Afterpay’s (APT) business model is pretty simple.  APT assumes the role of the credit provider and buys from the merchant the credit risk of the merchants paying customers. The merchant pays APT 4.2% of the transaction for this. Merchants love it as they receive their money upfront and in a credit world your average basket size goes up.

APT tries to make a profit by ensuring customers pay APT directly over 4x fortnightly instalments. APT has to hope that the cost of running their platform combined with the risk of not getting paid totals less than the payment fee it receives. If not it does not have viable business. So bad debts are an important key metric of this business.

Consider the chart from the FY2017 financials. The Income statement has a provision for bad debts of $8.2m or 1.45% of total underlying sales revenue of $561.2m. APT recovered $6.1m from their bad debts including late fees and it cost them $1m in recovery fees.

Adding these all together the net credit loss for the year was $3.1m or 0.6% of total underlying sales. So based on underlying sales their bad debt provision is 1.5%, recovery rate is 0.9% for a total credit loss of 0.6%.

Over the years I can not think of any provider of credit that has not had debts go bad on them. That is the nature of the lending business. In Australia many credit institutions have not had to face this bad credit cycle as we have not had a recession since the early 1990’s.

So I went back and had a look at bad debt levels in distressed parts of the cycle. The chart below was prepared by the RBA and was from a report on Banking credit losses between 1980-2013. The median credit loss ratio for the period was 0.34%. This is a blend of commercial, residential and personal lending.

We can see the credit loss ratio spiked to over 2% in the early 1990’s recession. It spiked by a factor of 6x from the median level. Similarly in the 2009 period the credit loss ratio spiked by a factor of 3x.

The following chart shows a quick breakdown of non-commercial lending between 2008-2013. Credit card debt runs around an average loss of 3% annually.

APT business model is slightly different to a credit card provider. APT makes its money by increasing the turnover of balances, it wants customers to pay quickly so it can recycle its capital. A credit card or a company like ZML wants the opposite. They would like their cycle to slow down so they can extract more revenue in the form of interest on outstanding balances.

APT loss provisioning at 0.6% is much lower than credit card providers at 3%. This I would guess has a lot to do with APT wanting velocity of capital turnover which assists in eliminating riskier credit as well as not having the drag of interest expense capitalising onto the principal as is the case with credit cards.

We do not know APT’s loss provisioning in the case of a turn in the credit cycle. Maybe APT customer default will be somewhere between mortgage lending and other personal lending. But we do not know.

The paradox is, the longer the debt cycle takes to turn the larger the receivables APT will have on their books if it keeps on growing the way it has. Therefore the bigger the credit loss when it occurs.

Imagine if Australia did have another recession. I know it sounds crazy. Credit default ratios could easily spike by a factor of 5x from the median level. Let’s say for APT they went up by a factor of 3x from the current bad debt level of 1.5% to 4.5%. Recovery remaining constant at around 0.9%. This would give an implied credit loss ratio of 3.6%.

The following table is a bit detailed but the key take outs are what happens to cumulative profit in 5 years. This is all hypothetical stuff but I really want to show that if you believe that this form of lending has no credit risk then buyer beware.

I have assumed that they grow customers at 3,000 a day for the next 4 years. By FY2021 they would be earning $208m in Merchant Revenue. Look what happens if Bad Debts triples on this level of underlying sales. In FY2021 (our assumed recession year) they would lose $85m.

One bad year of lending wipes out a good majority of the last 5 years. This becomes clearer when you chart cumulative profit:

Now I am not suggesting we should not invest due the possibility of a recession, in fact I own APT. But it is important to know that providing credit does have underwriting risk.

Ultimately you can not commoditise credit risk. If you make poor underwriting decisions you will pay for these. If APT have priced this underwriting too low then, like every lender in history, they will need to recapitalise their balance sheet.

Visa and Mastercard run an open loop network but more importantly they facilitate a transaction between merchant, customer and credit provider. They do not take the credit risk of the customer, rather they let other financials institutions take this risk and they charge all the players the privilege to use their payment network. They have a capital light business model which has a massive network effect.

That is why Visa and Mastercard business models are so damn good. They take a clip of the trade but they do not have to warehouse the credit risk, let that be someone else’s problem.

APT takes on the credit risk, so be careful not to confuse extrapolation with prediction when you are modelling the possibilities of bad debt provisioning.