車美仕 (KMX) 2017 Q4 法說會逐字稿

Good morning.

My name is Victoria, and I will be your conference operator today.

At this time, I would like to welcome everyone to the CarMax Fiscal 2017 Fourth Quarter Earnings Conference Call.

(Operator Instructions) Thank you.

I would now like to turn the call over to Katharine Kenny, Vice President, Investor Relations.

Thank you, and good morning.

Thank you for joining our fiscal 2017 fourth quarter earnings conference call.

On the call with me today, as usual, is Bill Nash, our President and Chief Executive Officer; and Tom Reedy, our Executive Vice President and CFO.

Before we begin, let me remind you that our statements today regarding the company's future business plans, prospects and financial performance are forward-looking statements that we make pursuant to the safe harbor provisions of the Private Securities Litigation Reform Act of 1995.

These statements are based on management's current knowledge and assumptions about future events that involve risks and uncertainties that could cause actual results to differ materially from our expectations.

In providing projections and other forward-looking statements, the company disclaims any intent or obligation to update them.

For additional information on important factors that could affect these expectations, please see the company's annual report on Form 10-K for the fiscal year ended February 29, 2016, filed with the SEC and soon to be replaced with our new 10-K.

(Operator Instructions) I also want to alert you to the fact that Richmond is having pretty lousy weather this morning.

So we're hoping we stay online and plugged in.

But if not, we have, hopefully, a back-up plan.

We'll see what happens.

Bill?

Okay.

Thank you, Katharine, and good morning, everyone.

As usual, I'll start off by reviewing key highlights for the quarter and then turn the call over to Tom, who will cover financing.

I'll conclude with an update on our growth and strategic initiatives.

As you read in our press release this morning, our used unit comps for the fourth quarter increased by 8.7%.

In total, used units grew by 13.4%.

Total used unit comps were driven by strong improvement in conversion as well as by a modest increase in store traffic.

As you know, we track market share data on a calendar year basis, and our data shows that in our comp markets, we increased our share of 0- to 10-year-old vehicles by approximately 2% in calendar 2016.

Our core business remained very strong as we estimate used unit comps for our non-Tier 3 customers grew 15.3%, which is the highest comp we have recorded for our non-Tier 3 business in many years.

But we again saw a headwind from Tier 3 sales.

Similar to last quarter, we believe there were a number of reasons for our growth in unit sales.

Strong store execution continued to benefit from enhancements to the online customer experience that we've made throughout the year, including capabilities like the new online finance prequalification product.

Website performance continues to support sales.

Total web traffic grew 3% over the prior year's quarter.

As we've previously said, one of our goals for the website is to drive more leads, which we continued to do this quarter.

Gross profit per used unit was $2,134 compared to $2,109 in the fourth quarter of last year.

This is the 24th consecutive quarter that we've managed to gross -- we've managed to a gross profit dollar per unit of between $2,100 and $2,200.

We continue to achieve this level of dollar GPU despite fluctuations in used vehicle prices and a challenging competitive environment.

The growing supply of off-lease vehicles and its impact on our business seems to be a hot topic.

As we've previously said, we continue to believe that increasing supply is a good thing for CarMax's business.

We believe lower prices will be beneficial to sales as cars are more affordable for our customers.

We have consistently demonstrated that we're able to manage our dollar gross profit per unit regardless of the pricing environment, which includes times of sudden price decline and other periods of high off-lease volume.

We believe that our inventory management system is a significant competitive advantage, especially during periods of fluctuating prices.

Our wholesale units declined by about 1% in the fourth quarter.

The growth in our store base and a higher buy rate were more than offset by lower appraisal traffic.

We believe that late tax refunds also impacted appraisal traffic.

As we have discussed in previous quarters, we continued to see a lower supply, 7- to -- a lower supply of older 7- to 9-year-old vehicles that correlates to the years of decline in industry new vehicle sales during the recession.

As this new bubble moves into older vehicles, we would expect wholesale unit sales to normalize.

Gross profit per wholesale unit decreased to $938 compared to $1,005 in last year's fourth quarter.

We believe this was due to several factors, including the delay in tax refunds; the decrease in 7- to 9-year-old vehicles, which are some of our more profitable wholesale vehicles; and a tough comparison from last year's fourth quarter, when wholesale gross profit per unit was unusually high from a historical perspective.

A few other topics before I turn the call over to Tom.

As a percentage of our sales mix, 0- to 4-year-old vehicles was approximately the same as last year's fourth quarter at 77%.

As a percent of sales, large and medium SUVs and trucks rose by over 3 percentage points to 28% in this fourth quarter.

Now on SG&A, expenses for the fourth quarter increased 15.4% to $385 million.

This growth was due to a variety of factors.

First, it reflects the 13%, or 20-store increase in our base since the beginning of the fourth quarter of last year.

It also incorporates a $12 million increase in share-based compensation expense, which relates, like every quarter, to nonexecutive compensation units that are settled in cash.

Just as a reminder, share-based compensation expense is driven by the change in CarMax's stock price during this quarter versus the change in the previous year's quarter.

In this case, it was based on an increase in price of about $7 in this fourth quarter versus a decrease of $11 in last year's fourth quarter.

Some other factors that drove SG&A included the increase in variable cost due to our higher level of sales, spending related to our strategic initiatives and higher advertising costs.

Remember that our advertising expense last year was higher in the third quarter and lower in the fourth quarter due to the timing of our new brand launch.

While we reported deleverage of approximately $39 per unit, this includes a $61 per unit impact of share-based compensation expense.

Now I'll turn the call over to Tom.

Thanks, Bill.

Good morning, everybody.

In the fourth quarter, we saw the same trends in customer flow that we've been discussing all year.

We continue to experience increases in credit applications from customers at the higher end of the credit spectrum and less applications at the lower end.

We believe that Tier 3 volume in the fourth quarter was also impacted by a delay in tax refunds, which as you may have seen from IRS data, was about 10% lower year-over-year in late February.

This mix continued to drive opportunities for CAF and growth in sales where customers paid cash or brought their own financing to the table.

CAF net penetration increased to 43% compared to 42% in last year's fourth quarter.

Net loans originated in the quarter rose 13.2% year-over-year to $1.4 billion.

This was due to a combination of both CarMax sales growth and despite some tightening of credit standards versus last year, higher penetration driven by the better credit mix.

These are partially offset by lower average amount financed.

As you saw in today's press release, we have included information on financing penetration by channel, including CAF, Tier 2, Tier 3 and other.

Penetration is shown prior to 3-day payoffs in order to highlight the mix of financing utilized at the point of sale, and we hope this will be useful information for you.

We again saw strong performance by our Tier 2 partners, despite lighter application volume.

Penetration grew slightly to 18.2% versus 17.9% in last year's fourth quarter.

Third-party Tier 3 sales mix was 9.4% of used unit sales compared to 14.5% for the same period last year.

Consistent with last year, CAF's Tier 3 activity remained at less than 1% of sales.

The decrease in Tier 3 penetration resulted from the factors we've been discussing: lower application volume and credit tightening as well as the delay in tax refunds.

Despite the headwind from Tier 3, we had comps of 8.7% and 5.4% in the past 2 quarters.

This means we've more than offset the missing sales to Tier 3 customers with sales at the higher end of the credit spectrum, which are more profitable for CarMax.

CAF income fell 10% to approximately $83 million compared to the fourth quarter of fiscal 2016.

While average managed receivables grew by 11.5% to $10.5 billion, the provision for loan losses increased and the portfolio interest margin decreased modestly.

For loans originated during the quarter, weighted average contract rate charged to customers was 7.4% compared to 7.5% a year ago and 7.3% in the third quarter.

Total portfolio interest margin was 5.7% of average managed receivables.

This compared to 5.9% in the fourth quarter of last year and 5.8% in the third quarter.

As you can see from our last securitization, the cost of funds has grown due to increasing benchmark rates.

We have tested and implemented increased APRs to preserve our interest margin to the extent the market will let us; and as always, we will continue to test rates and origination strategy to optimize income for CAF and for CarMax overall.

The ending allowance for loan losses at $124 million was 1.16% of ending managed receivables compared to 1.10% last quarter and 0.99% in last year's fourth quarter.

Most of the sequential increase from 1.10% to 1.16% was driven by an update in the assumptions we used to construct our loss allowance.

In the fourth quarter, our loss experience was largely as expected when we booked the allowance at the end of Q3.

However, each year-end, we review the assumptions we use to project future losses.

Our loss projection methodology is based on historical experience, and this year, we have incorporated heavier weighting on more recent data points, which resulted in approximately $5 million being added to the allowance.

A couple of things to remember here.

In the years leading up to fiscal 2017, our loss experience has been quite favorable and approximately 10% of the loss allowance relates to our Tier 3 program as is expected.

This you know -- as you know, is funded separately from our core securitization program.

As we have consistently demonstrated in other areas of our business, we believe that CAF is also nimble in reacting to changes in market conditions.

We believe we are originating a highly financeable, profitable portfolio, and that the current level of loss allowance remains consistent within our range of expectations, given our origination strategy and our portfolio mix.

Turning to our capital structure, during the fourth quarter, we repurchased 1.5 million shares for $101 million.

And for the full year, we repurchased 10.3 million shares at a cost of $558 million.

At the end of the year, we had $1.6 billion remaining in our authorization.

And now I'll turn the call back over to Bill.

Thanks, Tom.

During the fourth quarter, we opened 4 stores, including 2 in new markets: Mobile, Alabama and Albany, New York; and 2 in our existing Los Angeles market.

During the year, we opened a total of 15 stores and had 173 stores open at the end of fiscal 2017.

In fiscal 2018, we currently plan to again open 15 stores.

Lastly, we were excited to open our first 2 stores in the Seattle market.

Also during the first quarter, we will open a store in Pensacola, Florida.

Of the 15 stores we plan to open this year, 6 are in metropolitan statistical areas, or MSAs, having populations of 600,000 or less, which we now define as small markets.

It's important to note that on average, we would expect smaller MSA stores to sell less vehicles than their midsize and large MSA counterparts.

We also announced in our press release that we plan to open 13 to 16 stores in fiscal 2019.

Now let me take a few minutes to update you on several of our initiatives to advance our online offerings and the tests we are conducting to continue to enhance the customer experience.

Last quarter, we talked to you about our new online financing capability to help customers get prequalified for a loan.

It was available in all stores for the entire fourth quarter, and we are pleased with the results so far.

The feature is resonating well with customers and contributing to increased leads, which we believe generate incremental sales.

In the fourth quarter, we kicked off a test of an online appraisal offering in Charlotte.

This is a new digital solution for our customers who are interested in getting an appraisal value by submitting information online without having to come into the store.

Customers are engaging well with the product and we will continue to focus on refining and testing.

In regard to search engine optimization, we've made substantial improvements in order to capture the full opportunity to drive customers to our website through their online search.

Nonbrand SEO visits, which are searches that don't include the name CarMax, have more than doubled since June of last year.

This contributed to our overall web traffic and lead growth.

We are proud of these results and we will continue to invest in both innovation and execution in the coming fiscal year.

We are confident that all of our initiatives, both online and in-store, will ensure CarMax continues to lead the industry and deliver an exceptional car-buying experience.

We realize that there's a lot of noise right now in the market and concerns about the macro environment, but we are confident about the track that we're on and the future outlook for the CarMax business.

Now I'll open up the call for questions.

Victoria?

(Operator Instructions) Your first question comes from the line of Matthew Fassler with Goldman Sachs.

Thanks for all the detail that you offered up, particularly on credit.

I have one question related to that.

So the highest decibel level in the marketplace seems to relate to the apparent decline in used car prices based on the recent NADA data.

I'll try to make this one question, but is what you're seeing in the marketplace consistent with some of those headlines?

And to what degree does the adjustment in the provisioning reflect assumptions on recovery rates in sync with accelerated used car declines?

Matt, I'll start with your last question.

The adjustment in the allowance for loan losses doesn't include any prognostication or reaction to recoveries.

In specific, we base our loss allowance on -- with real experience and on some historical data.

So the adjustment that we made was really refinement in the assumptions that we use around that relating to the part that is based on historical data, and we've elected to weight recent experience a little bit heavier than longer-term experience.

So that -- I think that's question one.

And with regard to wholesale recovery rates, all else equal, a weaker rate does put some pressure on loss experience.

But I think as we've said in the past, there are several factors that come into play.

And the customers' willingness and/or ability to pay is a much more powerful driver than what we were recovering on the losses.

Anecdotally, if we're looking at a $15,000 car loan, this difference between recovering 50% and getting $7,500 and flipping 3 percentage points and only getting $7,050 is very de minimus relative to being able to keep that customer in the car and not lose the entire deal altogether.

And since no one can predict the future, we're making our credit decisions with the knowledge that we have and the knowledge that recoveries can fall within a range in the future.

Our next question comes from the line of Brian Nagel with Oppenheimer.

So I think my question is probably going to follow-up on Matt's question, that a lot of focus on the finance business and specifically the provision rate.

So the question right there is, you took provision up again here in the fourth quarter.

You gave some commentary around that.

But any -- can we better contextualize that as how we should think about, at least the puts and takes for the provision rate going into 2017?

Recognizing you don't give guidance, but just how should we think about how that number is likely to progress through 2017?

And then related to that, I guess is my follow-up, is we've taken provisions up.

And you think you've made it very clear that losses is measured by various metrics are still very -- within your comfort zone.

But has there been any commensurate tightening in lending standards on the part of CAF?

Sure, sure.

I'll hit your questions in order there, Brian.

As far as looking forward at the provision, by definition, the provision is what we believe losses to be over the next 12 months.

And it is our best estimate at the time given the information we have; very difficult to speak to the future.

Now as I mentioned, we conducted an examination of how to construct a loan allowance, and our methodology relies on both what's going on currently and historical experience, because you can't really predict the future.

The fact that we've moved to a heavier weighting on recent experience is likely to result in the model being a little bit more reactive to changes, in both directions.

That's -- I'm not saying that's good or bad, but we've always got to be careful when we're talking about the loan allowance, because one period isn't always telling on what's going on in the marketplace over time.

So it's really a balance between kind of speed of reacting to what's going on currently and the risk of whipsawing back and forth by overreacting in the long run.

I guess that's as much color as I can give you on that.

And as far as CAF and our credit standards, we're always looking at pockets where we can expand, where we can contract.

We have taken some deliberate measures over the past year to tighten credit, to improve what we expect losses to come out at.

And as we always do, the folks down in Atlanta will be continuing to look at that on an ongoing basis.

Your next question comes from the line of Sharon Zackfia with William Blair.

I have a non-finance-related question.

All right.

Thanks, Sharon.

I know you'd be excited.

So more on the SG&A side.

Obviously, there was a lot of stock comp increase this year, and you talked through a little bit about that.

I'm just wondering, first, kind of what you think a good ballpark number is for dollars in stock comp for fiscal '18.

I mean, I'm assuming it's not going to go up $40 million again, but maybe you can correct me.

And then secondarily, do you still think, like a mid-single-digit comp, you can kind of hold SG&A on an ongoing basis?

I know there was a stairstep increase in digital initiatives and so on.

We're starting to lap some of that, so I'm trying of think about this next year, what kind of a comp do you need to hold that SG&A?

Okay, Sharon.

So on your first question, the stock-based comp, where do we think that's going.

We don't know where it's going.

If it did go up another $40 million, that would probably be a good thing for us.

But we have no idea where that's going to go.

And as far as the leverage on SG&A, we've said we need mid-single digits to comp.

I would still hold to that.

I would say we probably need to be at the higher end because of investments that we're putting into the business.

But we still would stand by that range.

And just to clarify, higher end to leverage SG&A?

Or just to hold SG&A as a percent of sales?

To leverage, to leverage SG&A.

And sorry, one more follow-up on the stock comp.

I know you can't predict it because of where the stock might be, but if you -- if the stock was the same price today, I mean, is it going -- is it pretty static?

Or I'm just trying to figure out from a grant perspective, what's going on.

Yes, I think it'd be pretty static, but...

There will be some growth in headcount and pay at the company because we are putting on 15-ish stores a year.

Your next question comes from the line of Scot Ciccarelli with RBC Capital Markets.

First, just hopefully this is a housekeeping item and doesn't count against my questions, but the vehicle finance penetration rates, what's the difference between the 43% Tom talked about and the 48.4% in the release?

Is that just a difference in the 3-day payoffs?

Yes, exactly, Scot.

I guess I wasn't clear enough in my remarks.

We wanted to give you a flavor for what financing channels were enabling sales at point of sale, but we're all going to continue to disclose CAF's net penetration.

And the different (inaudible) represents those payoffs.

All right, got it.

And then I am going to switch back to the finance piece.

When you guys look at the change in your latest securitization funding rate, do you have any estimate or gauge how much of that was the increase that we saw in terms of short -- the increase or change in short-term rates?

And how much was for maybe investors looking for a bit more yield as they've seen loss rates increasing?

Obviously, that increases their risk profile.

Yes, I think the data available is public, but if you look at the spread on 2017-1, I think you'll see that benchmarks were up about 56 basis points, and spreads were virtually -- spreads across -- over benchmark were virtually the same as they were in prior year.

So we have seen -- and I'll hit on this, because this is probably your next question, Scot, is we did -- we have seen some escalation in benchmark rates, so that represents an increase in the cost of funds for the securitization market.

And as I mentioned during the quarter, we did test and then roll out rate increases, which were pretty much 50 basis points across the board for CAF.

Got it.

And is the plan to just kind of maintain your spread at that point?

Or you have to be careful about the elasticity for the customers, right?

I guess I'm trying to figure out go forward.

That's exactly right, Scot.

It's a balancing act between preserving that spread and preserving the best overall income for CarMax.

So as I've said many times before, we can only do what the market's going to allow us.

So one thing that we're very good at, at CAF is testing things quickly.

And we have the ability to do that by kind of running different champion/challenger strategy.

And so when we see rates increase, we can test rates.

We'll look to see how many people stick with CAF versus 3-day payoffs, but we'll also look to see how that impacts our conversion at CarMax.

Because we've got to bear in mind that we don't want to upset people over a finance offer if we're charging more than what the market would bear.

Also, the guy that comes in on Saturday afternoon and buys a car from CarMax utilizing CAF and then pays off is better than the guy that comes in then goes to his credit union on Monday and might end up somewhere else.

So there's a lot of factors that we have to look at to make sure that we're optimizing both CAF and CarMax.

Your next question comes from the line of Craig Kennison with Baird.

With web traffic up 3% and a nearly 9% comp conversion, clearly appears to be very healthy.

To what extent can you attribute that strength in conversion to recent improvements in the digital experience online?

Craig, it's hard to pinpoint exact amounts that are driving them.

In my opening remarks.

I talked about several different factors.

Some of this improvement on the customer experience.

Some of the improvement that we've made for website and SEO.

And I would also add to that, we had -- and Tom talked about that in his remarks, we had a higher mix of high-credit customers.

And as you know, high-credit customers convert better than lower-credit customers.

So that had an impact on it as well.

I think that inventory availability had an impact.

So there's a lot of different factors and it's hard to quantify how much the web is driving it versus all these other factors, including just better execution at the store level.

And then a point of clarification.

I think you said your share improved 2 percentage points.

Can you give us the actual market share estimate that you consider?

It's actually -- it improved 2%.

And again, this is comp market share growth.

And the comp markets, we're still in the range of total share about 4% to 5%, which is similar to where we were last year.

Your next question comes from the line of John Murphy with Bank of America.

This is Liz Suzuki on for John.

The extended protection plan revenues increased pretty substantially.

And in the press release, you mentioned favorable adjustments to the reserve for cancellations.

Can you just quantify what that favorable adjustment was?

And whether it should impact EPP revenues in the quarters ahead.

There's a lot of things that will go into impacting EPP revenues in the quarters ahead, so I can't state how it will affect.

But as you saw, we're up about $13.6 million in EPP revenue and margin, because it's the same thing.

The majority of it was due to the combination of kind of sales growth, pricing changes and our penetration.

We saw penetration drop a little bit, but that's mainly expected due to the higher credit quality mix of customers.

Those folks that bring their own financing, which you saw grow from 20% to 24% this quarter, are less likely to attach on those extended products.

The remainder of it, which is about $5 million, was due to year-over-year changes in the return reserve activity.

Great.

And just one follow-up on what you mentioned about those getting financing elsewhere.

Are those mostly people paying in actual cash?

Or are they getting finance from, like, local credit unions and other sources?

We don't have visibility to that actual data.

But my hunch would be majority of them are doing finance -- are financing in some way, shape or form.

But it represents both people who come to the table with cash, or their own money, and people who close the transaction with CarMax or one of the partners and then redo it later.

Got it.

But either way, you wouldn't have any credit risk in those loans.

No, absolutely not.

Your next question comes from the line of Mike Montani with Evercore.

I wanted to first ask, in the press release you mentioned that there was tightening in credit standards early in the year.

Can you just expand and elaborate a little bit on that in terms of what you're seeing from the various lenders and how that's trended right up until current day?

And the related thing was, can you talk a little bit about, Tom, normalized provision rates and spreads and how quickly we should expect those to be realized in the business as we look ahead?

Sure, I can talk to the first one.

As we've been talking about through the year, we did experience some tightening with our third-party Tier 3 lenders.

And if you remember, that was in the middle of the first quarter of this year.

Since then, their behavior has been relatively consistent.

I can't speak to how they behave and what their appetite is going forward.

But since middle of the first quarter of last year, it's been relatively consistent behavior from those partners.

As I mentioned, our Tier 2 partners have been very successful and more aggressive.

They're doing more with less, and we're very happy with their performance.

As far as normalization, that's going to be based on whatever the marketplace will bear.

From the perspective of a loss ratio or a expected allowance, at 1 -- remember, at 1.16%, that includes our Tier 3 activity, which is, as we've talked about before in our releases and our disclosure, is about 10% of the overall allowance and represents 1% of the portfolio.

So I mean, if you're kind of doubling the reserve as a proxy for expected cumulative net loss, we're still very comfortable that we're generating a portfolio that's within the range of expectation that we target in order to be able to access the securitization market.

Now there's no hard and fast rules.

This is a range that we target.

And there's been times when it's been at the high end and the low end.

And I think we'll keep an eye on how things are performing.

We maintain close dialogues with the rating agencies and we'll continue to originate a portfolio that we can fund.

From a spread perspective, normal is going to be whatever the market will bear.

We're still looking pretty good relative to kind of the decade of the 2000s as far as spread, even given some increases in interest rates.

But as I've said before, when we see interest rates move, we're going to attempt to preserve that spread.

But if we -- if -- by testing, looking at 3-day payoffs, looking at conversion, that we determine that the market won't bear it, then we're going to be a market lender.

So I can't really tell you where the market for finance is going to go.

It's going to be dependent on the appetite of our competitors.

Is there a long-term normal for provisioning rate, though?

Of like, 1.5% or 2%?

I feel like you've given that number in the past.

I just wanted to see how you...

No, I don't think we have, because we've only been provisioning since the accounting change in 2010.

We've given a range of where we expect cumulative net losses to run, and that's kind of in the 2% to 2.5% range.

And as I said, you can kind of -- the provision is a 1-year look forward.

So.

Your next question comes from the line of John Healy with Northcoast Research.

I wanted to ask a 2-part question on financing.

I appreciate the color you put in the release regarding the CAF Tier 2 and Tier 3 in terms of penetration, but I was hoping you could comment on what the profile of those borrowers look like in terms of maybe the average credit score in Tier 2 and Tier 3 as well as CAF for the year in terms of originations made and maybe how that compares to what you've seen over the last couple of years.

And then as a follow-up to that, I just kind of wanted to ask, just when you are making your assumptions for recoveries on the loans originated in fiscal '17, how do those recovery amounts compare to the assumptions that you made on '15 and '16 recoveries?

If you had a $15,000 loan, is it $7,500 you're expecting in recovery, in the example that you talked about in the past?

Or is it 4% lower because you're expecting the used car market 3 years from now to be 4% lower than it was this year?

Yes, let me answer your second question first, and that is, there's not nearly that much precision around recoveries.

As I mentioned I think a couple questions ago, we originate loans based on the knowledge that in the future, we won't know what recoveries are; and we have a range of recoveries that we've experienced over time.

And so we go into the equation of contemplating a range of recovery and hopefully it falls within that range.

So there's really not much to be specific on that front.

As far as the credit profile of the CAF customer versus Tier 2 versus Tier 3, in general, they're going to -- CAF is going to be at the very highest level, because we take the first look and nothing goes down to Tier 2 or 3 until after we've either declined them or asked for conditions on their financing.

But CAF does -- if you look at our securitizations, and we do lend across a pretty broad spectrum, our average FICO this quarter is a 707 for the portfolio.

So but we do a lot of stuff higher that and some stuff lower than that as well.

So you -- I mean, you can get a flavor for that.

But we do, do a relatively broad spectrum of credit.

But we're not looking only at FICO scores as the measure for whether we'll approve a person and what we'll charge them as far as rate.

And so in Tier 2 and Tier 3, I guess, Tier 3 is what you'd call deeper subprime.

And generally, that's going to be kind of 550 below; and then Tier 2 is kind of a gap in between.

But it's very hard to give any specifics on that, because everybody's got their own methodology.

That's why we have multiple lenders in various spaces so we can maximize the sales for CarMax.

No, I appreciate that.

And the reason I ask that is just I think there's so many different definitions of subprime, so many definitions of prime.

And I just wanted to get to try an update on where those stood.

So I appreciate that.

And you're right.

Some people would say subprime is below 630.

Your next question comes from the line of Brett Hoselton with KeyBanc.

Two questions that are related to one another, and again, financing is a popular topic here.

First of all, your adjustments for provisions.

Was that more a change in your recovery expectations in terms of used car values?

Or was it more in terms of default rates?

That's question #1.

Question #2, can you give us an update as to the subprime pilot?

Sure.

I'll start with the first one.

As I mentioned when I described the allowance, it was a result of a change in the assumptions we used behind the scenes.

I'm not -- I can't go into -- I'm not going to go into detail on it, but it was -- we're using more recent data and weighting that heavier in the historical portion of what we utilize to build the allowance.

Recovery rates, as I said when we originate loans, we go into that knowing that there can be a range in the future when the loss may indeed happen, so we don't have any prescription on what they're going to be.

And your second question is on the Tier 3 test.

I think it's steady as it goes.

We're comfortable with the portfolio.

It's not doing anything unexpected as far as its performance.

And again, we're originating roughly 5% of the Tier 3 volume that CarMax does.

So as the customer applicant flow increases and decreases, our volume of the Tier 3 will increase and decrease with it.

We're not changing the way we approach it.

And we're comfortable participating in this manner for the foreseeable future.

We like -- as we said it originally, we really are interested in learning more about this space.

And even if it's just to have better information to manage our partners, I think it's helpful for us to have a toe in the water with this.

Is the pilot large enough that you can take it to the securitization market at this point?

Yes, we mentioned earlier in the year that we put a private securitization structure in place to fund this business.

If it were something that we decided to expand and continue for a longer period of time, we would also consider permanent funding for it.

But as far as the public market at this point, I don't know.

And I apologize, just one more follow-on, which is just, can you talk about the spreads and so forth?

Which spread?

At that level of the market, the subprime pilot.

What I can tell you is it's a much higher APR and it's a much higher expected loss.

I mean, you can kind of extrapolate from the data that we've given regarding the size of the allowance set that is comprised of Tier 3, that we're expecting -- you expect roughly 10x the losses in Tier 3 that you do it in CAF.

Charging...

No, I apologize.

I was actually thinking about the change versus maybe 6 months or a year ago.

It's not material.

Your next question comes from the line of Mike Levin with Deutsche Bank.

So first off, looks like you did a very nice job in holding in your retail GPUs despite lower used values.

Just wondering if you can kind of discuss how you think your ability will move from here with the expectation that lower used prices will continue to flow through the market.

Is there some point at which we'll necessarily kind of take a step down in terms of those GPUs below $2,100?

Or should we expect some kind of margin -- percentage margin expansion back to kind of where you were a few years ago?

Yes.

So obviously, the retail GPUs are an important thing for us to focus on.

We've been focused on, we'll continue to focus on them.

I think in my opening comments, I talked about the fact that we've been able to maintain these -- this $2,100 to $2,200 range for the last 24 consecutive quarters.

During that time period, there's been lots of fluctuations in price.

I would even go back to, if you to think about the recession, when our average selling price went down by about 20%, we were still able to maintain GPUs during that time.

So it's hard for me to say or think of a situation where we can't maintain the GPU.

Now that being said, we do constantly test to see if giving up GPU will drive more sales and does it make sense for the overall business?

If we saw a scenario where that elasticity was different and benefited CarMax, we would certainly do that.

We have not seen that.

And we will continue to focus on hitting that target.

Got it.

That's really helpful.

And then just quickly on -- you've been testing rate increases to kind of maintain your collateral spreads, but I'm just wondering, as we're expecting to see a higher mix of credit quality come in to used vehicle market and you're going to get lower customer rates from then, is that going to kind of hamper your ability at all to raise customer rates overall, with that mix of credit coming in?

I think as I said, we've -- during the quarter, we increased rates pretty much 50 basis points across the board.

If you look now, you see that the best rate that you can get at CarMax is about 2.45% versus 1.95% several months ago.

And to answer your question, it really just depends on what's coming through the door and what the market will bear.

As I said, we've got to provide competitive offers to what other people can achieve outside the CarMax system for our customers in order to keep them happy.

And as rates move, we're very cognizant down in Atlanta of the spread and maintaining profitability of CAF.

And as we see things move, we'll constantly look for opportunities to optimize it.

I guess more of what I mean is, because your credit quality is increasing, does that necessarily mean that your collateral spreads will decrease because your customer rates are going to skew towards a lower number?

As I said, it depends.

It probably depends on a lot of factors.

If you put that in isolation, the answer is yes.

But you also expect significantly lower losses on that from that set of customers.

Well, and I think to Tom's point also, it depends on what the competitive -- what the competition looks like.

As they move their rates for the highest-tier customer as well.

Yes.

I don't want to give any forward-looking, but you could hypothesize that as we get more and more high-credit customers in there with lower loss expectations, you would see that, all else equal, if we're getting a changing environment, you would see the portfolio expected losses start to drift down, which means we might be leaving something on the table at the other end, and we're constantly looking for pockets -- as I said, for pockets where we can expand and contract as makes sense.

So we still will do that to the greatest extent possible.

Perfect.

And did you quantify how much of the increase in loan loss reserves was attributable to the change in methodology?

Virtually all of it.

And let me clarify that, too.

It's the increase vis-à-vis Q3.

It's from 1.10% to 1.16%.

Not year-over-year.

Obviously, we've had negative loss experience during the course of the year that's incorporated -- that has already been built into that over the first 3 quarters.

Your next question comes from the line of Ali Faghri with Susquehanna.

So on the same-store sales, you've had a meaningful acceleration in recent quarters despite a sizable drag from lower Tier 3 volumes.

Could you remind us when you lapped that headwind?

And also help us quantify if there is any impact on your sales from delayed tax refunds in the quarter?

And maybe potentially whether you saw some of that pressure moderate as those refunds started to catch up at the end of February?

Yes, so on the first question on when we lapped the Tier 3 headwind, that would be after the first quarter of this year.

That's when we started seeing the tightening by the partners.

So not yet.

So yes, not yet.

And then I'm sorry, what was your second question?

What the impact of the delayed tax refunds were in the quarter, and whether you saw some improvement maybe at the end of the quarter as those started to flow back into the market.

It's really hard to quantify how much it impacted the quarter or the month.

We do know, and I think Tom talked about it in his introductory remarks, that I said by the end of February, year-over-year, there were probably about 10% less refunds.

We think that it had an impact, but it's really hard to quantify the degree.

It's really not something -- over a longer period of time, it's really -- it all comes out in the wash.

Your next question comes from the line of James Albertine with Consumer Edge.

And 48 minutes into a call, let me just say congratulations, which I think would be in order for 8.7% comp.

Most retailers would kill for that right now.

And I'm surprised we're not making a bigger deal about that and we're having sort of a CAF 101 lesson.

Thank you, Jamie.

So I just want to make that statement.

The question I had here is in the constitution of comp sales.

We've talked in prior quarters about the fact that the supply of late model is littered with maybe more cars than the market's demanding and not enough trucks.

Your comp sales surprised us, certainly to the positive.

Is that a reflection of a more balanced supply environment on a car and truck side?

And sort of related to that, trucks are going to be more expensive generally.

And is there a way that you could sort of pivot even higher or at least resist the trend that the NADA is calling out with respect to broader used vehicle declines, given the constitution of trucks in your mix?

Yes.

I think the fact that the inventory, I think we had -- I think industry estimates are saying about 3 million cars -- rental cars came off-lease this past year.

They're estimating another 3 million or 4 million next year.

I think that availability of the inventory certainly plays into -- or is a component, as I said earlier, in our comp sales.

There are a lot of things that I think play into the reason our comp sales are where they are today.

Our average selling price this quarter was down, but it's a little bit of a 2-sided story.

One, it was driven up by the fact that our mix in trucks, large SUVs, that went up a little bit, caused it to go up.

But then, the overall acquisition price went down.

So that offset any increase in that -- increase in the truck inventory.

I think as we go forward, I think it'll continue to be a positive for CarMax as far as whether it changes up the mix.

I think the leasing data would tell you that a lot of leases on trucks and large SUVs, so over time I would expect that to feather back into the marketplace.

So -- and that could be some push to make average selling prices higher, but it remains to be seen.

Just to voice this.

And we get concerns that new vehicle sales have pulled forward used vehicle demand, and that you're going to have to flex margins per unit to reflect -- combat that.

Clearly, the fourth quarter, I think, represents otherwise.

So really congratulations there.

Thanks, Jamie.

Your next question comes from the line of Bill Armstrong with CL King.

So we've been hearing a lot about consumers with negative equity in their trade-ins.

And I was wondering if you can maybe discuss if you're seeing that trend with the customers coming into your stores.

And to what extent that might influence your ability to offer them financing.

Any influence on loan-to-value ratios?

And in general, how that might be affecting your business overall?

Yes, Bill, I would just tell you we haven't seen anything that's worth mentioning.

It just hasn't played out to any type of meaningful impact for us at this point.

Would you see it maybe more on the lower end of the credit spectrum versus maybe your higher end?

I would be speculating.

Like I said, I think overall for us, we haven't seen -- it just hasn't been an issue for us.

So I can't really say if it's a little bit more impactful on the lower end or the higher end.

It's just overall hasn't been a meaningful number for us at all.

Yes.

As we've mentioned, Bill, the theme throughout the year has been really one of volume of credit apps at that lower end not showing up as much.

Or at least not pulling the trigger on applying for credit, so it's hard to divine what's the driver of that.

It could be that they do have more negative equity and are less confident and therefore not coming in and applying.

But that would be a hypothesis.

Right, okay.

Makes sense.

And then just a quick follow-up.

Your tax rate was a little bit lower, your effective tax rate in the fourth quarter.

Anything to call out there?

And what should we be maybe modeling going forward in terms of the tax rate?

Yes, I'm glad you asked that.

There's a couple of things to comment on there.

And probably, the go-forward one is the bigger issue.

In any given period, state tax items can cause the rate to move, and particularly at year-end, when you're truing up.

And this year, we had a number of things fall favorably from that perspective on the true-ups.

We also, as required, reserve for uncertain tax positions.

And from time to time, these positions become certain, whether it's because they're resolved or the statute of limitations goes away.

And this year, we had a number of items that timed out from that perspective as well.

So there were some favorability on the state side that caused that change in rate.

On a go-forward basis, I think because of the FASB guidance regarding stock-based compensation, which I'm sure many of you are familiar with, that new treatment is likely to cause significant volatility in effective tax rate and then consequently net income versus pretax income.

And then you can go look at the guidance yourself.

But in effect, what it -- on a go-forward basis, tax benefits or deficiencies from option exercises will now flow through the income statement in the tax line rather than flowing through the balance sheet in equity.

What that means is, in any given year, changes in stock price, changes in option exercise behavior will have an impact on your tax rate.

And just in certain other potentially significant level of volatility in that tax rate and then, like I said, consequently net income.

So good stuff happened this year.

Going forward, it's very hard to give you any kind of guidance what the tax rate would be.

Right.

But that wouldn't change the actual cash tax you're paying.

That's -- this is just...

No, not at all.

This is just the accounting reflection of where that tax shield or cost runs through.

But it definitely will impact reported net income and EPS, so you may want to consider that as you think about things.

Your next question comes from the line of Rick Nelson with Stephens.

This is Nick Zangler on for Rick.

Just digging into that same-store sales number, you guys are obviously making some strong unit sales.

And this is coming despite the credit tightening in Tier 3 lenders.

But to be clear on the dynamics, if not for the tightening by the Tier 3 lenders, would total company same-store sales be stronger and therefore you're currently missing sales?

Or are those sales being picked up by CAF or more likely the Tier 2 lenders?

Or even outside financing, which would then obviously contribute to that 15.3% comps report exclusive of the Tier 3?

Yes, it's hard -- it will be hard to say for sure.

But in an all-equal environment, I think if we had not seen such a decrease in the Tier 3, same-store sales could have been higher.

But keep in mind, as far as, are other people picking this up, what we've talked about even today and previous this year is a lot of that is driven by the fact that those customers just aren't coming in the door.

So it's not like they're coming in the door, we're not getting them and somebody else is getting them.

We're just not seeing them.

And the dynamics that Tom just spoke to a little bit, but it really is a hypothesis on our side as far as why they're not coming in.

So whether they're just not coming in, going to other people, I don't think that's necessarily the case.

I think we've seen industry numbers that support that this is a phenomenon that goes beyond CarMax.

Yes, I'd just add a little bit of color there with regard to the Tier 2, Tier 3 phenomenon.

There's clearly, as we've said, a decline in the number of applicants in that -- at that very low end.

But I think our Tier 2 partners are doing a great job in their approach to credit.

And a small portion or some portion of the decline in Tier 3 may have been picked up by Tier 2. But it's impossible to say what.

Yes.

And let's not forget, I mean, while we're happy to have a Tier 3 sale, they're also our least profitable sales.

And the fact that our core business is growing at such a good rate, those are way more profitable sales for us.

So we're very pleased with that.

Understood.

And then there's also been some discussion around -- or within automotive retail, some weakness in some markets due to immigration concerns given the rhetoric that's coming out of DC.

I'm curious if you saw this play out at all in the fourth quarter.

Or if you're seeing anything related to that early in the first.

Yes, not.

I mean, there's really nothing to comment there for us.

We don't get into market-specific dynamics.

Your next question comes from the line of David Whiston with Morningstar.

Just want to go back to pricing first, kind of an open-ended question for you, just generally speaking, 30,000 feet level.

Why shouldn't shareholders be very worried about this?

Is it just a matter of getting a better SG&A leverage than your competitors?

Are you talking about our margin?

Yes, just the decline on the overall market on used vehicle prices going down, there's a balancing up between GPU and SG&A leverage and I was just...

Yes, average selling prices.

We don't -- I mean, we target the GPU.

As average selling prices go down, we can still maintain GPUs and pass those savings along to customers.

And as I spoke to earlier, we've demonstrated this in multiple periods when there's been high fluctuation in pricing.

So -- and then on top of that, we'll continue as an organization to look at opportunities to take costs out of the system, both through cost of goods sold and through SG&A.

And when you take cost of goods sold down, you have the option of passing that along in the form of lower prices or you could take some of that to margin.

And we will look at that as we continue to make progress.

Okay.

And on digital, did I hear right that non-CarMax searches, those people have -- it's led to a double in those people going to your site.

Is that right?

Yes, so non-brand, so searches that folks do that don't include CarMax, since June since we rolled out the website, those searches are more than double.

And that's important, because we already do well on the CarMax brand when people search that.

But it's important because a lot of folks will start their search with things like Honda Accord or Honda or Accord, which don't speak to CarMax.

So making progress on that, we feel is a very good thing, because that's what most people search.

And we feel like we have a lot of upside and a lot of work that we can still to do to get better at that.

And not being a web advertising expert, I mean, can you briefly talk about how you can do that beyond just spending more with Google to be higher up in the search results?

Well, remember, what we're talking about is SEO, search engine optimization, versus search engine marketing.

Search engine marketing is where you pay for the ads.

Search engine optimization, the way you -- it's not necessarily that you're paying for ads.

The way we do it is through this year, we've done a couple different things.

One, it started with the website.

Since the new website.

We’ve also redesigned how the search engines crawl our website and make it so it's more efficient for them to crawl.

We've focused on keywords.

Again, keywords that aren't related to CarMax, and building out pages.

And so that when the search engines search your website, they see keyword and are like, okay, CarMax must be an expert on that, so we're going to bring them into the organic search content.

That's another way we've really increased our SEO, is building relevant content that, again, when the search engines look at your website, it's relevant to the questions that people are asking.

So those are the levers that we've pulled and we will continue to pull to get the SEO or the organic search going for our customers.

Your next question comes from the line of Chris Bottiglieri with Wolfe Research.

Sort of a clerical question.

Given that new -- the financing mix that you provided on the call, on the press release, does that 18.2% for Tier 2, is that comparable to the 15.7% you reported last quarter?

Chris, we're looking at it.

Yes.

Hold on, Chris.

It doesn't sound (inaudible).

Well, you have Tier 2...

Last year, we...

Yes, it's Tier 1, not Tier 2 (inaudible).

Yes.

Third party.

Okay.

Yes, there's a little disconnect there.

So it is not quite consistent with the 15% reported last year.

We had some -- we have some other programs that are higher-credit, referral-type programs that are not CAF, defined as Tier 1 not CAF.

We've now bucketed them with Tier 2. It represented about 1.5%, so that's the disconnect.

Got you, so if I take the last year's cadence with 17.4% -- I mean in '16, 17.4%, 16.2%, 15.7% and now 18.2%.

You're saying I need to adjust the prior 3 quarters to this new methodology?

Specific for Tier 2.

The 18.2% includes about 1.5 points that the others don't.

That the other -- so the first 3 quarters don't include that 1.5 points.

Right.

Okay.

And then what's going on -- why is other picking up so much?

Like, what's the take away there?

And how do we interpret that?

I think in general, people with higher credit and more wherewithal to get funding migrate more to others.

So people -- that's credit unions and that's also people with cash.

So as we -- it's not surprising, as we've seen double-digit growth at the very highest end of the credit spectrum, kind of the 700, 750 and above, and shrinkage at the very lowest end, to assume that some of that's just flow that higher-credit people who have access to other funding are buying more cars.

Got you.

Okay.

And then just a follow-up on the financing mix here, so you added a nice new disclosure to your securitization where you give the mix by FICO score.

It only goes back to 2012, though.

Can you give us a sense what your mix of scores, sub 630, were back in like 2006, 2007?

So we can compare where you are versus last cycle?

That's some really old data that I don't -- I can't put my fingers on.

I mean, if you go -- you can get an idea for it.

But what I can say is the spectrum of credit that we lend to has not changed dramatically over time, I mean there's fine-tuning here and there.

During the recession and into kind of 2010 time period, we significantly dialed back on what we were doing to ensure that we had a highly financeable portfolio.

And we were successful at having our partners step up and take more volume.

From 2011 kind of forward, we've been moving back up to what we've historically done.

I can kind of give you that color.

But as far as the details, not really.

Just one last one and maybe just from memory here.

Is there may be like a basis point spread I could plot your mix by APR?

Say, 1000 basis points spread to the 3-year treasury to get to what that mix might have been.

Is that a fair way of looking at it?

I'm sorry, I didn't understand your question.

If I was to look at your mix by APRs historically, what do you think the appropriate risk spread is to the treasury for subprime?

Or what you would charge your customers over time...

That's going to depend on the environment.

And we're always fine-tuning.

We don't talk about that kind of detail.

Your next question comes from the line of Adam Jonas with Morgan Stanley.

This is Linda in (inaudible).

So there's a thesis out there that your GPU is not sensitive to a decline in used car pricing and that you can manage to a dollar value per unit on used cars regardless of what the used ATP is.

Do you subscribe to that view?

Yes, that is the view out there.

And actually, that's what we've demonstrated being able to do for many years now.

Okay.

And now do you think the credit worthiness and financial strength of your used car customers is in any way impacted by a potential decline in used car values?

Do I think the used car prices -- the used car price decline...

The creditworthiness and -- yes, and financial strength of your used car customers is in...

We have no way of knowing if those are correlated or not.

Okay.

And finally, is your business sensitive to the credit worthiness of your customers?

Well, I can address that.

I mean, there -- we strive to have a credit offering for customers across the spectrum.

And as we've talked about many times before, CarMax Auto Finance participates at the higher end of that credit rate.

And we take a look at everything.

If we don't approve, we pass it down.

And we have a pool of very strong, very savvy Tier 2 lenders that all take a look at those customers and compete for it.

And we think having a portfolio of those lenders adds sales and benefit to CarMax.

And then we have a Tier 3 offering.

So we attempt to have a finance offering in place for any customer that would come through the door, so as to insulate ourselves those kind of changes.

That said, as Bill mentioned earlier, higher credit quality customers tend to convert more readily than low credit quality customers, perhaps just because -- by virtue of the fact that the offers they're getting are more attractive.

The Tier 3 offers are typically less attractive offers.

So it's -- like I said, it's impossible to correlate those two things, but we think we do everything we can to manage in whatever environment will rear its head.

That's all the time we have for questions today.

I'd like to turn the call back over to the presenters for any closing remarks.

Thank you, Victoria.

Just a couple of things.

One, if you didn't get to ask a call, please feel free to call -- yes.

If you didn't get to ask a question during the call, please feel free to call IR, call Katharine, and we'll talk to you off-line.

I also just want to -- given the nature of the questions in this whole session on auto finance, I just really want to reiterate Tom's -- one of Tom's earlier statements, which is like the rest of the business, we've consistently demonstrated where CAF can be nimble and react to different changes in market conditions.

We feel good about the lifetime loss we're shooting for at this point.

And so we feel good about the CAF business.

I want to thank all of you for joining the call and your interest in CarMax today.

I also especially want to thank our associates nationwide who are continuing to work hard to deliver the incredible customer experience every day.

We were recently honored as a Fortune 100 best companies to work for, on that list for the 13th year in a row.

That is absolutely a testament to our associates.

They're out there driving what's possible every day for each other, for the customers, for their communities.

They are CarMax.

They are our differentiator.

I'm proud of their accomplishments.

I'm proud to work beside them.

And I look forward to a great new year.

Talk to you next quarter.

Again, thank you for your participation.

This concludes today's call.

You may now disconnect.