Japanese R&D Profitability: Industry Production Function Estimation Using Panel Data

Introduction

In the Japanese manufacturing industry, research and development (R&D) investment as a share of operating profit (that is, R&D intensity), was about 50% in the 1980s but grew to some 65% in the 1990s, and nearly 100% after 2008. As companies undertake investment in R&D according to the level of operating profit, it is important to gauge its contribution to profitability to plan and implement appropriate managerial strategies.

R&D is not naturally associated with effective management and cost control. However, this does not mean cost control is unnecessary. Cost management pertaining to this area needs to be strategic and integral to profit management, with a focus on profit contribution. Conventionally, nonfinancial measures such as the number of patent applications and acquisitions, the new product sales ratio, and the number of reports on research finding have served to assess R&D outcomes. However, none of these directly shows the contribution of R&D to profit.

The purpose of this study is then to assess accurately the current state of the profit contribution of R&D using industry-level accounting information for Japanese manufacturers. In this regard, there are two main ways to estimate quantitatively the contribution of R&D to profitability (hereafter, R&D profitability). The first is a case study approach relating to a specific innovation. This method calculates profitability from the cost required for the innovation and the profit gained by the innovation and has the advantage of basing its calculations on detailed data and information. However, one drawback is the lack of generalizability, as it tends to emphasize successful cases of innovation.

The second approach is the use of a production function framework to estimate quantitatively the profitability of R&D using large samples of firms, which has the merit of any findings being more generalizable. Prior studies using this method include Griliches (1979, 1980, 1988), Griliches & Lichtenberg (1984), and Lichtenberg & Siegel (1991).

Griliches (1979, 1980) estimated the rate of return on R&D expenditures based on empirical data from 1957 to 1965 for the entire American manufacturing industry and six industries. Griliches (1988) again conducted empirical analysis for 911 manufacturing companies in 1966 - 1977. Griliches & Lichtenberg (1984) formulated a model to estimate the spillover effect of R&D on other industries by the Cobb-Douglas production function. They tried to quantitatively grasp the movement of technology flow between industries in such a way that R&D expenditures built into products flowed into purchasing industries through intermediate goods or investment goods. Lichtenberg & Siegel (1991) added variables on raw materials to the simple Cobb-Douglas type production function used by Griliches (1979, 1980) to capture the role of intermediates traded between the business units of each company. They aimed to estimate the rate of return on R&D expenditures at the business level of each company.

All of these analyses incorporate a variable representing the stock of technological knowledge, along with others representing the labor force and capital stock and possible explanatory variables, typically in a Cobb–Douglas production function specification. All of these studies also estimate R&D profitability using industry-specific data for the US manufacturing industry, specifying either sales or value-added as the explained variable. Subsequently, Ravenscraft & Scherer (1982) and Sougiannis (1994) expanded upon these pioneering studies by adding the physical capital stock and R&D spending to their production function specifications, activity costs for advertising, and sales promotion as explanatory variables, and profits as the explained variable. All employed ordinary least-squares (OLS) methods in their estimations.