Report 10. Is Consumer Ability To Repay Predictive Of Actual Repayment Of Storefront Payday Loans? BY RICK HACKETT 1

Transcription

1 Report 10 n o n P r i m e W H I T E P A P E R Is Consumer Ability To Repay Predictive Of Actual Repayment Of Storefront Payday Loans? BY RICK HACKETT 1

2 I S C O N S U M E R A B I L I T Y T O R E P A Y P R E D I C T I V E O F A C T U A L R E P A Y M E N T O F S T O R E F R O N T P A Y D A Y L O A N S? Introduction In a recent White Paper, we analyzed the ability-to-repay (ATR) of a large sample of storefront payday borrowers, using the CFPB s recently proposed method of computing residual cash flow available to repay a loan. In that paper, we divided borrowers into three classes: cash-flow insolvent without a loan, cashflow solvent but having insufficient cash to make the required single payment, and sufficiently cash-flow solvent to afford the loan repayment while meeting other obligations and basic living expenses. Only this last group would qualify for a loan under CFPB requirements. DO THOSE WITH THE ABILITY-TO- REPAY APPEAR MORE LIKELY TO ACTUALLY REPAY? THE ANSWER IS: NOT MUCH. We then asked the question: is there any difference in actual repayment behavior amongst the three groups? In other words, do those with the ability-to-repay appear more likely to actually repay? The answer is: NOT MUCH. REVISION HISTORY: This report was revised February 15, 2017, to correct a computational inconsistency found in an appendix table. That revision did not affect report conclusions or key tabulations. A comparison of old and revised computation results is set forth in Table S-1 1. Rick Hackett is a Special Policy Consultant to nonprime101, and was formerly the Assistant Director for Installment and Liquidity Lending Markets at the CFPB. This report is based on statistical analysis by Heather Lamoureux, Research Associate for nonprime101. 2

3 n o n P r i m e W H I T E P A P E R The Dataset Our source for storefront payday loans and borrowers is described fully in nonprime101 Report #7-B. For this study, we randomly selected 90,000 loans and borrowers from approximately 15.6 million storefront payday loans made in We appended to those records the debt service obligations of the borrowers at the relevant times, as reported in both a nationally recognized credit bureau and in Clarity records. We had two sources of income information for this computation, each with alternative values for income. The master storefront loan dataset had income values at the time of application for the loan we tested and, in most cases, income values from other applications seen in Similarly, Clarity production data showed income values from other credit applications, both at the time of the loan being tested and at other times in k LOANS AND BORROWERS FROM APPROXIMATELY 15.6 MILLION STOREFRONT PAYDAY LOANS MADE IN Of the original set of borrowers, Clarity production data contained income information from 2013 loan inquiries (often multiple inquiries) for nearly 85,000 borrowers. Because we had multiple income values for many consumers, we used both a median value for the year 2013 and a value that corresponded with time of the application for the loan being tested (called last seen income in our report). Similarly, the master storefront dataset provided a last seen income value for 79,000 loans and a median income value (based on multiple applications) for more than 76,000 borrowers. Results for all four income values are shown in Appendix Tables 1-3. We then applied the CFPB s proposed methodology to compute residual income after payment of debt service obligations and used the income remaining to cover a new loan payment after paying basic living expenses (as defined by the CFPB proposal). CFPB defines basic living expenses as those necessary for the borrower s health, welfare and ability to produce income, including health and welfare expenses of dependents. We translated this into expenses for shelter, food, transportation, communication, medical care and dependent child care. Where the consumer reports included debt payments for shelter (a mortgage payment) or an auto loan, we used those values. In all other cases, we proxied expenses based on data from the Bureau of Labor Statistics (BLS) and the U.S. Census Bureau, both sources endorsed in CFPB s proposal. The BLS data is segmented based on income and age of the consumer, and we used those segmentations. I S C O N S U M E R A B I L I T Y T O R E P A Y P R E D I C T I V E O F A C T U A L 3 R E P A Y M E N T O F S T O R E F R O N T P A Y D A Y L O A N S?

4 I S C O N S U M E R A B I L I T Y T O R E P A Y P R E D I C T I V E O F A C T U A L R E P A Y M E N T O F S T O R E F R O N T P A Y D A Y L O A N S? We first computed the count and percentage of borrowers who had negative residual income after paying pre-existing debts and basic living expenses. For those who had a positive residual income, we then computed the count and percentage whose residual income exceeded the required payment on the payday loan in question. For monthly payroll consumers, we compared loan payment to monthly residual income. For semi-monthly, biweekly and weekly payroll borrowers (whose new loan payments are due at the end of a less-than-monthly payroll period), we used a corresponding percentage of residual income (e.g., a semi-monthly payroll borrower has half of their monthly residual income available to make a payday loan payment due at the end of the payroll cycle). This methodology does not take into account the lumpiness of some payments, such as rent, and therefore may overstate the number of consumers who can make a payday loan payment, meet their other expenses, and have money left over. Accordingly, we provided information on the mean and median cash available after making the payday loan payment. THIS METHODOLOGY DOES NOT TAKE INTO ACCOUNT THE LUMPINESS OF SOME PAYMENTS, SUCH AS RENT, AND THEREFORE MAY OVERSTATE THE NUMBER OF CONSUMERS WHO CAN MAKE A PAYDAY LOAN PAYMENT, MEET THEIR OTHER EXPENSES, AND HAVE MONEY LEFT OVER. Having defined various groups by their level of cash flow solvency relative to the payment due on the test loan, we then reviewed lender records on whether that loan defaulted and tabulated the results in Appendix Table 4. 4

5 n o n P r i m e W H I T E P A P E R Results We found two unexpected outcomes. First, the relative level of defaults (percentage of total loans that defaulted) was relatively consistent across all groups. We expected significantly fewer defaults among those with more cash available to repay a loan, as compared to the average. We summarize the results in Table 1. Table 1 PERCENTAGE OF DEFAULTS BY CASH FLOW SOLVENCY BEFORE AND AFTER LOAN PAYMENT Clarity Median Income % Storefront Median Income % Clarity Last Seen Income % Storefront Last Seen Income % Average Rate Cash Flow Negative Before Loan Cash Flow Negative After Loan Cash Flow Solvent with Loan While the absolute values of default rate vary across the different income values, the relationship of default rates is consistent: cash flow solvency (ATR) is not highly predictive of actual payment behavior. Those with adequate cash flow perform slightly better, for most income values, than those who are cash flow insolvent - whether with or without the loan payment. The reduction from average default rates ranged from 3.7% to 5.5%, except in the case of Clarity-observed median income values, where, those who are cash flow negative without the loan actually perform better than those who are cash flow positive with the loan. The ATR analysis is not predictive at all. 5.5 % AT BEST, CUSTOMERS WITH PASSING ATR CASH FLOWS HAD A 5.5% LOWER THAN AVERAGE DEFAULT RATE. I S C O N S U M E R A B I L I T Y T O R E P A Y P R E D I C T I V E O F A C T U A L 5 R E P A Y M E N T O F S T O R E F R O N T P A Y D A Y L O A N S?

6 I S C O N S U M E R A B I L I T Y T O R E P A Y P R E D I C T I V E O F A C T U A L R E P A Y M E N T O F S T O R E F R O N T P A Y D A Y L O A N S? Our second unexpected outcome revealed a sampling effect in our study method. The absolute value of average test loan defaults in our computations (generally 15%-23%) is typical of first loan behavior, and is above the average per-loan default rate across the entire sample, which is closer to 4%. The source of the anomaly was the random sampling method for borrowers and loans whose data would be supplemented with prime credit bureau data. In order to avoid duplicating borrowers, the sampling method looked for loans associated with borrowers who had no prior loan in the sampling year (2013). As a result, the sampling method oversampled the first loan in the year (but not necessarily in the entire five-year sample). As discussed in prior work by nonprime101 and in CFPB reports, payday loans tend to fall in sequences and defaults tend to occur either on the first loan or the last loan in a sequence. The typical first loan default rate is on the order of 20%, consistent with the test loans we sampled. 4.6 % -5.2 % RANGE OF AVERAGE PER-LOAN DEFAULT RATES FOR ALL 2013 LOANS TO SAMPLED BORROWERS. INITIAL LOAN SUBSAMPLE SHOWS TYPICALLY HIGHER FIRST LOAN DEFAULT RATES. In order to test the sensitivity of our results to this sampling effect, we reviewed the per-loan default rate for all loans taken out by our test borrowers in the sample year. This approach would allow us to see whether other loans in the sequences that began with our test loan behaved in a manner inconsistent with our initial results regarding the predictive value of ATR. We could also see whether the annual per-loan default rate of these borrowers was in line with the overall sample. The results are also shown in Appendix Table 4. The overall per-loan default rates for all loans in 2013 for the three borrower groups are even closer together than the first-loan default rates. ATR cash flow status produces at most a 1.6% difference in default rates (e.g., lowering default rates from 5.1% for insolvent borrowers to 3.5% for solvent borrowers, using the Median- Storefront Income Measure). We also see that the absolute value of mean per-loan default rates is in line with other observations of this metric in this and similar datasets, suggesting that our borrower/loan selection did not skew the results. In summary, we find that borrower behavior (paying or not paying a loan) is not strongly predicted by the ability of the borrower to make the payment, at least as computed under proposed CFPB ATR standards. 6

7 n o n P r i m e W H I T E P A P E R Observations There could be many reasons for the lack of correlation between available cash flow and likelihood to repay. Some that occur to us include: Our analysis does not account for lumpiness of expenses (like rent), and that characteristic may skew the results. Note, however, that monthly payroll borrowers do not have that mismatch issue, and their results are consistent with other payroll cycles. Notably, the CFPB proposal would require a lender to account for lumpiness issues by running the cash flow analysis for 30 days after the loan is paid in full, an effort we could not simulate. A lender s access to alternative payment sources (e.g., a bank account) many incent borrowers to prioritize paying the loan notwithstanding cash flow shortages affecting other expenses. Given the lack of credit reporting of payday loans and the cost inefficiency of formal collection procedures for small loan amounts, borrowers may elect not to pay loans that they could possibly repay. There is no effective sanction for non-payment. The ATR analysis does not take into account use of proceeds. Other studies have suggested that a majority of payday loans are used to cover recurring (basic living) expenses. The loan itself replaces a basic living expense with another debt. If that is the case, the ATR analysis over counts that expense, once in the BLS expense model and a second time as the loan payment. We do not have the data to subsample loans based on use of proceeds. Other factors, such as volatility of income and expenses may make any static analysis of available cash flows insufficient to predict a consumer s financial situation, even two or three weeks later. We do not have the data to test any of these hypotheses. Our results suggest, however, that they should be tested before the ATR proposal becomes final law. I S C O N S U M E R A B I L I T Y T O R E P A Y P R E D I C T I V E O F A C T U A L 7 R E P A Y M E N T O F S T O R E F R O N T P A Y D A Y L O A N S?

8 Appendix Table 1 CFPB RESIDUAL INCOME ATR ANALYSIS - LAST SEEN INCOME CLARITY VS. STOREFRONT VALUES CLARITY MEDIAN INCOME MEASURE 1 YEAR CLARITY LAST SEEN INCOME MEASURE FUNCTION TOTAL* MONTHLY BI-WEEKLY WEEKLY TOTAL* MONTHLY BI-WEEKLY WEEKLY Total A=B+C 84,668 25,321 45,647 10,166 84,771 25,338 45,711 10,185 Negative Residual Income- Before Loan B' 26,635 11,727 11,770 1,986 33,030 15,739 13,792 2,384 Positive Residual Income- Before Loan C' 58,033 13,593 33,877 8,180 51,741 9,625 32,007 7,810 Negative Residual Income- After Loan B 56,629 16,675 28,609 7,812 60,606 19,818 29,423 7,828 Positive Residual Income- After Loan C 28,038 8,646 17,038 2,354 24,165 5,520 16,288 2,357 Percentage of Borrowers with Positive Residual Income- After Loan D=C/A 33.1% 34.1% 37.3% 23.2% 28.5% 21.8% 35.6% 23.1% Mean $605 $945 $488 $206 $630 $972 $567 $263 Median $383 $670 $344 $143 $390 $614 $387 $174 all 2013 by same (84,668) borrowers (incl. above loan sample) E 773, , ,588 84, , , ,046 84,647 Per Borrower All Loans F=E/A All Loans to Borrowers who had Positive Residual Income G 262,942 73, ,844 19, ,519 49, ,902 21,300 Per Borrower With Positive Residual Income Originally-- All Loans H=G/C Percentage of All Loans in 2013 to Borrowers Which Had Positive Residual Income I=G/E 34.0% 33.9% 38.7% 23.0% 31.1% 22.9% 38.6% 25.2% Appendix Table 1 Continued on next page 8

9 Appendix Table 1 Continued EFFECT OF 30-DAY COOLING OFF PERIOD CLARITY MEDIAN INCOME MEASURE 1 YEAR CLARITY LAST SEEN INCOME MEASURE FUNCTION TOTAL* MONTHLY BI-WEEKLY WEEKLY TOTAL* MONTHLY BI-WEEKLY WEEKLY Keep C 28,038 8,646 17,038 2,354 24,165 5,520 16,288 2,357 All Loans to Borrowers who had Positive Residual Income in the 84,668/84,771 sample G 262,942 73, ,844 19, ,519 49, ,902 21,300 Lost to "Cooling-off" (Post-ATP) K 72.2% 59.5% 77.3% 74.9% 73.7% 60.1% 77.4% 75.5% of Loans Kept Post-ATP and Passing the "Cooling-Off" Test L 73,201 29,617 38,315 4,875 63,327 19,790 38,108 5,216 Percentage of all Loans in 2013 to Borrowers Lost to ATP and Cooling-Off M=[E-L]/L 90.5% 86.3% 91.2% 94.2% 91.8% 90.8% 91.3% 93.8% Percentage of All Loans in 2013 to Borrowers Remaining after ATP and Cooling-off N=L/E 9.5% 13.7% 8.8% 5.8% 8.2% 9.2% 8.7% 6.2% Average Number Loans Remaining for Borrowers with Positive Residual Income after Accounting for Cooling Off Period P=L/G *all pay periods may not sum to total some missing pay frequency I S C O N S U M E R A B I L I T Y T O R E P A Y P R E D I C T I V E O F A C T U A L 9 R E P A Y M E N T O F S T O R E F R O N T P A Y D A Y L O A N S?

10 Appendix Table 2 CFPB RESIDUAL INCOME ATR ANALYSIS - MEDIAN INCOME CLARITY VS. STOREFRONT VALUES CLARITY MEDIAN INCOME MEASURE 1 YEAR MEDIAN INCOME STOREFRONT MEASURE 1 YEAR FUNCTION TOTAL* MONTHLY BI-WEEKLY WEEKLY TOTAL* MONTHLY BI-WEEKLY WEEKLY Total A=B+C 84,668 25,321 45,647 10,166 76,861 22,917 41,797 9,658 Negative Residual Income- Before Loan B' 26,635 11,727 11,770 1,986 28,351 14,147 12,013 2,191 Positive Residual Income- Before Loan C' 58,033 13,593 33,877 8,180 46,020 8,604 29,783 7,499 Negative Residual Income- After Loan B 56,629 16,675 28,609 7,812 53,910 18,316 27,817 7,777 Positive Residual Income- After Loan C 28,038 8,646 17,038 2,354 20,462 4,601 13,980 1,881 Percentage of Borrowers with Positive Residual Income- After Loan D=C/A 33.1% 34.1% 37.3% 23.2% 26.6% 20.1% 33.4% 19.5% Mean Positive Residual Income- After Loan Mean of D $605 $945 $488 $206 $602 $913 $547 $253 Median Positive Residual Income- After Loan Median of D $383 $670 $344 $143 $377 $576 $375 $171 all 2013 by same borrowers (incl. above loan sample) E 773, , ,588 84, , , ,886 79,351 Per Borrower All Loans F=E/A All Loans to Borrowers who had Positive Residual Income G 262,942 73, ,844 19, ,429 38, ,746 15,858 Per Borrower With Positive Residual Income Originally-- All Loans H=G/C Percentage of All Loans in 2013 to Borrowers Which Had Positive Residual Income I=G/E 34.0% 33.9% 38.7% 23.0% 28.5% 20.7% 35.8% 20.0% Appendix Table 2 Continued on next page 10

11 Appendix Table 2 Continued EFFECT OF 30-DAY COOLING OFF PERIOD CLARITY MEDIAN INCOME MEASURE 1 YEAR CLARITY LAST SEEN INCOME MEASURE FUNCTION TOTAL* MONTHLY BI-WEEKLY WEEKLY TOTAL* MONTHLY BI-WEEKLY WEEKLY Keep C 28,038 8,646 17,038 2,354 20,462 4,601 13,980 1,881 All Loans to Borrowers who had Positive Residual Income in the sample G 262,942 73, ,844 19, ,429 38, ,746 15,858 Percentage Lost to "Cooling-off" (Post-ATP) K 72.2% 59.5% 77.3% 74.9% 70.4% 54.0% 74.6% 73.2% of Loans Kept Post-ATP and Passing the "Cooling-Off" Test L 73,201 29,617 38,315 4,875 57,762 17,860 35,467 4,251 Percentage of all Loans in 2013 to Borrowers Lost to ATP and Cooling-Off M=[E-L]/L 90.5% 86.3% 91.2% 94.2% 91.6% 90.5% 90.9% 94.6% Percentage of All Loans in 2013 to Borrowers Remaining after ATP and Cooling-off N=L/E 9.5% 13.7% 8.8% 5.8% 8.4% 9.5% 9.1% 5.4% Average Number Loans Remaining for Borrowers with Positive Residual Income after Accounting for Cooling Off Period P=L/G *all pay periods may not sum to total some missing pay frequency I S C O N S U M E R A B I L I T Y T O R E P A Y P R E D I C T I V E O F A C T U A L 11 R E P A Y M E N T O F S T O R E F R O N T P A Y D A Y L O A N S?

12 Appendix Table 3 CFPB RESIDUAL INCOME ATR ANALYSIS - MEDIAN INCOME CLARITY VS. STOREFRONT VALUES LAST SEEN INCOME CLARITY MEASURE LAST SEEN INCOME STOREFRONT MEASURE FUNCTION TOTAL* MONTHLY BI-WEEKLY WEEKLY TOTAL* MONTHLY BI-WEEKLY WEEKLY Total A=B+C 84,771 25,338 45,711 10,185 79,192 23,820 42,971 9,871 Negative Residual Income- Before Loan B' 33,030 15,739 13,792 2,384 31,498 15,143 13,142 2,389 Positive Residual Income- Before Loan C' 51,741 9,625 32,007 7,810 47,692 8,676 29,828 7,482 Negative Residual Income- After Loan B 60,606 19,818 29,423 7,828 58,875 19,465 29,124 7,756 Positive Residual Income- After Loan C 24,165 5,520 16,288 2,357 20,317 4,355 13,847 2,115 Percentage of Borrowers with Positive Residual Income- After Loan D=C/A 28.5% 21.8% 35.6% 23.1% 25.7% 18.3% 32.2% 21.4% Mean Positive Residual Income- After Loan Mean of D $630 $972 $567 $263 $601 $929 $551 $257 Median Positive Residual Income- After Loan Median of D $390 $614 $387 $174 $369 $563 $374 $171 all 2013 by same (84,668) borrowers (incl. above loan sample) E 774, , ,046 84, , , ,185 82,577 Per Borrower All Loans F=E/A All Loans to Borrowers who had Positive Residual Income G 240,519 49, ,902 21, ,106 38, ,003 18,914 Per Borrower With Positive Residual Income Originally-- All Loans H=G/C Percentage of All Loans in 2013 to Borrowers Which Had Positive Residual Income I=G/E 31.1% 22.9% 38.6% 25.2% 27.7% 19.2% 34.6% 22.9% Appendix Table 3 Continued on next page 12

13 Appendix Table 3 Continued EFFECT OF 30-DAY COOLING OFF PERIOD LAST SEEN INCOME CLARITY MEASURE LAST SEEN INCOME STOREFRONT MEASURE FUNCTION TOTAL* MONTHLY BI-WEEKLY WEEKLY TOTAL* MONTHLY BI-WEEKLY WEEKLY Keep C 24,165 5,520 16,288 2,357 20,317 4,355 13,847 2,115 All Loans to Borrowers who had Positive Residual Income in the sample G 240,519 49, ,902 21, ,106 38, ,003 18,914 Percentage Lost to "Cooling-off" (Post-ATP) K 73.7% 60.1% 77.4% 75.5% 73.5% 58.9% 77.2% 75.4% of Loans Kept Post-ATP and Passing the "Cooling-Off" Test L 63,327 19,790 38,108 5,216 53,359 15,906 32,647 4,656 Percentage of all Loans in 2013 to Borrrowers Lost to ATP and Cooling-Off M=[E-L]/L 91.8% 90.8% 91.3% 93.8% 92.7% 92.1% 92.1% 94.4% Percentage of All Loans in 2013 to Borrrowers Remaining after ATP and Cooling-off N=L/E 8.2% 9.2% 8.7% 6.2% 7.3% 7.9% 7.9% 5.6% Average Number Loans Remaining for Borrowers with Positive Residual Income after Accounting for Cooling Off Period P=L/G *all pay periods may not sum to total some missing pay frequency I S C O N S U M E R A B I L I T Y T O R E P A Y P R E D I C T I V E O F A C T U A L 13 R E P A Y M E N T O F S T O R E F R O N T P A Y D A Y L O A N S?

14 Appendix Table 4 #1 Median Income Clarity Measure 1 Year Test Loan % All Loans for Borrowers from #1 in Loans % #2 Median Income Storefront Measure 1 Year Test Loan % All Loans for Borrowers from #2 in Loans % Total 81,133 12, % 739,427 38, % 74,372 17, % 660,212 30, % Negative Residual Income- Before Loan Positive Residual Income- Before Loan Negative Residual Income- After Loan Positive Residual Income- After Loan Percentage of Borrowers with Positive Residual Income- After Loan 25,483 3, % 229,567 11, % 28,351 7, % 231,717 11, % 55,650 8, % 509,860 26, % 46,021 10, % 428,495 18, % , % 476,480 25, % 53,910 13, % 464,783 23, % 28,038 4, % 262,947 13, % 20,462 3, % 195,429 6, % 33.1% 35.6% 26.6% 28.5% * These sample sizes are different than previous samples due to missing values used in the calculations. #3 Last Seen Income Clarity Measure Test Loan % All Loans for Borrowers from #3 in Loans % #4 Last Seen Income Storefront Measure Test Loan % All Loans for Borrowers from #4 in Loans % Total 81,234 12, % 740,188 38, % 73,542 12, % 666,453 32, % Negative Residual Income- Before Loan Positive Residual Income- Before Loan Negative Residual Income- After Loan Positive Residual Income- After Loan Percentage of Borrowers with Positive Residual Income- After Loan 31,914 5, % 276,136 15, % 29,386 5, % 252,377 13, % 49,320 7, % 464,052 23, % 44,156 7, % 414,076 19, % 57,069 9, % 499,669 28, % 53,233 9, % 465,489 24, % 24,165 2, % 240,519 10, % 20,309 2, % 200,964 7, % 28.5% 31.1% 25.7% 27.7% 14

15 Table S-1 #3 LAST SEEN INCOME CLARITY MEASURE #4 LAST SEEN INCOME STOREFRONT MEASURE TEST LOAN DEFAULT % 2013 LOANS DEFAULT % TEST LOAN DEFAULT % 2013 LOANS DEFAULT % ORIGINAL REVISED ORIGINAL REVISED ORIGINAL REVISED ORIGINAL REVISED 2013 LOANS DEFAULT % Total Sample (non-zero) 14.8% 15.1% 5.1% 5.2% 23.0% 23.2% 4.5% 4.6% Negative Residual Income- Before Loan Positive Residual Income- Before Loan Negative Residual Income- After Loan Positive Residual Income- After Loan 13.3% 13.9% 4.8% 5.1% 24.7% 24.7% 4.9% 5.1% 15.1% 15.7% 5.3% 5.3% 22.3% 22.3% 4.3% 4.3% 14.5% 15.5% 5.1% 5.3% 25.3% 25.3% 4.9% 5.1% 14.5% 14.5% 5.1% 5.1% 17.7% 17.7% 3.5% 3.5% #3 LAST SEEN INCOME CLARITY MEASURE #4 LAST SEEN INCOME STOREFRONT MEASURE TEST LOAN DEFAULT % 2013 LOANS DEFAULT % TEST LOAN DEFAULT % 2013 LOANS DEFAULT % ORIGINAL REVISED ORIGINAL REVISED ORIGINAL REVISED ORIGINAL REVISED Total Sample (non-zero) 14.8% 15.1% 5.1% 5.2% 14.1% 16.6% 4.8% 4.9% Negative Residual Income- Before Loan Positive Residual Income- Before Loan Negative Residual Income- After Loan Positive Residual Income- After Loan 15.3% 15.8% 5.3% 5.7% 14.6% 17.5% 5.1% 5.4% 14.0% 14.7% 5.0% 5.0% 13.1% 16.0% 4.5% 4.6% 15.5% 16.5% 5.4% 5.6% 14.7% 18.0% 5.1% 5.4% 11.9% 11.9% 4.4% 5.4% 11.0% 12.9% 3.9% 3.9% I S C O N S U M E R A B I L I T Y T O R E P A Y P R E D I C T I V E O F A C T U A L 15 R E P A Y M E N T O F S T O R E F R O N T P A Y D A Y L O A N S?