Manheim Conference Call Part 2: The Used Vehicle Market

Includes: GM, KMX, TM
by: Jerry Marks

I would say that we really don't have a good handle on what the past level of used vehicle sales have been year to year. There are popularly cited sources out there that have them down substantially in 2006. I don't think they were down that much. They were probably off at the most 2%, on the dealer side. And private transaction volumes were up.

The way I read it. . . (pause). Of course we haven't seen the public dealers report yet. But CarMax (NYSE:KMX) did in February, and they had strong sales. Certified pre-owned sales have continued to be strong. If you look at auction volumes, the market is going as expected.

- Thomas Webb, Manheim's Chief Economist 1Q07 Conference Call

Manheim Conference Call Part 2: An imaginary flower shop and the irony in the reported used vehicle decline (that the "solution" may be causing the "problem")
Yesterday I said Manheim's chief economist referred to the new vehicle sales environment as easier to address than the used vehicle environment. The reason he said that the used vehicle sales environment is more difficult to discuss is not because it is less predictable, "it is more predictable" he said. But the problem comes in trying to figure out what the prior year used vehicle sales really were.

You see today's opening quote.

Now I think it is important to explain a few things about reported vehicle sales. Every month you hear me discuss U.S. light vehicle sales. General Motors (NYSE:GM) sold 90 vehicles for every 100 it sold last January. Or Toyota (NYSE:TM) sold 150 for every 100 it sold last March you might hear me explain. Most headlines will simply read: GM's sales were down 10%, and Toyota's were up 50% (obviously I am exaggerating the sales figures a bit).

But you want to know a little secret? GM or Toyota did not sell a single one of those vehicles. The vehicle sales figures you hear me discuss every month will probably not even count in GM or Toyota's first quarter sales (revenue) figures. They were probably reported as revenues in the fourth quarter. And the reason is because GM and Toyota likely built the vehicles and sold them to their dealers a couple months ago. Well, I guess we are in the second quarter now, but hopefully you get the point. So when you hear me discuss April or May vehicle sales, those vehicles were likely recorded as revenues for the automakers in the first quarter.

The monthly new vehicle sales figures you hear reported therefore are what happened at GM or Toyota's dealerships (so the sales results you hear coming out of the public dealer groups). The dealers periodically (daily, weekly, monthly) report their sales to the manufacturer usually electronically. Although I am told that sometimes sales are still manually sent in via fax machines.

Are there problems that happen with the reported figures? Sure. Sometimes the paper work associated with a sale isn't filled out correctly and so a vehicle that was sold in one month actually slips into the count the next month. And I am sure dealers can fill me in on all sorts of other problems that happen with the reported figures. But we're talking about the law of large numbers. So even though there may be a slight problem with the data, as long as it is gathered in a consistent manner, it really does not become a big problem when folks like you or I are simply trying to understand if underlying demand (sales) are growing or shrinking.

Think about it this way. Imagine I owned a flower shop. These were really rare and expensive flowers, and so I only sold 100 a year. Most people bought them around Christmas time. But no matter how good of care I took, every once in a while I ended up selling a "dud." This tended to happen with about 10 of the flowers every year. And sadly I could not tell which ones were the "duds" prior to them leaving the incubator in my shop. So instead, I offered a money back rebate if the flowers wilted within two weeks of purchase.

Since I sold the bulk of my flowers around Christmas time, about half (5) of the flowers would be returned (in their wilted state) prior to New Year's. But even though the flowers (if they were to wilt) would do so within a few days, since I gave people a couple weeks, some (the other 5) would generally wait until after New Year's to bring in the flowers for the rebate.

Now I close my books December 31, 2006. So when I record my sales for the end of the year, I take the 100 flowers I sold, less the 5 that were returned. Oh, and the same thing happened last year, so I have another 5 that were returned in January. So my "net sales" for the year are 90 flowers. It is a pretty stable business, so year in and year out, this is what ended up happening, always ending up with about 90 flowers being sold a year. Similarly, this is what I think happens with reporting problems with new vehicle sales. Sure some sales that happened in one month may end up being reported the next month, but over time (and across such a large dealer base) it balances out.

But let's continue on with the flower shop example. Suppose I noticed that some of the people that brought in the flowers after December 31st really weren't bringing me wilted flowers. Instead, they had just kind of destroyed or not taken good care of the flowers? But I didn't want to argue with a customer, so I just gave them the refund. In 2006, however, I made a change. I required customers to bring the flowers back to the shop within 3 days of purchase if they wanted a refund.

You know what happened? I had 8 people return the flowers the week before New Year's. And no one bring in a flower the week after New Year's. Remember my accounting system? I sold 100 flowers, I had 8 people return flowers in December, and don't forget about the five from January that spilled over from using the old system (where people could bring back the flowers within 2 weeks).

So from a reporting standpoint, if I just looked at my reported sales for the year, I sold 100, and had 13 returns, so "net sales" of 87. The first time I would have reported lower sales in several years versus the 90 I usually sell. In reality, this holiday season I only had 8 people return flowers instead of 10 (although they all did it all in the last week of December). So my net sales really were 92 instead of my usual 90, an increase. But on a reported basis, I am showing a decline (year over year).

Starting to notice something? A better, more efficient system that is actually improving my results, makes things (on a reported basis) at first appear like sales are dropping. This is what I think is happening in the used vehicle market.

Registration Data Actually Getting Better
Unlike the system I described above where car dealers report in their new vehicle sales every month to a manufacturer, used vehicle figures are determined by registration data.

You know how every year you need to "register" your vehicle with the state you live in (the little sticker that goes on the license plate?) And of course if you buy or sell a vehicle, you need to buy or transfer that little sticker.

Well, why do you need to register your vehicle? In the state of Florida, here is the reason the department of motor vehicles specifically gives to answer the question: "why must I register my motor vehicle?"

In the State of Florida, a motor vehicle is required by law to be registered within ten days of the owner either becoming employed, placing children in public school, or establishing residency. Registering your motor vehicle goes hand in hand with the titling process.

And I think you would get a similar response at most state department of motor vehicle offices. It is a way for the state to understand who is working, living, or sending their kids to school under their taxes. Well, it's actually (or can be) a lot more than that even, but hold on, we don't want to get ahead of ourselves.

So there are a number of groups: Experian, CNW Market Research, Polk, and I forget some of the others. All of these groups go around to the various Department of Motor Vehicle departments and combine (aggregate) the data to try to show (month to month) and year to year the number of used vehicles that have been registered, by state and in total nationwide.

Even the Federal Highway Administration (FHA) "aggregates" the data from the state Department of Motor Vehicles (although I think they purchase some of it from these aggregators like Polk). But here is what the FHA has to say about annual vehicle registration data (for their Highway Statistics 2000)

The annual vehicle registration date varies among the States. Although many States continue to register specific vehicle types on a calendar year basis, all States use some form of the "staggered" system to register motor vehicles. The "staggered" system permits a distribution of the renewal workload throughout all months. Most States allow pre-registration or permit "grace periods" to better distribute the annual registration workload.

In order to present vehicle registration data uniformly for all States, the information is shown as nearly as possible on a calendar-year basis. Insofar as possible, the registrations reported exclude transfers and re-registrations and any other factors that could otherwise result in duplication in the vehicle counts.

I won't even go into the part where they talk about how "registration practices for commercial vehicles differ greatly among the States," which can also skew the data.

Because it is the last sentence above that I really want you to concentrate on. The one that talks about how the FHA tries their best to exclude transfers, re-registrations and other factors that could result in duplication in the vehicle counts.

There seems to be a growing push to reduce this "bad data" or duplications in the vehicle counts. Consider the following letter sent by Ann McCartt, Vice President of Research for the Insurance Institute for Highway Safety on March 9, 2006 to Jacqueline Glassman, Acting Administrator for the National Highway Traffic Safety Administration and Norman Mineta, the Secretary of Transportation at the time:

The Institute and other highway safety organizations rely on national and state-level driver's license and vehicle registration data to track progress in highway safety. Based on the evidence presented in this comment and discussions with (Federal Highway Administration) FHWA, we believe the current and historical license and registration data are not reliable measures of driver and vehicle exposure. The use of these data may lead to erroneous conclusions about the success or failure of highway safety initiatives. To rectify the situation, we recommend that the driver licensing agency in each state be designated as the contact agency for provided data, that the instructions for defining teenage licensed drivers be revised to ensure intermediate licenses are included, and that an effective system of quality control be instituted to ensure reliable data are produced. We urge the U.S. Department of Transportation to move expeditiously.

And on the state level, I think you are seeing a real push for better/more efficient data collection. 3M is even trying to sell the idea of radio frequency (rf) identification technology to "electronically identify vehicles and validate the indentity, status, and authenticity of vehicle data." Now there might be some "big brother" concerns with this technology, but it certainly would help address many of the concerns/issues Ms. McCartt raised above.

We shouldn't kid ourselves on why the push is probably really happening (at the state level). The real incentive is better tax collection dollars. But whether it is the Colorado Department of Motor Vehicles replacing their old "green screen" distributed data processing system with a new state of the art system (granted it has encountered some glitches early on). Or Massachusetts beginning to experiment with Electronic Vehicle Registrations (for businesses at least), you can see a growing push for more automated and better registration systems.

On top of that, throw in CarMax's proprietary inventory management system, lanelogic, AmericanAuto Exchange, FirstLook, VAuto, Ebay Motors, and just a host of new technologies that may be helping the vehicles to sell right the first time. Instead of the old system where you probably had a lot of vehicles that would get tossed from dealer A to wholesaler A, to dealer B, to Auction house A, to dealer C, to wholesaler B, to dealer D, to maybe by then retail customer Fred. All along the way creating a ton of transfers and re-registrations that increased the odds of duplication in vehicle counts. Now the vehicles may be going to the lots where they might sell right the first time. Just imagine if dealers could end the bucket system and price the vehicles so they sell right out of the gate?

In other words, unlike in the past where you had a lot of duplication in vehicle counts occurring in the market place. Investments into better technologies, as well as systems and processes both at the dealer and state department of motor vehicle level are probably making the data better (so less duplication in the data).

Used vehicle market, welcome to my flower shop!

Here lies the irony. The reduction in duplicated vehicle registrations means the data is probably getting better. Unfortunately, when we are trying to compare the data to previous year results that have more duplication in the data, what you end up with is a greater efficiency appearing at first like sales are dropping.

Used vehicle market, welcome to my flower shop
First of all, I need to say that I have no proof this is happening. It is only the best explanation I can think of as to why such a historically stable market now appears to be acting so volatile. Especially, when other data sources (like public dealership groups and certified used sales) are running contrary to the aggregated registration statistics.

I don't know if this data aggregation collection anomaly (abnormality) takes a year, three years, or maybe five years to work through. I can only tell you that I think we need to be very careful when we hear people tell us how "used vehicle sales" are down so much. Because I think it really means the used vehicle market is becoming more efficient.

So in addition to the public dealer results (that obviously don't get caught with "duplicative data problems") I also think we should watch the certified used vehicle sales trends. Because those sales (from what I understand), are reported by dealers to the automakers just like they do with new vehicle sales. And March certified used sales were pretty good. . .

Dealers made their traditionally strong certified sales push in March and achieved record success, according to Autodata Corp. The company reported that certified sales came in at an all-time high, reaching 161,938, up 7.7 percent from the second strongest selling month set March 2006.

- Jennifer Reed, Subprime Auto Finance News (April 11, 2007) titled: "March Certified Sales Set New Record."

How profitable is the used vehicle market?
So maybe used vehicle sales are not the problem. The real question comes down to how profitable the market is right now? For that, I thought I would end today's piece with some of the last comments made by Tom Webb on the conference call Tuesday (because I think it says it all).

On the retail side, there has been some erosion in gross margins, they've become more competitive. They are fighting for business harder than in past because they recognize reduction in new vehicle volumes that are coming.

Actually I would change the above statement to: "because they recognize or are reacting to the reduction in new vehicle volumes and profits that have already come." But that is just my opinion.

In any case, Mr. Webb concluded with probably the most important point. . .

So there's pricing pressure there in terms of the retail market. In terms of achieving growth. But as any good dealer knows, if they turn their inventory, even just a little bit quicker they have more than made up for their loss of gross.

Have a great weekend.