The sample analysis
Unsecured US consumer instalment loans originated on the Lending Club platform, 36 and 60 month terms, from the public accepted-loans table. It went through the same pipeline as an uploaded tape: mapped to the standard schema, standardised, then run through every analysis below.
- Loans
- 886,837
- Original balance
- $13.1BSum of original balance.
- Weighted average coupon
- 13.6%Interest rate weighted by original balance.
- Charge-off rate (count)
- 17.2%Loans charged off at the snapshot over all loans.
- Charge-off rate (balance)
- 18.1%Original balance of charged-off loans over total original balance.
- Net loss rate
- 10.6%Gross loss (charged-off original balance less principal received) less recoveries, over total original balance.
- Recovery rate
- 12.4%Recoveries over gross loss on charged-off loans.
- Still current
- 6.3%Share of loans with status current at the snapshot, by count.
Composition
Stratification of the book at origination. Each strat shows the share of loans and the share of original balance in each class.
Composition by grade
Share of loans (count-weighted) and of original balance (balance-weighted).
-
Note
Grades C and B hold 53.7% of original balance.
Composition by term
Share of loans (count-weighted) and of original balance (balance-weighted).
-
Note
60-month loans are 30.0% of loans but 40.8% of balance: the longer term carries the larger tickets.
Composition by purpose
Share of loans (count-weighted) and of original balance (balance-weighted).
Top 8 by balance; the rest, and any class the tape itself labels other, folded into Other.
-
Note
The largest purpose, Debt consolidation, is 61.8% of original balance.
Composition by region
Share of loans (count-weighted) and of original balance (balance-weighted).
Top 10 by balance; the rest, and any class the tape itself labels other, folded into Other.
Composition by home ownership
Share of loans (count-weighted) and of original balance (balance-weighted).
Top 6 by balance; the rest, and any class the tape itself labels other, folded into Other.
Composition by income verification
Share of loans (count-weighted) and of original balance (balance-weighted).
Composition by credit score
Share of loans (count-weighted) and of original balance (balance-weighted).
Credit score at origination, lower-inclusive bands.
Composition by debt-to-income
Share of loans (count-weighted) and of original balance (balance-weighted).
DTI in percent at origination, lower-inclusive.
Composition by loan size
Share of loans (count-weighted) and of original balance (balance-weighted).
Original balance in bands of 10k, lower-inclusive.
Composition by origination quarter
Share of loans (count-weighted) and of original balance (balance-weighted).
Charge-off by segment
Charge-off rate at the snapshot for each strat, against the book rate of 17.2% by count and 18.1% by balance. Segments above 1.5x the book rate on more than 2% of balance are flagged watch; above 2.0x, concern.
Charge-off rate by grade
Loans charged off at the snapshot over loans in the class, count-weighted. Book: 17.2%.
Status at the snapshot: less-seasoned classes understate lifetime charge-off.
-
Concern
Grade F loans charge off at 38.4%, 7.0x the 5.5% of grade A loans, while paying 3.3x the coupon (23.6% against 7.3%).
Charge-off rate by sub-grade
Loans charged off at the snapshot over loans in the class, count-weighted. Book: 17.2%.
Status at the snapshot: less-seasoned classes understate lifetime charge-off.
Charge-off rate by term
Loans charged off at the snapshot over loans in the class, count-weighted. Book: 17.2%.
Status at the snapshot: less-seasoned classes understate lifetime charge-off.
-
Note
60-month loans charge off at 25.0% against 13.9% for 36-month loans (1.8x), on 40.8% of balance.
Charge-off rate by purpose
Loans charged off at the snapshot over loans in the class, count-weighted. Book: 17.2%.
Status at the snapshot: less-seasoned classes understate lifetime charge-off.
Charge-off rate by region
Loans charged off at the snapshot over loans in the class, count-weighted. Book: 17.2%.
Status at the snapshot: less-seasoned classes understate lifetime charge-off.
Charge-off rate by income verification
Loans charged off at the snapshot over loans in the class, count-weighted. Book: 17.2%.
Status at the snapshot: less-seasoned classes understate lifetime charge-off.
Charge-off rate by credit score
Loans charged off at the snapshot over loans in the class, count-weighted. Book: 17.2%.
Credit score at origination, lower-inclusive bands. Status at the snapshot: less-seasoned classes understate lifetime charge-off.
Charge-off rate by debt-to-income
Loans charged off at the snapshot over loans in the class, count-weighted. Book: 17.2%.
DTI in percent at origination, lower-inclusive. Status at the snapshot: less-seasoned classes understate lifetime charge-off.
-
Watch
Outside the risk-grade ladder, 1 segment charge off above 1.5x the book rate of 17.2% while holding over 2% of balance: loans with DTI 35+ at 26.7% (1.55x, 2.3% of balance).
Charge-off rate by origination quarter
Loans charged off at the snapshot over loans in the class, count-weighted. Book: 17.2%.
Status at the snapshot: less-seasoned classes understate lifetime charge-off.
Loss and recovery
How much a charge-off costs once principal received and recoveries are netted off.
Gross loss, recoveries and net loss by grade
Gross loss is original balance less principal received on charged-off loans; net loss is gross loss less recoveries.
Charged-off loans only, as of the snapshot.
-
Note
Charged-off loans lost $1.58B gross; recoveries returned $195.3M (12.4%), leaving $1.38B net, a loss given default of 58.5%.
Loss given default by grade
Net loss over the original balance of charged-off loans, balance-weighted.
Gross loss is original balance less principal received on charged-off loans; net loss is gross loss less recoveries.
-
Note
Loss given default runs from 45.2% (grade A) to 70.8% (grade G), a 1.6x spread flatter than the 7.8x spread in charge-off frequency between the same grades. Grades with fewer than 30 charge-offs are left out of the comparison.
Vintage curves
Cumulative charge-off by months on book. The tape has no charge-off date, so the last payment date stands in for the default month; the actual charge-off typically follows some months later, so the curves lead the true timing.
Cumulative charge-off by months on book, by grade (count)
Charged-off loans over loans originated, count-weighted.
Grades beyond the sixth are folded into one series (F–G) to keep the chart to 6 lines. The tape has no charge-off date, so the last payment date stands in for the default month; the actual charge-off typically follows some months later, so the curves lead the true timing. Seasoning measured to the snapshot, 2019-03-01 (per the dataset metadata). Each month's increment uses only cohorts observed that long; a curve stops once under 10% of its loans are observed. 670 charged-off loans have no default or payment date and are placed at month 0.
-
Note
By month 36 on book, grade F–G has charged off 36.3% of loans against 5.4% for grade A; the gap is 14.8 points at month 12 and 30.9 points by month 36.
Cumulative charge-off by months on book, by grade (balance)
Charged-off original balance over original balance originated, balance-weighted.
Grades beyond the sixth are folded into one series (F–G) to keep the chart to 6 lines. The tape has no charge-off date, so the last payment date stands in for the default month; the actual charge-off typically follows some months later, so the curves lead the true timing. Seasoning measured to the snapshot, 2019-03-01 (per the dataset metadata). Each month's increment uses only cohorts observed that long; a curve stops once under 10% of its loans are observed. 670 charged-off loans have no default or payment date and are placed at month 0.
Cumulative charge-off by months on book, by origination year (count)
Charged-off loans over loans originated, count-weighted.
Only the 7 most recent origination years are drawn. The tape has no charge-off date, so the last payment date stands in for the default month; the actual charge-off typically follows some months later, so the curves lead the true timing. Seasoning measured to the snapshot, 2019-03-01 (per the dataset metadata). Each month's increment uses only cohorts observed that long; a curve stops once under 10% of its loans are observed. 670 charged-off loans have no default or payment date and are placed at month 0.
-
Note
The 2015 cohort runs worst, 17.4% charged off by month 36 against 12.6% for 2010.
-
Note
Across the book, 5.6% of loans have charged off by month 12 and 12.3% by month 24 (default month by proxy).
Cumulative charge-off by months on book, by origination year (balance)
Charged-off original balance over original balance originated, balance-weighted.
Only the 7 most recent origination years are drawn. The tape has no charge-off date, so the last payment date stands in for the default month; the actual charge-off typically follows some months later, so the curves lead the true timing. Seasoning measured to the snapshot, 2019-03-01 (per the dataset metadata). Each month's increment uses only cohorts observed that long; a curve stops once under 10% of its loans are observed. 670 charged-off loans have no default or payment date and are placed at month 0.
Yield against loss
Whether the extra coupon on riskier grades pays for their losses.
Yield against loss by grade
WAC is balance-weighted. Interest received and net loss are cumulative to the snapshot, over original balance, balance-weighted; not annualised, before servicing and funding costs.
-
Note
Grade G (0.9% of balance) has received interest of 38.7% of balance against net loss of 30.1%: the coupon covers the losses so far.
Interest received less net loss, by grade
Interest received and net loss are cumulative to the snapshot, over original balance, balance-weighted; not annualised, before servicing and funding costs.
-
Note
After net losses, grade D returns the most, 11.8% of original balance to date; grade A the least at 7.5%.
Concentration
How much of the balance sits in a few regions, purposes or names.
Concentration of balance
Share of total original balance, balance-weighted.
-
Note
The five largest regions (CA, TX, NY, FL, IL) hold 41.8% of original balance, CA alone 14.5%.
-
Note
The largest single loan is $35k, 0.0003% of the book: single-name risk is negligible.
Herfindahl-Hirschman index of balance
Sum of squared balance shares in percent, 0 to 10,000.
Read against the usual antitrust thresholds: under 1,500 unconcentrated, 1,500 to 2,500 moderate, above 2,500 high. 10,000 / HHI gives the equivalent number of equal-sized classes.
-
Note
HHI by region 535 (unconcentrated, equivalent to 18.7 equal-sized classes out of 51); by purpose 4,448 (highly concentrated, equivalent to 2.2 equal-sized classes out of 14).
Data quality
Completeness and validity of the mapped fields.
Missing values by standard field
Share of rows where the field is null, count-weighted.
-
Note
1 field is more than 5% null: employment_length_years 5.1%.
Validation checks
Rows failing each check. Snapshot for date checks: 2019-03-01 (per the dataset metadata).
-
Watch
1 check fails: dti outside 0–100 (9).
-
Watch
Leakage warning: current_balance, interest_paid, last_payment_date, principal_paid, raw_status, recoveries, status, total_paid are observed after origination. The metrics here use them to measure outcomes; any model that predicts default must exclude them.
Caveats
What to keep in mind before quoting any number above.
About the data
- The report covers the 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015 vintages: 886,837 loans issued Jan 2008 to Dec 2015.
- 2007 is excluded (603 loans across 7 issue months): incomplete or not seasoned 36 months at the snapshot.
- 2016 is excluded (434,407 loans across 12 issue months): incomplete or not seasoned 36 months at the snapshot.
- 2017 is excluded (443,579 loans across 12 issue months): incomplete or not seasoned 36 months at the snapshot.
- 2018 is excluded (495,242 loans across 12 issue months): incomplete or not seasoned 36 months at the snapshot.
- Status is as of the snapshot, taken as the latest last-payment date in the file (Mar 2019). Loans issued in 2017 and 2018 would be mostly current at that date, so their charge-off rates would not be comparable to seasoned vintages.
- The tape has no charge-off date. Vintage curves place each charge-off at the last payment date, which leads the actual charge-off by several months.
- Credit score is the low end of the FICO range at origination. Original balance is the requested loan amount (loan_amnt).
- Statuses 'Default' and 'Charged Off' both map to charged off; 'In Grace Period' maps to current; the two 'Late' buckets to late.