Empirical Research in Innovation Dongmei Li University of South Carolina
Roadmap Literature review Data on innovation Empirical methods Asset pricing implication of innovation Corporate finance Data on innovation Input: R&D Output: Patents Empirical methods Portfolio sorts/factor regressions Fama-MacBeth regressions
Asset Pricing Implication of Innovation R&D intensity Chan, Lakonishok, and Sougiannis (2001): R&D/sales does not predict returns R&D/market equity predicts significantly higher returns Explanations: over-extrapolation
Asset Pricing Implication of Innovation (cont’d) Li (2011): interaction between R&D and financial constraints Motivation: two asset-pricing puzzles Financial constraints and stock returns R&D and stock returns Findings (theoretical and empirical): Among constrained firms, risk increases with R&D-intensity Among R&D-intensive firms, risk increases with financial constraints
Asset Pricing Implication of Innovation (cont’d) R&D efficiency (Hirshleifer, Hsu, and Li 2012) Motivation: limited attention Measurement: patents or citations generated per dollar of R&D Results: R&D skills (Cohen, Diether, and Malloy 2012)
Asset Pricing Implication of Innovation (cont’d) Innovative originality Hirshleifer, Hsu, and Li (2014)
Asset Pricing Implication of Innovation (cont’d) Capital investment in innovative capacity (IC) Kumar and Li (2015) Motivation: Investment anomalies Behavioral explanations: underreaction to empire building (Titman, Wei, and Xie 2004); overreaction to investment (Cooper, Gulen, and Schill 2008) Real options (Carlson, Fisher, Giammarino 2004, 2006) Q-theory (Li and Zhang 2010)
Asset Pricing Implication of Innovation (cont’d) Examples of IC investment: building research infrastructure, purchasing R&D equipment, buying patents Uniqueness of IC investment: helps firms generate options with uncertainty We study dynamic implications of IC investment on Expected returns Future investment Profitability
Asset Pricing Implication of Innovation (cont’d) R&D growth Eberhart, Maxwell, and Siddique (2004) R&D spillover
What drives innovation? Financial constraints and innovative efficiency Heitor, Hsu, and Li (2014)
Data on Innovation Input — R&D expenditure Compustat Data quality is higher after 1975 FASB No. 2: R&D reporting rules are standardized Missing R&D is generally equivalent to zero or very small amount of R&D
Data on Innovation (cont’d) Output — patents, citations, company/security identifiers (gvkey, permno) NBER https://sites.google.com/site/patentdataproject/Home http://eml.berkeley.edu/~bhhall/patents.html https://iu.app.box.com/patents
Empirical Methods — Portfolio Sorts References: Fama and French (1992, 1993) Timing: Portfolio formation: end of June of year t (based on characteristics in year t-1) Holding period: July of year t to June of year t+1 Purposes: Avoid look-ahead bias: make sure sorting variables are observable to investors at formation (financial reporting lag) Rebalance once a year (reduce transaction cost)
Measure of Abnormal Returns Time-series regressions of portfolio excess returns on factors returns Intercepts: alphas (returns that cannot be explained by existing factor models) Slopes: factors loadings (quantity of risk)
Empirical Methods — Fama-MacBeth Regression References: Fama and MacBeth (1973) Conduct monthly cross-sectional regressions of individual firms’ returns (July of year t to June of year t+1) on lagged variables Compute average and t-statistics of the time-series slopes from regressions above Benefits: Control more variables Address cross-sectional correlation Allow “more” samples