In a perfect world, different pairs of warranty metrics would be plotted on all six sides of a cube, which readers could rotate at will to examine whichever comparisons they want.

On a flat web page, however, with only two dimensions to work with, we can make only so many comparisons. Yet it really does take multiple comparisons to make sense of the trends in a given industry.

So first we'll look at the amount of claims paid by the major computer manufacturers over time. Then we'll compare their claims payments to their warranty reserve balances, and contrast that with their accrual rates. And finally, we'll compare their claims and accrual rates over time.

Warranty "Market Share"

In the first of six charts below, we've dropped all but the current top four vendors from the chart. And rather than simply drawing the rising mountain of claims payments in dollars, we've plotted each company's percentage share of the overall total. We did this to spotlight the incredible rise of Apple Inc.

By the third calendar quarter of 2012 (corresponding to Apple's fiscal year's fourth quarter), the company's share of the industry's total claims payments (nearly $1.5 billion altogether) had climbed above 33% of the total for the first time. This is an incredibly swift rise, considering that its share didn't cross 10% until late in 2009.

In contrast, Hewlett-Packard's share of the total, which had flirted with 60% in 2008, fell below 42% for the first time in the most recent fiscal quarter, which for them included the months of May, June, and July 2012. At this rate, in two or three more quarters, Apple will be the computer industry's largest warranty provider.

Apple passed IBM in late 2009 and passed Dell in mid-2010. On the way, it also passed GE, Cisco, Caterpillar, Deere, Navistar, and Boeing. Now all that's ahead of America's fourth-largest warranty provider is HP in the computer industry, and GM and Ford in the auto industry.

If Apple continues to grow and HP stands still, Apple is likely to surpass both HP and Ford within a year and become America's second-largest warranty provider.

On the accrual side, Apple already holds that title, based on the latest data. In the third quarter, GM accrued $919 million; Apple accrued $558 million; HP accrued $542 million; and Ford accrued $318 million. Caterpillar was fifth with $273 million, though we still haven't heard in 2012 from GE, which is likely in the same ballpark as Cat. No other U.S.-based company was above $250 million per quarter.

America's Largest Warranty Provider?

Therefore, given that claims are a backwards-looking metric and accruals are a forward-looking metric, it can be argued that Apple has already taken the second-place slot away from HP and Ford. In product revenue terms, it's also now pushed past HP and Ford, and was within $4.3 billion of catching GM as well.

In Figure 2, we've taken some of the data used in last week's newsletter and added in data for Dell, to make it into a chart of accrual rates and reserve capacities for the Big Four of the U.S. Computer OEM business.

The reason we left Dell out last week is its less-than-stellar marksmanship: we had to exclude five outlying Dell data points to stay within these axes. By the way, four of the five outliers were from 2003 -- before Dell learned how to shoot straight. But more on that later.

The Dell and IBM data points are mixed in with each other, suggesting they have similar warranty expense rates, generally close to the middle (the September 2012 industry average, as pinpointed by the dotted lines, is 2.06% and 10.6 months). And all four companies generally keep their reserve balances close to what they spend over the previous 10 to 12 months.

But what's most obvious from the data points in Figure 2 is that Apple and HP are at the opposite extremes of the industry. Apple is primarily below 2.0%, while HP is mostly between 3.0% and 4.0%. IBM and Dell are somewhere in the middle. And the middle, given the weight of Apple's rising sales, has now shifted to just above two percent.

The biggest drawback of Figure 2 is the lack of any sense of time. There are 39 data points for each company, covering the 39 quarters between the start of 2003 and the third quarter of 2012. But there's no way to tell which of the data points is oldest or newest.

In Figure 1, there is a definite sense of time to the data, but there's still no comparison to revenue. Apple may be paying out almost as much as HP, but there's no way of telling if claims are a larger or smaller percent of the company's total hardware revenue. And actually, Apple does sell more hardware, but pays out much less per unit for warranty work.

Comparisons to Sales Revenue

In Figures 3 through 6, we're providing both a sense of time and a relative comparison to revenue. Each company's quarterly report for claims paid and accruals made are compared to product revenue totals, yielding a pair of percentage figures. Over time, as rates change, these points form the red and green lines seen in the figures below.

In the computer industry, four percent is high and one percent is low. And though the figures of one company shouldn't be compared to another company's unless both companies count warranty expenses the same way, we can still make some general conclusions.

In Figure 3, it's clear that Apple has cut its warranty expense rates considerably since 2005 and 2006. It's also clear that costs have been rising faster than revenue since 2009. And it's clear that Apple is at the low end of our one-to-four-percent scale.

In Fact, Apple is the only top computer OEM to ever report either a claims or an accrual rate below one percent. But keep in mind that over the years, Apple has become less and less a computer manufacturer, with mobile phones, tablets and portable music players accounting for a much larger share of product revenue than Macintosh (see Figure 4 in the September 6, 2012 newsletter for more).

Extended Warranty Sales

In addition, Apple has aggressively pushed its own brand of extended warranty, which on the one hand has gotten it into trouble with European regulators, and on the other hand (we suspect) has allowed Apple to reduce its product warranty costs by providing it with another expense account to draw upon.

Hewlett-Packard also pushes its own brand of extended warranty, and also has gotten in trouble for it (in Australia, however). HP also has a changing product line, of which personal computers is just one part. However, HP's product line scales more towards the higher end, with models such as the Superdome, NonStop, and ProLiant -- hardly the domain of consumers. And it also has an extensive printer lineup, as well as calculators, scanners, and scanner/printer/fax all-in-one units, some of which generate above-average rates of warranty expense.

We mention this because while HP's warranty expense rates have been less volatile than Apple's, they've also been generally higher as well. In Figure 4, one can see a generally declining expense rate from 2003 to 2007, then a jump in 2008 and 2009, followed by another multi-year decline. But even with these improvements, HP's lowest expense rates are still higher than Apple's highest.

The reasons for this include the difference in their respective product lines and the average duration of their product warranties. In addition, we suspect that HP includes a more thorough accounting of all its partial and indirect aftermarket expenses (such as training and call center costs) as warranty.

Similar Product Lines?

IBM and Dell might be better comparisons for HP, but even in those cases, HP is prone towards higher warranty costs than its closest peers. However, HP is also prone towards more stable and predictable warranty costs. In Figures 5 and 6, it's clear that both Dell and IBM have seen their warranty expense rates spike, for whatever reason. And in Figure 4, it's clear by the relative lack of space between the red and the green lines that HP has aggressively managed its estimates so that accruals keep pace with changes in claims rates.

Dell also has aggressively pushed its own brand of extended warranty, and because the vast majority of its sales are made through direct channels, it has the greatest opportunity to close the sale of these service contracts. And, at least until 2007, its extensive mixing of basic and extended warranty accounts allowed it the luxury of deciding which label to use when processing a claim. But since 2007, the accounts have been kept quite separate, so this is no longer an option.

In Figure 5, Dell has a spike in its accrual rate at the end of calendar 2005, which caused its accrual rate to suddenly rise above 5.2%. And then, just as quickly, it returned to "normal." Most of that was attributable to problems in the OptiPlex product line, and a $307 million charge the company was forced to make to pay for it.

Rising Warranty Costs

In Figure 6, IBM's claims rate rose above 4.3% in the first calendar quarter of 2006, and then only gradually declined through the rest of that year and the next. More worrisome is the gap between claims and accrual rates, which remained above one percent until 2008. It's as if nobody told the finance department about the increase in claims payments, and nobody tried to talk them out of reducing the company's accrual rate.

Going back to Figure 2 for a second, it's obvious that HP is the better marksman from the tighter clustering of its data points. From Figure 4, it's obvious from the relative lack of space between the red and green lines that HP also is better at adjusting its accruals based on changes in claims.

Dell, on the other hand, is more consistent over time. Except for that late 2005 OptiPlex spike, most of its data points have been between two and three percent. While IBM, Apple, and HP have had both upward and downward trends over the past decade, Dell has little if any trend at all since 2005. And while HP is now enjoying some of its lowest warranty expense rates of the past decade, Dell is still in the middle of its seven-year range.

SEC Investigation

By the way, that little flat section in Dell's data is the result of an accounting scandal which directly impacted the company's warranty costs -- forcing it to retroactively restate multiple years of data. After that was done, we had to average the warranty costs of the fiscal year that ran from January 30, 2004 to January 28, 2005, to make them equal over four quarters.

We don't want to spend too much time rehashing the causes and effects of this scandal. As Dell's own internal investigation found at the time, "We did not design and maintain effective controls to ensure the completeness, accuracy, and timeliness of the recording of accrued liabilities, reserves, and operating expenses, primarily related to our accrued warranty obligations."

The outcome, we suspect, is that since 2007 the company has done its utmost best to maintain those effective controls. And as a result, since 2007 it has remained the most consistent warranty marksman of all its U.S.-based competitors.