Identifying Technology Flows and Spillovers Through NAICS Coding of ATP Project Participants

NISTIR 7280 - Identifying Technology Flows and Spillovers Through NAICS Coding of ATP Project Participants

INTRODUCTION

Project Overview

The Advanced Technology Program (ATP) provides cost-shared funding to private company research and development (R&D) projects that promise significant commercial payoffs and widespread benefits for the nation. ATP project selection focuses on generic technologies developed by upstream producers that enable downstream producers to improve the quality of their products or reduce their costs. To facilitate tracing the flow of ATP-enabled technologies and their spillover benefits, a methodology, documented in this report, was developed to identify each project participant's own-industry North American Industry Standard Classification (NAICS) code and the NAICS code of the downstream use-industries for each separate proposed commercial application for all ATP projects that started between January 1999 and July 2003.¹ The motivation for this project was twofold. First, NAICS coding enables matching of ATP projects with external data that use NAICS codes (e.g., the economic Census, National Science Foundation R&D expenditure data, and Compustat data). This matching in turn will facilitate research on project impacts and economic outcomes. Second, NAICS coding provides evidence that ATP project selection focuses on projects with high spillover potential.

ATP invests in risky, challenging technologies with the potential to deliver significant national economic benefits. Between 1990 and September 2004, ATP awarded 768 projects involving 1,511 project participants. Of the 768 awards, 550 were led by a single company, and 218 were joint ventures. To date, a total of $4.4 billion of high-risk research has been funded, consisting of $2.27 billion from ATP and $2.1 billion from industry. Small businesses lead 66 percent of all such projects.

This report has two parts. The first part documents the methodology developed to assign NAICS codes to each of the ATP participants and commercial applications, using ATP's Business Reporting System (BRS) data.² The authors identified the own- and use-industry NAICS codes by cross-checking ATP project descriptions with industry data from Compustat and Hoover's databases and the Internet. After an initial coding of each commercial application, the exercise was repeated to ensure that consistent decisions had been made regarding the assignment of NAICS codes across the various commercial applications.

The second part of this report is an analysis of the ATP project portfolio using a January 1999—July 2003 data sample. The analysis consisted of examining factors, identified by Jaffe (1996), that possess high spillover potential. The findings suggest that ATP selects projects with high spillover potential. In addition, the data were organized into the five different technology areas as defined by ATP. These technology areas include biotechnology, electronics, chemicals and materials, manufacturing, and information technology. The data provide a snapshot of the ATP technology portfolio, as contained in appendix B.

Economic Studies Involving the Trace of Technology Flows

ATP selects projects with broad economic impact; therefore, ATP tries to measure how R&D performed by its project participants flows through the U.S. economy. ATP has conducted many standard benefit-cost analyses.³ These studies examine individual projects on a case-by-case basis. They follow the Mansfield et al. (1977) method of examining innovations at the company level and interviewing company officials and customers about the associated benefits resulting from the innovation. These studies provide precision in measuring the benefits of individual projects. However, they fail to provide a sense of the overall technology flows (defined in box 1-1) occurring throughout the entire ATP portfolio. In addition, the cost of conducting in-depth benefit-cost studies makes them impractical to use for each and every ATP project.⁴

Another method, used by economists to study the effects of R&D performed, looks at how technology flows occur from upstream developers to downstream users. Early discussions of technology flows can be found in Brown and Conrad (1967) and Terleckyj (1974). These authors were interested in measuring the impact of R&D, mostly performed by manufacturing industries, on downstream industries that utilized the capital goods in their production process. To identify these technology flows, they used the Input-Output tables produced by the Bureau of Economic Analysis to allocate R&D expenditures from origin to using industries. An Input-Output table lists the expenditures by industry that it uses to make its products. For example, almost one out of three of the inputs in the agricultural sector are manufactured goods. Improvements in those manufactured goods used by the agricultural sector may eventually flow into the agricultural sector, thereby, raising the productivity of that sector. This is the main thrust behind analyzing R&D expenditures using Input-Output tables.

Box 1-1. A Note on Terminology: own- and Use-Industries
The literature uses the terms own-industry and use-industry, as well as upstream and downstream industry, to convey the path by which goods and services flow through the economy. Typically, economic studies using the Input-Output tables employ the terms own- and use-industry because the Input-Output tables are not based upon the concept of upstream/downstream but of use and make tables. Other economists use the terms upstream and downstream especially in the spillover literature. For the purposes of our study, we explicitly use the terms own- and use-industry to describe our data. However, conceptually, we think of the economy as a circular flow of goods between three types of producers; upstream, midstream, and downstream, which we describe in more detail below.

For the purposes of this analysis, the term own-industry is used to refer to each participant's primary industry. Use-industry refers to the primary industry in which the commercial application will be employed. The terms upstream and downstream industry illuminate the structure of ATP own- and use-industries in relation to the overall structure of the U.S. economy.

Instead of using the Input-Output tables, Scherer (1982) looked at patents by companies performing R&D as a proxy for their innovative output. His group analyzed over 15,000 patents produced by the firms that accounted for 74 percent of total U.S. R&D expenditures for the year 1972. They identified the use-industries from each patent and discovered that two out of three possessed one to three use-industries. For the first two-thirds patents where a use-industry was identified, he allocated the R&D expenditures from those origin industries directly to the use-industries. With the remaining third, R&D expenditures were allocated across the use-industries, based upon the percentage that each use-industry consumed of the origin industry's output, as determined by the Input-Output tables. He used these numbers to estimate the short-run effect of R&D expenditures on using industries. Scherer (2003) revisited the exercise by using the same 1972 data on R&D expenditures, but this time he allocated the R&D expenditures using variations of the Input-Output tables.

The motivation behind both exercises was to determine how R&D performed in the upstream industries might affect productivity in the downstream industries. In general, using both the 1982 and 2003 methods, Scherer found that allocating a portion of upstream R&D expenditures to downstream using industries explained some of the observed downstream productivity increases. It would be possible to perform a Scherer-type analysis on the ATP projects using the NAICS data.⁵

Popkin (2003) used SIC/NAICS codes to develop a method to estimate the potential spillover effects of ATP projects resulting in successful commercialization.⁶ Popkin calculated total spillover potential by summing the intra-industry shipments involving the own- and use-industries for ATP project participants. For example, he showed where one ATP project participant's own-industry was printed wiring boards. The ATP participant proposed a commercial application in the communications equipment industry which would be the use-industry. Popkin asked, "What potential impact might improvements in the printed circuit board industry have on the communications equipment industry and vice-versa?"

The logic behind his method is, the more an industry uses an input to make a final product, the greater the impact improvements in that input will have on the final product. In addition, improvements in the use-industry may spillover into the own-industry. In the printed wiring board case, about 9 percent of the inputs that the communications equipment industry uses come from the printed wiring board industry. Also, how much an industry uses, as its own input, would be impacted by improvements within the industry. For example, 20 percent of the printed wiring board's inputs are printed wiring boards. This is known in the literature as the "diagonal" since it is a column that goes across diagonally in a square Input-Output table matrix.

He suggested that improvements in the communications equipment industry (in this case, the use-industry) would result in spillovers within the communication equipment industry. The impact of these spillovers would be related to the amount of communications equipment used by the communications equipment industry (the diagonal) and the amount of communications equipment used by the printed wiring board industry (a.k.a. the amount of use-industry needed to make own-industry).

The sum of these four measures is the measure for total potential spillovers. Popkin argued that the likelihood of spillover impact was directly related to the sum of the four measures and suggested the methodology be used to analyze all ATP projects.

The research presented in this paper identifies industry codes for all the proposed commercial applications of projects still in progress. We then examine whether ATP selects projects with high spillover potential. Popkin examined already completed ATP projects, identified the commercialized product (if there was one), and classified the own- and use-industry SIC code for the entire project. These data may be used to perform a similar analysis once commercial outcomes are known.

Report Organization

This section introduces and defines industry classification terminology and provides a brief overview of industry classification in spillover analysis. Section 2 reviews the concept of economic spillovers and how coding facilitates the identification of projects with high spillover potential. Section 3 contains a brief history of ATP efforts to capture SIC and NAICS code data through its data collection system and describes the process of NAICS coding by the authors of this report. Section 4 contains a detailed analysis of the research findings. Section 5 concludes and provides suggestions for further research.

____________________
1. Some commercial applications may be tied to multiple NAICS codes. However, since these data will be used in economic studies, having multiple observations for a single piece of data may prove to be problematic.

2. The BRS is a set of survey instruments used to capture data on ATP project participants over the life of the ATP funding and up to six years after funding ends; see section 3 for more information.

3. See the following examples for benefit-cost analysis: Pelsoci (2003), White and Gallaher (2002), and Austin and Macauley (2000). Fora literature review of ATP Economic Assessment Office studies through 2000, see Ruegg and Feller (2003).

4. ATP generates status reports on every ATP project. A status report is a mini-case study of the project about three to four years after the completion of the ATP project.

5. Polenske and Rockler (2004) performed an analysis of the effects of a single project involving the automobile industry and its impact on the U.S. economy including both direct and indirect effects.

6. See Popkin (2003), pp. 3-5, for a review of the economic spillover literature.

Return to Table of Contents or go to next section.

Date created: May 25, 2006
Last updated: June 7, 2006

ATP website comments: webmaster-atp@nist.gov / Technical ATP inquiries: InfoCoord.ATP@nist.gov.

NIST is an agency of the U.S. Commerce Department
Privacy policy / Security Notice / Accessibility Statement / Disclaimer / Freedom of Information Act (FOIA) /
No Fear Act Policy / NIST Information Quallity Standards / ExpectMore.gov (performance of federal programs)