Captains of industry know the competitive advantage that a healthy research and development operation provides. Last fall, Microsoft Corp.'s chairman, Bill Gates, loudly trumpeted Microsoft's plans to boost R&D spending by 20 percent this year, even though his company cut R&D spending by 1.6 percent in 2002. Chairman and CEO Jeffrey R. Immelt of General Electric Co. (ranked 40 among the top 100 in total R&D expenditure in 2002) has stated simply that GE's growth depends on increases in R&D spending. Putting the company's money where his mouth is, Immelt boosted GE's R&D spending in fiscal year 2002 (FY '02) by a healthy 12 percent, or US $235 million.

With a company's future at stake, it's no wonder that investors, corporate managers, and other decision-makers keep close tabs on the R&D spending of companies they're involved with and of their competitors. R&D watchers need look no further than the IEEE Spectrum list of the Top 100 R&D Spenders in 2002 to gauge the possible trajectory of future technological innovation for these companies.

Some firms, such as Advanced Micro Devices Inc. (ranked 83 among the top 100), increased R&D spending last year despite lower sales. Other firms, such as Microsoft (ranked 10), decreased R&D spending despite sales growth, a trend that Gates, true to his word, has reversed through the first two quarters of 2003 by increasing R&D spending by 8.8 percent. Overall, for FY '02, aggregate R&D spending has tracked sales and moved slightly downward from totals for the year before. Industry by industry, there are some stark differences, however, with significant increases in spending by companies in the Automobiles & Components sector and large decreases by companies in the Technology Hardware & Equipment category.

2002 trends

The past year was difficult for many of the large R&D spenders. Certain sectors, such as Automobiles & Components with firms like Toyota Motor Corp. (ranked 5), recorded increases in sales as well as in R&D. In stark contrast, the Technology Hardware & Equipment sector, with firms such as Sun Microsystems Inc. (47) and Cisco Systems Inc. (17), had significant declines in both R&D and sales. But though the Semiconductors & Semiconductor Equipment sector had a sharp decline in sales of 8.5 percent, its R&D fell only 1.1 percent. On the other hand, despite gains in sales for the Software & Services sector, its R&D spending decreased [see "Changes in Sales and R&D Percentages by Industry,"].

Faced with all these seemingly contradictory metrics, analysts rely on a factor called R&D intensity, the ratio of R&D spending to sales, to compare one company with its peers. This metric also comes in handy when looking at the many companies that do business in several industry sectors at the same time and when comparing those sectors against each other.

For example, Johnson & Johnson (ranked 12), listed as a pharmaceutical and biotechnology company, does a substantial part of its business in medical equipment and consumer health care, where typical R&D intensities are lower. For FY '02, Johnson & Johnson's R&D intensity for its pharmaceutical segment was 15.7 percent, but it was only 6.6 percent for its consumer and medical devices and diagnostics segment. Overall, its R&D intensity was 10.9 percent, which is less than the 12.9 percent for the pharmaceutical and biotechnology industry but greater than that of the typical company in the health care equipment industry.

How to read the numbers

Companies balance the amount of R&D information they disclose publicly between two competing objectives: convincing investors that their R&D spending is sufficient and effective and not revealing potentially sensitive data to their competitors. Because the second objective is also of great interest to investors, publicly disclosed R&D numbers are often quite limited. When interpreting R&D spending, keep the following caveats in mind:

• U.S. law requires U.S. and non-U.S. companies traded on U.S. stock exchanges to report R&D expenditures in each reporting period. Public companies report R&D spending quarterly on their 10-Q form (6-K for non-U.S. firms) and annually on their 10-K form (20-F for non-U.S. firms).

  In practice, many companies, including GE, do not report this data quarterly, apparently without any repercussions. GE contends that publishing quarterly data is not meaningful because of its short-term volatility. Some stock analysts, however, are concerned that R&D may be used to "manage" earnings; not reporting them quarterly can mask a company's financial results.

• Some R&D analysts claim that while overall industrial R&D spending has increased in recent years, the mix between research and development has shifted away from basic research to later-stage development. While the accounting profession recognizes the distinction between research and development, it does not require companies to report research separately from development spending.

• The R&D spending metric measures only the amount of money a company invests in R&D, not the impact that spending might have on future revenues or profits. Nor does it measure the efficiency of spending. For example, many firms are increasing their share of R&D in developing countries to take advantage of lower labor costs and government subsidies. This practice can drive efficiencies, but it remains to be seen whether the effectiveness of these operations is equivalent to those in "higher cost" countries where R&D facilities had been set up previously.

• Companies that have recently acquired a firm are required to report the R&D figures of the purchased firm separately. It is reported as in-process R&D. Such R&D should not be lumped together with regular R&D expenses because its cost is calculated in a different way. In-process R&D includes allocations of goodwill (the difference between the price paid for and the book value of the company acquired) from business combinations. R&D expense covers the actual costs incurred by the company. In other words, combining in-process R&D with R&D expenditures lumps together apples and oranges. Worse, comparing them across time or companies can skew how you view a company or even an entire industry sector.

From year to year, in-process R&D can be a source of great volatility in R&D reporting. For example, Amgen Inc. (ranked 60) recorded a remarkable 375 percent R&D spending increase, from $865 million to $4.1 billion, but most of that was due to an FY '02 in-process R&D of nearly $3.0 billion from firms it acquired. Subtract in-process R&D and Amgen still recorded a robust but more modest growth of 29 percent. Unless otherwise specified, in-process R&D has been removed from the data presented here.

Remember, too, that Spectrum's list of Top 100 R&D Spenders is a snapshot taken at a particular time. Companies can and occasionally do restate their financial results, including their R&D spending, and tax treatment of R&D spending differs from country to country. For instance, in the United States, research spending is subsidized by a special tax law called the Research and Experimentation Credit, which allows firms to claim a tax credit equal to 20 percent of the amount by which its qualified R&D expenditures exceed a base amount.

Also, some service-oriented companies thought of as cutting edge, such as Electronic Data Systems Corp. and Computer Sciences Corp., didn't make the list because they report little or no R&D spending even though they clearly advance technological innovation.

And, finally, the data presented in the R&D scoreboard here is based on the latest annual reports available from Standard and Poor's, the financial information company in New York City. Most companies have based their results on their FY '02 statements, three firms have completed their FY '03 totals, and in one case only FY '01 data was available.