The impact of firm and industry characteristics on technology licensing.
It has been argued that scientific and technological knowledge is subject to significant complications that inhibit markets from attaining socially optimum outcomes. For instance, anecdotal evidence shows that large companies in the United States, Western Europe, and Japan ignore much of their patented technologies, which could be licensed or profitably sold (British Technology Group 1998). The inefficiency of the market for technology is caused by a number of impediments. Arrow (1962) argues that preventing knowledge from being appropriated is the major obstacle to the efficient market for technology. Once an idea is disclosed to a potential buyer, it is possible for that buyer to use the information without paying for it. As a result, a potential licensor is reluctant to disclose the core of a technology, and this leads to a typical market failure. Studies on contracts and transaction costs have elaborated on the causes and effects of moral hazard and asymmetric information in exchanging knowledge through arm's length transactions. These problems make the underlying contracts incomplete (Caves, 1996; Hart, 1995; Menard, 1996; Salanie, 2002). On the other hand, evolutionary economics (Nelson and Winter, 1982) and management theory have given a lot of attention to the organizational aspects of the innovation process, showing that often the requisite capabilities and routines are difficult to exchange through the market (Teece, 1977). Cognitive limitations in the transfer of technology to another context require extensive adaptations and costs (Arora and Gambardella 1994).
Even so, markets for technology have become increasingly important in recent decades, especially in technology-intensive industries, as increasing competition through globalization, accelerating rates of technological change, and outsourcing and collaboration have become pervasive. For instance, a recent study by Anand and Khanna (2000) reports that licensing accounts for about 20% to 33% of all inter-firm alliances (depending on the sector) in high-tech sectors such as chemicals, biotechnology, software, computers, and electrical and nonelectrical machinery. Thompson Financial's SDC database used in this paper lists more than 10,000 publicly announced licensing agreements during the 1990s.
What factors affect technology holders' incentives to license? Are there differences in licensing activities across firms, industries, and technologies? What causes such differences? How do firms' operational, organizational, and primary industry characteristics affect business managers' decision-making on technology licensing? The author addresses such questions.
This paper studies the validity of the factors that might affect the incentives of companies to license out their technology. Empirical analysis is provided with the help of a unique panel data set describing licensing arrangements in publicly traded companies in the United States. Specifically, we explore inter-sectoral differences and similarities in the determinants of licensing across three high-technology industry clusters: information and communication technology (ICT), biotechnology, and advanced materials. (1)
A company's incentive to sell its technology to prospective competitors is driven by two principal effects on the profits: the revenue effect and the rent dissipation effect (competition effect) (Arora, et al. 2001). The revenue effect is driven by the profits that will accrue to the licensor in the form of licensing payments (i.e., a fixed licensing fee or royalty) by licensees. The licensor firm essentially increases its aggregate market share of products produced with its own technology by adding a licensee who pays licensing payments. On the other hand, licensing generates a negative rent dissipation effect on the profits of the licensor, arising from the profit erosion caused by increased competition from the licensees. Technology licensing may represent current opportunities to enter into new market and to service new customers (Mitchell and Singh, 1992). Companies that cannot innovate may be able to produce products and compete with the licensor if they receive the rights to use the technology. The trade-off between these two opposite effects determines the technology owner's licensing behavior.
Transaction costs theory provides another perspective on licensing. Transaction costs refer to the negotiating, monitoring, and enforcement costs accruing to participants in a deal. According to transaction costs theory (Williamson, 1979), terms and types of alliances depend on the level of uncertainty and opportunism surrounding the transaction. The greater the level of uncertainty and opportunism, the more controls are placed on a transaction. The intensity of the market use of technology will then be determined by the relative costs of carrying out a transaction under each organizational structure. For instance, the decision regarding the sale of technology will depend on contract enforceability, i.e., the degree to which knowledge may be appropriated (the strength of patent protection) (Teece, 1987). Arrow (1962) and Merges (1998) also argue that the transaction costs of technology licensing are negatively correlated with the strength of patent protection. To counteract the licensee's incentive to shirk payment in the case of weak intellectual property rights (IPR) protection, the licensor's costs for monitoring and enforcement will increase. Entrepreneurs, therefore, will organize the least costly and most efficient transaction of technology.
We hypothesize that various characteristics of the firm's primary operating industry affect the principal effects of licensing. In addition, the manager's licensing decision is presumed to be determined to a large extent by a company's operational and financial characteristics. In this study, the probability of selling technology licenses is explained by variables that represent the firm's characteristics (sales, R&D intensity, capital investment, profitability, prior licensing experience) and the firm's primary industry characteristics (concentration, sales growth, market size, the propensity to receive patents).
This study focuses on companies in three high-tech industry clusters: information and communication technology (ICT), biotechnology, and advanced materials. These technologies have "infrastructural" characteristics: they have penetrated throughout the economy during the past two to three decades, dramatically altering the basic meaning of high technology. Rather than referring to the output of R&D-intensive industries, high-tech now refers to a style of work applicable to just about any business (Branscomb and Florida 1998; Porter 1998). The penetration of these so-called general purpose technologies has gradually shifted the locus of high technology production from exclusively manufacturing to a combination of manufacturing and service industries. Technology is rapidly transforming the products of both sectors (Hauknes 1998; Leech, et al. 1998; OECD 2000). ICT, biotechnology and advanced materials, because of their infrastructural nature, have become major technology suppliers in the market for technology.
The sample firms for this study are drawn from COMPUSTAT by Standard & Poor's (i.e., publicly traded companies in the U. S.) to obtain financial information necessary for the analysis. Among firms operating in 1999, we chose companies publicly traded in the U. S. in all six years from 1994 to 1999. Table 1 shows the frequency distribution of final sample companies in each industry.
To locate and obtain the licensing history of these sample firms, this analysis uses the Securities Data Company (SDC) database of Thomson Financial. SDC database records of all publicly announced strategic alliances worldwide recorded in Securities and Exchange Commission filings, newswires, the press, trade magazines, professional journals, and the like. SDC provides information on contract type (e.g., joint venture, licensing agreement, joint marketing and manufacturing agreement, joint development or production, etc.), description of the deal, the date of agreement, and identities of participants (e.g., company name, Standard Industrial Classification (SIC) code of primary business, nation, parent companies, etc.). For the analysis, we read through the description of all licensing agreements to ensure that each deal was related to technology transfer or exchange of technology, licensing of a new product, or process technologies and designs, and to confirm the direction of technology transfer (i.e., licensor, licensee). We also include a few licensing deals that were accompanied by other types of agreements, such as joint ventures or joint marketing and research, since this inclusion does not create an obvious bias. However, licensing deals referring to the termination of licensing agreements and litigation settlements of past licensing deals were not counted.
As shown in Table 1, about 26.5%, 22.8%, and 9.6% of the sample firms in each sector, respectively granted licenses to others at some point during 1994-1999. The rest of sample firms, who sold no licenses during this period, are included as a control group. The highest proportion of companies involved in licensing deals in their respective technology clusters were computer and office equipment, drugs and pharmaceutical, and chemicals and plastics.
This study employs a random-effects probit model to estimate the probability that a firm will license its technology. Since the purpose of this study is to examine inter-sectoral differences and similarities in the determinants of licensing, the model is tested over three high-technology industry clusters separately for six years (1994-1999).
* Dependent variable
Licensing is coded as 1 if firm i grants at least one technology license to another firms in period t (t = 1994-1999). Otherwise it is 0.
* Independent variables
1. Firm characteristics
Patent stock is firm i's patent stock at period t. (2) Patent-intensive firms may have more technologies than they can exploit internally. They may also want to exploit peripheral technologies. They may license in order to establish and dominate an existing standard. This variable controls such possibilities. A positive sign is expected.
Logsale is the log of sales of firm i at period t. This variable is a proxy for the firm size. It may be rather risky for small firms to commercialize their inventions or innovations on their own because they often lack a sales network and have cash flow constraints. Therefore, licensing technologies to others would be the best strategy for small companies, since they can avoid some risk and still extract profits from proprietary technologies by collecting royalties and licensing fees. Small firms are expected to license technologies more actively. The anticipated sign is negative.
Capital/asset is the capital investment over total invested assets, and R&D/Sale is the R&D expenditure over sales amount (i.e., R&D intensity) of firm i at period t. These variables represent the technological capabilities and rate of implementation of innovations. A higher level of capital and R&D investment is usually associated with a higher possibility of new invention or innovation. Thus, capital and R&D intensive companies may have more technologies available internally and increasingly try to supplement income from the active management of intellectual property, frequently involving technology sales. Positive signs are expected.
Income/asset is the net income over total invested assets of firm i at period t. Although the technology holder has different strategies, such as quantity restrictions and exclusive territories to minimize the negative rent dissipation (competition) effect of licensing, an entrant licensee is nevertheless a potential threat to the licensor. For firms that are generating higher rates of return, however, this dampening effect on licensors' profits would be relatively minor. Therefore, the expected sign is positive.
Experience_dummy is coded as 1 if firm i has licensed one or more technologies during the previous five years; otherwise it is 0. Former licensing experience would help licensors lower the transaction costs of licensing, such as gathering information about prospective licensees, negotiating with licensees, writing contracts and enforcing them. Since companies would tend to license out more in the presence of low transaction costs, the anticipated sign is positive. This dummy variable also controls for an unobserved firm-specific effect, since firms who sold licenses before may be qualitatively different from those who have never sold.
2. Industry characteristics
Industry growth is the percentage change in total sales of the primary industry of firm i at period t. The higher the growth rate of industry output, other things being equal, the less an entrant's supply will depress industry price and output (Orr, 1974). Therefore, firms' licensing incentives will increase in high-growth industries since the rent dissipation effect of licensing can be reduced. A positive sign is expected.
Concentration is the collective market share of the four leading firms in the primary operating industry of firm i at period t. (3) Low concentration implies that the firms already have many competitors (lower market power) in their primary product market (Caves, 1970). Therefore, it will be less costly for a licensor to create another licensee in a low concentrated market since the marginal rent dissipation effect is small. The anticipated sign is negative.
Industry patent is the propensity of the primary industry of firm i to receive patents at period t. (4) This variable partly reflects the strength of an industry's intellectual property rights protection. Since strong IPR protection can solve the problem of idea appropriation and help reduce the transaction costs of licensing, a positive sign is expected.
Market capacity is the total sales of the principal industry of firm i at period t. (5) A technology holding firm can raise market share and revenue indirectly with its technology by allowing licensees to supply products and receiving appropriate licensing payments from them. Given that it is usually easier to increase market share in larger markets, larger market capacity should be attractive to technology holders. The expected sign is positive.
Year dummy variables are Y1994, Y1995, Y1996, Y1997 and Y1998. These are included to control for potential year-specific macroeconomic effects.
Descriptive statistics are presented in Table 2.
Results and Discussions
Table 3 reports the estimation results for information and communication technology (ICT) (model 1), biotechnology (model 2), and advanced materials (model 3). Coefficient on patent stock enters positively and significantly in biotechnology and advanced materials. Experience_dummy maintains its expected sign and significance level across all industries. Logsale, however, generates an unexpected positive sign and is statistically significant in ICT and advanced materials. R&D/sale and industry growth enter with significant positive signs only in ICT. Consistent with expectations is the negative sign of concentration, although it is significant only in ICT. Industry patent has a significant positive impact on the propensity to license out technology only in biotechnology. Capital/asset and income/asset fail to produce statistically significant results, and market capacity did not perform well as an explanatory variable for licensing.
One of the unexpected findings was that large established firms actually license out more technology than small farms. We expected a negative association between firm size and licensing activities since small firms may be better off avoiding the risk of marketing their inventions and profiting by selling technology licenses instead. A possible explanation is that large companies tend to have more proprietary technologies to license than small firms due to their better financial capability for innovations. Also, large firms may worry less about potential competition with licensees. They are confident of winning in the potential product market competition thanks to their superior market power. In biotechnology however, company size has no statistically significant impact on technology licensing. The biotechnology industry, including pharmaceuticals, comprises a relatively large number of small firms similar to research labs. Small research-oriented firms often earn their profits through licensing arrangements with more established firms in commercializing a new technology (Gans and Stern 2000). Due to financial constraints, small firms often cannot even attempt to market their inventions without assistance from larger companies. Therefore, small firms are pressured to license technologies and may grant as many licenses as larger companies, especially in biotechnology.
The stock of technical knowledge (patent stock) does not necessarily exert a strong positive influence on licensing incentives across all sectors. It even enters with an opposite sign in ICT industries. These results are somewhat puzzling, but the findings support anecdotal evidence that some companies ignore a large fraction of their patented technologies that could be licensed or profitably sold.
In fact, there seems to be big differences among sectors. While the sign of its coefficient is inconsistent with expectations in others, R&D intensity demonstrates a positive sign with strong statistical significance in ICT, indicating that the propensity to license technology rises with technological capabilities. In addition, firms have more incentive to license in rapidly growing industries only in ICT. Licensees often create additional competition for the incumbent technology owners. The adverse effects of such competition should be less important when the industry is fast growing, other things being equal.
Industry propensity to receive patents--a proxy of the strength of the intellectual property protection system in the industry--has a significant positive effect on technology licensing only in the case of biotechnology. It is, of course, well understood that the uniform IPR (intellectual property rights) legislation has different effects on industries with very different characteristics. The strength of an IPR protection system, as perceived by the members of an industry, varies extensively across industries according to their nature and history of business practice. For instance, Cohen, et al. (2002) distinguish between "complex" versus "discrete" product industries on the basis of whether a new product comprises numerous separately patentable elements or relatively few. They argue that complex technologies may be easier to invent with weak IPR, whereas simple technologies can be better protected with strong IPR. ICT and advanced materials will typically be good examples of the former, while the biotechnology cluster, like drug industry, is typically a good example of the latter. Our findings hint that IPR play a more important role in biotechnology and strong IPR protection can be the solution for the appropriation of ideas problem in the technology market, enhancing the set of candidate technologies for sale.
We found some similarities across industries as well. Prior licensing experience raised the probability that firms would engage in technology licensing as a licensor across all three technology clusters. This suggests that transaction cost considerations are clearly an important determinant of managers' licensing incentives. Also, industry concentration has a negative impact on the propensity to license out technology in all industries, although it is significant only in ICT. Low concentration among licensors would raise the incentive of a licensor to create another licensee, since this would affect the licensor's profit less than in a less competitive market. Market capacity did not seem to be a significant determinant in all sectors, perhaps because it is a very crude proxy, measuring revenue effects of technology licensing.
A technology alliance can be a powerful weapon in the strategic manager's arsenal of options (Chart and Heide, 1993). Where the appropriation and adaptation of technological advances takes a central role, technology licensing has become increasingly important for the competitive strategy of firms in high technology industries. As Anand and Khanna (2000) observed, technology licensing is one of only a few significant methods of technology transfer between firms and one of the most commonly observed inter-firm contractual agreements today.
This paper investigated what determines firms' incentives to license out their technology and whether there are cross-industry differences as well as similarities across three high-technology clusters: information and communication technology, biotechnology, and advanced materials.
We found that the firm's characteristics, such as the company's stock of technological knowledge (patent stock), prior involvement in technology licensing, size and R&D intensity, as well as the firm's primary industry characteristics such as concentration, sales growth and the propensity to receive patents, are key factors that influence managers' licensing decisions. The results also suggest that there are significant inter-sectoral differences as well as similarities in determinants of the propensity to transfer technology through licensing agreements.
Table 1. Frequency Distibution of Sample and Licensor Firms, by Industry, 1994-1999 Percent of Number licensor (c) of firms among Technology sample sample, by cluster (a) Industry (b) firms industry ICT Computer & office equipment 122 43 (35.2%) Electronic & equipment 235 48 (20.4) Communication 86 10 (11.6) Computer & data processing 240 80 (33.3) services Total 683 181 (26.5) Biotechnology Drugs & pharmaceutical 206 77 (37.4) Agricultural chemicals 16 2 (12.5) Measuring & controlling devices 116 13 (11.2) Medical instruments & supplies 153 20 (13.1) Total 491 112 (22.8) Advanced Chemicals & plastics 115 16 (13.9) Materials Petroleum & coal products 36 0 (0) Rubber products 68 5 (7.4) Total 219 21 (9.6) Notes: (a) The author refers to CorpTech's (Corporate Technologies, 1999) classification. (b) The industry definitions follow the Standard Industrial Classifications (SIC); the classification is based on information obtained from the public sources used to collect the data. (c) Sum of exclusive and nonexclusive licenses; includes cross licenses. Table 2. Descriptive Statistics ICT Biotechnology Advanced Mean Mean Materials Variables (Std. Dev.) (Std. Dev.) Mean (Std. Dev.) Licensing .0913 .0733 .0289 (.2880) (.2607) (.1676) Patent Stock 62.3036 13.3460 37.3711 (503.4017) (75.0263) (262.7969) Logsale 1.8693 1.3878 2.5844 (1.0633) (1.0920) (1.0911) Capital/Asset .0611 .0462 .0679 (.0598) (.0510) (.0512) R&D/Sale .2068 2.8748 .1018 (1.6016) (24.6563) (1.6834) Income/Asset -1.1219 -.2034 .0037 (.7548) (.8779) (.2925) Experience Dummy .1259 .1914 .0868 (.3318) (.3935) (.2816) Industry Growth 12.6743 3.5594 4.9053 (4.5929) (1.9824) (3.8628) Concentration 29.2982 9.4346 13.069 (10.4202) (2.240) (5.966) Industry Patent .1861 .3142 .3606 (.1030) (.1295) (.2072) Market Capacity 147562.4 250420.3 270752.2 (70908.96) (106821.7) (103840.4) Table 3. Random-Effects Probit Estimate 1 2 3 Licensing ICT Biotechnology Advanced Materials Patent Stock -.0000641 .0015 *** .0008 *** (.0001) (.0006) (.0002) Logsale .4159 *** .0839 .4139 *** (.0638) (.0604) (.1333) Capital/Asset .4461 -.1945 .0386 (.8521) (1.0178) (2.4117) R&D/Sale .0451 ** -.0095 -2.2803 (.0205) (0072) (4.1141) Income/Asset -.0067 -.0087 .2941 (.0752) (.0654) (1.0691) Experience_Dummy 1.3752 *** 1.4237 *** 1.1323 *** (.1638) (.1342) (.2389) Industry Growth .0404 *** -.0355 -.1932 (.0130) (.0403) (.2899) Concentration -.0267 *** -.1173 -2.1738 (.0099) (.7293) (2.6884) Industry Patent .0652 3.4562 * -2.8477 (.8078) (1.8152) (3.0084) Market Capacity 3.27e-07 1.92e-06 .00003 (1.16e-06) (.00002) (.00004) 1994 -.0652 .5652 2.9405 (.1975) (.6695) (2.9641) 1995 .1053 .5438 2.0588 (.1983) (.3453) (2.4345) 1996 -.0695 .0533 .6888 (.1731) (.3320) (.8309) 1997 .3688 ** .2998 .0417 (.1608) (.2678) (.6530) 1998 .1765 (.1730) .4510) (.1340) (.1964) (.4518) Constant -2.961 *** -.9236 11.4977 (.4010) (3.1935) (15.7402) N 4059 2946 1314 Wald Chi Square 158.06 179.36 94.77 Notes: 1. *** significant at 1 %; ** significant at 5 %; * significant at 10 % significance level. 2. Standard errors are in parentheses.
Acknowledgements: The author thanks Professor Nicholas Vonortas and the Center for International Science and Technology Policy at the George Washington University for their research support, and participants at the SAM 2004 International Business Conference for their helpful comments.
(1.) See also Kim (2004) for the analysis on ICT.
(2.) PATENT STOCK = [I.sub.it] + (1-[rho]) PATENT [STOCK.sub.1], where [I.sub.it] is the number of patent granted to the firm by U.S. Patent Office in a particular year from 1969 to 1999, PATENT [STOCK.sub.1] is firm i's patent stock at period t-1, and p is the depreciation rate, which is taken to be 15%. Fifteen percent is frequently taken as the rule of thumb in knowledge depreciation in the empirical literature. There is little difference with higher values (20%, 30%) in our experiment. The value of the patent stock depreciates because of newer inventions by the same owner or others, developments in complementary technologies, and ultimately the expiration of legal rights. We assume the initial patent stock is zero. The assumption over initial stock makes no practical difference due to the long period of depreciation.
(3.) The 4-firm concentration ratio is used (U.S. Census Bureau, 1997).
(4.) Total number of patents in a firm's primary two-digit SIC industry divided by total R&D expenditures in that industry. Both numbers are constructed by aggregating over firms in COMPUSTAT (net of firm's own patents and R&D).
(5.) Value of shipments and receipts (millions of dollars) are used (Annual Survey and Economic Census, U.S. Census Bureau).
Arora, A., Fosfuri A., and Gambardella, A. (2001). Markets for technology: The economics of innovation and corporate strategy. Cambridge, MA: The MIT Press.
Arora, A., and Gambardella, A. (1994). The changing technology of technological change: General and abstract knowledge and the division of innovative labour. Research Policy, 523-532.
Arrow, K. J. (1962). Economic welfare and the allocation of resources for invention. In R. R. Nelson (Ed.), The rate and direction of inventive activity: Economic and social factors. Princeton University Press.
Branscomb, L. M., and Florida, R. (1998). Challenges to technology policy in a changing world economy, In L. M. Branscomb and J. H. Keller (Eds.), Investing in innovation. Cambridge, MA: The MIT Press.
British Technology Group (BTG). (1998). IPR Market Benchmark Study.
Caves, R. E. (1970). Uncertainty, market structure and performance: Galbraith as conventional wisdom. In J. W. Markham and G. F. Papanek (Eds.), Industrial organization and economic development. Houghton Mifflin.
Caves, R. E. (1996). Multinational enterprise and economic analysis (2nd ed.). Cambridge, MA: Cambridge University Press.
Chan, P. S., and Heide, D. (1993). Strategic alliances in technology: Key competitive weapon. SAM Advanced Management Journal, 9-17.
Cohen, W., Goto, A., Nagata, A., Nelson, R., and Walsh, J. (2002). R&D spillovers, patents and the incentives to innovate in Japan and the United States. Research Policy, 1349-1367.
Gans J. S., and Stern, S. (2000, Winter). Incumbency and R&D incentives: Licensing the gale of creative destruction. Journal of Economics & Management Strategy, 485-511.
Hart, O. (1995). Firms, contracts and financial structure. Oxford: Oxford University Press.
Hauknes, J. (1998). Services in innovation--innovation in services, final report, si4s project (erb-soel-ct-96-1015). Targeted Socio-Economic Research Program, DG XII, European Commission.
Kim, Y. (2004, September). An analysis of the incentives to licensing in U.S. information technology. The Journal of American Academy of Business, 31-36.
Leech, D. P., Link, A. N., Scott, J. T., and Reed, L. S. (1998). The economics of a technology-based service sector. [Planning report, 98-2], National Institute of Standards & Technology, Gaithersburg, MD: NIST.
Menard, C. (1996). Why organizational matter: A journey away from the fairy tale. Atlantic Economic Journal, 281-300.
Mitchell, W., and Singh, K. (1992). Incumbents use of pre-entry alliances before expansion into new technical subfields of an industry. Journal of Economic Behavior & Organization, 347-372.
Nelson, R. R., and Winter, S. (1982). An evolutionary theory of economic change. Cambridge, MA: The Belknap Press of Harvard University Press.
Organization for Economic Cooperation and Development. (2000). Knowledge management in the learning economy. Paris: OECD.
Orr, D. (1974). The determinants of entry: A study of the Canadian manufacturing industries. Review of Economics and Statistics, 58-66.
Porter, M. E. (1998, Nov.-Dec.). Clusters and the new economics of competition. Harvard Business Review, 77-90.
Salanie, B. (2002). The economics of contracts (5th ed.). Cambridge, MA: Cambridge University Press.
Standard & Poor's. (1999). COMPUSTAT (North America), Retrieved at the Gelman Library, The George Washington University.
Teece, D. J. (1977). Technology transfer by multinational firms: The resource cost of transferring technological know-how. Economic Journal, 242-261.
Williamson, O. E. (1979). Transaction cost economics: The governance of contractual relations. Law Economics, 233-261.
Young Jun's research interests include technological strategic alliances, managerial economics, econometrics, and industrial economics.
Young Jun Kim, The George Washington University
|Printer friendly Cite/link Email Feedback|
|Publication:||SAM Advanced Management Journal|
|Date:||Jan 1, 2005|
|Previous Article:||Toxic convertibles: catalysts of doom or financing of last resort?|
|Next Article:||Is eBay for everyone? An assessment of consumer demographics.|