I'm Director of Investor Relations. Joining me for today's call are Rob O'Hare, our Senior Vice President of Finance, and Ben Director, a Senior Manager on our Strategic Finance team. The purpose of today's session is to give analysts and investors a better understanding of the financial and accounting drivers of our business. We are not providing any updates to our outlook today, nor are we disclosing any historical financial information beyond what has already been publicly disclosed. Today's presentation will have three main sections. First, a walkthrough of our revenue less transaction costs or RLTC. We will also discuss potential approaches to forecasting RLTC and the main drivers of RLTC. Second, we will walk through the components of our operating expense base and the drivers for each. Within this section, we will also discuss the drivers of share-based payment expenses.
Lastly, we will wrap up by discussing some final considerations to take into account when modeling our business. In order to make this session more interactive, we will pause for questions and answers after the RLTC walkthrough and again at the conclusion of the session. We estimate that the session overall will last about an hour, depending upon the amount of Q&A. As a reminder, today's session is being live-cast, and both the replay and presentation will be available on our investor relations website after this meeting. Before we begin, I would like to remind everyone that today's discussion may contain predictions, estimates, or other information that may be considered forward-looking statements. These forward-looking statements are subject to numerous risks and uncertainties, including those set forth in our filings with the SEC. Actual results may differ materially from any forward-looking statements that we make today.
These forward-looking statements speak only as of today, and the company does not assume any obligation or intent to update them, except as required by law. In addition, today's call may include non-GAAP financial measures. These measures should be considered as a supplement to, and not a substitute for, GAAP financial measures. Reconciliations to the most directly comparable GAAP measures can be found in our earnings supplement slide deck, which is available on our investor relations website. In our RLTC walkthrough, we will cover the interplay between revenue and transaction costs and how these two come together to generate RLTC. We will also suggest two approaches to modeling RLTC, discuss the main drivers of RLTC, and address potential challenges to forecasting RLTC. You could consider two approaches to forecasting RLTC.
Under the bottom-up approach, in which your goal is to forecast individual revenue and transaction cost line items, you would begin by forecasting GMV by product. Next, you would also need to forecast GMV by funding channel to arrive at an on versus off-balance sheet mix. Once you have both the product and funding segmentation of GMV, you could then forecast revenue and transaction costs to build up your revenue and transaction cost estimates. Under the second top-down approach, you would simply forecast GMV and consider various drivers that would shift RLTC as a percent of GMV up or down compared to prior periods. This latter approach has the advantage of being simpler, though it is less detailed. Overall, we believe that the two approaches complement each other. Before we delve into the two approaches, we will first discuss the main drivers of product mix and GMV growth.
We generate GMV from three main types of products: interest-bearing installment loans, 0% APR monthly installment loans, and Pay in 4 loans, or transactions which are split into four bi-weekly payments paid over six to eight weeks. Over time, our product mix has shifted towards interest-bearing installments and away from 0% APR products. In part, this is due to the addition in late calendar year 2021 of a large enterprise partner that generally only offers monthly interest-bearing products. We previously generated a meaningful amount of 0% APR GMV from Peloton. We observed a significant reduction in GMV with Peloton as the pandemic subsided. Our GMV growth is influenced by both structural and temporary drivers, which we have listed here. We encourage you to take these drivers into account when forecasting GMV.
If you decide to take a bottom-up approach to estimating RLTC, it is also important to estimate the mix of RLTC that is retained on-balance sheet versus sold off-balance sheet. This is because funding decisions impact both revenue and transaction costs line items. We disclose the portion of loans sold during any given period in our Form 10-Q and Form 10-K. As shown here, within any given quarter, we have sold loans equivalent to roughly 35%-55% of GMV. The actual mix is driven by funding market conditions and our GMV growth rate relative to funding capacity additions. Once you forecast total GMV by product, as well as the amount of GMV retained on-balance sheet versus loans sold, you can begin forecasting individual revenue and transaction cost line items.
We monetize our business through four revenue streams: network revenue, interest income, gain on sale from loans that we sell, and servicing revenue. Due to the previously mentioned change in product mix, our revenue mix has also shifted over time. We will discuss this mix shift as well as the drivers for each revenue line item on the next slides. The two main drivers of network revenue are, first, overall GMV, and second, the product mix between pay-in-full, 0% APR monthly loans, and interest-bearing products. We discuss the impact that product mix has on network revenue as a percent of GMV on the next slide. The network revenue that we earn as a percent of GMV, also known as the Merchant Discount Rate or MDR, varies significantly by product type.
For example, long-term 0% APR loans have a volume-weighted average MDR above 10%, while interest-bearing loans have a volume-weighted average MDR of approximately 2.5%. As a result, product mix shifts can have a significant impact on overall MDR. In general, we have seen two clear trends in recent periods. First, like-for-like pricing within any given product has been stable over time. Second, overall GMV mix has shifted towards interest-bearing loans, which have a much lower average MDR. As a result, overall MDR has gradually declined due to product mix shift. You should take this into account when you forecast network revenue, but also consider that our lowest volume-weighted average MDR by product is approximately 2.5%. On this slide, we show our interest income over time and as a percentage of average loans held for investment.
The main drivers of interest income are the proportion of loans funded with on-balance sheet, warehouse, and ABS securitizations, and the blended APR of those retained loans. The mix of interest-bearing GMV versus 0% APR and pay-in-full loans will determine the blended APR of the portfolio. Similar to interest income, gain on sales of loans also reflects funding decisions we make after a loan is originated, namely, whether we decide to retain the loan on balance sheet or sell it to a third party. As an example of a way to model gain on sales of loans, you could take your estimate of off-balance sheet GMV and then make some assumptions about the gain on sale as a percentage of loans sold.
Factors that impact the actual gain on sale as a percentage of loans sold include funding market conditions, the interest rate environment, the average APR of the loans, and more. Our smallest revenue stream is servicing revenue. We earn servicing income for loans held off balance sheet that we continue to service. The main driver of this revenue stream is the absolute size of our platform portfolio held off balance sheet. To model servicing revenue, you can estimate the off-balance sheet portfolio and multiply it by some assumption about servicing revenue as a percent of off-balance sheet platform portfolio. Your estimate of off-balance sheet GMV should directly impact your estimate of the total off-balance sheet portfolio. With that, we have covered our four revenue streams and will now move down the P&L to transaction costs.
As we generate GMV in revenue, we have four transaction cost streams: loan loss provisions, processing and servicing expense, funding costs, and loss on loan purchase commitment. We will discuss the main drivers of each of these expense streams on the next slides. Overall, transaction costs as a percent of GMV have been largely stable over the past two years, but we have slightly increased in recent quarters due to the higher interest rate environment and the higher proportion of loans held on the balance sheet. Before we discuss our provision for credit losses, we would like to describe how we set our loan loss allowance and how this ultimately flows through to our provision expense on the income statement. The allowance we hold for loan losses is set at the time of the loan origination, based upon the credit risk of the transaction.
We believe that loan loss allowance as a percent of loans held for investment is the best measure of the credit risk that we are taking. As loans mature over time, some portion of them become delinquent. After 120 days of delinquency, we charge off the remaining delinquent loan balance against the allowance set aside for loan losses. As a result, the provision for credit loss expense is an output of the allowance that we set at the end of the period, plus the charge-offs during the period, less the allowance at the beginning of the period. In other words, the provision expense is downstream of our allowance. As discussed, the realized provision expense reflects loan loss expense associated with new and existing loans retained on the balance sheet.
While provision expense is generally correlated with the growth of retained loans, significant swings in expense can occur on a period-over-period basis. When modeling provision for credit loss expense, you could therefore consider beginning with an assumption about our end-of-period allowance rate as a percent of loans held for investment and solving for the provision expense. You can also check this figure against GMV, less loans sold during the period. GMV is by far the main driver of our processing and servicing expense. A secondary factor is the mix of GMV coming from a large enterprise partner, which has a revenue share agreement. If the mix of GMV originating from this partner increases, as it has over the past two years, processing and servicing expense will increase as a percent of GMV.
The main driver of funding costs is the dollar amount of average warehouse funding debt plus on-balance sheet securitization notes. We have a mix of fixed and floating-rate debt facilities. Our warehouses are floating-rate facilities, mostly tied to SOFR, while our ABS securitizations are structured with fixed borrowing costs. Our average funding cost is a function of benchmark interest rates plus credit spreads. Over the past year, both have increased due to increases in benchmark rates, plus increased volatility in the funding market, which has led to higher credit spreads. Loss on loan purchase commitment is an accounting concept which applies primarily to 0% APR loans that are originated by a bank partner like Celtic Bank. We have a contractual obligation to purchase all loans originated on our platform by our bank partners at par.
When we purchase a 0% APR loan at par, we book a loss on loan purchase commitment to reflect the GAAP discount required to reduce the loan's carrying value down to reflect the appropriate fair value at the time of origination. The main driver of loss on loan purchase commitment is therefore the amount of 0% APR loans originated in a given period, multiplied by some assumption about the markdown required at the time of origination. As an alternative to forecasting individual revenue and transaction expense line items, you could consider a top-down approach to forecasting RLTC. Using this approach, you would forecast overall GMV and then make an assumption about RLTC as a percent of GMV to produce your RLTC forecast. At a high level, there are two categories of factors that influence RLTC as a percent of GMV.
The first category includes Affirm and BNPL industry-specific dynamics. These include the impact of our pricing initiatives, product and merchant mix shift, and product efficiency improvements. Overall, we estimate these factors were a slight tailwind during the first three quarters of fiscal year 2023. The second category includes macroeconomic factors such as consumer spending trends, funding costs, and the consumer credit environment. Holiday buying seasonality is another important consideration, especially when forecasting quarterly results. In total, we estimate that macroeconomic factors were a significant headwind to RLTC as a percent of GMV during the first three quarters of fiscal year 2023. To recap, we have outlined two approaches you could consider when forecasting RLTC. Under the first approach, you begin by forecasting GMV by product and also by on versus off-balance sheet mix.
You estimate each individual revenue and transaction cost line item to arrive at an estimate for RLTC. Under the top-down approach, you forecast total GMV and focus on the factors that could shift RLTC as a percent of GMV, higher or lower versus prior periods. Overall, you should take into account several considerations regarding RLTC. First, macroeconomic factors have been a significant headwind to RLTC as a percent of GMV during the first three quarters of fiscal year 2023. Second, Affirm-specific factors have been neutral to a slight tailwind during that same period. Third, our pricing initiatives, particularly 36% APR, will take several quarters to achieve maximum benefit. Finally, RLTC, as a percent of GMV, is highly seasonal, with the second fiscal quarter being seasonally lowest.
Well, we've covered a lot of material, we promise that the next section will be shorter. With that, we'd like to pause here and begin a question and answer session for questions specifically relating to RLTC. For those of you participating in the webcast, you can either raise your hand and ask your question live or submit the question via the Q&A box. You can also email us at ir@affirm.com. We will pause for approximately one minute to collect questions.
Great, everybody. Thanks again for joining. We're going to go ahead and begin the Q&A session for RLTC. If you have any questions, please submit them via the Q&A box, or alternatively, you can email us at ir@affirm.com. We're going to begin with one of the questions submitted via the Q&A box. James Faucette from Morgan Stanley asked: How predictable has the mix of products become for you internally? How long is your visibility into when or which merchants may offer 0% , Pay in 4 or interest-bearing products? Rob, do you want to go?
Sure. Yeah, thanks for the question, James. you know, for the most part, merchants don't make significant changes to their financing programs on sort of a month-to-month basis. We do have some large merchants that will use things like 0% offerings on a promotional basis, and we have done some work internally within the portfolio to move some programs that we're paying for only to be Adaptive Checkout, meaning it's a Pay in 4 offering and typically a monthly installment offering alongside. There's been some structural changes that we've made internally, but for the most part, it's pretty constant. Obviously, we have a wide variety of merchants that are growing at different rates, the mix will fluctuate a bit given that dynamic as well.
Great, thanks for the question, James. Let's turn to one of the live questions. Eugene from MoffettNathanson, would you like to go ahead and ask your question?
That did it also. I didn't, I didn't know which way is better. Can you talk a bit about the best way to forecast charge-offs? Because I'm thinking about the credit allowance calculation, and I appreciated the point that you guys made that, you know, best way is to think about the allowances percent of loans and then kinda calculate the expense from that. The other piece is charge-offs, if I'm thinking about it correctly, and I was just hoping you guys can speak to that a little bit.
I think just given the nature and the timing of when charge-offs occur on our platform, they happen 120 days after the first delinquency. For the most part, delinquencies tend to be front-end weighted within a loan, so relatively close to the origination date. I would look, I would sort of build a cohorted view of on-balance sheet GMV and then time lag it by something like 150, 180 days. I think you can see a bit of a clear relationship if you were to sort of play with the charge-offs that have been reported versus the GMV that was originated.
Hmm. Got it. Okay, thank you.
Great. I think we'll go ahead and turn it over to Mike from Goldman Sachs. Mike, are you there? Going once, going twice. Great. Any other questions about RLTC, ladies and gentlemen? If not, we can go ahead and move on to operating expenses. You will have a chance again at the end of the operating expense section, at the end of the conclusion section, to ask additional questions if you would like. You can ask about RLTC or anything else you'd like, but for now, I would like to focus on just RLTC, unless you have any questions there. Looks like Eugene raised his hand one more time, so let's try that.
I'll throw in one more high-level question. Can you talk a little bit about, and I think it's pretty simple, I just wanted to confirm. When the share of the loans you keep on off balance sheet changes, how does that impact RLTC? In other words, when you're keeping more on the balance sheet, does that automatically, all else being equal, make RLTC higher?
Yeah.
5% from JV. Yeah.
Maybe, it's a bit of a nuanced answer. Thanks for the question. You know, I think you should think about it in terms of maybe RLTC in the period of GMV origination. Then I think you should also think about, because we're, for the most part, originating, you know, 12-ish month loans, you should also think about RLTC on more of a lifetime or a horizontal basis. When we keep the loans, because we're not sharing the economics with a loan buyer, we that's obviously a more profitable activity for us. In the long run, retaining more loans will increase our RLTC margin or RLTC as a percentage of overall GMV.
However, given the timing and the mechanics of the accounting for retained loans, we do book what is one of the largest expenses against a loan, which is putting that allowance, that initial allowance on, and that'll flow through as provision expense. That dynamic does create a bit of a J-curve, where in the period of origination, a monthly installment loan of, say, 12 months, for example, would potentially be negative RLTC in the quarter of origination, even though it's likely to be a very profitable loan, you know, if held to maturity over a 12-month period.
Got it. Okay, thank you.
Great. I think we had a question from... Let's see if I can find him again. John Coffey from Barclays. Go, John. John, are you there? If not, I think he also submitted the question via email. We also received a question via email from a different questioner. For this, how does the 2026 senior convertible note buyback impact forecasted share count? Really a question more for the second section, but, Rob, do you mind answering that? How does it impact the forecasted share count, or does it?
Sure. I typed in an answer in the Q&A section, but, just for folks on the call or on the phone, we really haven't forecasted share counts that assume a conversion of the convertible bonds. Like, the conversion price for the bonds is roughly $215 per share, we're not anticipating today that those bonds would convert into shares. There would be about, I think, 8 million shares against the bonds, based on the original issuance size. We have bought back a bit of the face value in the March quarter. Short answer is, they're not anticipated to convert and therefore aren't included in any of the share count forecasts.
We got another question via email. Again, this is from John Coffey at Barclays: Can you walk through the quarterly cash flows from 0% APR loans as well as interest-bearing loans?
Sure. Again, I think the, the cash flows are going to depend a bit on the term length of the loan, right? We, we do range on the monthly installment side from three months on the short end, to 60 months, on the long end. Just taking, maybe a 12-month example for both 0%, and interest-bearing, you know, in a 0% loan, right, it's 0% APR to the consumer, obviously. All of the cash compensation that comes to Affirm is coming from the merchant relationship. In terms of the cash flow, obviously, there would be the cash outlay for the loan itself, so the loan amount would go out the door to the merchant.
We would then net the Merchant Discount Rate, the fee that we collect from the merchant, against that principal balance. We would settle that net. Then, you know, we would book an allowance against that loan. Again, that's a non-cash charge for us, so not necessarily a cash flow hit. Then we would just collect principal repayments from the borrower over the 12 months of term. So assuming the loan pays off, we would just collect the full principal amount. You know, if you were to take a cohort view, there would be some losses, typically in a loan cohort, so that would just reduce the amount of repayments that we would collect from the consumer. The other sort of cash charges that would show up as we're making those...
As we're collecting those repayments from the consumer, would be things like payment processing costs, right? Whether the consumer pays us back with an ACH bank transfer or if they use a debit card, right, those all have different cost profiles for each of the repayments that are made. Then also, you know, any sort of customer support burden. You know, we do run call center support for our borrowers, so we allocate a charge to the loan book for those activities as well. Those tend to be, you know, sort of semi-variable. They're a bit more fixed and, you know, sometimes difficult to allocate down to a loan-by-loan basis, those would be some of the other transaction costs that would show up.
That all assumes, of course, that we're holding the loan to maturity. If we weren't holding the loan to maturity, in a 0% example, Oh, sorry, the piece that I missed is that we hold the loan at fair value, there's also an upfront non-cash expense that we call loss on loan purchase commitment, which Zane walked through. That would help build a discount that would set the fair value for the loan. If we were to sell that loan to a third-party loan buyer, the gain on sale would be booked against that reduced fair value.
Sorry, the other sort of expense that I missed on the loan that's held to maturity would obviously be the financing cost for the loan itself. We typically see advance rates for on-balance sheet financing that's about $0.80 on the dollar against the principal. There would be a charge for the funding costs against, you know, a subset of that balance. For an interest-bearing loan, a lot of the mechanics are similar. Typically, we would receive a Merchant Discount Rate upfront in cash, so that would be a cash inflow. Again, that's netted against the settlement to the merchant when funds are disbursed for the loan amount at the purchase time.
Then we would hold the loan to maturity, we would receive a bit different than the 0% example. We would receive both cash and principal payments from the borrower. Those are the way our loans work. There's typically a monthly amortization. It's a fixed payment amount, so the first repayment has a larger proportion of interest than principal, and then the last repayment would have more principal than interest, or the way that a mortgage works. The balance would pay down on a month-by-month basis. We would have funding costs against the principal balance that's outstanding, similar to the 0% example.
You know, again, assuming, depending on what you assume for loss rates, the loss rates would show up in on a cash basis, they would show up in reduced repayments. On an accounting basis, The expectation for losses against the loan would be all booked upfront as a non-cash charge.
Great. Thanks for that very comprehensive answer.
I think we'll have time for maybe two more questions, so we'll try Mike again at Goldman Sachs. If Mike isn't available, we'll go to John Hecht at Jefferies. Mike, your line is open if you want to ask it. Looks like he's not there. Okay. John Hecht from Jefferies?
Guys, question about getting a little echo, so I apologize if I sound a little discombobulated. It's the gain on sale margins. Obviously, one component of modeling that is the mix of loans sold versus retained. The other is going to be the gain itself, which is predicated on interest rates and, interest rate, benchmark interest rates, interest rates in the marketplace, interest rates on the loan, and so forth. Taking all those considerations, h ow do you think we should account for that in the model?
Yeah, I mean, I think we showed the trends in terms of gain on sale versus, I believe we showed it against the loan sold, the average loan sold balance. You know, so obviously there has been a compression as we've seen rising rates, John, to your point, right? The yields that our loan buyers need to achieve have gone up, and as a result, the gain on sales pricing needs to go down for them to achieve those yields. I would say, the offset that we expect to help sort of alleviate that downward trend is really sort of getting more economic content into the loans themselves.
The way that we expect to do that is through the continued rollout of this 36% APR cap that we've been speaking about for the last couple of quarters in our shareholder letters. If there are, you know, higher coupons or more economics in the loans, that will help us sort of maintain or even raise loan sale pricing as we sort of reset to this new rate environment.
Do you generally use the forward curve as the basis for your guidance when considering gains?
Yeah, that's right. you know, it's not, I would say because these are, you know, bespoke, bilateral agreements, it's not, you know, it's not always a truly floating rate assumption for the gain on sale, but it's informed, of course, by the forward curve, and also informed by just the relationships and the color that we're getting from our capital team, which owns, you know, those relationships with the loan buyers internally for Affirm.
It looks like we've answered all the questions we had in the queue. As a reminder, if you would like to ask additional questions, you will have an opportunity to do so at the conclusion of today's session. We'll leave another 15 minutes or so for Q&A.
Now that we have covered RLTC, we will explore the three components of our operating expense base and the drivers for each component. Because we tend to look at expenses on a non-GAAP basis internally, we will begin with non-GAAP expenses, but we will later touch on the reconciliation to GAAP expenses. Our two largest operating expenses are tech and data analytics, and general and administrative. Due to our efficient go-to-market motion, sales and marketing expenses are relatively low as a percent of GMV and have declined in absolute terms in recent quarters.
We have exhibited the most operating leverage on the sales and marketing in G&A lines, in part because these expenses are not directly correlated to GMV or the number of active consumers. Tech and data analytics spend has scaled more directly with GMV. This is because some of the tech and data spend is related to infrastructure spend, such as cloud computing providers, as well as consumer credit reports that scale with our loan application volumes and the number of active consumers. As many of you know, we have substantial GAAP operating expenses tied to share-based payment expenses. These share-based payments are primarily tied to two categories. First, employee stock compensation expenses, including our CEO Value Creation Award. Second, enterprise warrant and commercial asset expenses associated with equity and warrants that we provided to two enterprise partners as part of our commercial agreements with them.
We believe that the accounting for share-based payments is one of the more misunderstood parts of our business. Under GAAP, share-based payments are added back to shareholder equity. The main impact of share-based payments is dilution, rather than any change in overall shareholder equity. With that in mind, we do not expect to see substantial declines in shareholder equity over the coming years. Since the third fiscal quarter of 2021, our shareholder equity has actually increased from $2.4 billion-$2.5 billion, despite us incurring substantial net losses. In our most recent quarter, shareholder equity slightly increased sequentially, even as we recorded a $206 million net loss. Another complicated part of our business is our enterprise warrant and commercial asset expense.
These expenses stem from commercial agreements we have with two enterprise platform partners. As part of the commercial agreements, the partners agreed to accept Affirm as a payment option, and separately, also agreed to a certain exclusivity period. We remain in the commercial agreement period for both partners, even though our exclusivity with one partner ended in January. For both partners, in exchange for entering into the commercial agreement, we effectively granted the partner a block of our shares, known as a commercial asset. Under GAAP, we take the fair value of the commercial share grant at the time of the issuance and amortize it over the life of the commercial agreement. This is a fixed expense over the life of the agreement. For the enterprise warrants, we take the fair value of the warrant at the time of grant and amortize it as certain performance-based conditions are achieved.
As a result, the enterprise warrant expense is more variable in nature. Importantly, we expect total enterprise warrant and commercial agreement expenses to decline substantially as we exit fiscal year 2025, or approximately two years from now. This should lead to a substantial reduction in our GAAP share-based payment expenses. The last component of our share-based payment expense is employee stock compensation, including our CEO Value Creation Award. We would like to emphasize that we set stock-based compensation based upon an annual share dilution target. We do not target a certain level of GAAP stock-based compensation expense. One component of stock-based compensation expense is the CEO Value Creation Award that we granted at the time of our IPO.
This award consists of stock options with a $49 stock price, strike price, excuse me, that are earned when our share price exceeds certain thresholds for a 90-day period. Importantly, additional stock options under the CEO Value Creation Award will only be earned if the 90-day trading day VWAP of our share price exceeds $132. We have reached the conclusion of our financial model information session. Before we go to Q&A, we would like to mention some final considerations that you might take into account when modeling our business. You could consider focusing on forecasting RLTC rather than other income statement or balance sheet items. This is because RLTC has a significant impact on our operating income and can be complex to model. When forecasting RLTC, you could consider using both bottom-up and top-down approaches.
You could also consider taking both the Affirm-specific and macroeconomic factors that affect RLTC as a percent of GMV into account. Second, we walked through some of the recent drivers of our non-GAAP operating expenses. In general, we have demonstrated good non-GAAP sales and marketing and G&A expense leverage. On the other hand, tech and data analytics expense has scaled to a greater extent with GMV and customer growth, and therefore we have delivered less operating leverage on this expense line item. Third, with respect to our share-based payments, we target a certain amount of annual share-based dilution. We do not target a certain amount of GAAP share-based compensation expense. Additionally, the incremental vesting of the CEO Value Creation Award options is subject to high share price thresholds, with the lowest threshold being $132.
Finally, we have a substantial amount of share-based payments associated with our enterprise warrants and commercial asset agreements. We do expect these expenses to decline substantially towards the end of fiscal year 2025.
Thank you all for joining today. We hope you found the session informative, and we will now move to our final question and answer session. For those of you participating in the webcast, you can either raise your hand or ask your question live, or submit the question via the Q&A box. You can also email us at ir.affirm.com. We will pause for approximately one minute to collect questions.
Great. Well, thank you again. We're back to our Q&A session. We did receive a few questions via email, and we also have at least one attendee with his hand up. The first question we received via email was from Mike at Goldman Sachs. He asks: First, can you please remind us what types, loan types are self-originated versus originated via bank partners? Second, can you talk about the impact of self-originated loans at a loss that are accounted for as a contra revenue to MDR? Is that a material amount?
Ben, did you want to take that one, actually?
Yeah, I'd be happy to. On the first piece, just around which type of loans are self-originated versus bank partners, basically anywhere we legally have to go through a banking partner. Essentially, in our U.S. business, all of our installment lending comes through our originating bank partners. All of our Pay in 4 loans. Those are the shorter, kind of six to eight-week term length, zero interest loans are actually self-originated out of our own fully owned subsidiaries. All of our international lending is done by a self-originated arrangement. In terms of the second piece around kind of which are recognized as contra revenue, that is largely immaterial right now. It's basically only on the Pay in 4 and our international volume, which is quite a bit smaller.
You can think of that somewhat as related to what we would need to do for the loss on loan purchase commitment if we were to originate ourselves.
We received a question in the Q&A box from [Dick Shagira]. Is it fair to look at the interest income from an interest margin perspective, meaning interest income minus funding costs as a percent of GMV or loans? Rob, do you want to take that one?
Sure. I think that's totally a fair way to look at the business. We tend to look at the business on a bit more comprehensive basis, where w e factor in all of the consumer interest, but also the fees that we receive, either from loan buying partners or from the merchants as well. We tend to look at the total revenue against the total transaction costs, but looking at sort of how the yield or the net yield has trended, where you're looking at consumer interest against the funding cost, I mean, that's completely appropriate as well.
Great. Thank you. Looks like we have a question from John Coffey again, I think it was. Let me see if I can find him. John, your mic is unmuted now, or it should be able to unmute at least, and ask your question.
Yeah. Are you able to hear me this time?
Yeah, yeah.
All right, great. Perfect. I was just wondering if you could walk me through your thought process about what goes on the balance sheet, what goes off. Is it a simple decision? I just kind of wonder what factors go into that decision.
Sure. Yeah, this is a question that we get quite often, so thanks for asking it here. You know, I would say that there does tend to be, sort of a predetermined mix, that ranges by loan product type, I would say. The easiest example is, you know, a Pay in 4 loan that is six or eight weeks long in duration, probably has a three or four week weighted average life. We typically don't sell those loans at all, and so they tend to stay on balance sheet because, there's not much capital required to support a pretty sizable loan book, right? The loan, a Pay in 4 loan would turn over, you know, roughly 17 times a year.
We tend to keep those loans on balance sheet, and they sit in a warehouse facility. You know, I would say on the opposite end of the spectrum, long-dated, 0% APR loans from a consumer. You know, some of our programs go out to 48 or even 60 months on the 0% side. We tend to sell the majority of that volume because the only compensation that we're typically receiving is from the merchant that's collected up front. It doesn't make sense for us to sort of hold those loans to maturity over a three or four or five-year basis. We tend to sell a very high proportion of that, the long-dated 0% loans.
I would say, you know, for our third loan product, interest-bearing, it's gonna be a function of sort of the forward flow or the loan buying capacity that we have. We typically know, where a loan's gonna go, you know, at the time of origination. We're not pooling, loan originations and then making an allocation decision later. We have a pretty regular allocation process to our loan buyers, and, you know, we keep a close eye, obviously, on the capacity that's available as the loan balances that our loan buyers are holding, you know, amortized down on a daily or monthly basis. It's a pretty mechanical process in terms of allocating a subset of the book, on the interest-bearing side to loan buyers. We do that, you know, on a weekly basis.
John, did you have any other questions?
No, that was it. Thank you very much.
Okay, thank you. We'll pause for maybe another minute, see if there's any last-minute questions. Looks like Mihir Bhatia, let me find where your hand is here. Give me one second, Mihir. Apologies. Mihir, your line's open. Go ahead. You need to unmute yourself.
I did. Okay. Sorry. I just wanted to Rob, can you clarify that last answer a little bit? I guess the part... If you know at the time of origination where the, I guess, where the loan is going, how does that, I mean, I guess what I'm struggling a little bit with is, it seems like there are times when you know, whether you pull a securitization or what have you, because market conditions are volatile, and we get that, right?
Mm-hmm.
What, in one of those situations, what happens? Like, you need it seems like you knew where the loan was going, but then you had to pull it. Does that result in, like, any like, what is the impact or something like that?
Sure. Yeah. You know, for the most part, Mihir, and thanks for the question. For the most part, when we're choosing to launch a securitization, we're typically pulling loans out of a warehouse and then pledging them to that new securitization trust, right? You know, that's one of the reasons why we don't do a securitization, you know, every week, right? We need to sort of pool the right type of assets for our various ABS shelves to make sure that we have critical mass to sort of get to, you know, the minimum required deal size. Sometimes we will build a bit of extra capacity if we have an ABS deal on the horizon. We'll build a bit of extra, or sorry, capacity is not the right word, but we'll hold some more loans in on-balance sheet warehouses as we're preparing to move those loans into an ABS deal.
Right. I guess, like, the question is just, when you make the origination, you have a good idea that eventually this loan is gonna end up in a securitization versus this is going to a forward flow partner, this is going on balance sheet. Is that the right way to think about it?
That's right. Yeah. I mean, you know, if you think about our various ABS shelves, right? We have the A, B, and C deals, which tend to be primarily interest-bearing, but also have a little bit of split pay, and a small amount of 0% loans as well. Then we have the longer-dated, off-balance sheet securitizations, our Z and X deals, which are both longer-term interest-bearing and then longer-term 0% offerings. We typically have some loans sitting in warehouses for each of those loan types. If we know, like I said, that a deal is on the horizon, we'll go a little bit more balance sheet heavy leading up to the deals.
In the event that, you know, a deal is pulled, which did happen last calendar year, we typically keep, you know, ample capacity in the warehouses such that it doesn't put any, you know, undue stress on the warehouses themselves.
Understood. Thank you.
Great. Thank you for the question, Amir. We received a question via the question and answer box. The attendee asks: Could you help us get a better sense of the non-GAAP tech and data analytics spend that is headcount versus non-headcount related? Also, to what extent, I believe the question is asking, what extent are those two sources correlated to GMV?
You know, I don't have the exact split of the dollars that are headcount versus the dollars that are non-headcount. I don't have that at hand. Apologies. I would say, of our three expense line items, tech and data analytics is the line item that has the most non-headcount spend in it, at least in recent history. There are things like our cloud services infrastructure, which is variable with GMV and with applications, and some other sort of infrastructure and data costs that sit in there, too, that tend to be more variable than fixed.
Great. Thanks, everyone. let's go for a live question now. I think we have David from Wedbush. I'll go ahead and unmute. go ahead and unmute yourself, David, if you can.
Yep, thanks. Not sure if you're answering questions such as this, but curious about the gain on sale margin. Looking on slide 14, it looks like it's been coming down over the past few quarters. Can you talk about, you know, how we should think about gain on sale margin pressure going forward? I guess timely with the Fed announcement today, it looks like there could be, you know, a few more hikes in the dot plot. Could you talk about any potential pressure you could see on the gain on sale margin from that?
Yeah, we received a similar question earlier. I would say that, you know, the repricing efforts that we're doing on the consumer side to move the APR cap from 30%- 36% should help us get more economic content into the loans that we're selling, and that should help us either maintain or in some cases, increase the gain on sale pricing that we're receiving from third-party loan buyers. You know, at the end of the day, our loan buying partners are solving for a yield on their side, if there's more economic content in the loans via a higher interest, that allows us to sort of keep a bit more and sort of increase or maintain the gain on sale pricing.
That, that's another area where the repricing work that we're doing, you know, should show up. There should be a benefit there, as well as obviously showing up for the retained loans in terms of more interest income flowing through the P&L as well.
In terms of the rate environment, you know, I think, obviously, the reference rates and the. That's one input on the loan buyer side. I think they're also looking at things like credit losses, and so, as we've been running with reduced credit losses, that helps our loan buyers make their yields. By managing credit effectively, you know, and I think there is some obviously correlation or inverse correlation between interest rates and unemployment and in turn, you know, credit performance. If we're able to continue to manage credit the ways that we have historically, that'll also help us maintain our loan pricing or even potentially improve it, given the last couple of quarters' results.
Great. Thanks very much.
Thanks for the questions, David. We received a few more questions in the Q&A box, the first one is from Matthew O'Neill at FT Partners: With the merchant repricing, how can we effectively model the impact of the progression of interest-bearing volume on the 36% APR cap? Are there any other components to the initiative that will provide a material impact?
Hopefully, I'm tracking the question. I think the question is basically: When will we start to see the benefit of the consumer interest changes that will occur as we move the APR rate cap from 30%-36%? You know, I would encourage you to look at the table that we included in our last shareholder letter. I think it will take a bit longer for the loan book to fully reflect the higher rate caps. We showed a table with sort of two lines. One is the percentage of originations that are running under the new 36% cap.
Obviously, that's the part of the business that's going to see the increase first, but it's going to take a while for those new originations that we're doing, to show up fully in the back book or the platform portfolio. Eventually, those two lines will converge, but it will take. You know, we've got a weighted average life of roughly five months, so it'll take a while, for the impact to show up. We also have a ways to go in terms of getting the majority or the remainder of our interest-bearing portfolio up to the 36% cap. We're at about half, as of the end of the March quarter, so some more work to do on that front as well.
Thank you. We also received another question, again, via the Q&A box. Actually, there's two questions. I think they're closely related. The attendee asks: Do you anticipate there to be a margin uplift once rates start to go down next year or after? Related to that, would you expect the APR to adjust down soon after that? I think presumably, meaning consumer APRs. Will those also begin to adjust downward as a result of a lower interest rate environment, or will they stay elevated?
Yeah, I don't want to get into guidance beyond the end of fiscal 23. In terms of, you know, sort of the elasticity of the APRs, in the context of the macro rate environment, we are trying to establish the new APR caps at 36%, really as a cap. It should be at Affirm's discretion to reduce rates to the consumer if we deem that appropriate. I think it's going to be a function of where rates are, but it's not something that we would expect mechanically our consumer APRs to step down if rates were to start to decline.
Great. Thank you for that, and thank you for the questions. It looks like we've reached all the questions, unless there's any last-minute ones in terms of people raising their hand or submitting them via the Q&A box. With that being said, thank you all again for joining us today. If you have any feedback, questions, we certainly welcome it. This presentation is already available on our website. A replay of this session will also be available on our website shortly. Of course, we're always here if you need our help, so you can reach us at ir@affirm.com. Thanks again for your interest in Affirm today, and we hope to see you again soon in August when we report earnings. Have a good day.
Thanks, everyone.