Corporate Finance Corporate Investment and innovation Yanzhi Wang

Eberhart, Maxwell and Siddique (2004) Firms experience long run abnormal return after their unexpected R&D increases Firms who unexpectedly increase their R&D exhibit abnormal operating performance post to R&D increases High–tech firms outperform than low-tech firms do 2

Purpose and Hypothesis Efficient market hypothesis (EMH) has been tested by the investigation on long run stock return since Ritter (1991). Considerate papers have documented the long run return drift after corporate events. (e.g., IPO, SEO, repurchase and so on) These corporate events also support the idea that managers “time” the market to take advantage of a “window of opportunity” when the firms’ stocks are mispriced. 3

Purpose and Hypothesis Rather than the corporate financial decision, R&D represents an investment decision, which should be no clear timing motive for managers. Previous study: Chan, Lakonishok and Sougiannis (2001) argue that R&D cannot predict the stock return. Based on Danel and Titman (2001), if investors indeed misreact to the intangible information such as R&D increase, then we should observe significant long-term abnormal stock return following the R&D increases. 4

Purpose and Hypothesis Return True return at event Time Event 1yr 2yr 3yr 4yr 5 No underreaction Underreaction- type I Underreaction- type II Increases in R&D exp.

Data Their data consists in the sample with changes in R&D expenditure from Compustat during 1951 to The finalized sample includes 8,313 firm-year observations whose R&D and R&D intensity both change over 5%. “R&D increase event” is based on the accounting announcement. Long run return is started four month after that “R&D changing” fiscal year-end. 6

Data- Some Sub-samples 5-year sample: To exclude the R&D multiple increases during each 5-year horizon. High/low tech sample: High or low-tech firms based on the high-tech categories from Chan, Martin and Kensinger (1990). High/low growth sample: Market-to-book ratio greater (smaller) than unity. 7

Data- Sample Distribution 8 Computers, software, and electronic product Healthcare, medical equipment, and drugs

Result- Return by Factor Model Long run stock return debate: Fama (1998), Lyon, Barber and Tsai (1999), Loughran and Ritter (2000), Mitchell and Stafford (2000). Calendar time series regression (factor model) is the least power in detecting the long run stock return. Either Fama and, French (1993) or Carhart (1997) factor model is examined in this paper. 9

Result- Return by Factor Model (since 1951) 10

Result- Return by Factor Model (since 1974) 11

Result- Return by Zero-portfolio Run the factor model using “abnormal monthly return”, where abnormal monthly return is computed by sample firm’s return minus matching firm’s return. Method 1:match the industry, size, profitability. Method 2:match the industry, size, B/M, momentum. 12

Result- Return by Zero-portfolio 13

Operating Performance Abnormal operating performance is to subtract matching firm’s operating performance. Here they adopt the profit margin as the measure of the operating performance. Method 1: match the current profit margin within the same 2-digit SIC industry. Method 2: match the B/M, size and momentum. 14

Operating Performance 15

Eberhart, Maxwell and Siddique (2008) Though R&D can increase equity value by increasing firm value, it can also increase equity value at the expense of bondholder wealth through an increase in firm risk because equity is analogous to a call option on the underlying firm value. Shi (2003) finds that the net effect of R&D is negative for bondholders. This paper reexamines Shi’s (2003). They find that Shi’s (2003) results are sensitive to the method of measuring R&D intensity. When we use what we argue is a better measure of R&D intensity, we find that the net effect of R&D is positive for bondholders. Second, when we use tests that Shi (2003) recognizes are even better than the ones that he uses, we find even stronger evidence of the positive effect of R&D on bondholders. Third, this paper documents a negative relation between R&D increases and default risk.

R&D and bondholder wealth Bondholder wealth can also change in response to an R&D increase, and this change may not be in the same direction as the shareholder wealth change. Though shareholders and bondholders both benefit from a rise in firm value, shareholders benefit at the expense of bondholders from an accompanying rise in firm risk because stocks are analogous to call options (implicitly sold by the bondholders) on the underlying firm value. Therefore, the benefit to shareholders of R&D that previous studies report may just reveal the effect of a wealth transfer from bondholders to shareholders, not any benefit to the entire firm value (i.e., the sum of stock and debt values).

Finding of Shi (2003) Shi (2003) examines the relation between R&D and bond risk measures (i.e., bond ratings and bond risk premiums) for newly issued bonds. Using a series of cross-sectional regressions of the initial bond ratings and risk premiums on R&D intensity and several control variables, he examines the net effect of the potentially higher firm value (again, beneficial to bondholders) versus the potentially higher firm risk (again, detrimental to bondholders) associated with R&D. He finds evidence that a higher R&D intensity corresponds to worse bond ratings and higher risk premiums and concludes that the net effect of R&D is negative for bondholders.

Reexamination When this paper uses Shi’s (2003) intensity measure of R&D divided by the market value of equity, they find results qualitatively similar to his. When this paper measures R&D intensity as the ratio of R&D to sales or R&D to assets, however, it is found that a higher R&D intensity is associated with better bond ratings and lower spreads, ceteris paribus. The problem with the ratio of R&D to the market value of equity is that the denominator incorporates the market’s expectations of the R&D value, and this inverts the true relation between R&D intensity and bond ratings (and risk premiums).

Data The paper uses the SDC new debt issue database to identify newly issued bonds over the 1990–1998 time period. Following Shi (2003), they restrict this sample to the five R&D intensive industries: chemicals and pharmaceuticals (SIC 28), machinery and computer hardware (SIC 35), electrical and electronics (SIC 36), transportation vehicles (SIC 37), and scientific instruments (SIC 38). The final sample consists of 218 bonds issued by 72 firms for which they have complete bond and financial information.

Summary statistics

SUR analysis

Abnormal stock and bond return

Analysis of changes in default prob. How do they measure the default probability? What is distance-to- default scores?

Chen, Chen, Liang and Wang (2014) We examine how R&D incoming spillovers affect long-run firm performance following firms’ R&D increases. We use a stochastic frontier production method to capture R&D incoming spillover effects. Firms reaping more benefits from R&D investment made by other firms experience more improvement in profitability and more favorable long-run stock performance in the post-R&D- increase period. The evidence also shows that firms experiencing more R&D outgoing spillover effects tend to underinvest in R&D.

R&D spillover A research and development spillover occurs when privately owned firms are unable to fully appropriate returns from their R&D investment. Jaffe (1986) shows that when the potential R&D spillover pool increases by 1%, profits of other firms increase by 0.3%. Arrow (1962) and Jones and Williams (1998) argue that R&D spillovers cause R&D investment to deviate from its optimal level. In this view, R&D spillovers promote underinvestment in R&D. Cassiman and Veugelers (2002) suggest that R&D spillovers can be seen from two perspectives: i) incoming spillovers, which assess a firm’s ability to take advantage of innovations created by other firms; and ii) appropriability of a firm’s own R&D, which evaluates a firm’s ability to profit exclusively from its new technologies.

R&D spillover and stock performance R&D incoming spillovers might not have an immediate impact on firm valuation when there is an unexpected increase in R&D, because investors may have difficulty measuring the extent of the firm-specific incoming spillover effect. Firms have different incoming spillovers, depending on technology flows from specialized conferences or meetings, foreign direct investment flow from international channels, an R&D investment itself, and the location of R&D generator and receiver. This study attempts to measure the R&D incoming spillover effect and its association with the long-run performance of R&D-increase firms.

Stochastic frontier production function In a given year, we assume that firm i uses inputs including capital (K), labor (L), and research and development (RD) to produce output (Y), according to a Cobb-Douglas production function technique as follows: We follow the stochastic frontier production analysis of Aigner, Lovell, and Schmidt (1977) in including the two error terms. v i represents the symmetric disturbance and is assumed to be independently and identically distributed (i.i.d.) as N(0, ). v i is independent of the error term u i, which is assumed to be i.i.d. half-normal distribution |U|, given U~N(0, ).

Stochastic frontier production function The random variable u i captures the incoming spillover effect of firm i from other firms, and the white noise v i captures the impact of other random factors on the shocks of output. u i is non-negative because the spillover effect represents a positive R&D externality. We further interpret The incoming spillover effect of firm i can be decomposed into two terms. The first term relates the incoming spillover effect of firm i to RD j and s j, where RD j is the R&D level of firm j, and s j is the extent to which firm i absorbs the R&D spillover effect from firm j

Data and method We follow Eberhart et al (2004) and identify sample firms that meet five selection criteria: i) their ratios of R&D expenditures-to-sales are over 5%; ii) their ratios of R&D expenditures-to- average total assets (beginning plus end-of-year assets divided by two) are over 5%; iii) changes in their ratios of R&D expenditures-to-sales are over 5%; iv) changes in their ratios of R&D expenditures-to-average total assets are over 5%; and v) their percentage changes in R&D expenditures are over 5%. The final sample includes 7,554 firm-year observations

Summary statistics

Long-run abnormal returns

Earnings announcement abnormal returns To avoid a bad model problem or potential bias in the estimation of long-run return performance, we further examine abnormal returns for the short period in which quarterly earnings announcements occur ( La Porta, Lakonishok, Shleifer, and Vishny (1997)).

Operating performance regression

R&D outgoing spillover effects We note at the outset that R&D spillovers may result in an underinvestment in R&D because of the free rider problem. This underinvestment feature has to do with the appropriability of R&D results rather than incoming spillovers. A firm that is less able to appropriate R&D benefits would be more likely to have R&D outgoing spillover effects. Thus, R&D outgoing spillover effects could lead to firm underinvestment. We use the production function to estimate the outgoing spillover effect for each of our firm-year observations. We run a regression using the pre-R&D-increase time-series data: Thus, a 1 reflects how much a firm’s R&D investment spills over to its industry peers; a higher value indicates more outgoing spillover effects

Outgoing spillover and R&D underinvestment