Harmoney

[Cool Bear]

This will be a very 'wide' graph. Are you wanting by grade as in A, B,...F (50 bars) or as in grade A1, A2,...F5 (300 bars)?

Perhaps it needs to be plotted differently (just a line graph)?

Will wait till the combined data comes together and then have a look.

I was thinking of one graph for each cohort (a1 to f5) but if that is too time consuming, then one overall graph with just A to F as a start

[leesal]

This will be a very 'wide' graph. Are you wanting by grade as in A, B,...F (50 bars) or as in grade A1, A2,...F5 (300 bars)?

Perhaps it needs to be plotted differently (just a line graph)?

Will wait till the combined data comes together and then have a look.

I was thinking of one graph for each cohort (a1 to f5) but if that is too time consuming, then one overall graph with just A to F as a start

You'd probably want to run it as a line graph, with a separate line for grade (A, B... F) [imagine plotting by cohort would overlap too much]. Alternatively if using a bar graph, may need to run a separate graph for each cohort.

Thanks Myles Must be interesting being able to work all that data - just under 50% of the loans out there!

[myles]

Must be interesting being able to work all that data

Haven't had much time to play

But here is the latest:

Unique Loans: 21249

Total Loan Value: $432,596,325.00

Attachment 10047

Number of loans that defaulted for each bar has been added, which is a good addition

Just adding this one as it is significantly different from what was shown in the past:

d_payment_protect.jpg
Calculated by determining the minimum date where a loan has a PP and only included loans on and after that date. More like what most expect it to be.

[Cool Bear]

Haven't had much time to play

But here is the latest:

Unique Loans: 21249

Total Loan Value: $432,596,325.00

Attachment 10047

Number of loans that defaulted for each bar has been added, which is a good addition

Just adding this one as it is significantly different from what was shown in the past:

d_payment_protect.jpg
Calculated by determining the minimum date where a loan has a PP and only included loans on and after that date. More like what most expect it to be.

I wonder if the PP are taken up more by the borrowers in the risker grades and thus have more defaults. Could you do one showing % of PP taken by the various grades? If the percentages are about even, then we can safely conclude to stay away from PP loans. But if it shows that more borrowers in the risker grades take PP then that conclusion may not be valid.

[myles]

I wonder if the PP are taken up more by the borrowers in the risker grades and thus have more defaults. Could you do one showing % of PP taken by the various grades? If the percentages are about even, then we can safely conclude to stay away from PP loans. But if it shows that more borrowers in the risker grades take PP then that conclusion may not be valid.

Probably need to plot this differently but the details are shown:

Attachment 10050

The ratios are pretty even across all grades, but more PP taken up in mid risk grades.

[RMJH]

Great work Myles, adding the sample size is very helpful. Some interesting patterns there. Have you got graphs by 6m cohort too?

[beacon]

Number of loans that defaulted for each bar has been added, which is a good addition

Just adding this one as it is significantly different from what was shown in the past:

d_payment_protect.jpg
Calculated by determining the minimum date where a loan has a PP and only included loans on and after that date. More like what most expect it to be.

Thanks Myles. Part covers defaulting more than full covers - I thought it was an anomaly just in my much smaller dataset, but not so. How interesting!

[Cool Bear]

Probably need to plot this differently but the details are shown:

Attachment 10050

The ratios are pretty even across all grades, but more PP taken up in mid risk grades.

Great work Myles. Thank you.

I think we can safely conclude that PP loans are a greater risk than non-PP loans (although not to the extent that the earlier aggregate chart shows). With partial PP performing a bit worse than full PP.

With us having to pay HM for the sales commission even for fail PP loans, that makes the loss to us investors even higher. Another factor to take into account when investing manually.

Thanks again.

[myles]

The hard part is determining if the increased income out-ways the increased risk from PP's. I've never looked at the numbers but 'trusted' Harmoney's suggested 1% gain?

Need to correct what I said earlier too, now looking again at that last PP graph:

There is more PP taken up in higher risk grades in proportion to loans in that grade (not what I said before about mid grades - that was volume of PP not proportion)...needed to be plotted differently to highlight that...but the detail is there.

[humvee]

Probably need to plot this differently but the details are shown:

Attachment 10050

The ratios are pretty even across all grades, but more PP taken up in mid risk grades.

It is kind of surprising that defaults on payment protect loans is higher, as I would have thought a number of the causes of defaults on normal loans should not cause a default - but should instead cause a claim against payment protect on loans with payment protect