In general, medical devices and scientific equipment are among the types of products with low warranty costs and high reliability rates. However, among all the different types of machines and apparatus out there, it seems that anything that involves lasers or X-rays has much higher-than-average warranty costs. Or at least it used to: these manufacturers have also been among the best when it comes to warranty cost reductions.

We began with a list of 196 U.S.-based manufacturers of medical and scientific equipment that disclose their warranty expenses in their financial statements. Most of the companies do little but manufacture medical and scientific equipment, but a handful are primarily engaged in other industries. For example, La-Z-Boy Inc. and Tempur Sealy International Inc. are well-known in the furniture and bedding industries, but they also make medical recliners and hospital beds. The lasers made by Coherent Inc. have applications in everything from avionics to solar cell manufacturing. But they're also used for everything from DNA sequencing to molecular spectroscopy. So we'll include them in our group as well.

Over the years, we've noticed that there is a small group of companies within the list that have much higher warranty expense rates than the others. At first, we thought it had something to do with how some companies report their revenue, including sales of supplies and consumables in with their medical and scientific hardware, thereby artificially depressing their warranty expense rates (claims and accruals divided by product sales). But then we began to notice their product lines always seemed to have something to do with either lasers or X-rays.

We identified 44 companies whose products use lasers or X-rays, and sure enough, they tend to spend a higher percentage of their sales revenue on warranty expenses than the others. There are far more medical and scientific equipment companies whose products don't involve lasers or X-rays, and together they have far higher revenue, and far higher warranty expenses, than these 44 companies. However, the laser and X-ray equipment companies pay warranty expenses at a far higher rate.

Four Warranty Metrics

Let's go through the metrics that bring us to this conclusion. We started by collecting four essential warranty metrics from each financial statement: the amount of claims paid, the amount of accruals made, the amount of reserves held, and the amount of product sales. Using the claims, accruals, and sales metrics, we calculated two rates: claims as a percentage of sales, and accruals as a percentage of sales.

In Figure 1, we're detailing the claims payments by these 196 companies over the past 14 years. In 2016, they reported paying $762 million in claims, a decrease of about $14 million from 2015 levels. But that decrease was caused by the laser and X-ray companies, which spent $26 million less on claims in 2016 than they did in 2015. The other medical and scientific companies actually spent $12 million more on claims last year.

A big part of the reason for the contraction in laser and X-ray equipment claims may be the way the merger of Sirona Dental Systems Inc. and Dentsply International Inc. was accounted for. Sirona was a major supplier of dental office equipment and a top warranty provider, paying $26 million in claims in 2015 and $23 million in 2014. Meanwhile, almost half of Dentsply's revenue came from sales of consumables and supplies, and its warranty expense disclosures were incomplete and non-compliant with accounting rules.

The merger was completed in February 2016, so Sirona did not file any independent financial reports last year. Meanwhile, the financial reports of the now-renamed Dentsply Sirona Inc. continue to include only the beginning and ending balances of the company's warranty reserve fund, and no details on claims paid, accruals made, or what they did with the $8.7 million in warranty reserves they acquired from Sirona. So it's possible that the entire $26 million annual decrease is due to the virtual disappearance of Sirona from the statistics.

Elsewhere in the industry, there were several large but ultimately offsetting changes in annual claims payment totals. Dental and life sciences equipment maker Danaher Corp. paid out $34 million less in 2016, but medical device manufacturer Medtronic plc paid out $34 million more. Thermo Fisher Scientific Inc. paid out $4.5 million more, but Agilent Technologies Inc. paid out $3.0 million less and biotechnology toolmaker Illumina Inc. paid out $1.8 million less. La-Z-Boy paid $1.4 million more and Tempur Sealy paid $2.5 million more, but scientific instrument manufacturer Bruker Corp. paid out $3.9 million less.

Warranty Accruals

With warranty accruals, it was much the same story. Dentsply didn't disclose how much it accrued last year, while Sirona accrued $28 million in 2015 and $21 million in 2014. And while accruals by the laser and X-ray equipment group fell by $20 million last year, most of that is traceable to the non-compliance of Dentsply Sirona with existing accounting regulations.

The other group showed a $16 million increase in accruals made last year, but that still meant a $4 million reduction in accruals by the industry as a whole. Accruals by all 196 medical and scientific equipment companies fell from $769 million in 2015 to what turns out to be their second-lowest total of the past 14 years (2003's $738 million remains the lowest).

Despite the relatively small changes overall, there were some big changes in accruals among some of the larger companies in the industry. Danaher cut accruals by $29 million. Hospital furniture maker Hill-Rom Holdings Inc. cut accruals by $7.3 million, and Illumina cut accruals by $6.3 million. Bruker reduced accruals by $3.7 million, and imaging technology company PerkinElmer Inc. cut accruals by $2.0 million. For each of these companies, the reductions in annual accruals were significant, exceeding 10% of their 2015 totals. Hill-Rom, in fact, cut its accruals by more than 40%.

Danaher's sales also fell significantly, so its accrual reduction was more or less proportional. However, Bruker's and PerkinElmer's sales dipped only slightly, and Illumina's and Hill-Rom's annual sales actually rose, so their accrual reductions were caused by something other than merely changes in revenue.

And then there were some significant increases in accruals as well. Medtronic raised its accruals by $13 million. Thermo Fisher increased its accruals by $11 million. Boston Scientific Corp. boosted accruals by $10 million. Coherent raised its accruals by $5.4 million. And wheelchair manufacturer Invacare Corp. raised its accruals by $4.4 million.

For all of these companies, their increases exceeded 10% of their 2015 accrual totals. That's significant because unless there's been a change in product reliability or the cost of repairs and replacements, accruals should remain roughly proportional to sales.

However, of these five companies, only Thermo Fisher saw a roughly proportional increase in sales. Warranted product sales grew from $14.7 billion to $15.7 billion, a 7% gain, while accruals grew by 13%. For the others, there was a mismatch. Sales grew much faster than accruals at Boston Scientific. Accruals grew much faster than sales at Coherent. Sales actually fell at Invacare. And since Medtronic's fiscal year ends on April 29, we can't make timely comparisons, but its sales grew by 42% in the previous year, while its accruals grew by 25%, so it's actually doing fine.

Warranty Reserves

With this metric, the trend was reversed. Warranty reserves held by the laser and X-ray companies rose by $13 million in 2016, to a level of $110 million by year's end. For the other medical and scientific equipment companies, however, reserves fell by $91 million, to their lowest level of at least the past 14 years.

Companies in the medical and scientific instrument industry have traditionally kept roughly as much in their warranty reserve as they spend in a year paying warranty claims. That hasn't dropped appreciably in recent times. For instance, at the end of 2016 the ratio between reserves and claims stood at just under 11.83 months.

There are several reasons for the recent decline. First, companies such as Danaher, as they cut claims and accruals, inevitably are also reducing reserves. Second, companies such as Medtronic are not increasing accruals as fast as claims are rising, so reserves fall. And third, companies such as Hill-Rom are cutting accruals even though claims are rising, so reserves fall.

But then there's another group of companies that are literally disappearing from the roster. We already mentioned Sirona and Dentsply. But then there are others, such as St. Jude Medical Inc., which was acquired by Abbott Laboratories. While the St. Jude brand name continues, its $25 million in warranty reserves apparently disappeared. Abbott dismissed the very idea of compliance with accounting rules concerning warranty expense disclosures, stating in its most recent annual report that "Product warranties are not significant."

It's happened more than once. In late 2014, Becton Dickinson & Co. acquired CareFusion Corp., and $16 million in warranty reserves disappeared. In 2015, Thoratec Corp. was acquired by St. Jude Medical, which did indeed report $7 million in acquired warranty reserves in its 2015 annual report. But then along came Abbott Labs.

Partially offsetting these warranty reserve reductions and disappearances are a handful of warranty reserve increases. Thermo Fisher, for instance, reported $16.8 million in reserves acquired from Affymetrix Inc. and FEI Co., among others. Coherent acquired $12.6 million in reserves from laser tools company Rofin-Sinar Technologies Inc. and others, helping to drive its reserve fund balance up to $28.6 million by the end of last year. And then there were a few whose reserves just grew: Agilent, La-Z-Boy, and Varian Medical Systems among them.

Warranty Expense Rates

Finally, with those raw dollar totals in mind, we get to the warranty expense rates. As was previously mentioned, unless there has been a change in product reliability or the cost or frequency of repairs and replacements, warranty accruals should remain approximately proportional to sales.

What we mean by that is if the cost of a product's warranty is $100, then the company should accrue $100 every time it sells a unit. Of course, there could be price changes, but if the product sells for $5,000, more or less, then its accrual rate would be two percent. If the price was cut by $250, the accrual rate would rise to 2.1%. And if its price rose by $250, its accrual rate would fall to 1.9%. But the warranty cost would remain at $100, unless there was a change in repair cost or reliability rates.

So what we've done in Figure 4 is to take the claims and accrual totals from Figures 1 and 2 and divide them by the corresponding sales figures. We've done this on a quarterly basis since 2003, so we actually have 56 measurements of the claims and accrual rates over the past 14 years. And since we've split the market into laser/X-ray and other, there are 112 pairs of measurements in the chart below.

The most obvious trend in the chart is the general decline in expense rates attributable to the laser and X-ray equipment companies. Back in 2003, they were putting two-to-three percent of their revenues towards warranty costs. Now, their warranty expense rates are more like one-to-one-and-a-half percent -- more or less half as much as they were paying 14 years ago.

Meanwhile, the rest of the medical and scientific equipment industry has kept its warranty expense rates relatively low. Their average claims and accrual rates rarely exceed one percent, and haven't done so at all since 2006. In fact, in recent years they've more typically remained close to 0.5% or 0.6%.

However, this means that while there has always been a gap between the warranty expense rates of the laser/X-ray and other groups, it has narrowed considerably in recent years. Back in the early years of this chart, the gap was as large as 2.2%. At the end of 2016, it was less than 0.5%. At this rate, if the trend continues, in five or six years there may be no difference between them.