by Brandon Shackelford[1]

In 2008, an estimated 92% of companies devoted all of their research and development efforts to one line of business, according to the Business R&D and Innovation Survey (BRDIS). The remaining 8% diversified their R&D spending across multiple lines of businesses, and these companies were the ones that tended to heavily invest in R&D. Companies reporting more than one line of business accounted for $107 billion (33%) of the $328 billion worldwide R&D expense for U.S. businesses. Survey findings also showed that 82% of companies with R&D expenses derived all of their worldwide sales from one line of business.

Companies typically start out offering a single type of good or service to their customers, but as these companies grow (organically or through acquisitions) they sometimes diversify into offering different types of products. Companies typically group related products together for both management and accounting purposes. This InfoBrief examines companies' R&D statistics in terms of these groups, commonly known as lines of business. Companies located in the United States that received BRDIS were asked to identify the lines of business in which they operated in 2008 and, for the first time, were asked to report their sales and R&D expenditures for these lines of business.

Comparing Line-of-Business Estimates to Industry Estimates

Most business R&D statistics are collected at the company level, not for each line of business within a company. At the company level, the company's total R&D is assigned to a single primary industry, even if there is evidence that the company has R&D in multiple lines of business.[2] For example, if a company has a total R&D expense of $100 million—$80 million in pharmaceuticals and $20 million in medical devices—the total R&D expense of $100 million will be assigned to the pharmaceuticals industry because pharmaceuticals is the largest component of its R&D expense. This can lead to a loss of precision in industry-level estimates if diversified companies have large amounts of R&D in secondary lines of business. Questions on the 2008 BRDIS were developed to help determine how much of an impact these cases have on industry-level R&D statistics.

Table 1 shows the worldwide R&D expense on an industry and corresponding line-of-business level. The former represents the total worldwide R&D expense for all companies where their total R&D was assigned to a single primary industry on the basis of their most important line of business. The latter is the total worldwide R&D expense on a line-of-business basis where R&D was allocated among multiple lines of business. Overall, 86% of the $328 billion worldwide R&D expense occurred in the primary line of business corresponding to the industry classification of the company's R&D. However, in some cases, the correspondence between the primary industry estimate and the related line-of-business estimate is not this large. Figure 1 illustrates the relationship between worldwide R&D expense estimates for the soap, cleaning compound, and toiletries industry and corresponding lines of business. In this case, only 55% ($1,933 million) of the industry R&D estimate ($3,486 million) occurred in lines of business corresponding to this industry, and the remaining 45% occurred in lines of business outside of this industry or was not allocated to a line of business. Similarly, almost one-third ($854 million) of the business code estimate was from companies where this code was not the primary line of business.

Larger R&D industries tend to have a greater correspondence between industry and line-of-business estimates than the industry presented in figure 1. The largest industry in terms of worldwide R&D expense is the pharmaceuticals and medicines industry (North American Industrial Classification System [NAICS] code 3254). In 2008, the worldwide R&D expense of companies classified in this industry was $74.4 billion, and 94% of this estimate was from R&D reported in lines of business related to the pharmaceuticals and medicines industry. Similarly, 96% of the estimate for the semiconductor and other electronic components (NAICS 3344) industry, the third largest industry in terms of worldwide R&D expense, occurred in lines of business corresponding to this industry.

There are notable exceptions that deserve mention. In absolute terms, the largest difference between the industry and corresponding line of business for worldwide R&D estimates is for the medical equipment and supplies industry (NAICS 3391). Companies in this industry reported that $6.3 billion (94%) of their $6.7 billion in worldwide R&D expense estimate occurred in related lines of business. Examining R&D expense on a line of business basis shows that companies in other industries reported $5.3 billion of R&D in lines of business related to medical equipment and supplies. The lines of business most frequently reported alongside medical equipment and supplies by diversified companies included pharmaceutical products (NAICS 3254) and electromedical apparatus (NAICS 334510, 334517).

Other large R&D industries with sizeable differences between the industry and corresponding line-of-business estimates include software publishers (NAICS 5112) and aircraft, aircraft engine, and aircraft parts (NAICS 336411–336413). In both of these cases, the industry estimate exceeds the corresponding line-of-business estimate by more than $4 billion, indicating that companies in these two industries conduct a significant amount of their R&D in lines of business outside of their primary industry. In cases such as these where industry and corresponding line-of-business estimates diverge by large amounts it may be helpful to analyze related industries as a group rather than individually. For example, the information provided above suggests that the R&D estimate for the group of medical products industries (medical equipment and supplies, pharmaceutical products, and electromedical apparatus) could be a more precise measure than the estimates of the component industries individually.

Of the other data items BRDIS collected by line of business, domestic R&D performance (the portion of worldwide R&D expense performed by companies in the United States) displays patterns very similar to worldwide R&D expense, with a large amount of overlap between industry and corresponding line-of-business estimates. Overall, 86% of the $233 billion domestic performance of R&D expense classified by primary industry in 2008 is estimated to occur in corresponding lines of business. There is much less overlap between industry and corresponding line-of-business estimates for worldwide sales, domestic sales, and domestic R&D performed by the company and paid for by others. For both sales variables, only 64% of the data for industry-level R&D are estimated to occur in related lines of business. For domestic R&D performed by the company and paid for by others, the correspondence between industry and line-of-business estimates is 60%.

R&D Intensity

The proportion of R&D to sales, or R&D intensity, is a commonly used indicator for comparing the relative importance of R&D across industries and among companies in the same industry. Calculating this indicator by taking the ratio of industry estimates of company-level R&D and sales can produce quite different results than the ratio of R&D to sales for line of business. Using industry estimates, the semiconductor and other electronic components industry (NAICS 3344) has the highest worldwide R&D intensity at 14.9% (table 2). However, when the line-of-business estimates for R&D and sales are used to calculate R&D intensity, the semiconductor industry no longer has the highest value for this indicator. The most R&D-intense industry on a line-of-business basis is the scientific R&D services industry (NAICS 5417)—at 46.6% it is a much higher value than the 9.0% that is calculated on an industry basis. Software publishers (NAICS 5112) and Internet service providers/Web search/data processing services (NAICS 518) are two other industries that have much higher R&D intensities when line-of-business estimates are used to calculate the indicator. This is an indication that some companies classified in these industries report large amounts of sales but relatively little R&D in ancillary lines of business.

Coincidence of Reported Codes

Some combinations of lines of business are much more frequently reported than others among survey respondents. These patterns can provide indications of lines of business with strong relationships within diversified companies. The lines of business most frequently reported together were codes 32600 (plastics and rubber products manufacturing) and 33200 (fabricated metal products manufacturing), followed by codes 51120 (software publishers) and 54150 (computer systems design and related services) (table 3).[3] The line of business for activities in software publishers was also frequently reported alongside data processing, hosting, and related services (code 51820). Almost one-third of the companies that reported this last code also reported activity in software publishing (table 3). These relationships in lines of business within companies in BRDIS give researchers a new opportunity to explore industry dynamics and the effect of diversification on R&D, innovation, and productivity.[4]

Data Sources and Limitations

The target population for BRDIS consists of all for-profit, non-farm businesses with five or more employees in the United States, with a particular focus on those that perform R&D in the United States. For 2008, a total of 39,553 companies were sampled, representing 1,926,012 companies in the population. Because the statistics from the survey are based on a sample, they are subject to both sampling and nonsampling errors.

In 2008, two forms were mailed to companies for BRDIS, the complete form (form BRDI-1) and an abbreviated form (form BRDI-1A). The complete BRDIS form asked companies to identify all the lines of businesses in which they operated using a list of business codes and subsequently asked them to report their sales and R&D for each line of business. The abbreviated form asked companies to identify only their single largest line of business in terms of R&D. The complete form was mailed to companies that were known to have performed over $1.836 million of R&D in recent survey years (4,834 companies in 2008). Because of the design of the abbreviated form and item nonresponse on the complete form, not all R&D can be allocated to a line of business.

Overall, $10.7 billion (3%) of the total worldwide R&D expense in 2008 and $1,092 billion (10%) of the total worldwide sales in 2008 was left unassigned.

The BRDIS forms used the more generic term "business activities" instead of "lines of business." BRDIS asked companies to report their business activities using a set of business codes, which were based on the 2002 NAICS codes. In some cases these codes may not be sufficiently detailed to capture the full diversity of activities a business may have. For example, only one line-of-business code is provided for software publishers (51120), but a company may have more than one software business (for example, games, business applications, and operating systems).

For an overview of worldwide R&D data collected by BRDIS see NCSES InfoBrief, U.S. Businesses Report 2008 Worldwide R&D Expense of $330 Billion: Findings from New NSF Survey (NSF 10-322) at http://www.nsf.gov/statistics/infbrief/nsf10322/. The full set of detailed tables from the 2008 BRDIS will be available in the report R&D and Innovation in Business: 2008 at http://www.nsf.gov/statistics/industry/. Individual detailed tables from the 2008 survey may be available in advance of publication of the full report. For further information, contact Raymond Wolfe.

Notes

[1]  Brandon Shackelford is principal consultant at Twin Ravens Consulting, Austin, TX. For further information, contact Raymond M. Wolfe, Research and Development Statistics Program, National Center for Science and Engineering Statistics, National Science Foundation, 4201 Wilson Boulevard, Suite 965, Arlington, VA 22230 (rwolfe@nsf.gov; 703-292-7789).

[2]  Each company was classified into the business code that accounted for the largest amount of total domestic R&D performance using data from question 2-14 plus question 4-5 (http://www.nsf.gov/statistics/question.cfm#13). If no business code data was available for a company's domestic R&D performance, the industry code used for sampling was used. The industry classification assigned to companies for sampling was based on payroll of U.S. establishments.

[3]  This analysis took into account all business codes reported by companies irrespective of whether R&D was reported in the code. Some companies perform R&D in only a subset of their lines of business.

[4]  Chang H, Fernando GD, Srinivasan D, Tripathy A. 2011. A re-examination of diversification and firm productivity. Paper presented at the 2012 meeting of the American Accounting Association Management Accounting Section, 15 Aug 2012. Available at http://ssrn.com/abstract=1909930.