1. The latest claims rates for Navistar is taken from the end of 2005, because the company is two years behind in its SEC filings. Beazer Homes is now six months late with its SEC filings. Black & Decker Corp.; Eaton Corp.; Exide Technologies; General Electric Co.; Goodyear Tire & Rubber Co.; Honeywell International Inc.; Illinois Tool Works Inc.; Motorola Inc.; and Xerox Corp. have chosen to reveal their warranty spending only once a year, so their claims rates were taken from the end of 2006. All other figures are current as of Sept. 30, 2007.

Overall, 45 of the 100 companies managed to reduce their warranty claims rates in the year between Sept. 30, 2006 and 2007. And 44 saw an increase. The remaining 11 are either late with their SEC filings or have decided (against FASB FIN 45 accounting rules) to file their warranty accounting figures only once a year (usually in March).

At the very bottom is Microsoft, which of course very publicly in the summer had to radically increase the funding for its Xbox 360 game console product line. This is also a very good example of how we calculate claims rates. We don't take a company's total revenue -- for Microsoft that figure would be $13.7 billion in the third calendar quarter of 2007. Instead, we take the revenue of just the division that makes the warranted hardware -- a much more manageable $1.9 billion in this case. And we use the $86 million in claims to calculate a 4.5% claims rate. That's on the high side, but not as high as some of the companies that have also made the news for their warranty crises in 2007.

In Tables 2 and 3, we're looking at the other side of the warranty accounting ledger. Claims are what a company must pay now to repair products sold in the past. Accruals are what a company sets aside now to finance predicted future claims. Therefore, if claims are what happens to you, accruals are what you predict will happen. And while you can't control the weather, even if you're the weatherman, you most certainly can control the weather forecast.

Quality Signals Through Accruals

In this way, the accruals rate signals what a company believes it will cost to repair product in the future. If a company believes quality is increasing, it will reduce its accrual rate. If it expects an Xbox-like shock in the future, or if labor rates for warranty work are soaring, it will radically increase the accrual rate. Changes in the accrual rate can signal changes in expected quality or repair cost, or both, especially when compared over multiple years.

The claims rate suffers from a lag between when a product is sold and when it must be repaired. For instance, a product sold in the fourth quarter of 2006 might need a warranty repair in the middle of 2007. So while we measure current claims against current sales, what we should really be doing is time-shifting the calculation so that current claims are compared against 2006 revenue. But that gets messy because companies don't announce the age of the products they're repairing -- only the cost. So time-shifting may actually introduce more uncertainty and errors than it removes.

Also, in industries such as housing, where revenue is now shrinking, the claims rates are disfigured by the sales declines, because homes sold last year are being repaired this year. If sales have fallen by half in the interim, that alone could double the calculated claims rate, even if there's been no change in quality, repair cost, or incidence per 1,000 units.

The accrual rate, however, suffers no such time lag. When a product is sold, it's up to the company to predict how much future claims will cost it, and to set that amount aside in a special fund -- the warranty reserve -- so it will be available when the time comes. In other words, if at any point sales should completely cease, there should always be enough in the warranty reserve to pay off all future warranty claims, because all that money should have been set aside at the time of sale.

The accrual rate is therefore not disfigured by changes in sales volume. If sales go down, less is set aside. If sales go up, more is set aside. So if a company believes a $100 product will cost it $3 in future warranty claims, it will set aside 3% whether it sells 10 or 10 million units.

Reasons for Changes in Accrual Rates?

In Table 2, we're looking at declines in accrual rates. It Table 3, we're looking at increases. Sometimes, the reason for the change is known, such as in the case of IBM (sold off its laptop division) or Apple (the battery crisis has passed). Sometimes it's unknown, or is just unannounced to the public. But here's how to do your own fact checking: compare Tables 2 and 3 to Table 1: If the claims rate has also changed in the same direction as accruals, the company is adjusting its set-asides to match actuals. If they don't match, the company is predicting a peak or trough before the tide actually turns.

Source: Warranty Week from SEC data   

Take, for instance, Fleetwood Enterprises Inc. The recreational vehicle maker managed to cut its claims rate from 4.9% to 2.8%, so of course it's also cutting its accrual rate. But then there are the unexplained cases, such as Briggs & Stratton, which cut its accrual rate from 2.4% to 1.9% despite an increase in claims.

In Table 3 the list of top ten accrual increases includes warranty newsmakers such as Microsoft, Palm, and BorgWarner, but it also includes anomalies such as Tyco International. The reason Tyco made the list is because it spun off a division that generated less warranty activity than the divisions that remained behind. And because we calculate these percentages for the company as a whole, it appears on our list as a 45% increase.

Source: Warranty Week from SEC data   

In Tables 4 and 5, we're looking at the annual change in the ending balance of the warranty reserve. In cases such as Tyco's, the spin-off resulted in a 33% reduction in the size of the parent company's warranty reserve -- from $245 million on Sept. 30, 2006 to $164 million on Sept. 30, 2007. IBM and Eastman Kodak also were making divestiture-related adjustments.

For companies such as Centex, the accrual rate is more or less stable, but sales are falling, therefore the warranty reserve is contracting in proportion. However, in other cases, such as Quantum and NCR, there doesn't seem to be any good or bad news supporting the reduction in in reserves. Claims and accrual rates are both increasing, while sales are falling, and it just so happens that the math works out that way.

Source: Warranty Week from SEC data   

In Table 5, we're looking at the ten largest increases in warranty reserves. Microsoft is of course the leader in this regard, having boosted its warranty reserves by almost a billion dollars in June of last year. By September it spent some and added some more, making the year-on-year comparison $857 million as of Sept. 30, 2007 and what we're estimating to be around $10.75 million as of Sept. 30, 2006. That's a 7,872% boost!

To put it another way, Microsoft boosted the size of its warranty reserve from roughly three months of coverage in 2006 to almost 30 months by 2007. That would be prudent given that its Xbox 360's warranty coverage jumped from 90 days to three years in the same period. We're not listing warranty reserve capacities here today, but it's another useful metric when comparing one company against itself over time.

Source: Warranty Week from SEC data   

Taken together, the five tables provide a snapshot of how American manufacturers have managed their warranty spending over the past year. For those keeping score at home, only IBM hit the trifecta of making all three top ten decline lists. BorgWarner and Microsoft made all three top ten increase lists. Tyco and Thermo Fisher also made three lists, but in conflicting directions. And 35 other companies made at least one top ten list.

When the remainder of the 2007 data arrives in March, we'll compare not only the past year, but also the past five years of warranty spending. For while it's difficult and unwise to directly compare companies against each other, it can be highly revealing when a company is compared against itself over such a long period of time.