Download to read offline
Uploaded byFlorida Atlantic University
Samuel Kortum and Josh Lerner 2000 RAND Journal of Economics 31(4) Assessing the Contribution of Venture Capital to Innovation
The document examines the relationship between venture capital (VC) and innovation from 1965 to 1992, highlighting that VC funding accounted for 14% of U.S. innovative activity by 1998. Using a stylized model, it finds a strong positive correlation between VC and patenting, suggesting that VC plays a significant role in fostering innovation despite constituting a smaller fraction of corporate R&D. The study also addresses challenges in methodology and the impact of the legal environment on patents.
More Related Content
Similar to Samuel Kortum and Josh Lerner 2000 RAND Journal of Economics 31(4) Assessing the Contribution of Venture Capital to Innovation
More from Florida Atlantic University
Samuel Kortum and Josh Lerner 2000 RAND Journal of Economics 31(4) Assessing the Contribution of Venture Capital to Innovation
- 1. Assessing the Contributionof Venture Capital to Innovation. The RAND Journal of Economics, 2000 31(4) Samuel Kortum and Josh Lerner 88 Citations Shiva Indukoori 1 88 Citations
- 2. Introduction • This paperexamines the relationship between innovation and VC. • Earlier models had simple regression models Ignoring the arrival of technological opportunities. • The authors present a stylized model of the relationship between venture capital, R&D, and innovation. Shiva Indukoori 2
- 3. Data • 1965 to1992 • Patents: US Patent and Trademark Office (USPTO) • Venture funding: Venture Economics • Industrial R&D expenditures: Collected by the US National Science Foundation (NSF) • Number of observations: 560 • Number of industries: 20 Shiva Indukoori 3
- 4. VC Disbursement andR&D Expenditure • Disbursements: VC funding in Drugs, office and computing, and communication equipment industry (54%). • The comparable figure for R&D expenditures is 39%. Shiva Indukoori 4
- 5.
- 6.
- 7. Methodology • The productionfunction approach is used to assess the contribution of venture capital. • Authors say that this article should be seen as a first cut at quantifying venture capital's impact on innovation. Shiva Indukoori 7
- 8. Concerns • Impact ofLegal environment for patents • Quality of Patents Shiva Indukoori 8
- 9. Legal environment forpatents • In 1982 creation of a centralized appellate court for patent cases coincided with an increase in the rate of U.S. patent applications. • Dummy control variables for each year are employed. Shiva Indukoori 9
- 10. Patent Quality Indicator •VC may spur patenting with no impact on VC-backed companies’ innovation. • They may do this to impress potential investors or to avoid expropriation of their ideas by these investors. • Patents of 122 venture-backed and 408 non-venture-backed companies were compared. Shiva Indukoori 10
- 11. Patent Production Function DependentVariables • Patented innovations : USPTO Independent variables • Venture funding: Venture Economics • R&D Spending: U.S. National Science Foundation (NSF) Shiva Indukoori 11
- 12. Patent Production Functionwith Venture Capital P = Number of patent applications filed by U.S. inventors in each industry and year R = Privately funded industrial R&D V = Venture disbursements ρ = degree of substitutability between R and V as means of financing innovative effort b = Role of venture capital in the patent production function. b>0 i = industry t = year α/ρ = returns to scale (% change in patenting brought about by a 1% increase in both R and V) u = shifts in the propensity to patent / technological opportunities Shiva Indukoori 12 Pit = (Rρ it+bVρ it)α/ρ uit
- 13. ‘b’ captures roleof VC in innovation Shiva Indukoori 13 If b=0, the equation reduces to Standard or Reduced Form: Pit = Rα it uit For any b>0, Venture funding matters for innovation Pit = (Rρ it+bVρ it)α/ρ uit
- 14. ‘ρ’ measures degreeof substitutability between R and V as means of financing innovative effort As ‘ρ’ goes to zero, the patent production function approaches Cobb Douglas Production Function. Shiva Indukoori 14 Pit = Rit α/(1+b) +Vit αb/(1+b)uit For ρ = 1, the equation reduces to Pit = (Rit+bVit)αuit Pit = (Rρ it+bVρ it)α/ρ uit
- 15. Non-Linear Least SquaresRegression • Dependent variable: Number of ultimately successful patent applications filed by U.S. inventors in each industry and year. • Independent variables • Privately financed R&D in that industry and year • Venture disbursements in dollar value or number of firms in the industry receiving venture backing. • Control variables • Logarithm of the federally funded R&D in the industry. • Year (dummy variable to control for differences in the propensity to patent). Shiva Indukoori 15 Pit=α ln(Rit+bVit)+lnuit
- 16.
- 17. Linear (OLS)Regression Analysis ShivaIndukoori 17 ln Pit=α ln Rit + αb(Vit/Rit)+ln uit • Dependent variable: Logarithm of the number of (ultimately successful) patent applications filed by U.S. inventors in each industry and year. • Independent variables • Privately financed R&D in that industry and year • Venture disbursements in dollar value or number of firms in the industry receiving venture backing. • Control variables • Logarithm of the federally funded R&D in the industry. • Year (dummy variable to control for differences in the propensity to patent).
- 18.
- 19. Instrument Variable RegressionAnalysis Assumptions • Innovations, on average, translate into patents in a proportional manner. • Marginal costs of innovating • Production function for innovations(I) in each industry(i) and time period(t) is Iit = (Rit+bVit)αNit = Hα itNit Where H is the total innovative effort and N is the shock to the Invention Production function. Shiva Indukoori 19 ln Pit=α ln Rit + αb(Vit/Rit)+ln uit
- 20.
- 21.
- 22. Patent to R&DRatio • In 1979, the U.S. Department of Labor clarified the Employee Retirement Income Security Act, a policy shift that freed pensions to invest in venture capital. • This shift led to a sharp increase in the funds committed to venture capital. • Technological changes. • The impact of venture capital on the patent-R&D ratio, rather than on patenting itself is estimated to make the causality problem disappear Shiva Indukoori 22
- 23.
- 24.
- 25. Findings • VC fundingaccounted for about 14% of U.S. innovative activity by 1998. • Though VC averaged less than 3% of corporate R&D from 1983 to 1992, it was more than 8% of industrial innovations in the decade ending in 1992. • Results are robust to different measures of venture activity, subsamples of industries, and representations of the relationship between patenting, R&D, and venture capital. • Strong association between venture capital and patenting exists suggesting a positive and significant effect of VC. • Corporate R&D and venture funding are highly substitutable in generating innovations. Shiva Indukoori 25
- 26. Questions • The processby which projects are chosen ex-ante is the key source of advantage, or is it the monitoring and control after the investment is made? • Why industrial R&D managers have not adopted some of the same approaches to financing innovation. What barriers have limited the diffusion of the venture capitalists' approaches? • Is it possible to disentangle the distinct effects of the rise of venture capital from other R&D management innovations? Shiva Indukoori 26
- 27.
Editor's Notes
- #2 Impact of VC on VC Backed Firm’s Innovation this is the other of the paper by Hirukawa on who is first Samuel Kortum, Boston University and NBER; [email protected], Josh Lerner, Harvard University and NBER; [email protected]. old (2000) in terms of data sources and to study the nature of VCs but models are interesting as they are more of quant and economics. Image challenging to read and no marking
- #3 2. This model overcomes the limitations of this approach in earlier literature. 3. Production function is the stylized model
- #4 DATA CHALLENGES USPTO does not classify by industry. data doesn’t allow a clean division between R&D financed by corporations and R&D financed by venture capital organizations. 1992 to 2000 ?
- #5 Disbursements: VCs in concentrated industries.
- #6 Log scale represents fundraising and Disbursements (VC in concentrated Industries). First leap: Late 70s and Early 80s. Second Leap: Second half of the 1990s, there was another leap in venture capital activity, which emerged as the dominant form of equity financing in the United States for privately held high-technology businesses.
- #7 This is year wise distribution of patents, R&D Expenditure, and VC Disbursement Patent Applications declined from the early 1970s to the mid-1980s, but then rose sharply. The ratio of venture capital to R&D jumped sharply in the late 1970s and early 1980s, and fell a bit thereafter.
- #8 Patents are the measure of Innovation
- #9 The authors had 2 major concerns about the quality of the study.
- #10 To control for changes in either the propensity to file for patents or for these applications to be granted.
- #11 VC may be interested in patents for reasons other than the firm’s innovation A sample of 530 Firms in Middlesex county of Massachusetts.
- #12 The standard form is the reduced form of production function. PATENT INNOVATIONS U.S. patents are issued to U.S. inventors by industry and date of application by USPTO
- #14 b>0 implies venture funding matters for innovation. If b = 0 it reduces to the standard form shown earlier b represents the role of VC in innovation. Standard form if there is no VC or no impact of VC. Year and Industry are the two variables
- #15 Standard form if there is no VC or no impact of VC. COBB DOUGLAS PRODUCTION FUNCTION Y(K,L) = ALα Kβ
- #16 The analysis is tabu Restricted Equation is Pit=α ln(Rit+bVit)+lnuit Where R and V are perfect substitutes
- #17 The results suggest that VC funding matters. In the unconstrained equation, the magnitude of b is substantial. A likelihood ratio test overwhelmingly rejects the special case of b equal to zero (with a p-value of less than .005). R&D and venture capital are highly substitutable, with the point estimate of [Rho] close to one. The results for the restricted equation are shown in 2nd 4th columns . Restricted Equation is Pit=α ln(Rit+bVit)+lnuit Where R and V are perfect substitutes
- #19 Table 2 and Table 3 suggest that the linear specification is more conservative in its implications for the potency of venture funding. Table 3 presents regressions employing the linear specification. Implied potency is positive in both Table 2 &3 ln Pit=α ln Rit + αb(Vit/Rit)+ln uit
- #20 Instrument Variable V/R Ratio In addition to the direct expenditures on R&D and venture disbursements, study assumes that there are associated indirect expenses. These might include the cost of screening opportunities, recruiting managers and researchers, and undertaking crucial regulatory approvals to sell the new product
- #21 V/R ratio (αb) is the instrumental variable with Alpha fixed at 0.20 and 0.50 Venture capital is αb 15 times @ alpha 0.20 and only 5 times @alpha 0.50
- #22 Panel B uses instruments both for venture funding relative to R&D and for R&D itself.
- #24 The results are consistent with the findings in Tables 3 and 4. VC funding is significantly more potent than corporate R&D.
- #25 This presents Univariate comparisons in 3 categories. Substantial differences between 122 venture-backed and 408 non-venture-backed firms: Venture firms are more likely to patent, have previous patents cited, and engage in frequent and protracted litigation of both patents and trade secrets. All tests of differences in means & medians in these three categories are significant at least at the 5% confidence level, as well as when regression specifications r employed. These findings help allay fears that differences in the propensity to patent drove our findings in Sections 3 and 4. At the same time, it is important to acknowledge that while the firm-level analysis allows us to examine whether the innovative behavior of venture-backed and non-venture-backed firms differs on measures other than patent counts, it does not allow us to address endogeneity issues as in the industry-level analysis.