In this third and concluding part of our series on warranty estimates, we'll take a look at the warranty cost projections and average ratios of several segments of the building trades.

In recent years, it's been common to read about a homebuilder or a supplier that went bankrupt and was unable to honor its warranties as a result. The reason is that little or no money was set aside after the closing to finance predictable warranty claims. Those expenses were paid out of cash on hand, and so when the cash dried up, the company declared bankruptcy and stopped all warranty work.

Fortunately, most manufacturers are better prepared than that. They usually set aside a credible amount of cash at the time a product is sold, establishing a warranty reserve that they then draw upon to pay claims costs.

In the past few newsletters, however, we've found that while a company may be writing warranty liabilities that last 24, 36 or even 48 months, the amount of cash they're setting aside wouldn't typically last that long. And so, like many pension funds, we're finding that some warranty reserve funds are under-funded liabilities. And some industries are more under-funded than others.

On the other hand, we've also found that some industry segments are quite organized and predictable with their warranty accounting and estimation. Like a marksman, they have good aim, and the data points produced by their key ratios look like a rifle target after a sharpshooter's done with it.

Warranty Calculations

In each of the four charts below, we started out working with four warranty metrics and ended up with two ratios. The first metric we collected was the dollar amount of product sales for the period in question. The second and third metrics were the amount spent on warranty claims, and the amount set aside as warranty accruals, during the corresponding sales period. And the fourth metric was the balance in the warranty reserve fund at the end of each of the past 38 quarters (9-1/2 years).

We repeated these data collections for each of the quarters between early 2003 and the middle of 2012. Then we grouped the companies by the types of product they sold. If they sold more than one type of building material, for instance appliances and furniture, they were counted in each category into which they belonged.

Then we took the four basic metrics and combined them to create two ratios. The amount set aside as warranty accruals was divided by the dollar amount of product sales for the period in question. This created a percentage that is labeled in the charts below as the accrual rate. For instance, if a company sells $10 million worth of carpeting in a quarter and sets aside $100,000 in warranty accruals over the course of those same three months, its accrual rate would be 1.0%.

Then we took the ending balance in the warranty reserve fund, and divided it by the amount being spent per month on warranty claims. This created a metric we're calling the warranty reserve capacity, expressed as the number of months of claims coverage in the reserve fund. For instance, if the reserve fund contained $9 million and the company was spending $500,000 a month on warranty claims, the reserve capacity would be 18 months.

The charts below track two key assumptions that a company has to make: how much money should we set aside to pay future warranty claims, and are we keeping enough in the reserve given the duration of our warranties? In each industry, and for each product line, the answers will be different.

And of course, there's no one right answer for all the companies in a given industry. But by showing the averages of all the companies in a given industry or industry segment, we're tracking the best estimates of all those companies combined. So what we're really doing is charting the warranty estimates of all a given industry segment's members as a scatter plot.

Warranty Accounting Theory

A company is supposed to set aside enough funds at the time a product is sold to cover all the warranty liabilities the company expects for that product. So if the builder of a $200,000 home expects $2,000 in warranty expense over the life of that new home's warranty, they're supposed to set that amount aside at the time of sale. They may not spend it all. They may spend more. But they're supposed to estimate the cost at the outset and accrue that amount.

If the warranty were to last for 24 months and if sales remained flat, the result of that accrual estimate would cause the company's accrual rate to remain flat at 1.0%. And if claims were paid at a constant rate of $250 per home per quarter for two years, the claims rate would also be flat at one percent.

After a while, the warranty reserve fund would contain exactly 24 times as much as was spent in a month, which would give it a reserve capacity of 24 months. And this would be fortunate, because the company's warranties would expire after that amount of time, just as the last of the accruals were spent on each home.

Warranty Accounting Reality

If, for whatever reason, the company stopped building homes, there should be exactly enough money in the warranty reserve fund to pay all claims until the last warranty expired. In the homebuilding industry, this is a very real concern. In a bankruptcy proceeding, the warranty is a liability, and the homeowner is a creditor. And if the reserve fund is empty, it's an unfunded liability. And the creditor is probably out of luck.

However, the basic theory remains valid: At any given moment, there should be enough money in the warranty reserve to pay all remaining claims for products already sold. And if a company's products carry a 24-month warranty, the capacity of that warranty reserve fund (its balance divided by claims paid per month) ought to be somewhere close to 24 months.

In the figures below, we're measuring the real world fluctuation of a given product type's warranty metrics. The value of the vertical axis should be close to its average warranty duration. The value of the horizontal axis should be close to its expected warranty cost.

In Figure 1, for roughly 57 past and present homebuilders, we've charted the average accrual rates and reserve capacities for each of the 38 fiscal quarters since 2003. And as is obvious, there is no real pattern to the data. Accrual rates range from 0.9% to 1.3%, and reserve capacities range from 12 to 27 months.

The thin dotted lines represent the industry's averages as of June 2012: a 1.06% accrual rate and 21.83 months of reserve capacity. Most of the other data points, however, are below and to the right of this center point, implying that average accrual rates are usually higher and reserve capacities usually lower than they were in June 2012.

In Part 1 & Part 2 of this series of newsletters, we found other industry segments with equally chaotic patterns. For instance, in the November 15 newsletter we found a wide dispersion of data for the recreational vehicle industry. There are actually 11 companies that we count as both homebuilders and RV makers (they sell manufactured homes, prefabricated homes, double-wide trailers, etc.). And both the site-built and the mobile home industries have been hard-hit by the recession. So perhaps the lack of a recognizable pattern in their warranty data is caused in part by their financial challenges?

HVAC Warranty Metrics

If they all looked like Figure 1, we'd have to say there's no pattern in the warranty data for any industry. However, in Figure 2, it's clear that at least the makers of heating, ventilation, and air conditioning systems are pretty reliable when it comes to making their warranty estimates. And while the appliance makers let their accrual rates roam over a wide range, they're pretty good at keeping the ratio between their warranty reserves and claims payments between 9 and 18 months.

To get the data in Figure 2, we tracked 37 past and present HVAC system manufacturers and 36 appliance manufacturers. Notably, there was no overlap between the two groups, with companies such as United Technologies Corp. and Johnson Controls Inc. in the HVAC category and companies such as Whirlpool Corp. and Jarden Corp. in the appliance category.

Even knowing nothing about the industries, the data in Figure 2 suggests that the warranties are longer in the HVAC segment, though the warranty work on appliances costs more as compared to their purchase price. In other words, the maker of a $5,000 HVAC system may expect an average of $40 of warranty work per unit, while the maker of a $2,000 refrigerator may also expect $40 worth of warranty expense per unit. For one, the warranty accrual rate would be 0.8%, while for the other the accrual rate would be 2.0%.

The clustering of the data around the vertical axis suggests that the HVAC manufacturers are keeping roughly 24 months' worth of warranty reserves on hand at any given moment. This is significantly shorter than the average duration of the warranties they issue. The appliance makers, meanwhile, are keeping their warranty reserve capacity closer to 12 months. Given the recent trend towards shorter appliance warranties, this is probably sufficient.

Building Materials

Next, we'll look at the suppliers of three other types of building materials. In Figure 3, we're tracking 20 suppliers of electrical and plumbing fixtures, 30 makers of furniture, and 45 manufacturers of other types of building materials such as roofing, windows, flooring, doors, and wall covering.

There is no overlap between the members of each category. However, as can be seen by the jumble in Figure 3, there is quite a bit of overlap between the data points for these three segments. What this suggests is that while the products are different, the warranty cost metrics really aren't. Whether it's a door or a faucet, it's probably going to require an accrual rate around 0.7% or 0.8%, and a warranty reserve balance that's 12 to 15 times the average monthly bill for warranty work.

In other words, they're not all that different. However, each segment has a few outliers in different directions. The building material companies have allowed their warranty reserve capacity to grow beyond 24 months on occasion. The furniture makers have set aside more than one percent of revenue on occasion. And the fixture makers have let their reserve capacity rise a little high at the same time that they let their accrual rates fall a little low.

For the most part, however, all these manufacturers have clustered their warranty estimates in the same area. In June 2012, their collective average was 0.74% accruals and 16.8 months of reserves. That's the approximate center point for accrual rates since 2003, though it's much higher than the typical range for warranty reserve capacity. For that metric, 12 to 15 months is the more typical range.

Power Equipment

In Figure 4, we've charted the warranty reserve capacity and accrual rates of 67 manufacturers of electrical power generating equipment (alternating current) and 26 manufacturers of battery storage and direct current equipment.

There's an overlap of 11 companies between the two groups, but as can be seen in the figure below, there's really no overlap between their warranty metrics.

The makers of alternating current power equipment tend to have higher rates of accrual, and higher levels of warranty reserves. They also tend to cluster around 1.0% to 1.5% and 17 to 21 months, though there are a few data points outside that range.

The makers of batteries and other electrical storage systems tend to accrue less, though their accrual rate averages are spread across a much wider range. So are the multiples between their monthly claims cost and warranty reserve balances.

In fact, what seems to be the case is that the AC equipment makers are more organized and predictable, while the DC equipment makers are more disorganized and unpredictable. A big reason for that pattern may be the age and maturity of the technologies involved, however. The AC power grids are leveraging more than a century of technology, while the DC power companies making electric cars and mobile electronics charging systems and batteries are still experimenting.

The Best Marksmen

Besides the AC power equipment industry segment, the most organized of this week's groups is the HVAC segment in Figure 2. The least organized are the homebuilders (Figure 1), appliance (Figure 2) and the DC power equipment makers (Figure 4).

In the automotive sector, the most organized segments were the truck, mining equipment and farm equipment companies. The least organized were the RV makers, bus manufacturers, and emergency vehicle makers.

In the high tech sector, the most organized segments were the computer OEMs, the LAN and datacomm equipment makers, and the medical & scientific equipment manufacturers. The least organized were the makers of computer peripherals, the data storage companies, the makers of semiconductor manufacturing equipment, and the makers of satellite and microwave gear.

In general, the data points of the best organized fell along a horizontal line, meaning that while their average accrual rate varied, the ratio between claims and reserves remained steady.

In contrast, a vertical bunching would mean that accrual rates remained steady while the ratio between claims and reserves varied. This would be less than ideal because while accrual rates can fall for very good reasons (increased reliability, lower repair costs, etc.) the average duration of a company's warranties is likely to remain the same for long periods of time. So what a vertical bunching would expose is a constant liability that was sometimes more fully funded and sometimes less than fully funded.

Fully-Funded Warranty Liabilities?

In only a few cases did we see warranties that were close to being fully funded, however. In the automotive industry, for instance, the warranty reserve capacity was almost always less than 24 months, even though the warranties lasted for 36 or even 48 months.

In the high tech sectors, however, even though the warranty reserve capacities were generally lower, the warranty liabilities were also closer to fully funded. For instance, with the computer OEMs, the LAN and datacomm equipment makers, and the medical & scientific equipment manufacturers, the warranty reserve capacity was usually very close to 12 months. But so is the average warranty duration.

In the building trades, the HVAC manufacturers set the best example, keeping their accrual rates in a range of 0.7% to 1.0%, and their warranty reserve capacities between 20 and 30 months. Their warranties tend to be longer, so a higher warranty reserve capacity is worth having.

The bottom line is this: a product warranty, like a pension fund, is a future liability for a manufacturer. The warranty liability is supposed to be fully funded by the accrual made at the time of sale. If that were the case, then the ratio between monthly claims cost and warranty reserves would approximate the average length of a company's product warranties.

In some industry segments, product warranties are only a partially-funded liability. The sad fact in these segments is that reserves would not be sufficient to pay claims if a company suddenly ceased operations. So the warranties are promises that would then have to be broken.

If extended warranties were funded this way, people would go to jail. For while it's common for manufacturers to go bankrupt and to then take their warranty liabilities with them to the bottom, it's equally common to see an extended warranty program go into runoff when a retailer goes broke, with reserves being used to pay all remaining claims until the last contract expires.