Harmoney

[myles]

The change from 1 to 2 looks something like:

Grade	Change
A1 - A2	78% drop
B1 - B2	74% drop
C1 - C2	3% rise
D1 - D2	9% drop
E1 - E2	69% rise
F1 - F2	4% drop

So I'd say only A and B are showing significant drops from 1 to 2?

[RMJH]

The change from 1 to 2 looks something like:

Grade	Change
A1 - A2	78% drop
B1 - B2	74% drop
C1 - C2	3% rise
D1 - D2	9% drop
E1 - E2	69% rise
F1 - F2	4% drop

So I'd say only A and B are showing significant drops from 1 to 2?

I'd agree but the interest charged is somewhat significantly different.. Possibly just statistical noise or bias in our sample. I won't be changing my filters for it.

[beacon]

One thing I've not noticed before is that the early loans ~ pre mid 2017 were being paid off earlier - around week 20, while ~post mid 2017 loans are being paid off at around 30-35 week mark. Could be something to do with scorecard changes or borrower selection or even a shift in grade selection by lenders. Just interesting...

Could be change in Harmoney sales focus and strategy due to Commerce Commission actions/ruling...

[beacon]

No (I think), exactly why I produced the Expected Defaults graph. The Hazard Curve (i.e. Harmoney Hazard Curve), is plotting defaults in a month vs total loans (all loans from all time). ... I'm just not sure [mine] is the Hazard Curve?

The way I'm reading it, your graph is a hazard curve too. It shows that 42% of the defaults in our data pool occurred within the first 15 months of the loan. Harmoney hazard curve (to july 17) stated that almost 60% of the defaults that occurred were within the first 15 months of the loan. Harmoney hazard curve was plotted on maturity approx two-thirds of ours (time wise), which explains some of the improvement in percentage.

I'm getting the distinction you are making, as well as the goal you are aiming at, through Survival rate computation. But how to plot it, given that survival rate gets impacted not just by defaults but by early repaids too? Hmm...

[beacon]

The return wasn't too different from the overall average though - mainly due to the set containing over 60% A and B grade loans. In contrast the default minimiser set contained 60% at CDE grade and turned in a lower default rate and higher return.

Thanks leesal, very interesting. Also your work on max vs non-max loans, which was supplemented by additional insights from Myles, humvee, Cool Bear and RMJH.
Also interesting was the "grade ceiling overflows" from A limits into the B1 and B2, and B limits into C1, C2 and C4. So, not a ceiling after all, more like a rough guideline...

[beacon]

If you liked that then you might like these even more - same but with grade as well pdf runs to 67 pages though...

Some interesting observations emerged as I played around with your trivariate tables today, Myles. For example, while all except 50-59 year olds were high risk when buying new vehicles, all new vehicle buyers in AB grades were cool to loan to (despite small bases by age groups within grade).

Full owners in A grade taking holidays were cool, but in C were high risk. Etc.

While the above 60 wanting loans to buy a new vehicle were not at all cool overall (2 defaults /21 total), we may have to wait until the next data pool to analyze them by grade.

Thanks again for your time and effort in producing and sharing these nevertheless.

[myles]

Not sure if this has been discussed before so just putting it out there - it's really a question: Is a loan paid off in a shorter period, higher, lower or the same risk/value as a loan paid off over a longer period?

Some numbers relating to variants and their average age in months (Paid Off​ loans only):

Variant	Description	Loans	Avg Mths
Age Band	30-39	2852	9.82
Age Band	40-49	3744	10.13
Age Band	20-29	1200	10.58
Age Band	50-59	3039	10.75
Age Band	Above 60	948	11.33
Loan Purpose	Purchase Boat	42	8.31
Loan Purpose	Clear Overdraft	307	8.51
Loan Purpose	Funeral Expenses	116	8.53
Loan Purpose	Legal Fees	38	8.86
Loan Purpose	Computer	60	8.91
Loan Purpose	Medical Expenses	150	9.17
Loan Purpose	Education Expenses	210	9.20
Loan Purpose	Home Improvements	1743	9.36
Loan Purpose	Holiday Expenses	1191	9.96
Loan Purpose	Household Items	385	10.06
Loan Purpose	Purchase Caravan	32	10.37
Loan Purpose	Purchase New Vehicle	141	10.46
Loan Purpose	Loan to Family Member	150	10.49
Loan Purpose	Debt Consolidation	4539	10.56
Loan Purpose	Business Cash Flow	436	10.61
Loan Purpose	Wedding Expenses	236	10.69
Loan Purpose	Tax Bill	57	11.12
Loan Purpose	Purchase Used Vehicle	679	11.22
Loan Purpose	Other	1271	11.91
Residential Status	Owned - Paying Mortgage	5721	9.88
Residential Status	Fully Owned - No Mortgage	211	10.43
Residential Status	Living with Parents	543	10.51
Residential Status		5	10.61
Residential Status	Renting	4119	10.66
Residential Status	Boarding	533	10.84
Residential Status	Other	497	12.00
Residential Status	Supplied By Employer	154	12.14
Term Months	60	8398	9.73
Term Months	36	3385	11.91

The fact that, on average, a 60 month term loan is paid of in less time than a 36 month term loan is just nuts? Perhaps it's human nature to take the extra time and perhaps most don't realise the extra Interest paid...

[Cool Bear]

My view is that early repayment does not affect the returns or risk. If you accept a certain risk/returns and if the loan is repaid early then you can always invest the proceeds into a similar loan.

[RMJH]

Under the original fee structure, early repayment hurt. Not so much now but there is still a re-investment lag. Just to confirm. Your figures represent the average age at which early repaid loans were repaid rather than average loan life? I think that must be the case.

[RMJH]

Forgot to say, I think the remainder of the original cohort will become a bit more risky with these early repayments as they are prior to the default peak and thus the bad loans are less diluted.