Harmoney

[Cool Bear]

Good to have that gem of information

Here goes. Shows a not insignificant difference in default between frequency between max and non max loans. There is a slight mitigant, that max loans (particularly at the mid grade) stay current longer. Potential for more interest payments needs to be weighed up against default frequency.



Attachment 10089

Thanks Leesal

From your chart, it does looks like the default rates of the max loans are twice the non max loans (except for F). So it is significant indeed.

As for the max loans being more current, that would be because the non max loans may be paid off early to rewrite as max loans.

Thanks again (and for the colours too

[myles]

I'd be a little cautious on those Max Loan values.

The maximums for each grade in the unique.csv are below. There are a couple of rogue (high) values that look almost like there is some flexibility in the ceiling value?

If I put in a few previous dates I get what looks like more than a couple of different sets. Not 100% sure, but it looks to me like those maximum loan amounts have changed quite a bit???

A1|79500.0
A2|79500.0
A3|77475.0
A4|77475.0
A5|77475.0
B1|79500.0
B2|77475.0
B3|55500.0
B4|55825.0
B5|55500.0
C1|55500.0
C2|56950.0
C3|45675.0
C4|55500.0
C5|44500.0
D1|34400.0
D2|34325.0
D3|34400.0
D4|33525.0
D5|33525.0
E1|22525.0
E2|22525.0
E3|23050.0
E4|22525.0
E5|22525.0
F1|11525.0
F2|11525.0
F3|11200.0
F4|11525.0
F5|10500.0

[Cool Bear]

I'd be a little cautious on those Max Loan values.

The maximums for each grade in the unique.csv are below. There are a couple of rogue (high) values that look almost like there is some flexibility in the ceiling value?

If I put in a few previous dates I get what looks like more than a couple of different sets. Not 100% sure, but it looks to me like those maximum loan amounts have changed quite a bit???

A1|79500.0
A2|79500.0
A3|77475.0
A4|77475.0
A5|77475.0
B1|79500.0
B2|77475.0
B3|55500.0
B4|55825.0
B5|55500.0
C1|55500.0
C2|56950.0
C3|45675.0
C4|55500.0
C5|44500.0
D1|34400.0
D2|34325.0
D3|34400.0
D4|33525.0
D5|33525.0
E1|22525.0
E2|22525.0
E3|23050.0
E4|22525.0
E5|22525.0
F1|11525.0
F2|11525.0
F3|11200.0
F4|11525.0
F5|10500.0

I think the loan limits are what the borrower gets in the hand. The extras are the loan fees and payment protect which we lenders fund. Although that does not explain the "overflow" of the A limit to the B1 and B2, as well as the B limit to C1, C2 and C4

[RMJH]

Good to have that gem of information

Here goes. Shows a not insignificant difference in default between frequency between max and non max loans. There is a slight mitigant, that max loans (particularly at the mid grade) stay current longer. Potential for more interest payments needs to be weighed up against default frequency.



Attachment 10089

Big thank you Leesal (and Myles). Given the picking skills of some of those in our data group there could be bias in the sample but that said your analysis does seem to confirm common sense logic that those taking every penny that they can get are (over a large sample) more likely to be in an unsustainable financial situation.

Any thoughts on the pattern showing letter grade 1's seem to have higher defaults that letter grade 2's in most grades?

[Cool Bear]

I think the loan limits are what the borrower gets in the hand. The extras are the loan fees and payment protect which we lenders fund. Although that does not explain the "overflow" of the A limit to the B1 and B2, as well as the B limit to C1, C2 and C4

Myles, sorry for my above post which just sort of repeat what you said. I did not read your post properly - early signs of senior moments.

Either HM is flexible in their ceiling values (and interest rates) or the grade presented to us was a mistake? I remember, about two years ago, a case where the interest rate charged for one loan was not correct and was from another grade. I did not follow up on that with HM and just accepted that their system is far from perfect.

[Wsp]

Has anyone looked at time of year and defaults? There has previously been speculation on this forum that loans taken out just before and after Xmas have a higher default rate.

[myles]

Has anyone looked at time of year and defaults? There has previously been speculation on this forum that loans taken out just before and after Xmas have a higher default rate.

Good question (I've been the main speculator )

First chart shows the calendar month of when loans start vs default rate. There is a definite bias to the later months of the year. June stands out, not sure why - school holidays? Perhaps not as driven by Christmas as I suspected? However, I had a quick look at grade level, and there were some grades that showed more distinct ramp up to Christmas, in particular B Grade.

Attachment 10093
Worthwhile considering what month defaults start in as well (not when the loan starts). Fairly flat with a bit of a bump in May? This could be influenced by the timing of when Harmoney started up i.e. there have been more cycles in some months than others. Slight bias in the beginning of the year influence from post Christmas debt challenge maybe...

Attachment 10092
Not as clear cut as I speculated

[myles]

Just re-looking at that first Calendar Month Started chart - when I started, I dropped $100K into Harmoney in the months of March, April and May - bloody good luck I guess?

[myles]

Any thoughts on the pattern showing letter grade 1's seem to have higher defaults that letter grade 2's in most grades?

Had seen that at some stage and didn't know what it was about, so I took the time to plot it

It appears it only occurs in A's and B's, and with their volume, they influence the whole when looking at the entire data set. Why it occurs, I have no idea:

Attachment 10095
If you're picking A's or B's, A2 and B2 certainly look like a good pick...

A bit of a ripple in the Harmoney matrix? Grading system's got something wrong in those areas...

[RMJH]

Had seen that at some stage and didn't know what it was about, so I took the time to plot it

It appears it only occurs in A's and B's, and with their volume, they influence the whole when looking at the entire data set. Why it occurs, I have no idea:

Attachment 10095
If you're picking A's or B's, A2 and B2 certainly look like a good pick...

A bit of a ripple in the Harmoney matrix? Grading system's got something wrong in those areas...

Thanks Myles. Doesn't it also occur in D's and F's? I wonder if bumping up is used as a sweetener by the sales force.