Docstoc

Auto Loan Credit Model_1_

Document Sample
Auto Loan Credit Model_1_ Powered By Docstoc
					                                  Auto Loan Credit Model
                    Jason Riggs, Samit Shrivastava, Yelena Veretennikova
Introduction

Auto loans are among the most common loans as the ever-expanding automobile market attracts
millions of new customers. Banks and loan providers want to tap into this large pool of
customers, but worry that the risk of these customers may be too high. This brings these
financial institutions to modelers who can build credit models to identify the Good and Bad
among the prospective customers and provide them with a solid analytical and logical base to
approve loans. The credit model built for ABC Savings and Loan will predict the risk of default
based on the customer data provided and generated when savings and loan customers apply for
an auto loan at the bank.

Customers submit their application to banks and credit providers for the given amount of loan.
As part of application process, banks want to have some information such as personal
information, job profile and credit history etc. about their prospective customers. This
information is used to decide the suitable candidate who could repay the loan granted to him/her.

In the current model, we made no specific assumptions such as a poor financial market, currency
fluctuations across the globe, etc. We believe that ABC Savings and Loan will provide loans to
every customer who fulfills their basic criteria. However, in special circumstances, some riders
and supplementary clauses will come into effect in decision-making process. This report is an
effort to provide an overview of the model we built using existing data to predict a Good
customer for our client ABC Savings and Loan. We define a Good customer as one who would
not default on an auto loan.                           S&V BOOK RETAIL VALUE
                                        BKRETL          Frequency Percent Cumulative Cumulative
Data                                                                       Frequency  Percent
                                       $1 TO $2400             153    2.19        153      2.19
The data set used to build this      $2401 TO $3000            218    3.11        371        5.3
                                     $3001 TO $3400            223    3.18        594      8.48
credit scoring model consists
                                     $3401 TO $3800            258    3.68        852     12.17
of actual customer data              $3801 TO $4000            153    2.19       1005     14.35
provided by the client. A total      $4001 TO $4500            472    6.74       1477     21.09
of 56 variables were collected       $4501 TO $5000            539     7.7       2016     28.79
for each customer for a total        $5001 TO $5500            579    8.27       2595     37.06
                                     $5501 TO $6000            650    9.28       3245     46.34
of 14,042 applicant records.
                                     $6001 TO $6500            543    7.75       3788      54.1
The data set covers an 18            $6501 TO $7000            554    7.91       4342     62.01
months period between                $7001 TO $7500            455     6.5       4797     68.51
January 1998 and June 1999.          $7001 TO $8000            460    6.57       5257     75.08
This data comprises personal         $8001 TO $9000            682    9.74       5939     84.82
                                    $9001 TO $10,000           385     5.5       6324     90.32
information and other data
                                   $10,001 TO $11,000          250    3.57       6574     93.89
such as banking history,           $11,001 TO $12,000          178    2.54       6752     96.43
current and past credit               OVER $12,000             250    3.57       7002       100
                                                                                      Page 1 of 14 
 
history, credit defaults, etc. An example of one of these variables is shown in the table to the
right. The table shows the frequency distribution of the book retail value of the vehicle for
which the loan is sought. This variable and the others like it are independent variables that help
to identify the Good customers. Good customers are the ones who have not defaulted on a loan
during the 18 months outcome period stated above and this is the dependent variable. To check
the accuracy of the current model, we have split the data in to two smaller datasets called
“Training” (7,013 records) and “Validation” (7,029 records) randomly and each of them consists
of almost 50% records. The training dataset was then used to build the model, and the validation
dataset to score the model.

Method

A diagram of the method used to build the model and evaluate its effectiveness can be seen in
Figure 1 of the Appendix. A brief description of each step of the method used is outlined below.

    •   In order to build our model, we needed to create dummy variables. In order to create
        accurate dummy variables, we wanted to divide each variable into 2%-10% frequency
        bins. We accomplished this using the Proc Freq procedure in SAS.

    •   Once each variable conformed to the 2%-10% rule, we created crosstabs with each of the
        independent variables versus the binary dependent variable GOOD. This was done
        within SAS using a variant of the Proc Freq procedure and then outputting the table as an
        HTML file. An example of one of these crosstab tables can be seen in the Appendix.

    •   We then analyzed the crosstabs and created dummy variables for each of the data
        groupings within each of the independent variables. The output of this process can be
        seen in Table 1of the attached Appendix. The most efficient way to carry out this dummy
        creation is to print the crosstab. The ratio of good to bad is then evaluated and similar
        groups that are within 0.1% of each other are generally grouped together. Larger changes
        may be grouped together as well if this makes sense based on what you know about each
        of the variables. Once the dummy variables have been determined by analysis of the
        crosstabs, SAS is then easily used to create the dummy variables within the dataset. The
        outcome of the dummy creation process is later repeated using the validation dataset.

    •   Regression analysis is done with all the dummies. After the first analysis, the worst
        dummies with highest p-values greater than 0.9 are removed. More such dummies with
        higher p-values are removed. When dummies look significant, fine-tuning is done by
        adding some dummies back in regression analysis. In the current analysis, thirteen
        iterations were done to finalize the model for K-S testing. Please see Table 2 in the
        Appendix for the specific iteration cut-offs.
           



                                                                                       Page 2 of 14 
 
    •               After finalizing the training dataset, K-S test is run first on training dataset, then on the
                    validation dataset, and then on both datasets combined. K-S value of validation dataset
                    should be within 10% of that of training dataset. After getting the desired K-S test values
                    for both datasets, a scorecard is prepared. This scorecard attempts to provide a
                    quantitative measurement of the likelihood that a customer will default on an auto loan
                    given a certain characteristic of the customer (i.e. the model year of the car the customer
                    purchased, the age of the customer, the down payment, etc.). Please see Tables 3-5 and
                    Figures 2-4 in the Appendix for the K-S test results. Please see Table 6 for our
                    Scorecard.

Results

The K-S test is based on the maximum difference between the Cumulative Good and Cumulative
Bad. The difference between the Good and Bad for training and validation datasets should not
be more than 10 % where K-S test value for training dataset must be greater than K-S test value
for validation dataset. In our model, after running regression and fine-tuning the variables, we
ran the K-S test and observed the values after regression 9. In the final model, the K-S test value
for training and validation datasets are 0.2545 (25.45 %) and 0.2333 (23.33 %) respectively
giving difference of 8.33 %.


                              Difference between the K-S Test Results for Validation and
                                                  Training Datasets


                         14.00%   12.84%                                  12.60%
                                                             11.93%
                         12.00%                10.95%                                   10.38%
        Percent Change




                         10.00%                                                                       8.33%
                          8.00%
                          6.00%
                          4.00%
                          2.00%
                          0.00%
                                     9           10            11           12            13            14
                                                        Regression No


Figure 1: Difference between the K-S Test Results for Validation and Training Datasets

Based on the results of the K-S Test, we recommend a cutoff score of 650-699. At this cut-off, in
the training dataset, 64.12 % of Good applicants scored above 650, however, 38.66 % of Bad
applicants also scored above the benchmark value of 650. The acceptance rate of our model with
training data and all data is 55.7 % and 57.16 % respectively. However, it also produces a bad
rate of 22.96 % and 23.63 % for these datasets. However, these Bad Rates are far better than the
current Bad Rate of 32 %, when no model is used to judge a Good or Bad person before giving
                                                                                                      Page 3 of 14 
 
the loan, reflecting robustness of our model. These results provide bank/credit provider a fair
basis to accept or reject someone’s loan application.

To monitor the effectiveness of our model, we have included a monitoring report in Table 7 of
the Appendix. We would like for ABC Savings and Loan to fill in the empty cells in the
monitoring report three months after ABC begins to use our model.

Conclusion

We have successfully produced a model that acceptably predicts the default risk of auto loan
applicants utilizing commonly available data. The model was built with historical applicant data
and then validated by KS testing. The current model accepts over half of loan applicants but
among those accepted nearly a quarter will default. The model should now be analyzed in
conjunction with a profitability model to make sure it enhances returns. The model should also
be verified periodically with new data in light of changing customer behavior and industry
practices. Future modeling refinements remain possible and the goal would be to further
increase the acceptance rate and decrease the bad rate of funded loans.




                                                                                       Page 4 of 14 
 
Appendix

Figure 1: Method Diagram

                                    Obs: 14,042                                                                      Obs: 14,042
                                    Var: 37                                                                          Var: 20
                                                         Data1                                        Xtra1




                                                                       Editor_Assignment2.sas


                                                                                                   Obs: 14,042
                                                                                                   Var: 55
                                                                                   Three




                                                                         Training_validation.sas


                                                                                                                     Obs: 7,029
                                    Obs: 7,013                                                                       Var: 56
                                    Var: 56                Train                                      Valid



                                                               Pcrosstabhtml.sas                      DummyCr8_rev.sas


                              Pfreq.sas                   Prepare crosstabs in Excel to
                                                            display Good/Bad Ratio

                                                                                                                                   Obs: 7,029
                            Pformat_rev.sas               Determine neutral groups and                                             Var: 237
                                                                 dummy splits                                 Validx
    Confirm variable fits
     the 2%-10% rule
                            Papplyfmt_rev.sas                  DummyCr8_rev.sas
                                                                                                     Regression 9- 14.sas


                                                                DummyContent.sas
                                                                Verifies that dummies                  Scoredummyp.sas
                                                                were created.

                                                                                                                                   Obs: 7,029
                                                  Obs: 7,013                                                                       Var: 238
                                                                                                                 Scrvalx
                                                  Var: 237               Trainx



                                                                                                         Modifyscrdata.sas
                                                                 Regression 1 - 14.sas


                                                                                                       Bgscorexgoodhtml.sas
                                                                   Scoredummyp.sas


                                                  Obs: 7,013
                                                  Var: 238                                              reg24_10_goodvalx -
                                                                        Scrtranx                        reg24_14_goodvalx



                                                                   Modifyscrdata.sas



                                                                 Bgscorexgoodhtml.sas



                                                                   reg24_1_goodvalx -
                                                                   reg24_14_goodvalx                                 K-S Test


                                                                                                                                                   
                                                                                                                                      Page 5 of 14 
 
Table 1: Crosstab BKRETL x GOOD

A Crosstab of BKRETL by GOOD is shown in the figure below. An illustration of the dummy
creation process has also been included for this variable. As shown below we selected a our
neutral variable to coincide with the grouping of data where the ratio of good to bad is roughly
equal.

           Frequency
            Col Pct
                      Table of BKRETL by GOOD
    BKRETL(S&V BOOK RETAIL      GOOD(PERF. (NOT CHARGED OFF))          Total
            VALUE)                   0               1                         Good/Bad Dummy
                                                                                 Ratio  Variable
          $1 TO $2400                             84                 69 153
                                                 3.7              1.46           0.395    BKRETL1
         $2401 TO $3000                           95               123 218
                                               4.18                 2.6          0.622
         $3001 TO $3400                         100                123 223
                                                 4.4                2.6          0.591
         $3401 TO $3800                         114                144 258
                                               5.02               3.04           0.606    BKRETL2
         $3801 TO $4000                           59                 94 153
                                                 2.6              1.99           0.765
         $4001 TO $4500                         189                283 472
                                               8.32               5.98           0.719
         $4501 TO $5000                         209                330 539
                                                 9.2              6.98           0.759    BKRETL3
         $5001 TO $5500                         196                383 579
                                               8.63                 8.1          0.939
         $5501 TO $6000                         238                412 650
                                              10.48               8.71           0.831    BKRETL4
         $6001 TO $6500                         164                379 543
                                               7.22               8.01           1.109
         $6501 TO $7000                         173                381 554
                                               7.61               8.05           1.058
         $7001 TO $7500                         130                325 455
                                               5.72               6.87           1.201    Neutral
         $7501 TO $8000                         111                349 460
                                               4.89               7.38           1.509
         $8001 TO $9000                         186                496 682
                                               8.19              10.49           1.281
        $9001 TO $10,000                          89               296 385
                                               3.92               6.26           1.597    BKRETL5
       $10,001 TO $11,000                         44               206 250
                                               1.94               4.36           2.247    BKRETL6
       $11,001 TO $12,000                         36               142 178
                                               1.58                   3          1.899
          OVER $12,000                            55               195 250
                                               2.42               4.12           1.702    BKRETL7
              Total                            2272               4730 7002                           

 

 
                                                                                         Page 6 of 14 
 
Table 2: Regression Analysis using Dummy Variables

    Regression   Variables   Variables
       No.       Removed       Used       Cut-Off >     R2
        1            0         182            -       0.1125
        2           13         169          0.90      0.1123
        3           11         158          0.80      0.1123
        4           23         135          0.60      0.1117
        5           18         117          0.45      0.1106
        6           50          67          0.50      0.0895
        7           12          55          0.50      0.0892
        8           11          44          0.20      0.0866
        9            6          38          0.20      0.0859
                                          BKRETL1
       10            2          36       VDDASAV2     0.0854
       11            1          35        HST03X1     0.0833
       12            0          35         T2924A     0.0851
       13            0          35        HST79X2     0.0851
       14            1          34        BKTIME2     0.0842




                                                               Page 7 of 14 
 
Table 3: K-S Test result for training dataset

    BGSCORE                               Bad                 Good          Cumulative Cumulative    Bad       Good                                            Difference
                                                                              Bad        Good     Percentage Percentage
    1000 or More                                      5               83             5         83      0.0022     0.0177                                                0.0155
     950 to 999                                       5               90            10        173      0.0043     0.0369                                                0.0326
     900 to 949                                      19              146            29        319      0.0125     0.0680                                                0.0555
     850 to 899                                      36              262            65        581      0.0280     0.1238                                                0.0958
     800 to 849                                     103              417           168        998      0.0724     0.2127                                                0.1402
     750 to 799                                     152              563           320       1561      0.1379     0.3326                                                0.1947
     700 to 749                                     234              657           554       2218      0.2388     0.4726                                                0.2338
     650 to 699                                     343              791           897       3009      0.3866     0.6412                                                0.2545
     600 to 649                                     382              605          1279       3614      0.5513     0.7701                                                0.2188
     550 to 599                                     339              489          1618       4103      0.6974     0.8743                                                0.1769
     500 to 549                                     288              314          1906       4417      0.8216     0.9412                                                0.1196
     450 to 499                                     190              165          2096       4582      0.9034     0.9763                                                0.0729
     400 to 449                                     134               62          2230       4644      0.9612     0.9896                                                0.0284
     350 to 399                                      51               29          2281       4673      0.9832     0.9957                                                0.0125
     300 to 349                                      29               18          2310       4691      0.9957     0.9996                                                0.0039
     250 to 299                                       8                2          2318       4693      0.9991     1.0000                                                0.0009
     200 to 249                                       2                0          2320       4693      1.0000     1.0000                                                0.0000
     150 to 199                                       0                0          2320       4693      1.0000     1.0000                                                0.0000
     100 to 149                                       0                0          2320       4693      1.0000     1.0000                                                0.0000
      50 to 99                                        0                0          2320       4693      1.0000     1.0000                                                0.0000
       0 to 49                                        0                0          2320       4693      1.0000     1.0000                                                0.0000

 

Figure 2: K-S Test graph for training dataset

                  K-S Test for the Training Data                                                    Bad Percentage             Good Percentage

                    1.0000




                    0.8000




                    0.6000
     Difference




                    0.4000




                    0.2000




                    0.0000
                             1000 950 to 900 to 850 to 800 to 750 to 700 to 650 to 600 to 550 to 500 to 450 to 400 to 350 to 300 to 250 to 200 to 150 to 100 to 50 to 0 to 49
                              or   999 949 899 849 799 749 699 649 599 549 499 449 399 349 299 249 199 149                                                       99
                             More




                                                                                                                                                              Page 8 of 14 
 
Table 4: K-S Test result for validation dataset

    BGSCORE                                 Bad                 Good         Cumulative Cumulative    Bad       Good                                          Difference
                                                                               Bad        Good     Percentage Percentage
    1000 or More                                       2               38             2         38      0.0009     0.0080                                              0.0071
     950 to 999                                        2               63             4        101      0.0018     0.0213                                              0.0195
     900 to 949                                       14              134            18        235      0.0079     0.0495                                              0.0416
     850 to 899                                       52              227            70        462      0.0307     0.0973                                              0.0666
     800 to 849                                       88              414           158        876      0.0693     0.1845                                              0.1152
     750 to 799                                      189              655           347       1531      0.1522     0.3224                                              0.1702
     700 to 749                                      282              811           629       2342      0.2759     0.4932                                              0.2173
     650 to 699                                      372              851          1001       3193      0.4390     0.6724                                              0.2333
     600 to 649                                      389              624          1390       3817      0.6096     0.8037                                              0.1941
     550 to 599                                      336              441          1726       4258      0.7570     0.8966                                              0.1396
     500 to 549                                      244              266          1970       4524      0.8640     0.9526                                              0.0886
     450 to 499                                      188              136          2158       4660      0.9465     0.9813                                              0.0348
     400 to 449                                       57               58          2215       4718      0.9715     0.9935                                              0.0220
     350 to 399                                       45               20          2260       4738      0.9912     0.9977                                              0.0065
     300 to 349                                       17                7          2277       4745      0.9987     0.9992                                              0.0005
     250 to 299                                        2                4          2279       4749      0.9996     1.0000                                              0.0004
     200 to 249                                        1                0          2280       4749      1.0000     1.0000                                              0.0000
     150 to 199                                        0                0          2280       4749      1.0000     1.0000                                              0.0000
     100 to 149                                        0                0          2280       4749      1.0000     1.0000                                              0.0000
       50 to 99                                        0                0          2280       4749      1.0000     1.0000                                              0.0000
        0 to 49                                        0                0          2280       4749      1.0000     1.0000                                              0.0000  

 

Figure 3: K-S Test graph for validation dataset

                  K-S Test for the Validation Data                                                   Bad Percentage            Good Percentage

                  1.0000




                  0.8000




                  0.6000
     Difference




                  0.4000




                  0.2000




                  0.0000
                           1000 950 to 900 to 850 to 800 to 750 to 700 to 650 to 600 to 550 to 500 to 450 to 400 to 350 to 300 to 250 to 200 to 150 to 100 to 50 to 0 to 49
                            or   999 949 899 849 799 749 699 649 599 549 499 449 399 349 299 249 199 149                                                       99
                           More




                                                                                                                                                             Page 9 of 14 
 
Table 5: K-S Test result for All dataset

    BGSCORE                               Bad                  Good          Cumulative Cumulative    Bad       Good                                             Difference
                                                                               Bad        Good     Percentage Percentage
    1000 or More                                      6             111               6        111      0.0013     0.0118                                                 0.0105
     950 to 999                                       7             139              13        250      0.0028     0.0265                                                 0.0237
     900 to 949                                      33             303              46        553      0.0100     0.0586                                                 0.0486
     850 to 899                                      90             469             136       1022      0.0296     0.1082                                                 0.0787
     800 to 849                                     177             778             313       1800      0.0680     0.1906                                                 0.1226
     750 to 799                                     347            1242             660       3042      0.1435     0.3222                                                 0.1787
     700 to 749                                     529            1561            1189       4603      0.2585     0.4875                                                 0.2290
     650 to 699                                     708            1527            1897       6130      0.4124     0.6492                                                 0.2368
     600 to 649                                     815            1346            2712       7476      0.5896     0.7918                                                 0.2022
     550 to 599                                     667             926            3379       8402      0.7346     0.8899                                                 0.1553
     500 to 549                                     508             557            3887       8959      0.8450     0.9488                                                 0.1038
     450 to 499                                     363             268            4250       9227      0.9239     0.9772                                                 0.0533
     400 to 449                                     203             128            4453       9355      0.9680     0.9908                                                 0.0227
     350 to 399                                      96              66            4549       9421      0.9889     0.9978                                                 0.0089
     300 to 349                                      39              13            4588       9434      0.9974     0.9992                                                 0.0018
     250 to 299                                       9               8            4597       9442      0.9993     1.0000                                                 0.0007
     200 to 249                                       2               0            4599       9442      0.9998     1.0000                                                 0.0002
     150 to 199                                       1               0            4600       9442      1.0000     1.0000                                                 0.0000
     100 to 149                                       0               0            4600       9442      1.0000     1.0000                                                 0.0000
       50 to 99                                       0               0            4600       9442      1.0000     1.0000                                                 0.0000
        0 to 49                                       0               0            4600       9442      1.0000     1.0000                                                 0.0000




Figure 4: K-S Test graph for All dataset

                  K-S Test for All Data                                                                       Bad Percentage              Good Percentage

                    1.0000




                    0.8000




                    0.6000
     Difference




                    0.4000




                    0.2000




                    0.0000
                             1000 950 to 900 to 850 to 800 to 750 to 700 to 650 to 600 to 550 to 500 to 450 to 400 to 350 to 300 to 250 to 200 to 150 to 100 to 50 to 0 to 49
                              or   999 949 899 849 799 749 699 649 599 549 499 449 399 349 299 249 199 149                                                       99
                             More




                                                                                                                                                              Page 10 of 14 
 
Table 6: Scorecard
                                            Model Scorecard
                                                                              Model     Actual
         Variable       Intervals                                             Points   Frequency
                        Special Cases                                            0         17.96%
                        Greater than 0 months but equal to or less than 17     -56         20.32%
     Average Age of Greater than 17 months but equal to or less than 59          0         50.88%
           Trade        Greater than 59 months but equal to or less than 83     35          6.99%
                        Greater than 83 months                                  49          3.85%
                        Total                                                             100.00%
                        Special Cases                                           0          11.86%
      Age of Oldest     Less than 84 months                                    0           54.85%
           Trade        Greater than or equal to 84 months                     17          33.29%
                        Total                                                             100.00%
                        Special Cases                                           0          88.95%
    Ratio of Balance to
                        0% to 40%                                              44           2.44%
     HC for All Open
                        Greater than 40%                                        0           8.61%
       Auto Trades
                        Total                                                             100.00%
                        Special Cases                                           0          55.51%
    Age of Oldest Bank Less than 72 months                                     0           30.85%
     Revolving Trade Greater than or equal to 72 months                        36          13.64%
                        Total                                                             100.00%
                        Less than $5,501                                       0           74.59%
     Base Wholesale
                        Greater than or equal to $5,501                        25          25.41%
           Value
                        Total                                                             100.00%
                        Special Cases                                           0           48.97%
    Ratio of Currently Less than or equal to 70%                                0           30.86%
    Satisfied Trades to
       Open Trades      70% to 100%                                            22           20.17%
                        Total                                                              100.00%
                        Special Cases                                           0           11.44%
    Number of Trades
                        Less than 2                                            0            72.26%
     Currently Rated
                        Greater than or equal to 2                             31           16.30%
         Satisfied      Total                                                              100.00%
                        Less than or equal to $1,200                           -50          48.35%
      Down Payment Greater than $1,200                                          0           51.65%
                        Total                                                              100.00%
                        Special Cases                                           0           11.43%
     Ratio of Satisfied Less than 20%                                           0           59.70%
      Trades to Total 20% to 65%                                               29           24.82%
          Trades        65% to 100%                                            75            4.05%
                        Total                                                              100.00%
                     Special Cases                                              0           11.44%
    Number of Trades 0 to 1                                                    -38          46.00%
    Never 90 Days or
     More Past Due Over 1                                                       0           42.56%
                     Total                                                                 100.00%


                                                                                       Page 11 of 14 
 
                                                                            Model     Actual
        Variable         Intervals                                          Points   Frequency
                         Special Cases                                        0          11.44%
    Number of Trades
                         Less than 11                                         0          76.73%
     Ever Rated Bad
                         11 or more                                          -53         11.83%
          Debt           Total                                                          100.00%
                         Special Cases                                        0           4.14%
Number of Inquiries      Less than 10                                         0          89.79%
 in Last 12 Months       10 or more                                          -81          6.07%
                         Total                                                          100.00%
                         Less than or equal to $1,500                        -18         38.11%
    Monthly Gross Pay    Greater than $1,500                                  0          61.89%
                         Total                                                          100.00%
                         13 to 22 years old                                 -140          3.19%
                         9 to 12 years old                                   -67         18.90%
                         7 to 8 years old                                     0          27.42%
     Age of Vehicle      3 to 6 years old                                     40         48.70%
                         0 to 2 years old                                     83          1.79%
                         Total                                                          100.00%
                         Less than $1,000                                    0           94.30%
                         Between $1,000 and $1,500                           17           3.70%
    Net Trade In Value
                         Greater than $1,500                                 58           2.00%
                         Total                                                          100.00%
                         Special Cases                                       0           64.78%
    Number of Open       Less than 1                                         0            9.45%
    Revolving Trades     1 to 2                                              34          20.86%
    With Balance > $0    Greater than 3                                      60           4.91%
                         Total                                                          100.00%
                         Special Cases                                       0           11.44%
       Number of         Less than or equal to 10                            0           83.43%
    Revolving Trades     Greater than 10                                     34           5.13%
                         Total                                                          100.00%

 Number of Trades Special Cases                                              0           11.44%
Rated 30 Days Past 0                                                         40          17.88%
 Due and More in Greater than 0                                               0          70.68%
the Last 24 Months
                   Total                                                                100.00%
                   Less than or equal to 42 months                           0           94.44%
   Term of Loan    Greater than 42 months                                    47           5.56%
                   Total                                                                100.00%
                   Special Cases                                              0          16.89%
Months Since Most
                   Less than or equal to 4 months                            -45         33.28%
  Recent 60 Days Greater than 4 months and less than or equal to 9 months    -27         22.52%
 Past Due or More
                   Greater than 9 months                                      0          27.31%
      Rating       Total                                                                100.00%
                   3 or less                                                 56          33.82%
 Number of Trades Greater than 3                                             0           66.18%
                   Total                                                                100.00%
                                                                                     Page 12 of 14 
 
                                                                          Model     Actual
        Variable      Intervals                                           Points   Frequency
                      Missing                                               0           0.17%
                      Less than 48 years old                                0          90.99%
      Customer Age
                      48 years old or more                                  40          8.84%
                      Total                                                           100.00%
                      Less than 5 months                                   -54         24.97%
      Time at Job in Greater than or equal to 5 months and less than 49      0         57.54%
         Months       Greater than or equal to 49 months                    19         17.49%
                      Total                                                           100.00%
                      Less than 5 months                                   -53         16.52%
    Time at Residence
                      5 months or more                                      0          83.48%
        in Months
                      Total                                                           100.00%




                                                                                   Page 13 of 14 
 
Table 7: Monitoring Report

    ACTUAL vs EXPECTED SCORE DISTRIBUTION

     Score    Expected Score   Actual Score
     Above    Distribution     Distribution   Difference
      >1000          0.83%
      >950           1.87%
      >900           4.27%
      >850           8.25%
      >800           15.05%
      >750           26.36%
      >700           41.25%
      >650           57.16%
      >600           72.55%
      >550           83.90%
      >500           91.48%
      >450           95.98%
      >400           98.33%
      >350           99.49%
      >300           99.86%
      >250           99.98%
      >200           99.99%
      >150          100.00%
       >50          100.00%
       >0           100.00%




                                                           Page 14 of 14 
 

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:0
posted:5/15/2012
language:
pages:14
fanzhongqing fanzhongqing http://
About