CaM - I normally wouldn't share things I've prepared for my own benefit but given interest in the topic I've attached a few workings from my files.
Below is a snapshot that includes Heartland's actual incurred net credit losses as a % of average interest earning assets. Within a banks P&L there is total impairments, which consists mainly of the sum of the ACTUAL receivables written off during a period (net of recoveries), and then the movement in provision which can be a bit chunky as things are provided for and released. Bad debt expense as a % of ave assets dropped 29bps, and in relative terms improved 57% (-.22%/-.51% less 1) on FY20 results. Actual credit loss expense in 1H FY23 improved on FY21 and FY22 (when conditions were extremely benign), and in a macro environment where interest rates and inflation are high and the economy contracted in the fourth quarter. I think this is a remarkable achievement all things considered.
Attachment 14523
In my view much of this is due to the improved credit mix of receivables. For instance, HGH wrote P2p personal loans via Harmoney (who have now discontinued facilitating P2P loans) with those receivables peaking at 5.3% of total gross credit exposure in 2020. When P2P loans were peaking, while they only represented ~5% of total receivables they contributed 30-40% of total heartland impairments during those periods. Personal lending is now down to 0.9% of total exposure. Reverse mortgages and mortgages which are relatively low risk have also increased in % of the book, from 27% at the start of 2018 and are now 35.3% now.
I saw a great bit of analysis years ago from a broker looking into the mix of HGH receivables but they've never updated it so I went back and recreated it, updated it and made it my own. It basically looks at each type of credit exposure, assigns a relative risk grade (IE how they compare to each other, but not in an absolute sense), and then comes up with a weighted average index. It shows it has improved and continues to improve. Some of the weightings are subjective but none the less its a useful framework. IE personal lending clearly if on a 1-5 scale is a 5, RM's a 1, mortgages a 2...rural more difficult to consider but conservatively assigned it a high/4 risk weighting. P2P lending I've shown as a 5 (as wanted to keep it 1-5) but in reality its probably more like a 7. It's all highly subjective but I thought it was a (semi) interesting framework, though I think the real story is simply just the % mix of personal lending vs RM and mortgage lending.
Attachment 14524
NZ reverse mortgages have a weighted average LVR of 19.7% (0 loans over 75% LVR) and an average borrower age of 78. Australia's are much the same: weighted ave LVR of 20% (only 3 loans over 75% LVR) with an average borrower age of 77.
Looking at the relatively small mortgage book (4.3% of total exposure), as best I can tell by looking at the HBL disclosure statement, even after last years decline in property prices, 99% of the mortgage book was below 80%. Only 1% was above 80%, and 0% above 90%. They don't provide information below the 80% so dont know what the weighted average LVR is.
So I personally believe HGH have made strides in improving the credit quality of their book, evidenced in the proportion mix of high risk and low risk lending, and credit losses / bad debt expense as a % of assets.
On 30 June 2022 HGH provided $8m towards a specific macro conditions provision. Last year HGH had total impairment expenses of 13.8 - consensus forecast is for that to rise to 18.7 in the current year. I'm not brave enough to wager a guess on what it'll ultimately be.