1What Should Drive an Innovation Strategy?Chuck Eesley (Stanford), Edward B. Roberts (MIT), Delin Yang (Tsinghua Univ.)Strategic Management SocietyOctober, 2009(with support of a Kauffman Foundation Dissertation Fellowship, the Tsinghua Univ. Alumni Association, and the MIT Entrepreneurship Center)
New Firms Undertake Search ActivityWhat enables (or constrains) the adoption of an innovation strategy?Search (Simon, 1957; Nelson, Winter 1982, Cyert, March 1963) Firm-Centric, Past (Levinthal & March, 1981; Katila & Ahuja, 2002)
 External environment
 Educ./Training – (Beckman and Burton, 2008, Burton, Sørensen, Beckman 2002, Burton and Beckman, 2007)Liquidity Constraints (Arrow 1962, Nelson 1959)Kortum & Lerner 2000; Hall 2005Effects of Public R&D (Romer, 1990; Stern and Porter, 2004)Direct vs. indirect effects (Goolsbee, 1998; Henderson, Jaffe, Trachtenberg, ’98)
 Bush 1945, Aghion, Dewatripont, Stein 2008
 Grant-based vs. Contract-Based
 Short run vs. Long run effects (Mansfield, 1977)
David et al. 2000, Evenson, Kislev 1976, Adams 1990, Adams 1993, Kortum 1997Search: Environment, Initial Conditions, & StrategyExternal FundingContract R&DGrant R&DProductivity of applied R&DSupply of technical laborInput marketFounder characteristicsNew Firm (two searches)Product/market search
Funding searchR&D investment decisionsSurvival/Growth or ProductivityStrategy
Search ModelPr(Iit=1) = (HAitAit )/ C(Vit, i)		(1)A = total stock of knowledge/ideas, the ease of finding an innovation opportunity(Romer, 1990)HAt is a firm-specific component - ease of search technology space scaled by Search costs for funding 		C(V, )= 1/(V)		 			(2)= (0,1) - firm-specific component ability to raise fundingLevel of VC funding				V = ρθAt(3)ρ Proportion technological opportunities that are radical (Kortum & Lerner, 2000)θ = [0,1] financial frictions (information asymmetry/moral hazard)Public R&Dhuman capital, knowledge stock and the level of VC funding
Two Solutions to More Innovation in SocietyPr(It=1) = (HAtAt )/ [1/(ρθAt)] 		(4)  (1) existing firms doing more innovation  (2) new firms are created, a higher percentage of these innovate5
Public R&D Influence on Firm SearchH1: Grant based public R&D expenditures will (via indirect effects) result in greater knowledge spillovers and greater use of an innovation strategy (with a lag) H2: Grant-based public R&D expenditures will result (via direct effects in higher prices for research inputs) in lower use of an innovation strategy (contemporaneous)H3: Grant or contract-based public R&D expenditures will result in more scientists/engineers becoming entrepreneurs with a lag.H4: Venture capital funding will result in greater use of an innovation strategy. (Counter-hypothesis to hypothesis 1)
Ideal ExperimentExogenous shift in HAit , Ait or C(Vit, i)Proportion of firms adopting an innovation strategyThe effectiveness of government incubators, seed funding, …and other such policies for funding R&D deserves further study, ideally in an experimental or quasi-experimental setting. In particular, studying the cross-country variation in the performance of such programs would be desirable, because the outcomes may depend to a great extent on institutional factors that are difficult to control for using data from within a single country.							- Bronwyn Hall 2005tProportion of firms adopting an innovation strategyNo shiftt
Merged MIT and Tsinghua DatasetSimilar educational background, academic talent (engineering)Similar industries (electronics & software)2,067 + 330 firm observationsInnovation measuresPatents (foreign and domestic)Product/service available in the market 3 years ago (China)Importance of innovation, speed to market, low cost, other factorsDetailed fundraising dataUS and China data on public R&D expenditures, publications and venture capitalSources: OECD Science and Technology Indicators, 2008; Ministry of Science and Technology, China; China Statistical Yearbooks; SDC Venture Economics Database; Asian Venture Capital Journal; Dow Jones VentureOne; Thomson ISI Inflation and Purchasing Power Parity conversion process
Merged MIT and Tsinghua Dataset
Industry Breakdown
Capital and Lack of IdeasOnly 27% had never considered entrepreneurship.
Patents a Good Measure of Innovation Strategy?
MethodsDifferences-in-differencesestimationProbit Model Prob (innovation= 1) = Prob(Yt=1) = α + β1(funding)t + β2(science and technology funding)t + β3(human capital) + β4(business environment)t + β5(funding)t*(China) + β6(China location) + β7(science and technology)t*(China) + yeart + sector + η + φ +εtXi = Set of controls academic dept., region, education, work history, job type, Communist party, overseas educ. or work, family economic status.Include (τ + η + φ) grad. year, sector and Bachelor’s academic dept. fixed effectsProportional Hazards Test13
National LevelStandard errors are robust. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels (two-tailed). PPP GDP is used for China.  Log indicates that a log transformation was done to address the skewed distribution.  In parentheses, (t-1) and (t-6) indicate that the variables were lagged one year and six years, respectively
Individual LevelControls: Non-US citizen, Communist party, Gender, graduation year, founding year, Bach. Dept.Standard errors are robust. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels (two-tailed).
Regional Level (2004-2007)Expenditures are in billions of yuan, ratios and growth rates are percentages, all lagged 1 year. Controls for region and founding year are included.Standard errors are robust. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels (two-tailed).
Progress UpdateNational levelIndividual levelRegional levelFurther robustness checksAlternative definitions of innovationAlternative measures of R&D/funding environmentOther shifts – law/IP
Conclusion and ImplicationsInstitutional LevelTypes of institutional support needed for innovative, high growth firms
If evolutionary theory correct, larger impacts may be on new firmsIndividual LevelSuggestive of who to look for as cofoundersStrategyBetter understanding of environmental influences on search
Where to spend more time for early-stage, high tech founders

Eesley Comparing China US

  • 1.
    1What Should Drivean Innovation Strategy?Chuck Eesley (Stanford), Edward B. Roberts (MIT), Delin Yang (Tsinghua Univ.)Strategic Management SocietyOctober, 2009(with support of a Kauffman Foundation Dissertation Fellowship, the Tsinghua Univ. Alumni Association, and the MIT Entrepreneurship Center)
  • 2.
    New Firms UndertakeSearch ActivityWhat enables (or constrains) the adoption of an innovation strategy?Search (Simon, 1957; Nelson, Winter 1982, Cyert, March 1963) Firm-Centric, Past (Levinthal & March, 1981; Katila & Ahuja, 2002)
  • 3.
  • 4.
    Educ./Training –(Beckman and Burton, 2008, Burton, Sørensen, Beckman 2002, Burton and Beckman, 2007)Liquidity Constraints (Arrow 1962, Nelson 1959)Kortum & Lerner 2000; Hall 2005Effects of Public R&D (Romer, 1990; Stern and Porter, 2004)Direct vs. indirect effects (Goolsbee, 1998; Henderson, Jaffe, Trachtenberg, ’98)
  • 5.
    Bush 1945,Aghion, Dewatripont, Stein 2008
  • 6.
    Grant-based vs.Contract-Based
  • 7.
    Short runvs. Long run effects (Mansfield, 1977)
  • 8.
    David et al.2000, Evenson, Kislev 1976, Adams 1990, Adams 1993, Kortum 1997Search: Environment, Initial Conditions, & StrategyExternal FundingContract R&DGrant R&DProductivity of applied R&DSupply of technical laborInput marketFounder characteristicsNew Firm (two searches)Product/market search
  • 9.
    Funding searchR&D investmentdecisionsSurvival/Growth or ProductivityStrategy
  • 10.
    Search ModelPr(Iit=1) =(HAitAit )/ C(Vit, i) (1)A = total stock of knowledge/ideas, the ease of finding an innovation opportunity(Romer, 1990)HAt is a firm-specific component - ease of search technology space scaled by Search costs for funding C(V, )= 1/(V) (2)= (0,1) - firm-specific component ability to raise fundingLevel of VC funding V = ρθAt(3)ρ Proportion technological opportunities that are radical (Kortum & Lerner, 2000)θ = [0,1] financial frictions (information asymmetry/moral hazard)Public R&Dhuman capital, knowledge stock and the level of VC funding
  • 11.
    Two Solutions toMore Innovation in SocietyPr(It=1) = (HAtAt )/ [1/(ρθAt)] (4) (1) existing firms doing more innovation (2) new firms are created, a higher percentage of these innovate5
  • 12.
    Public R&D Influenceon Firm SearchH1: Grant based public R&D expenditures will (via indirect effects) result in greater knowledge spillovers and greater use of an innovation strategy (with a lag) H2: Grant-based public R&D expenditures will result (via direct effects in higher prices for research inputs) in lower use of an innovation strategy (contemporaneous)H3: Grant or contract-based public R&D expenditures will result in more scientists/engineers becoming entrepreneurs with a lag.H4: Venture capital funding will result in greater use of an innovation strategy. (Counter-hypothesis to hypothesis 1)
  • 13.
    Ideal ExperimentExogenous shiftin HAit , Ait or C(Vit, i)Proportion of firms adopting an innovation strategyThe effectiveness of government incubators, seed funding, …and other such policies for funding R&D deserves further study, ideally in an experimental or quasi-experimental setting. In particular, studying the cross-country variation in the performance of such programs would be desirable, because the outcomes may depend to a great extent on institutional factors that are difficult to control for using data from within a single country. - Bronwyn Hall 2005tProportion of firms adopting an innovation strategyNo shiftt
  • 14.
    Merged MIT andTsinghua DatasetSimilar educational background, academic talent (engineering)Similar industries (electronics & software)2,067 + 330 firm observationsInnovation measuresPatents (foreign and domestic)Product/service available in the market 3 years ago (China)Importance of innovation, speed to market, low cost, other factorsDetailed fundraising dataUS and China data on public R&D expenditures, publications and venture capitalSources: OECD Science and Technology Indicators, 2008; Ministry of Science and Technology, China; China Statistical Yearbooks; SDC Venture Economics Database; Asian Venture Capital Journal; Dow Jones VentureOne; Thomson ISI Inflation and Purchasing Power Parity conversion process
  • 15.
    Merged MIT andTsinghua Dataset
  • 16.
  • 17.
    Capital and Lackof IdeasOnly 27% had never considered entrepreneurship.
  • 18.
    Patents a GoodMeasure of Innovation Strategy?
  • 19.
    MethodsDifferences-in-differencesestimationProbit Model Prob(innovation= 1) = Prob(Yt=1) = α + β1(funding)t + β2(science and technology funding)t + β3(human capital) + β4(business environment)t + β5(funding)t*(China) + β6(China location) + β7(science and technology)t*(China) + yeart + sector + η + φ +εtXi = Set of controls academic dept., region, education, work history, job type, Communist party, overseas educ. or work, family economic status.Include (τ + η + φ) grad. year, sector and Bachelor’s academic dept. fixed effectsProportional Hazards Test13
  • 20.
    National LevelStandard errorsare robust. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels (two-tailed). PPP GDP is used for China. Log indicates that a log transformation was done to address the skewed distribution. In parentheses, (t-1) and (t-6) indicate that the variables were lagged one year and six years, respectively
  • 21.
    Individual LevelControls: Non-UScitizen, Communist party, Gender, graduation year, founding year, Bach. Dept.Standard errors are robust. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels (two-tailed).
  • 22.
    Regional Level (2004-2007)Expendituresare in billions of yuan, ratios and growth rates are percentages, all lagged 1 year. Controls for region and founding year are included.Standard errors are robust. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels (two-tailed).
  • 23.
    Progress UpdateNational levelIndividuallevelRegional levelFurther robustness checksAlternative definitions of innovationAlternative measures of R&D/funding environmentOther shifts – law/IP
  • 24.
    Conclusion and ImplicationsInstitutionalLevelTypes of institutional support needed for innovative, high growth firms
  • 25.
    If evolutionary theorycorrect, larger impacts may be on new firmsIndividual LevelSuggestive of who to look for as cofoundersStrategyBetter understanding of environmental influences on search
  • 26.
    Where to spendmore time for early-stage, high tech founders