsible for managing loans totalling billions of dollars both as cure loan instruments and as produces that recuire se~--r;cinc.
The decisions whether to rez3~n d=f=erent groups of loans or whether to sell them of' ~o other investors anc, on the other hand, whether to wurchase loan portfolios from other ins ti ration f or ownersh=p cr ser-r~.c;nc purposes are bottom line decisions that have the potential to of°ect the financial institution's profits and/or losses to the tune of tens or even hundreds of millions of dollars.
Hence, loan por~~olios are constantly examined by bank managers very carefully since different vintages of loans can perform quite differenti~r from one another.
For example, a portfolio of_loans representing mortgages granted in a particular locality during a particular time frame might be deemed to represent high quality loan instruments, as for example in the situation where the history of these groups of loans has shown that the rate of default for that group of loans has been extremely low and the interest rate on those loans is high compared to present interest rates. Conversely, another portfolio of loans granted in another region of the country which may have suffered economic decline may result at some future date in large rates of default.
Assuming further that these loans were issued at a low interest rate, it is not d_=ficult to understand that the particular "product" -- the portfolio of loans -- °aoulc be deemed to possess low vague and be a good candidate for being said. ?~lternativelv, a shrewd bank manager might see future value in a presently poorly per=arming loan portf olio and seek to buy a~ ins current low arice structure for its potential improvement. In the same vein, the "servicing" of such loans :nay be more di fficul and expensive due to higher defau'_t ins~ances. A bank might wish to sell off the ownership component of such a loan portfolio, or the serv_cing rights thereof, or both.
Sometimes, however, a financial institution which has a "servicing" subsidiary that is being underutilized may be willing to accept loan portfolios of servicing rights considered unattractive by other financial institutions.
In the prior art, bank managers entrusted with making the aforementioned decisions have often resorted to and relied on manual research and their intuition in ' their attempts to predict, manage and select loan portfolios for ownership and servicing purposes. The prior art approach has failed to provide a straightforward and easy to comprehend and administer system for assessing the past performance and future likely course of loan instruments.
SUMMAR'r OF THE INVENTION
Accordingly, i~ is an object of the present invention to provide a sys~em and method which improves the understanding of the pass perfor~aance cf lean portfolios.
It is another cbjec~ of t:~e invent=on to provide a sys tem :~hich enhances the obi 1 i t~r of f i nancia_ institution managers to choose which mer~gace and other debt instrument applications to under-write.
Yet another object of the present invention is to provide a system and method which enhances the ability of financial institution personnel to make decisions whether to retain or dispose of dil~erent groups of loans.
It is also an object of the present invention to provide a system and a method which is able to dynamically and automatically evolve loan underwriting criteria.
It is yet another object of the present invention to provide a dynamic underwriting model which is capable of being implemented in a general purpose computer.

It is also a further object of the present invention to provide a system and methodology which enables automatic processing of loan applications through a system that feedbacks infcrmation from a dynamic processor and which allows loan acceptance decisions tc be made automatically and rapidly.
The foregoing and other cb~ec~s of the invention are realized by a system and process which is tailored to analyze and selec~ lean portfolios for either continued or future investment by a financial institution. Eac:~ loan portfol io comprises a plural i ty of loan units and the system operates by separating the loan portfolios into a plurality of loan vintages, in a manner such that the loans included in each loan vintage 1~ have origination dates that are on average of the same age. The system of the invention produces an analysis of the past performance of loan portfolios, as well as an indication of the future performance thereof in two different formats.
As to past performance, the invention develops the loan vintages in a manner such that vintages of different years can be compared to one another meaningfully because the loan units in each of the different vintages are actually of the same comparative 25 ages. For example, when 1993 and 1994 loan vintages are -a-compared, the loans units that are being compared are or the same age to provide more meaningful comparisons.
This is referred to in the ensuing description as the Crus Classes analysis system. In one embodiment of the S Crus Classes system, output results are graphically depicted by means of a curve ;which represents the dif f er ence between the de 1 incruency ra tes o f leans i:~ the too yearly vintages. To improve the r='iabil=tw ef the results, an area of uncertainty is superimposed ever the IO difference to allow users to focus their analysis on those locations on the dif=erence plof which lies outside the area of uncertainty. This increases the reliability of the analysis and the ability to trust its results.
The area of uncertainty can be calculated as a ~1 and -1 15 standard deviation, but the actual size thereof is a matter of persona? choice.
The early warning system (EWS~ constituent of the invention is one of the systems and processes which predicts the percentage of the loans in a given loan 20 vintage which are likely to enter a "bad" state within a predefined forward looking time window, for example, the next two years. The prediction is calculated by using a logistic regression formula which has been developed in part on the basis of the analytic results obtained from 25 the Crus Classes analysis component of the invention.

_ g _ Finally, the so-called matrix link component ef the present invention is generally similar to the aforementioned early warning system in that it is a prediction tool. It differs from the early warning system in the resz~ec t that ~ ,. is capab 1 a o f f orecas tincr the percentage of loans tha ~ are li:ceiv to be bad a ' a date certain within the aforemen~ioned forward loeic_ng time window. In all cases, the results of the ana_-rsis can be graphically depicted by comparing vintages ~~ one ZO another, using various curves, bar charts and the ?i~Ce i_~.
a manner descr ibed her ein. For i._~nposing the intecr'_ ~y o.
the results it is desirable that the number of loan ~.mi~s in the analysis be large, preferably in the hundreds of thousands of loan units and preferably at least 50,000 loan units.
As a general note and definition applicable to throughout the present specification and claims, the tern "loan portfolios" means, includes and/or refers to booked loans, applications for loans for which underwriting decisions have to be made and the aforementioned loan servicing rights.
Other features and advantages of the present invention will become apparent from the following description of the invention which ref ers to the accompanying drawings.

BRIEF DESCRIPTION OF '~~iE DRAWINGS
Fig. 1 is an overall bloc! diagram of a dynamic underwriting system and method in accordance with the present invention.
Fig. L~ is an explanatory char which shows an example of the delinauency rates c~ various loans by vintage year for a g'_~ren loan pcr==olic .
F ig . 1B is a ~~ =_e czar ~ and exemp 1 ar c f an actual family of d=f==_rent ~~~es c_ 1 oa_~. groups and shows IO the rate of delinauenc_~ associate: therewith.
Fig. 2 il?ustrates a pr'_cr ar= method far assessing the profitability and per fcr-zance of a portfolio on a yearly vintage basis.
Fig. 3 illustrates a novel metucd of assessing the past perf ormance of l own nor tf o l ios based on yearly vintages in accordance with the present invention.
Fig. 4 is a graphic that ~~llustrates a method of the present invention involving assessing the relative value of different vintage loan por~folios.
Fig. 4A is a table which illustrates calculations performed to obtain data for the graphic of Fig. 4.
Fig. 5 illustrates a furt:'~er graphical method of the present invention for showing both the past performance and the future predicted performance of loan portfolios.
Fig. 6 is an explanatory graph provided for explaining how a portion of the graph cf Fig. 5 is obtained.
Fig. 7 illustrates the forma' of a roll rate delincuenc_~ table for one year ah'_ch .s used in the mat.:;x link analysis :nodule of the orssent inventicn.
Fig. 8 is a table which shcws the met:~odolocv l0 and ecruations used in_ forecasting bad rats probabi'_~-ies in the matrix link component of the present invention.
Fig. 9 is a plot cf the final results obtained with the matrix link component of the present invention.
Fig. 10 is a hardware/seftaare block diagram cf key components of the present invention.
DETAILED DESCRIPTION OF EMBODI?KENTS OF T~iE INVENTION
By way of background and introduction, the general environment for the method and system of the present invention can be better appreciated by initial reference to Fig. 1. As illustrated therein, home buyers and refinanciers 12 typically submit applications for loans to one or more financial institutions 14. These institutions include loan granting departments that decide whether or not to book given loans by applying i various credit screens, i.e. criteria. One screen may focss on the applicable LT~7 (loan to value) of a transaction, the D/~ (debt to income) ratio of the involved transaction and/or cn the credit history of the particular applicant.
Based on she aforemen~,ioned and other criteria, a decision is made tc accent or r~jec~ a particular loan appi~;cation. Each ican t:~a~ :pas been accented is added as another loan uni t to a la-ge oor-~y o l i o c= s _~n;lar families of loans, e.g. conf:,rmi:~g loans, jumbo loans, government loans, et~. d loan has typically a loan start date and a date by which the loan is expected to be fully paid up, as is t~mical of home mortgage loans. A loan that is issued for a =fixed amount and period of time is I5 known in the trade as a closed loan. These closed loans I6 are artificially split and treated as tao business securities or entities -- namely as a-"loan" entity and as a "servicing" right, as indicated at 32.
Each loan unit or instrument represents to the financial institution an opportunity to earn a profit on the differential between its cost of money and the amount of interest earned from the borrower. Another profit component is realisable from the servicing element of each loan entity. That is, a finite budget for labor and equipment use must be allocated when the loan is issued to service each loan over its life time. The banking trade has traditionally derived substantial revenues from the servicing of loan portfolios, to the extent that t.'~ev were able to service loans at a cost below the originally calculated service allocation. Consequently, banks and other financial institutions sometimes trade loan "servicing" contracts. These con'trac~s are rouZinelv purchased and sold in farce uni~s since trey represent income opportunities. For example, a Lank which lacks a l0 servicing department might contr ac t w~. t~: another bank t..
service its loans at a set, per loan pricing arrangement.
The bank that purchases the cont=act does so with the expectation of earning a profit on t.'~e project. If it develops later that a particular loan portf olio I5 experiences a large rate of defaults, the extra sere=cinc needed to collect funds on ~he loans might render the particular servicing contract unprofitable. In such a situation, the service organization might attempt to resell the service contract to another service 2o organization which might be interested in it, for example, at an increased service rate.
With further reference to Fig. 1, block 13 represents the department of the financial institution which makes the decision whether to retain or sell a 25 particular loan portfolio. Typically these loans are i sold in very large blocks, each containing thousands of individual loan units. Those loan units originating at block 12 which are retained by the given financial institution are represented by block 20. on the other hand, as indicated by tine block 22, a portion of the book of loans is sometimes said of' to investors and is secur=sized. Therefor=_, it wi_~. be appreciated that selling and purchasing loan aortfolios recuires careful examination of var'_ous loan product __nes to assess the==
viability, profitaaility and ==fated factors.
As already noted, another source of profit flows ~rom the servicing portion of the loans. Block 24 identifies the step which decides whet:~er to retain or sell the servicing component of a loan port_°olio. Those loans for which servicing is retained. are serviced at the bank which originated the loans as indicated at 26. The servicing of the balance of the loans procured at block 24 is contracted out to third parties for services as indicated at block 28. In addition, the servicing end 25 of the banking business is also able to purchase the servicing rights as indicated at 32.
As described, the banking industry distinguishes between ownership of loans and the servicing thereof. Loans that are owned by a given financial institution can be serviced by that institution's own servicing subsidiary or the servicing part can be contracted to third party servicing bureaus.
Indeed, not al' financial institution have loan servicing departaents. Conversely, a bank with a servicing organization can pu~Yc:~ase tze "ser-ricing" component associated with loans owned by other banks and render the se~~ricing thereon.
In any case, it is se?=-ev_dent that tae prof i is fr om earring inter es ~ on loan pcr-.f olics and from l0 the loan servicing 1_ne of business is heavily influenced by the performance of various loan groups vis-a-vis the default rate of these loans over the life of the loans, foreclosures, collection efforts, loan prepayment and the like. Loan portfolios which experience low default rates are easy to service and are highly profitable to financial institutions.
Traditionally, the decision to purchase, retain, sell or create loan portfolios demands critical analysis of the past performance of the loan portfolios under consideration. Moreover, such decisions invariably implicate assumptions and predictions as to how such loan portfolios will perform in the future. Not surprisingly, the decisions to book loans at block 14 typically depended on and reciuired analysis and consideration by highly skilled and experienced persons having very keen and sharp analytical powers to determine the potential profi~ability of loan portfolios being considered.
The present invention departs from the prior art by providing a dynamic under-~riting method and system 30 comprising several key components including an early warning system 32, a Crus Classes analysis sec~icn 34 and a mat=ix link 36, all to be described further on.
~'ssentially, the infcrmaticn obta;ned from the subsvs~ems 32, 3~ and 36 is des_gned to be applied, via °eedback line 33, to the decisicn box 14 in a manner which systemizes and provides a standardized approach to forming the decisions whether to book loans. The invention substantially increases the reliability, consistency and speed of the loan acceptance decision process. Further, the dynamic underwriting system 30 of the present invention can also be applied via feedback line 40 to the decisional box 32 which addresses the decisions at block 32 whether to purchase loan servicing rights of loans owned by other financial institutions.
Finally, the feedback line 41 provides feedback for forming the decisions identified in blocks 13 and 24.
The invention shall now be described with respect to the subcomponents of the invention, including the aforementioned subsystems 32, 34 and 36, beginning 2S first with the Crus Classes component 34.

Experience has shown that the past performance of a group of loans is often a key indicator of its future behavior. Therefore, the first step in the analysis process focuses on providing an improved analytical tool for the examination of the loans' past performance. This is the function provided by t:~e Crus Classes system 3.~. T_he Cyss Classes subsystem 3.:
essentia~.ly r epresents a f=esh approac:: tc the ana_nsis of the prior perfcrnanca or' already booked groups cf loans . The early warning s ys tem 3 2 is a f or.~ar d locking system which comprises a method and process that is able to predic t what portion of the over a 1 1 number of loans in a particular loan group will experience 90- day tardiness in payments by the bor=ower;s) thereof at anytime within a predefined time peried, fer example, the next two years. By way of example, i~ the system predicts that thirty loans out of a thousand in a given loan por'-rolio will experience 9D+ day delay in payment at any time within the next t-ao years, the EWS (Early Warning System) 32 will return the value .D3 to indicate that it expects 3% of the loans in the given group to gc 'bad" at least once during the predefined time period. Finally, the matrix link component 36 provides a more sophisticated analytical model, in that it not only assigns a probability to how many loans will enter a so-day default - is -but, moreover, calculates the expected number of loans in default on a specific future date.
Turning first to the Crus Classes system 3.~, a common technique used in credit risk management in the a mortgage indus~ry is tc group loans by common interTals of origination, e.g. annua;'_,r, to compare their performances. Fcr example, the mor~gace indus~iy m_ght typically wish to ana_yze the performance oy ~99~. v_~!'ace loans. ~7intage in this con~ext means all loans that have l0 bean originated in 199x. The class'_fvcatien cf loans into yearly vintages by the prior art has often resu'~ed in significant distor~ions cf ana~v~ical conclusions.
Unlike wines for which classification into yearly vintages makes sense, lumping a'_1 leans originating in 15 the same year into a same "vintage" distorts results because there are several exogenous fac~crs which affect how these loans perform and these factors intrinsically vary over time in a manner which can produce significant quarterly, and even monthly loan perfornance 20 fluctuations.
The present inventors have opted to use the term "Crus Classes" fcr its similarity to the wine industry. But Crus Classes, as used herein, differs from and departs from the prior art approach of grouping loans 25 by annual origination dates. The invention overcomes some of the statistical inaccuracies associated with the prior art s attempts to lump loans into yearly vintages solely on the basis of the origination of a loan in a given year.
Tradi tiona l vintage technicrues in the mor toacre indust=J allow bankers to gauge the cTUality of mcr~gages as they are "aging". However, the inventors have added certain s~atistica_ procedur=_s, such as hypothesis testing, used in the process cent=of manu~acturinc environment, that aglow the methcd of the invention to test for the statistical signi_icance of the diff_rences in performance among the "vintages". The result and benefits of the Crus Classes method to be described below is that it provides several advantages over the t,,rnical, prior art vintage analysis. For example, it incorporates a measure of dispersion. Further, it sets an analysis interval time shorter than a year to increase accuracy.
This produces several advantages over traditional vintage analysis: (1) it automatically adjusts the comparison to account for different numbers of loans and for different size loans; (2) the Crus Classes method also allows management to set the confidence intervals; and (3) it automatically adjusts the year-to-year comparisons for loans with different credit volatility.

However, prior to describing the spec:.f is features of the Crus Classes met:~od of the present invention it is wor~,~hwhile to introduce the following background inforaation. Mer~gage companies are vitally concerned °aith the performance ez the 1 oans they ser-rice and own. Active managemen~ of any lean port: olio recsuires the information needed t:, c'or'er ~ v cataccrize the pert crx~ance ef the uncier'_yr ing 1 cans . Gn an ex ~r eme 1 v broad and aide-sweeping ccmpa= ison, de'_nquencw r ats ar a generated and ccmpared to various c_asses of mor~cage loans and summarized on a naLiona= level.
It is common in t:he Indus trw f or dif f er ent financial institutions to share dais about the total number of loans serviced by them and the appropriate number of loans that are in some form of delincruencv.
Delinquency categories or "buckets'' range from the least serious, e.g. one payment past due, to the most serious category -- namely, in foreclosure. The following is an example prepared by the MORTGAGE BANKERS ASSOCIATION of a delinquency chart:
Total Loans1 Payment2 Payments3 PaymentsLoans Tots!
in Loans Serviced Past Past Due Past Due ForeclosureDelinquent Due or or or or or or Outstanding30 Days 60 Days 90 Oays F!C Total 2 5 of of of of of of 22.426,005733,330 163,710 141,284 230.988 1,269,312 6 Delinquent3.276 0.736 0.63,6 1.036 5.666 ~

Presenting delinquency perfor-.nance in this manner is helpful in quantifying total delinquency, but it reveals nothing about the endogenous factors contributing to the default performance of the underlying loans. These endogenous factors which affect the performance of loan portfol_os i.-~c'_ucie, but are net limited to, same for_n of ~ Per°ornance his~or yr cr ace of the loan, ~ remaining time to ma'urity, ~ loan amount, ~ interest rate, ~ borrower ' s crecit ;~cr rhiness , ~ geographic locations, and ~ underlying col'_ateral type.
However, the effect of "age" on the peryormance of loans is a main factor that mortgage originators use to discern whether a group of loans was (or is) "good" or "bad".
The vintage of the loan refers to the time when the given loan or family or group or set of loans has originated or has been placed on the books of the lending institution. In the mortgage industry, loans are categorized by year of origination, where all of the loans originated between January 1, 1996 and December 31, 1996 are referred to as '96 vintage loans. The yearly vintages and their corresponding delinquency rates are campared to each other to estimate relative peryormance and value. Fig. 1~ is an example of one such vintage chart listing multiple vintages. This is a snapshot taken at the last quarter of 1996. Theref ore, the oldes~
1995 vintage is twenty one months old_ In the figure, the 1995 curve shows a delincuency rate on the order of about 3~ for the 1995, t:~enm one mcnth oid vintage. Tn contrast, the 199 vintage csrwe spews a dei~.ncuenc~J rate approximately one ha'_. the see of the 199: v=ntage for 20 the same twenty one a~ont:~ vintage. T'~e mortgage defaul~
rates of Fig. 1~ s ignif icantl y of f ec t futur a loan performance. Indeed, a singl=_ percentage rise in the delinc,-uency rate represents many millions of dollars in losses to the typical financial institution which carries a very large port~olio of loans.
Fur+~her by way oz background, mortgage loan portfolios are quite heterogeneous, with many subtle and changing variations in the basic product characteristics and behavior. An important consideration in the methodology of the present invention is the evolvement of a model that preserves the heterogeneous nature of the mortgages. Therefore, the inventors have grouped the mortgages by various (endogenous) characteristics and made inferences about the relationship between each of these characteristics and the resulting level of default.

a large number of characteristics results in an extremely large number of combinations of groups to consider.
Fig. 1B illustrates the significance of maintaining proper distinction lines between various loan instriments based on origination. Thus, FiCT. 1B shows for a given financial ins~i~ution a total loan por~folio value of, for example, several billicns of dcllars, wita respect to which the cvera''_ rate of delinauencv is .36~
( at tier 11 or' the loan tr ee ef r ig . 13 ) . However , f or proper analysis the invention d'_vides that loan por:.~ol;o into loan types including "confox-ming", "jumbo" and "government" (originated) leans, as indicated at tier 13.
Note that the overall rate .860 of default is the weighed average of the delinquency ~a~e which varies from 1.25 f or conf orming loans , . 5 ~ a f cr j umbo loans , and 9 . 9 8 ~ f or government originated loans. Still further, each of the broad categories of conforming, jumbo and government loans are fur',,.her divided (at tier IS) into ARM (adjusted rate :mortgages) and fixed loans. Note the significant divergence in the rates of default. The same is true for the next subdivision (grouping) which hierarchically separates the third tier loan groups into low LTV loans and high LTV loans. For example, a government, ARM and low LTV loan at tier 17 has a rate of default of 0%, whereas a government, fixed and high LTV loan indicates (for the sample shown in Fig. 1B) a delinquency rate of 13.51°.
The invention applies the Crus Classes method on each node of the loan tr~e shown in Fig. 1B and ~-~.hen runs a hypothesis test to see if the performance of each year vintage is bettor, worse or statistically the same ( at a conf idence level of one s tandar deviation ) . I t is estimated that ther= are 308 di'ferent combinations and that it takes apprcximatelw loo megabytes of computer storage memory to analyze and graph the results for the model shown in Fig. 1B.
Analysis cf past performance of loan portfolios retxuires making a decis ion as to wha t cons ti Lutes a delinquent or "bad" loan, as for example for the purposes of creating a chart such as in Fig. 1A. In an embodiment of the invention which has been reduced to practice a first selection was to choose the definition of a "bad"
loan. It was chosen to represent a loan on which interest and principal payments were at least 9o days delinquent. That is, loans which are non-accruing or non-performing for a period greater than 90 days are deemed "bad".
Fur~~her, since the industry is accustomed to and prefers to refer to the "vintage" of a group of loans, for example 1993 vintage, 1994 vintage, etc., the WO 99/03052 PCTlUS98/13195 Crus Classes method 34 also produces and presents its results in terms of loan vintages. But, it groups and selects vintages differently than t:'~e prior art.
The difference in "vintage" selection can be appreciated from the matrix gables of Pigs. 2 and J. The abscissa axis 50 in Fig. 2 indicates the yearly quarter of origination, for example, March '9~, .Tune '9~, ezc.
The ordinate axis 52 ndicac=s the end quarter of a croup of loans, for exempla, June ' 95 , :~arc:~ ' 96 , Marc:~ ' 9~
.., etc. The mate ix data in Fig. 2 indicate the number of monz''~s that have elapsed fr om the guar ~er of or=cina tion to the end point. For example, a loan originating in March '93 is 36 months old in March '96 as indicated by reference numeral 54. Similarly, a loan originating in June '93 is 15 months old 'n June '94 as indicated by reference numeral 56.
The approach of the prior ar~ has been to select and aggregate as 1993 and 1994 loan vintages all of the loans between the bracket lines 58 and 60 for the respective years 1993 and 1994 . Carefully comnar; nrr t-~,A
precise ages of the 1993 and 1994 vintage loans reveals two aspects which may undermine and distor t the comparisons. First, the traditional approach reflected by Fig. 2 compares loans whose ages differ on average by twelve months. Indeed, some of the 1993 loans which are thirty six months old (see reference numeral 54) are twc years older than t.'~e tTaelve month old 1994 loans (reference numeral 55). The above approach skews the results considerably since t.'~e performance of loans is very age sensitive as can be appreciated from Fig. 13.
It is far more meaningful tc compare leans of the same age which originate in different. years. Therefore, i~ a far mot? relevan' tc be ab-a ~c ccmnare the perfcrmance of different loan vintages, as of the time ;,Then the~T wet=
at the same ages. For examp_e, in seeking ~o answer the cruestion: which loan vintage 1993 or X994 is better, it is more relevant to know and compare the comparative performances of the above noted loan v=ntages when each was, for example, tao years old. The Crus Classes methcd IS 34 of the present invention is at~le to do so.
With reference to Fig. 3, the present invention selects as the 1993 and 1994 loan vintages, those loans which are bracketed by the diagonally extending lines 62 and 64. In the selection method according to the present invention, the ages of the 1993 and 1994 loan vintages that are being compared are identical to one another.
Far example, for the year 1993 the ages of the loans vary between 6 months to 24 months and the same is true of the loans in the 1994 vintage year.

To further increase the accuracy of the campar isan, each yearly vintage is divided into four quarterly port~olios, i.e. =first, second, third and fourr.'~ guar ters. All of the 1993 portfolios of the same .. quarterly age are summed and di-rided by the total number of all the loans with the same r=_spectizre age. Ref errirc again to Fig. =, the gr~sent inventicn separately campar=s the 199 and .994 21 menth o_c loans, then the 21 mcnth old loans and so cn. '~':e actual mathematical IO calculations/analysis comparing loans originated in 199' and 1994 and the manner of calculating bad rates is shown below in Taples I, I= and T.aBLF I
I Age (Months)3 ~ 5 I 9 of Loans I n, I ~ I
(n) 1993 B~ rate I r, ~ r5 I ro I
Vintag(r) e STD Sort(r,*f1-r,)/n.)~ Surttr5(I-r5)/n~iI Sqrt(ro*(I-r,)lno) x of Loans( u, I y, ~ Na (N) 199a B~ ~ (R) Vintage~ R I R, - I ~ I

I STD SqrttR,*(1-R,)/N,)~ SarttR~'(1-R~/N,)Sqrc(Ro*(I-R4)Mo) DifferenceR~-r~ I R5-r5 (R-r) 1994-STD of STD-Syrt(rS*(1-STDs-Sqrtlr,~'(1-STD9-Sqrt(r9*(I-I993 (R-r) ~ ro)/no+It4*(I-R,)Mo) r~ r, /n )!n,+R.'(I-R,)/N,)) ,~R,S*(I-R~/N.~

Upper Bound+1*STD~ I +I*STDS +1*STD9 I

Lower Hound-1*STDz ~ -I*STDs -I*STD9 I

_ ~g _ T.aBLE II
I~~' ylonchs)( j 15 ,t of LuansI n" I n,s j n.9 ini 1993 B,~,D ate I r., I r I
i r ) , ~ ; ,, V'uttage -STD I Sqrtcr._~!1-r,=~/n,~~ Sarnr.,*~1-r,.;/n.5)~ Sqrt(r,s*!I-r,s;ln.~;
-of Loans ~ N., I N., I V.q (Nl 1994 B,~ ate I R., I R,< I R,g ~R) ' S i I SamR,*y-R,.,iu.:,, SurtIR,<'!1-R,SiiN.s,SuttlR,q*!!-R,,~,N.,~
D I

j DifferenceR.--r._ I R,a-r~< I R,9-r.R
~R-r) 1994-S~ .,t Si'D,:aartnr,:;-S'i'D.5-Surtcr,,:-S i'D.
.R-r; i -Sarnr, <1-1993 ~ - ~
, '/n.,-R.,'~!-R,__'iN..,."~mln,.-~..*~i-rZ.,'/N.x J , , y I r,q)ln,q-R.q*
1-R."/N.a, ~

j Uvver Bound:'S T D.-_ - _ "' _, _ I S ~ D.. ~ S'~'D.y Lower Sounda'S i'D..
; %STD.a I .
_ySLD

g T.~BL~ III
I .age I _. j =.~ ~ ,_ t~lonths) ,# of I n" j n,, j n~, Loans in) 1993 1 8,~ i ;,, I r-, j r, rate ~ r) V -image STD I Sumr.. -r,.,in,.;I Sartir,,~ll-r,,;in,,)j Sqrt(r-.'~I-r_;/n..j . ~

,# of I V-. ! N,' j V
Loans iNl 1994 ~ BAD I R.. I R,, R.,..
rate ;R) STD I SartIR"~~:-R.,jiN.,;~ SurtIR,_,"1-R,"IN,,)SurtfRr*(i-rT...,/V_;
I

Difference~ R..-r,. I R, -r, ~ R-r) j R..r r_..

199 S~,~,ii STD-;-Surtt:,,*';_~ Si'D=,-Surnr,,'~:-STD-.Surtvr-.!1-~R-r) 1993 ~ -,.'/n ,--R"x!1-R~ r ' Jn -R "!1-2 /n IN_ '/N -R
~ I r /N
, ) ! I
) ~!1 R
) ,. , ._., , , .., . ,, _ ..i ,, , ,~ , "
"
-, r -i Uoyer i ~ ! *5~.. j -1 *S i'D-, ~ 1 ~S T D-Sound .

Lower j -!'STD-. I -1'STD., -1'S T D-Bound Table contains the caculations and comparisons for 1993 and 1994 loans which have vintage ages of 3, 6 and 9 mcnths; Table II does the same for loans aged I2, 15 and la months; and Table III does so for loans that are 21, 2-I and 27 months. The first line in Tables I-III sets for~h the total number (n3, rlb, et~. ) of loans of the given v=ntage. The second line in the Tables sets for~-..h the bad rates (r3, ro, etc. ) for each vintage age. The third line calculates the standard deviation (STDj of the bad rate, using the indicated equations. The first three lines of Table I supply the relevant information and calculations For the 3, 6 and 9 month old 1993 vintage. The next three lines of Tabie supply the same infornati cn for the .994 vintage. The bottom four 1 i nes o= Tab 1 a = ca 1 cu 1 a to the d ~.~ f fer ences and compare the results for 199; and ~99.~ vintage years.
The last t~,ao lines calculate upper and lower bounds for the standard of deviation. The tap bcunds are plus one and minus one standard deviation, but management can set this based on their tolerance for default risk. These calculations constitute tine Crus Classes method 30 ef the invention whose effect can be appreciated from reviewing the analysis results plotted in Fig. 4.
That is the values calculated in Tables T_, II
and III above for the differences bet;aeen vintages 1994 and 1993 are plotted in the graph of Fig. 4 relative to a zero percentage base 71. The curve 70 represents the magnitude of the difference in the "bad" rates of loans of the same age. The value of the curve 70 equals r3-R3;
rb-R6; etc. shown in Tables I-III. One would be tempted to assume that the 1994 vintage performs better than the 1993 anytime the value of the curve 70 goes over zero percent and vise versa. However, such a made of analyzing the data would be subject tc reaching wrong conclusions due to statistical variations. To overcome this drawback, it is more significant tc ask whether a vintage that appears tc perform better does so in fact or whether i t merel y r ef 1 ec is a tempcrar~_r phenomenon . Tc answer the questicn, the invention uses a :zvoothesis testing technicue which a'_iows the anaivst 'c set a confidence interval ~rhich is adjustable to a'_icw for different corporate =isk teleranc' levels. Thus, the confidence interva; can be ecuated ~e the amount of risk tolerance management will accept in originating, purchasing, retaining or servicing loan por~folios.
These confidence intervals can also be used in product profitability and cacitai allocations. Management can then rank the vintages by produce, program, age and size.
To this end, the Tables presented above also calculate the standard deviation of the difference in performance and sets upper bounds and lower bound of +1 and -1 standard deviations for each quarterly vintage. These upper and lower limits which appear in the last two lines of Tables I, II, III are plotted in the form of curves 66 and 68 in Fig. 4. The area between the curves 66 and 68 is an area of uncertainty.

With this in mind, since the invention superimposes the curve 70 over the area of uncertainty, one can state with greater certainty which vintage performs better only in the areas outside the area of uncertainty. Thus, the graph of Fig. 4 shows that the 1996 loan vintages are "better" than corresponding X993 loan vintages f or loans the t era o' , .. _ and 2 .~ :non tzs old .
On the other hand, the 1993 v=ntage appears to be better for loans that are 3, and 30 months old. During other months, the resist is too close to conclude wit:n the chosen degree of cer tainty which vintage is better. The chart of Fig. 4 underscores the fallacy of the prior art in referring to yearly vintages as better o~-wcrse. One must be more specific as to time and other criteria, since relative performance changes dynamicall y wit:! time.
While the invention has been described above in relation to the consideration of vintages in yearly quarterly units, note that in the loan industry exogenous factors such as changes in economy, unemployment and inflation are time varying factors that vary greatly over an annual interval and therefore the system of the invention permits analysis based on the choice of any interval unit. The important thing to realize is that ir.
general, a new mortgage loan is more sensitive to small changes in delinquency performance than an older mortgage. This is shown by ~ridening of the confidence interval bands over time. So in essence, the application of the above described Crus Classes method corrects for this fact.
The invention also takes and adjusts the vintage rating based on quality comparisons for di'terent volatilities of default. In essence, using the svs~em lets the user to set policies -ai ~h r'saec~ tc volatilities of default which is another form of risk managemen t . This is new to the Indus tr v .
The confidence level in the assessment of the difference in quality between groups of loans decends to a certain degree on the sample size of the loans. =or small groups of loans, one will always be less certain or their performance. The real question is how much less certain. This is answered with the Crus Classes method.
The Crus Classes method also automatically adjusts the comparison for different sample sizes of loans in each node or product. This is evident iz the calculations in 2o the previously presented tables which always take into account the number of loans. An actual calculation that has been carried out to evolve the vintage comparison graph of Fig. 4 in accordance with Tables I, II and III
is presented in Fig.

As described above, t:~e Crus Classes method 34 delivers a comparison of two loan vintages either in t,'~e fora of a graph or tabulated data which permits one to get a sense of which vintages are performing better.
This information can then be used in making manual or automatic, computer generated yes/no decisions whether ~o originate, purchase, or tc maintain and sell various vintages of loan products or ser-ric'_:~g r fights as needec at tae decisional blocks ?3, 21 anc 32 of Fig. 1.
The basic premise of the C~us Classes analysis is that the future per °ormance cf t:~ese loan vintages will match the past pattern. This may not necessarily be true. To this end, t:~e earl:r warring system (EWS) 32 of the present invention furt.'~er enhances the loan analysis process by incorporating an application of behavioral scoring that has been speci=icallv designed to be used on closed end loans with longer maturities such as mortgages. The EWS 32 is able to statistically predict the probability that a group of loans will experience ZO credit performance problems during a future preselected time period, without waiting for that loan to season. In the case of mortgages, the time to season is typically three to seven years. The EW5 34 is intended to provide management with automated analytical tools which allow making decisions well in advance of the aforementioned t'zree to seven "seasoning" period.
By utilizing the EWS, a mortgage originator can perform portfolio analysis and ascertain which product type, program, type of under,~riting, proper ty trpe, type of customer, origination channel, etc. is at risk, without waiting for the mor~gages to ac~sall~.r matsre and entar default . '~he cr.~.y cons tr ain-. vs the amcunL cf dazes att ;ibutes that the mcr gage loan originator keeps on anv customer over time, which for the purcoses of the presant invention may be tyro years. The mer~gage originator can then dynamically adjust the flow of origination by altering any credit criteria derived from a particular attribute.
The EWS 34 constitutes the dynamic component of the underwriting concept of t:~e present invention. With this concept, the decision maker can estimate improvements in credit duality for each specific type or amount of change in a criteria, i.e. he or she can calculate the marginal contribution ef any attribute on record.
More specifically, the for~~rard looking feature of the EWS component of the invention attempts to forecast the likelihood the borrowers will enter a 90~
days past due delinquency on their mortgages. This condition -- the oc~~.irience of a 90-day past due delinquency -- is defined as a "bad" condition relative to any loan. The r.WS calculates the probabilities of bad conditions occurring by combining loan information with the credit bureau's current behavioral scorn for the given borrower. Iz other words, the ~wS combines the borr owes ' s cure ent :nor tgage s tat:~s ( def au 1 t s tatus and age ) wi th a forecas t ~:~at is base on the bore owes' s pertormanca on other obliga'ions and uses this information to forecast the bad condition. The EWS
systam makes three moor assumptions:
~ The future per=ormance pattern of defaults will be the same as in the past;
~ The future per=ormance depends upon the current loan charac~sristics and is dependent on past performance only through the credit bureau scores. Therefore, the EWS also carries all the assumption of the credit bureau's score that was used; and ~ The EWS employs a logistic regression mode?
to acc~,irately and sufficiently predict default behavior.
The aforementioned "bad" condition is a discrete (yes or no) event that occurs when and if an individual loan is at least once three payments past due at any point during a for~rard looking preset time per;od, for example, too years. Bad loans are assigned the value "1" and good loans the value "0". Hcw many times the loan enters "bad" is not considered in the EWS. ~ good loan is never t'~ree payments pas t. due over the aforementioned tac year time ~yame and therefor=_ is assigned a loan of a value "0" . Pr=f er abl y, ~ ~ crder tc provide reliable infcr_nation usinc ~:~e WvS svs'em, the underlying por=folio s:~~ouid have at leas i00, 000 loans IO and the loans should consis~ of di===_rent dist.ribuzions of ages, types, locations, et~. T_n an embodiment of the invention which has been reduced to practice the number of loans in the portfolio exceeded one million.
The EWS probability of loans entering the bad IS condition is developed cr calculates on the basis of looking bac7cward in time through a development period which may similarly constitute a tae-year time period.
The EWS formula considers the age of the loan at the beginning of development period; the credit bureau scare 20 for the loan at the beginning of the development period;
the delinquency status at the beginning of the development period; and the type of product, for example, whether a government or conventional or adjustable rate mortgage, etc.

The following logistic model has been applied to the underlying portfolio (government and conventional loans being considered separately). In the formula shown below, P is the probanility of a loan becoming bad at any time in the coming t-ao years Log(P/(1-i'))=A-fB,*.~GE)=8,*CD-83*D, -84*D.-3.;SCORE=B5*!~IO SCORE;.
In the equa;.ion, AGE is defined ~n categories of cruarters from 1-40. Therefore B, anc AGE ar_ 40 dimension row and column vectors respec~ively. SCORE is the mortgage score _-. cm a cr edi ~. nur'ae ra ring company such as the well known yquifax rating bureau, at the beginning of the two-,rear time period, i. e. Augus~. 1994.
Note, if no such score is available, the invention assigns the lowest possible value, namely 200. The Equif ax scoring scores vary from 200 to 1000. The dummy variables in the above equation are defined as follows:
1 if the loan is current ai the beginning of the tImC period;
CO=
0 otherwise.
1 if the loan is 1 month past due at the beginning of the time period;
2 0 D 1 - ~ 0 otherwise.
1 if the loan is Z months past due at the begittninQ of the time period;
D2=
0 otherwise.
NO 1 if the loan has no credit score available ai the beginning of the time period:
SCORE =
0 otherwise.

The coefficients A, B~, B,, B3, B1, BS and B6 are estimated by running the model over the underlying portfolio.
All of the forecasting is done at the individual loan level and then the results are aggregated into the portfolio of interast by defining or grouping certain loan characteristics (location, rate type, maturity, LT'7, etc.) to mace comparisons. Even though the invention presents the mean probability for a predicted group, the informat;cn conta_ned in the individual loan level data is preserved because the invention explicitly considers the dispersion around the mean of the bad rate over time.
To forecast the probability of an individual loan entering 90- day default during the predeter:~ined time period (i.e. the t~ao year time game), all one need do is insert the estimated coefficients and the characteristics of that individual loan into the logistic equation for "P" presented earlier.
The logistic model in the form of the aforementioned equation provides numerical results which are suitable of being graphically presented as shown in Fig. 5. The graph enables management to readily interpret and form decisions based on the predictions which it contains. In accordance with a further embodiment of the invention, the results are feedbac3t by the computer which then provides yes/no decisions based on predetermined default risk or profit criteria set by the operator.
Fig . 5 is a snapshot taken in 19 9 6 and den is ~s loan experience looking both backwards to the past two years and for-~ards over a similar two _rear time span.
The vertical solid bars 72 r=presen' the current mean (expose) bad rates nor a Dar~'cu;ar group over the past two years. In other wor3s, these bars show the mean bac rate percentages of a group of loans that originated i:~ a particular year. For example, the bar 72 for the loans originated in 1989 shows a bad rate or about 12.50. The score for the 1991 loans is just about 8~ whereas the bad rate for 1996 is rnaite low (under 50), reflecting the fact that this vintage of lcan has not yet matured sufficiently.
The hatched vertical bars 74_represent the forecasted mean bad rates (exante) for the same group of loans over the next two years. The value for the 1996 vintage is somewhere around 7o indicating an expected delinquency rate of 7% even though the past two-year performance had an actual bad rate of only about 2.50.
The curve 76 represents the expected bad rate curve that is obtained by modifying the forecasted bad rates by the risk ratio on nearby vintages, and this shall be explained more fully later on.
One should not places much emphasis on whether the curve 76 is above the solid bars 72, since this may merely reflect the normal pat~ern of seasoning fcr mortgages. Nor should one place much emphasis on she absolute height of each bar, since this may r=fl ec~
diff er ent expectations amone the varicus groups cr t-rt~es of loans that ar = being analyzed . Ins read , t:~e gr aph Cf IO Fig. 5 indicates three impcrtant benchmarks for revieTaing and forecasting the risk in the given loan portfolio.
First, note the jump which is indicated by the arrcw 84.
It represents a jump which occsrs when the difference between the hatched bars 7s and the solid bars 72 is I5 greater than one standard deviation above the his~crical age weighed performance for that vintage. The bigger the jump, the more serious the r~uality problem. This measure is particularly useful on younger vintages, e.g. the 1996 vintage to which the arrow 84 is pointed.
20 Second, the size of the portion of the hatched bars 74 which protrudes above the expected bad curve 76 indicates an unusual level of risk in the past or the future for that group of loans. Note the arrows 8o and 82 which indicates such conditions. Finally, the arrow 25 78 indicates a turning point which represents the point at which the first derivative of the expected BAD rate curve changes sign from positive to negative: i.e. the first time the bad rate drops as the loans age increases.
The younger the age at which the turning point occurs, the earlier the por tf olio' s cr ed i t per f ornance wi l l or has matured.
The manner In whic~. t:~e curie 75 of Fig.
derived may be bet=r under s";cd by r eviewing F i gs . o anc More specifical'_y, the curve 36 .s developed by IO taking a snapshot at a point in time looking at loans of different quarterly aces and asking haw many in each age group entered the 'bad" stage during the preceding predetermined time period, e.g. two years.
Initially, t_'~e ~dS ~4 develops for each 1.~ empir ical two-year per iod of per f ormance the bad rate curve as a function of age in 4uarters. See the quarterly bad rate curve 36 in rFig. 6.-Next, using a moving average, the invention smooths each curve to reduce the randomness of the Zd quarterly performance, thus obtaining the smooth curve 88 in Fig. 6. In fad, the slopes of the yearly bad rate curve are calculated as the percentage change in bad rates from one year to another year, which is known as the risk ratio. To improve the integrity of the results 25 and protect against possible statistical aberrations, at least eight two-year time periods are considered, with the means and standard deviations of the risk ratios calculated.
To find the point on line 76 for 1991, the invention uses t_'~e point on line 76 fer 1990 which has the expected bad rate for 1990. Mu~.tio?;ring this bad rate by the corresponding mean Yisk ra~'_o ~=rom Fig. 6 obtains the expected bad rape fcr X99= and its s~andard deviation. This expected 'cad rage is used for the ooin~.
on line 76 far 1991. However, __ p:~e =precast bad rate (bar 74) is within the expected bad ra'=, plus or minus one standard deviation, the invention juste substitutes the expected bad rate by the forecast. bad rate so that the line 76 passes through the top of bar 74.
The third in the triad of dynamic underwriting tools of the present inventions, the matrix link system 36, uses the information developed by the Crus Classes technique 34 and early warning system 32 to develop a probability based prediction of how many of a given set of loans will be "bad" at a selected future date.
More specifically, while the Crus Classes method 34 analyzes the past performance of loan vintages and the Ews system places a probability on a group of loans entering the bad state within a preset time frame, i.e. a window, the matrix link system 36 is designed to predict the default status per~ormance of a group of loans at a preset point in time within the window of operation of the EWS. For example, the EWS method is able to say with respect to a group of loans that 3v of those loans will enter a bad states (90+ days in arrears) at one time or another during a too year window. In comparison, the matrix link is designed to answer the cues tion how many loans will be non-accruing, i.e. iz the 90-- payment overdue state, at the end of the fist l0 quarter of the too year w=ndow cr nine months into the window and so forth, taking into consideraticn that some loans may enter t:7e bad state or exit therefrom due to prepayment or on account of having been matured, or by completing foreclosure.
To perform its functions, the matrix link system 36 predicts how many borrowers that enter a 9d-days delinquency state remain there, how many loans will return to "good" status and, finally, in which quarter over the predefined window, e.g. two years, will these transitions occur. Note that a particular loan can enter a bad state, return to a goad state or remain in a bad state. Loans may mature, gay off or complete foreclosure during the predefined window period and thus exit the Ioan portfolio being analyzed. In order to provide a quantitative measure of the transitions of loans between dif'erent states, the present inventors have developed a so-called delinauency transition matrix in a form which, in one embodiment thereof, appears as in the table in Fig. 7.
In the table of Fig. 7, use is made of a historical file spanning four years and including v_nzace years 1993-1996. The table shows the delinctuencv performance of a par~icul ar group of 1 pans . Further , t:~e table stratifies the loans by age of origination. Vote l0 that the table lisps separatel y the r esul is for t. r~~
diff erent types of loans , namely conf orning loans , ~ umbc loans and government loans. In each case, it shows the probability of a loan transitioning (a) from a bad sate to a bad state, (b) from a bad state to a good states; (c) exiting, i.e. maturing and therefore being dropped f=om tine sample of loans being considered, (d) from good to bad, (e) from good to good; and (f) from good to exit state.
Using the percentages listed in the delinquency ZO transition matrix in the Table of Fig. 7, one can then begin to canvert the information obtained through the EWs system 32 into the forecasts of how the Crus Classes vintages will perform at predefined time periods within the window. The table in Fig. 8 illustrates the formulae ZS evolved by the present inventors which are as follows.

First, the current age of the loan at the time or forecast is determined. In the table of Fig. 7, the age of the loans is indicated in years (1, 2, 3, 4...
10). However, in an actual implementation of the invention, the vintages have been stated and the formula is calculated in terns of quarters to obtain increased accuracy.
For each group of loans of a par~icula= ace, the invention uses a 3-mon~1 ~rans~tion matrix to l0 forecast three months for-~ra=::, a 5-month '--ansition matrix to forecast six months for-ward, a 9-mcnth transition matrix to forecas;. nine months forward and a 12-month transition matrix to forecast twelve months forward.
Based on the choice of data in the previous step, the invention calculates respectively looking forward three, six, nine and twelve months:
1. how many good loans and bad loans will exist from the portfolio;
2. how many good loans will turn into bad; and 3. haw many bad loans will remain bad.
From the above data, one obtains the classic "roll-rate" forecast which provides the (first component of the forecast. The above approach merely projects for.~ard the results that have already cccurred in the past, on the expectation that they wily repeat themselves. However, a greater benefit of the matrix link technique of the present invention comes from adding the additional information that is ccntained in and/or obtained by the early warning system 32.
To this end, the invention:
(a) Ca'cula~es an emoir-cal ratio obtained as -- the c~.imulative number c~ loans which are IO 90+ at each quarter (yCP) and divides __ by the number c_T.
loans that are 90+ at leas once durinc these four attar tars .
(b) From the EWS, the invention obtains or forecasts the "bad" rate for the t:~c-year window based on the EWS method 32.~
( c ) Us ing the ESvS , the invention forecasts the bad rate and the empirical ratio above as a new piece of information to adjust the classic "roll-rate" forecast. This is in essence what comprises the "matrix link" method 36. Fig. 9 provides the results of the matrix link in graphical form. In the example shown, the performance of different vintage loans is predicted one year forward in quarterly installments starting in 1997.

Far example, the plot 1.02 shows the two-year performance of 1995 vintage loans which range in age (months) from 0 to 24 months, as viewed looking backwards in time in 1997. The curve 104 shows the delinquency rate percentages predicted ror the next twelve months.
For example, when the groups of loans attain an age of 27, the delinquency raze can be read on the ordinate axis . The same is t~se f or t:~=s gr:.up of loans when then reach an age of 3 0 :nenths , .., monr.ia and 3 6 months . The reason that the curve 104 has a predicted value below the actual value is that the prediction in the matrix link uses a moving sum average ;which weighs down the actual sharp up-turn in the "bad" rate which has actually occurred toward the end of the curve 102.
The above remarks are also applicable to the curves 106, 108 which apply to the 1993 vintage, the c~,irves 110 and 112 which are applicable to 1994 vintage, and to the curves I14 and 116 which apply to the 1996 vintage.
ZO Note that the Crus Classes are less static than traditional mortgage vintage analysis. Therefore, the performance of the last three points of any vintage can still change somewhat, for better or worse.
The matrix link lines, i.e. the curve 104, 108, 112, and 116 also show where the inventors expect the last t'zree Crus Classes points to adjust over the next nine months.
The system of the present invention lands itself easily to being implemented through use of a general purpose programmable computer as i~lustrated in Fig. 10. Thus, the general purpose computer 124 communicates with a local database I22 which receives a wealth of statistical and specific information abcut various loans from diverse sources. for example, the scarce of the infcrnation may be a nationa'_ loan database 120 which is maintained by certain industry groups. The general purpose computer 124 has the usual complement of peripherals including an operator's console 126, Roil 120, RAM 130 and a hard disk 132.
The computer 124 operates under control of major soft~~aare blocks which perform the dynamic analysis 134 in a manner already described. The main software components are the software routines 136 which handle the development and analysis of the Crus Classes associated with the creation of the loan vintages. The early warning system block 138 calculates probabilities of loans going bad within a predetermined for~rard looking window. Finally, the matrix link saft;aare block 140 forecasts the probabilities that a fraction of loans will go bad within the window at a particular time.

The analytical results developed by these soft-aare subroutines or blocks 136, 138 and 140 are tabulated in the tabulation software block 150 and outputted through an output software block 152. The output can be in the form of a signal which drives a printer which generates a graphical representation of the results in the manner previcusly described.
alternatively, the output may supp~y the results to to console 126 for visua_ inspection by the operator.
Alternatively, the operator may program the computer _~*
via the console 126 to provide yes/no answers as to whether an investment should be made cr continued to be made in a particular loan portfolio, again as already described.
Although the present invention has been described in relation to particular embodiments theree~, many other variations and modifications and other uses will became apparent to those skilled in the art. It is preferred, therefore, that the present invention be limited not by the specific disclosure herein, but only by the appended claims.

Claims (40)

WHAT IS CLAIMED IS:

1. A process for analyzing and selecting loan portfolios, wherein each loan portfolio comprises a plurality of loan units, the process including the steps of:
separating the loan portfolios into a plurality of loan vintages in a manner such that the loans included in each loan vintage have origination dates that are on average of the same age and said origination dates of each loan portfolio are all within a first time interval and the respective time periods of the different loan vintages being mutually exclusive of one another;
counting a bad rate of the loans in each loan vintage by counting the loans in each loan vintage on which payments are in arrears for a time period greater than a second time interval during a time window having a third time interval, wherein the third time interval is substantially greater than said second time interval; and providing a comparison of the bad rates of the different loan vintages on a visually perceivable output medium, to allow comparing the bad rates of the different loan vintages.

2. The process of claim 1, in which the third time interval is about two years and the third time interval commences about two years prior to the date of calculation of the bad rates.

3. The process of claim 2, in which the first time interval is about three months.

4 . The process of claim 2, in which the first time interval is about one year.

5. The process of claim 4, in which the one year first time interval coincides with a calendar year.

6. The process of claim 1, in which the second time period is about ninety days.

7. The process of claim 1, in which each loan vintage comprises at least 50,000 loan units.

8. The process of claim 1, further including:
allowing an operator to set the durations of one or more of the first, second and third time intervals;

deploying a general purpose computer for enabling the computer to automatically provide "yes" or "no" comparisons as to whether a given financial institution should choose to invest in a particularly identified loan vintage.

9. The process of claim 1, including:
developing a projection of bad rates for a plurality of said loan vintages by calculating probablistic bad rates for said plurality of loan vintages during a forward looking window extending over a fourth time interval, the projected bad rates comprising an early warning system whose results are depictable on the output medium.

10. The process of claim 9, in which the step of calculating the probablistic bad rates for said loan vintages is carried out by using a logistic regression formula.

11. The process of claim 10, in which the logistic regression formula is Log(P/(1-P))=A+(B1*AGE)+B2*CO+B3*D1+B4*D2+B3*SCORE+B6*NO SCORE) wherein, P is the predicted bad rate, AGE is defined as the age of the loan vintage and B1 and AGE are row and column vectors of a dimension matrix of a predetermined size, SCORE is a mortgage score from a credit bureau and CO, D1, D2 and NO SCORE are dummy variables, and the coefficients A, B1, B2, B3, B4, B5 and B6 are estimated by running a model of the equation over a preselected loan portfolio.

12. The process of claim 11, wherein the value CO is assigned a value of one if the loan is current at the beginning of the third time interval and zero otherwise, the variable D1 equals one if the loan is one month past due at the beginning of third time interval and is zero otherwise, D2 is assigned a value of one if the loan is two months past due at the beginning of the third time interval and zero otherwise, and NO SCORE
equals one if the loan has no credit score available at the beginning of the third time interval and zero otherwise.

13. The process of claim 10, wherein the fourth time interval is of the same duration as the third time interval.

14. The process of claim 1, wherein the loan units are closed loan units.

15. The process of claim 14, wherein the loans are mortgage loans.

16. The process of claim 1, including separating all loan units into different groups based on type and thereafter carrying out said separation of said loan portfolios into separate loan vintages based on each type of loan.

17. The process or claim 16, wherein the types include conventional loans, jumbo loans and government originated loans.

18. The process of claim 9, including separating all loan units into different groups based on type and thereafter carrying out said separation of said loan portfolios into separate loan vintages based on each type of loan.

19. The process of claim 13, wherein the types include conventional loans, jumbo loans and government originated loans.

20. The process of claim 1, in which the output medium is a graphical chart.

21. The process of claim 9, wherein the output medium is a graphical chart.

22. The process of claim 20, which includes creating the graphical char by plotting a difference between the bad rates calculated for a preselected pair of loan vintages and including on the graphical chart an area of uncertainty.

23. The process of claim 22, wherein the area of uncertainty is selected as a +1 and -1 standard deviation of the difference in the bad rates for the preselected pair of loan vintages.

24. The process of claim 21, including plotting on the graphical chart the probablistic bad rates in the form of a first curve.

25. The process of claim 9, including developing a matrix link by calculating a bad rate of a preselected loan vintage as of a definite time within the fourth time interval.

26. The process of claim 25, including developing delinquency transition figures for selected ones of said loan vintages.

27. The process of claim 26, including developing the delinquency transition figures by counting the numbers of loans which have transitioned from (a) good to bad state; (b) bad to good state; and (c) a status as a loan.

28. The process of claim 27, further including counting the number of loans which have remained in a bad state and counting loans which have remained in a good state.

29. The process of claim 25, including calculating for at least one of the loan vintages matrix link results comprising a ratio of a predicted number of bad loans divided by a predicted number of new loans.

30. The process of claim 29, including plotting the ratio on the output medium.

31. The process of claim 9, including automatically providing yes/no decisions whether to invest in selected loan vintages using the early warning system.

32. The process of claim 25, including automatically providing yes/no decisions using the matrix link system.

33. A process for analyzing and selected loan portfolios, wherein each loan portfolio comprises a plurality of loan units, the process including the steps of:
separating the loan portfolios into a plurality of loan vintages;
developing a projection of bad rates for the plurality of said loan vintages by calculating predicted bad rates for said plurality of loan vintages during a forward looking window extending over a predetermined first time interval, the predicted bad rates comprising an early warning system the results of which are depictable on a visually perceivable output medium, a loan unit being included in the bad rates when payments on the loan unit are in arrears for a time period greater than a second time interval occurring during the forward looking window, the first time interval being substantially longer than the second time interval.

34. The process of claim 33, in which the step of calculating the predicted bad rates for said loan vintages is carried out by using a logistic regression formula.

35. The process of claim 34, in which the logistic regression formula is Log(P/(1-P))=A+(B1*AGE)-B2*CO-B3*D1-B4*D2-B6-SCORE+B6*NO SCORE;
wherein P is the predicted bad rate, AGE is defines as the age of the loan vintage and B1 and AGE are a row and column vectors of a dimension matrix of a predetermined size, SCORE is a mortgage score from a credit bureau and CO, D1, D2 and NO SCORE are dummy variables, and the coefficients A, B1, B2, B3, B4, B5 and B6 are estimated by running a model of the equation over a preselected loan portfolio.

36. The process of claim 35, wherein the value CO is assigned a value of one if the loan is current at the beginning of the first time interval and zero otherwise, the variable D1 equals one if the loan is one month past due at the beginning of first time interval and is zero otherwise, D2 is assigned a value of one if the loan is two months past due at the beginning of the first time interval and zero otherwise, and NO SCORE

equals one if the loan has no credit score available at the beginning of the first time interval and zero otherwise.

37. The process of claim 33, including depicting graphically in the form of a first curve an expected bad rate curve.

38. The process or claim 37, including producing a bar chart showing current mean bad rates and forecasted mean bad rates and superimposing the first curve over the bar chart.

39. The process of claim 33, including creating the first curve by creating a quarterly bad rate curve, smoothing the quarterly bad rate curve by averaging the values thereof with one another and further smoothing the curve by taking a risk ratio thereof.

40. The process of claim 39, including creating markers on the curves including markers which show the changes in the first curve at a positive to a negative slope transition thereof and markers which show jump paints of a predetermined size.