Homebuilders may not yet be seeing increasing sales, but at least they're not seeing increasing warranty costs. It may not yet be getting better, but at least it's not getting worse. That's our most optimistic opinion: the glass may be only half-full, but at least it's not leaking.

Last year, in newsletters published on May 19 and September 8, we searched in vain for the elusive "bottom" of the U.S. housing industry's recession in the sales and warranty expense data we collect from publicly-traded homebuilders. We still haven't found it.

Scraping the Bottom?

If anything, it looks like we've been scraping along the bottom for almost three years, with no upturn in sight. But at least it's not still declining. We now have all the third quarter 2011 data in hand, and some of the early fourth quarter figures as well, and we've identified a seasonal pattern in the recent data. Since 2009, the first quarter has always been the slowest of the year, in terms of not only sales, but also claims paid and accruals made. And then the second, third and fourth quarters are a little better. This pattern can be seen in the chart below, which tracks accruals made by homebuilders in millions of dollars per quarter.

Note that the peak came in 2005, in a quarter whose accrual total was more than $60 million higher than any that came before or since. This is much sooner than the peak in housing sales, which most experts say came in 2007. However, we should note that this week, we're tracking not only the traditional homebuilders, but also the makers of mobile homes and recreational vehicles. So that could be throwing off the trend.

The lowest total for quarterly accruals came in the first quarter of 2011, which is more recent than most estimates of the end of the recession. That first-quarter total was also lower than the first quarters of 2009 and 2010. However, each of these was the lowest of their respective years, which is what one would expect given the weather in January, February and March. No doubt, when the 2012 data becomes available in late May, it will also show a sequential drop. But will it be above or below 2011's $55 million low?

Accruals bear some watching, because the accrual rate is usually kept constant, so it rises and falls with sales. We're tracking a 14% decline in new home and RV sales from the third quarter of 2010 to the third quarter of 2011, however, and only a 3% decline in accruals. So perhaps there are other factors at work?

Worried About the Future?

It's possible that homebuilders are making excess accruals in anticipation of increased claims payments for problems such as mold removal or Chinese drywall replacement costs. But given the battering their sales volumes have taken in recent years, we suspect there's a much simpler reason: they haven't reduced accruals as fast as sales fell.

One reason is what happened with claims payments. In Figure 2, we're tracking the quarterly payments by some 57 homebuilders and RV makers from the beginning of 2003 to the third quarter of 2011. This time, the peak comes in late 2006, though once again the low water mark is in early 2011.

The reason for the time lag is, truthfully, the time lag. There's always a delay between when a house or RV is built and sold, and when it needs warranty work. So while the accruals are made at the time of sale, the claims are paid months or perhaps even years later.

Furthermore, in an industry where sales are in free-fall, the ratio between claims payments and sales revenue is going to get a bit wacky. That's because while the claims are generated by defects in past years' sales, the claims rate is calculated using the total for this year's sales.

Warranty Costs vs. Sales

The effect can be seen clearly in the data in Figure 3 below. Here, we've taken the claims and accrual totals from Figures 1 and 2, and divided them by the quarterly sales totals for all of the builders. Keeping in mind that the peaks in claims and accruals were in 2005 and 2006, one can see how the claims rate began to rise in 2006 and never came back down. This was because sales were falling, and builders were fixing old homes with new revenue.

The peak in claims as a percentage of sales didn't actually come until 2010, when the rate hit 1.9%. But notice how bumpy the green line for accruals has also been. It should be flat. But it goes up and down with the seasons, implying that homes built in the warm months are somehow better than those built in the cold months.

What we think we're seeing is how the homebuilders are modifying their accrual rates not to reflect their predictions of future costs, but to cover current costs. In other words, their accrual totals don't go up and down with sales. They go up and down with claims payments. And, as we've noted in past reports on homebuilder warranties, sometimes they cease making accruals for a quarter or two if they detect that they're making excess accruals.

Seasonal or Secular Trend?

As we said at the outset, we're still looking for the bottom of these recessionary curves in this particular industry. And based on what we see in Figures 1 and 2, we won't really know until May at the earliest if the 2012 totals are higher or lower than the early 2011 totals. The increases we're seeing in the second and third quarters are seasonal patterns. So even if the fourth quarter is up, as we expect it to be, it would probably still be part of that seasonal pattern, rather than a lasting trend.

We can't find much that's hopeful to say about the warranty reserve data we've seen so far, except that the balances haven't declined as fast as sales. And this suggests that an excess amount of reserves has accumulated.

As can be seen in Figure 4, the industry's aggregate reserve fund balance continues to fall from its $1.5 billion peak in 2006. As of September 2011, it stood at $723 million, by far its lowest level since at least 2003 (homebuilders didn't report any warranty metrics until 2003).

The warranty reserve fund balance is calculated by adding the accruals and subtracting the claims. Therefore, it should come as no surprise that since the industry's average claims rate has been above its average accrual rate since 2007 (Figure 3), the balance has been falling (Figure 4). Subtractions exceed additions, so the total declines.

Ratio Between Reserves & Claims

But there's another way to measure warranty reserves besides in dollars. We can also measure the level of reserves as a multiple of the claims payments per month. In other words, if a company has a $10 million warranty reserve balance and is paying claims at a rate of $1 million a month, the reserve would have a capacity of 10 months. If the balance grows to $12 million but claims payments fall to $800,000 per month, the reserve fund's capacity would rise to 15 months.

For all manufacturers of all kinds of warranted products, the ratio between claims and reserves has remained in a range of 16 to 18 months for almost nine years. In other words, the average manufacturer keeps as much in reserves as they pay out in claims in a year and a half.

In the homebuilding industry, the ratio has been a bit lower. From 2003 to 2006, it remained within a range of 12 to 15 months, as can be seen below in Figure 5. That was below the ratio for all manufacturers, but we suppose, was entirely appropriate for new home warranties lasting for one, two or ten years, depending upon the covered component.

But then something happened. The ratio began to rise, past a year and a half and even past two years. Was it Chinese drywall? Was it mold? We'd suggest it was a bit of those, but it was caused mostly by a failure to reduce accruals as fast as sales fell.

Since 2007, the homebuilders have kept proportionally more reserves than other manufacturers, while the reverse was true until 2006. And we think it's an accounting error, not a preparation for higher claims payments. Why? Because the higher claims payments, if they had ever come to be, would have taken the ratio back down below 18 months.

Yes, the claims rate is higher than ever. But that's because sales are lower than ever. As can be seen in Figure 2, the actual claims payments (in dollars) are at a low point. It's claims as a percentage of sales that's spiking. And that's not because of claims payments. It's because of sales, or more precisely, because of the lag time between sales and claims payments.

Excess Warranty Reserve Example

Let's give an example of this. In Figure 6 below, we've charted the claims and accrual rates for Toll Brothers Inc., a builder of luxury home communities with a presence in some 20 states across the U.S.

The shape of the curves in Figure 6 is more or less similar to the shape of the industry averages in Figure 3. The claims rate used to be below the accrual rate, and reserves used to accumulate. Now, the claims rate is higher, and reserves are being depleted. And that's what we're implying: that Toll Brothers is more or less a typical case for the homebuilding industry in general.

For five years, the company's accrual rate was more or less steady and the claims rate was almost always lower. Claims payments actually peaked at the end of 2005, but hefty sales volumes kept the claims rate down.

Claims were falling in 2006 and 2007, and then sales began to sink as well. But then in 2008, claims rose, and in sympathy, Toll Brothers raised its accrual rate, nudging it above 0.9% in 2008.

Rising Ratio Post-Recession

This was unnecessary. How do we know? The reason can be seen in Figure 7. In those "good" years of 2003 to 2007, reserve fund balances were rising, and the ratio between reserves and claims remained flat at around 24 months. But then the recession hit, and the ratio went haywire. In early 2011, as claims payments reached a new low, the ratio between reserves and claims hit a new high of 78 months.

Who needs six-and-a-half years worth of reserves to pay claims for one or two years? Actually, the typical new home warranty lasts for ten years, but only major structural problems are covered for that long. Heating and air conditioning systems and other mechanical equipment are typically covered for only two years. And everything else gets only a year's coverage.

Blend all those warranty durations together and you get exactly what the data in Figure 5 says is typical for the entire industry, at least before the recession: something between 12 and 18 months of capacity is what's needed in the typical homebuilder's warranty reserve fund. Anything more is excess.

Warranty Windfall?

Toll Brothers, however, has kept its ratio above 36 months for more than three years. The excess in the reserve fund, we believe, is therefore on the order of $12 to $20 million. In other words, Toll Brothers alone has between $12 and $20 million more in reserves than it needs to fund its warranty claims. And given that its warranty reserve fund balance has ranged between $42 and $46 million in 2011, we're saying that somewhere between 25% and 45% of Toll Brothers' reserves are excess, depending upon whether they aim for the 12-month or the 18-month level.

For a company with a $47.7 million operating loss in its most recent fiscal year, that's a large sum of money to have sitting around doing nothing. Granted, the company has been taking some of the excess funds out of its reserve fund in recent quarters. But the ratio between claims and reserves has remained stuck at 36 months for the past two quarters, covering the months of May to October 2011. We think Toll Brothers can take out lots more before imperiling its ability to pay claims.