New Home Warranty Report:
With products such as single-family homes, it's possible to measure warranty accruals not only per dollar of revenue, but also per unit sold. And then it becomes clear which companies have mastered their warranty cost estimation process and which are still making it up as they go along.
Warranty costs per new home sold are still generally rising. But new home prices are rising faster. Therefore, warranty costs as a percentage of sales are falling. But what matters more is whether warranty costs are stable and predictable over the long term, or if they halve then double from one quarter to the next.
Every fiscal quarter since the start of 2003, we have collected the warranty expense reports of all U.S.-based companies that include the data in their financial reports. Typically, we collect figures for claims paid, accruals made, warranty reserves held, and products sold. And then by dividing one metric by the other, we calculate claims as a percentage of sales and accruals as a percentage of sales.
This week, in order to take a closer look at the new home industry, we're going to add another metric and drop several others. Like passenger car manufacturers, homebuilders are very diligent about keeping track of the number of units sold, in this case counting house sales that are closed. And as with cars, the units range from low end to high end, varying in size, features, and price. But each builder's typical product mix doesn't change radically from one year to the next. Therefore, might it be possible to make some sense of their warranty costs per unit, in addition to warranty costs as a percentage of sales revenue?
If we're going to do this, we have to restrict the metrics we use. We can't use claims, because some of those costs come from homes sold last year, the year before, and in certain cases, up to a decade before the closing took place. So there's no timely correspondence between when a home was sold and when it generated claims costs.
The same goes for warranty reserves. The balance goes up and down for a variety of reasons, many of which have to do with the amount of warranty costs per unit. But there is a much more direct relationship between warranty accruals made and units sold, which allows us to measure cost per unit.
Warranty Accruals per Home
Theoretically, when a product is sold, the manufacturer sets aside an amount of money that's sufficient to pay its warranty costs. For a new home, the industry average is around $2,500. So every time a home is sold, the builder should set aside $2,500. However, for some builders, the average is closer to $3,000. For others, it's closer to $1,000. Whatever it is in each particular case, this is the amount a builder should set aside after each closing. It's their best guess of what a new home's warranty will cost them over the periods of coverage (typically, one year for the walls, two for the heat, and 10 for "major structural defects").
Averaged over the entire industry, the statistics look stable. But for individual builders, the accrual rate per home sold has ranged from $0 to $15,000. So what we're actually measuring is the deviation from the average, for both the individual builder and for the industry as a whole. And while this has some bearing upon the quality of a home, it has more to do with the quality of the estimates that prompt the accruals to be made.
In Figure 1 we're detailing the total amount of industry accruals made by some 55 builders over the past 14 years. We believe they represent somewhere between a quarter and a third of all new homes sold in the U.S., based on their count compared to industry statistics. In 2016, for instance, this group sold 173,000 homes, representing around 31% of the 561,000 new homes sold in the U.S. In 2015 they reported selling around 30% of the 501,000 new homes sold.
Last year, these builders reported making $520 million in warranty accruals, up significantly from $433 million the year before. As can be seen in the chart below, their peak year was 2005, and the bottom of the trough was in 2011, which corresponds precisely with industry sales figures. In other words, warranty accruals rise and fall proportionally with sales, more or less.
Figure 1
New Home Warranties
Accruals Made Worldwide by U.S.-based Builders
(in US$ millions, 2003-2016)
Figures are collected quarterly, and then are compiled into annual figures at the end of each year. We also have collected figures for the first quarter of 2017, but they're not yet quite ready for publication, because we're still awaiting an annual report from Cavco Industries Inc., the second-largest manufacturer of factory-built homes.
A year ago the company reported making about $6 million in accruals during the months of January through March. Without this year's figure for Cavco, we've already counted $101 million in accruals for the first quarter of 2017, which is up slightly from the first quarter of 2016.
Because we already have seven charts in this week's newsletter, we won't display a chart of new homes sold, but the raw data is easily obtainable from the U.S. Census Bureau. And we won't display a chart of dollar sales revenue either, since that data can be easily obtained from the financial reports of the builders. However, we will include a chart that takes the accrual totals detailed in Figure 1, and divides them by both units sold and dollar sales totals.
Homes Closed
For this metric, it's important that we count homes actually sold as opposed to homes built or homes ordered. However, each homebuilder uses a slightly different word or phrase to refer to homes sold. D.R. Horton, Shea Homes, and Meritage call them "homes closed." William Lyon Homes refers to the "number of homes closed." PulteGroup, Beazer Homes, and the Ryland Group simply call them "closings." Meanwhile, KB Home, Hovnanian, and M/I Homes call them "homes delivered." Lennar refers to "deliveries." Standard Pacific Corp. used to count "new homes delivered." MDC now calls them "new home deliveries," though it used to refer to "homes closed" until it switched to the new term in 2011. And NVR simply calls them "settlements."
In Figure 2, the accrual rate per home sold uses the scale on the left side, while the accrual rate per dollar of sales uses the right-hand scale. The rates have varied widely over the past 57 quarters, but let's agree that the long-term averages are very close to 0.8% of dollar sales and $2,500 per home sold.
Figure 2
New Home Builders
Average Warranty Accrual Rates
($ per home and % of revenue, 2003-2017)
The most notable recent trend in Figure 2 is the divergence between the two lines. Accrual rates per dollar have been generally declining since 2011, but accrual rates per unit have been rising. How is this possible? Rising home sales prices would do it, specifically, prices rising faster than expenses. For instance, last year's $2,500 on a $250,000 home made for a 1.0% accrual rate. But this year's $2,600 on a $325,000 home produces only an 0.8% accrual rate, because prices rose faster than expenses.
Seasonal Fluctuations
Home sales are very seasonal, but the accrual rates shouldn't be. If a home is sold in winter, its average warranty cost should be about the same as one sold in summer, given that its warranties will cover four, eight, or 40 seasons (depending upon the component being covered). And more importantly, if a home is not sold, there are no accruals made. Therefore, absent any change in product quality or repair costs, accrual rates should remain the same, both per unit and per dollar. Yet there is a noticeable seasonal effect in the data, which would disappear if we switched back to an annual format.
Many manufacturers try to smoothen their own data by reporting their warranty expenses for three, six, nine, and 12 months over the course of a year. Some of the builders do this as well, but most report expenses for three, three, three, and 12 months. Either way, they do seem to pack proportionally more expenses into that final year-ending quarter, which may contribute to the seasonal effect. But as we will show in these subsequent charts, there are some builders that keep their accrual rate per unit about as flat and steady as possible.
In Figures 3 through 7, we've detailed the accrual rates per home sold for 15 of the largest new home builders. We're not including any makers of factory-built homes, mobile homes, or recreational vehicles, though some of them are part of the statistics in Figures 1 and 2. So there's no reason for us to wait for Cavco, because we're looking at just site-built homes. Furthermore, with only minor except
ions, we're also looking at only the builders of single-family homes.In Figure 3 we're looking at four of the biggest publicly-traded homebuilders: D.R. Horton Inc.; Lennar Corp.; Meritage Homes Corp.; and Toll Brothers Inc. Just by the shape of the lines we can see that D.R. Horton is the best marksman among them, keeping its accrual rate per home extremely close to its $1,200 long-term average. In fact, it was only towards the end of 2005 that the company deviated more than $300 above its long-term average, an over-accrual that the company quickly corrected in 2006.
Figure 3
New Home Builders
Warranty Accruals Made per Unit Sold
(in US Dollars, 2003-2017)
Meritage, meanwhile, has stabilized its warranty accruals per home around the $2,000 level, though that average rose in 2016 and the beginning of this year. Lennar and Toll Brothers have always accrued more per home, but their averages are also creeping up in recent years.
There is a correlation between the average selling price of a home and its average warranty costs. So it's no surprise to see Toll Brothers and Lennar setting aside $4,000 per home while Horton sets aside only $1,200. As a percentage of dollar sales, their accrual rates are closer together. For instance, Horton accrues 0.5% of sales while Toll Brothers accrues 0.6%. But because they serve different markets with different-priced homes, the accrual rates per unit are also different.
Measuring Flatness
What matters more is the shape of the lines over time. The flatter the line, the better. Downward slopes are better than upward slopes, but horizontal is just fine. That means a company knows its warranty cost per unit precisely and doesn't play accounting games when deciding how much to accrue per quarter (banking a little extra when times are good, or cutting accruals when earnings need a little help, etc.).
Then again, there are other factors, such as the mold and moisture problems some builders have recently encountered, or the odor-emitting drywall imports they had to deal with a decade ago. And although the ideal is to simply count the closings and multiply by the average to calculate the accruals needed each quarter, that's not the way that some warranty finance teams work (even if they say they do in their methodology disclosures).
In Figure 4, we're tracking three more builders: KB Home; M.D.C. Holdings Inc.; and PulteGroup Inc. Again, flat lines are better than sine waves, and downward slopes are better than upward. And by that measure, MDC is most improved, though not lately, KB Home has a problem with rising costs, and Pulte still needs to dampen its seasonal tendencies.
Figure 4
New Home Builders
Warranty Accruals Made per Unit Sold
(in US Dollars, 2003-2017)
Notice also that in the fourth quarter of 2007, KB Home made no net accruals at all. In other words, the amount of accruals it listed in its annual report that year was the same as the amount it reported in the first nine months of the year. That doesn't mean that every one of the 8,132 homes it sold that quarter was built perfectly and therefore didn't need any accruals. We suspect it had more to do with the $929,000 net loss the company was also reporting that year, and the need to cut expenses so as not to make it worse.
Extreme Accrual Rates
In Figure 5 we can see that Beazer Homes made no quarterly accruals on three occasions, while during one scary quarter it set aside nearly $15,000 per home (the math is $54 million for 3,600 homes sold). We suspect that some or perhaps even most of that $54 million was for homes sold in previous years that were as of early 2005 suffering from moisture intrusion and mold problems. Indeed, Beazer Homes has not accrued anywhere near $54 million in any year since then, let alone a single quarter. So it set aside extra funds during a single quarter, which led to the spike that can be seen in Figure 5. In other words, it can be explained.
Figure 5
New Home Builders
Warranty Accruals Made per Unit Sold
(in US Dollars, 2003-2017)
Likewise, the 2009 spike for Hovnanian can be explained away by the coincidence of timing. But for Beazer Homes, it happened once and never again. For Hovnanian, it seems to happen every year or two. And for NVR, it happened twice in 2013 alone. Then again, for all three companies the past three years have been relatively stable, with accrual rates per home sold not rising or falling by much. Beazer Homes, in fact, completely lost its seasonal pattern in 2009.
Stability and Predictability
That kind of improvement in stability and predictability is even more apparent in Figure 6, where we're tracking three additional builders: M/I Homes Inc.; Shea Homes LP; and William Lyon Homes. For M/I Homes, it's as if a switch was thrown in early 2005, which eliminated all seasonality from its accrual rates, and which in the dozen years since has kept the company very close to its long-term accrual rate of $2,200 per home. That's an even more stable and predictable pattern than was seen with D.R. Horton, though it's at a higher level per unit.
Figure 6
New Home Builders
Warranty Accruals Made per Unit Sold
(in US Dollars, 2003-2017)
By the way, Shea Homes is no longer reporting its financials publicly. It did so from 2010 to 2014 only because it sold bonds on the open market that required it to file annual reports and quarterly financial statements with the U.S. Securities and Exchange Commission. When those bonds were redeemed, it stopped reporting its financials to the SEC, though it continues to build homes, and presumably, to also make warranty accruals, at rates that varied from $2,800 to $11,000 per home sold over some 20 quarters.
Post-Merger Stability?
And finally, we wanted to provide an update specifically on the Ryland Group and Standard Pacific, which merged in late 2015 to form the CalAtlantic Group Inc. In the chart below, Ryland stopped reporting its warranty expenses in the middle of 2015 after the closing of the acquisition, and Standard Pacific then changed its name to CalAtlantic. So there are really three companies tracked in the chart. Notably, for the past 18 months, the combined companies have managed to keep their accrual rate per home sold very stable, within a tight range of $1,500 to $1,700. That's better than either predecessor company was ever able to do on its own.
Figure 7
New Home Builders
Warranty Accruals Made per Unit Sold
(in US Dollars, 2003-2017)
It's a different way of looking at the warranty expenses of a given company. But it's only possible when the product is somewhat homogenous and unit sales are somewhat reliably counted by an outside source. That's possible with passenger cars and single-family homes. We're open to suggestions of other instances where it can be done.
However, with both passenger cars and single-family homes, we have to keep in mind the tremendous range of features and prices. Therefore, we're not in any way implying that setting aside $1,000 towards the warranty expenses for a $100,000 home is somehow better or worse than setting aside $4,000 for a $400,000 home. What matters more is the stability and predictability of accrual rates over time.
For instance, in the first quarter of 2017, Horton accrued $1,413 per home while Toll Brothers accrued $4,289 per home. Yet Horton's accrual rate was 0.5% of home sales revenue while Toll Brothers was at 0.6%. But that's because they're serving different markets, with different products and different prices. Rather than compare them to each other, it makes more sense to compare each company to itself over time.
Celebrating Predictability
In other words, it's not whose highest or lowest, but which lines are flattest, even if their slope is upwards because accrual rates are rising. Even in those cases, expenses per unit could be rising more slowly than selling prices, so accrual rates per dollar could actually be falling. Actually, that's what's happened in the entire industry since 2011, as was detailed in Figure 2.
meanwhile, in the preceding series of charts, it was clear that several builders, including D.R. Horton, Lennar, Meritage, Beazer, and M/I Homes, have in recent years kept their accrual rates per unit sold relatively stable and predictable. Others have not. And that seems to have less to do with the quality of the homes, and more to do with the quality of the financial experts making the predictions and deciding upon the accruals.