paper does stock liquidity enhance or impede firm innovation

We find that an exogenous increase in liquidity decimalization leads to a higher level of institutional ownership by transient and quasi-indexers which reduces innovation.. regulators ha

Does Stock Liquidity Enhance or Impede Firm Innovation?

This Draft: February 2011

Abstract

There has been much debate about whether stock liquidity enhances or impedes firm innovation Some fear that high stock liquidity hinders innovation (a long-term investment) by making managers myopic Others believe that high stock liquidity makes stock prices more efficient which mitigates managerial myopia and enhances firm innovation We document a negative relationship between stock liquidity and firm innovation productivity This relationship is more pronounced when management is less entrenched or when firm profits are low Both are consistent with high transaction costs insulating managers from pressures to maximize short-term profits (or stock price) To establish causality we show the negative relationship between stock liquidity and innovation productivity holds following an exogenous shock to liquidity (decimalization) We next examine the role of institutional investors We find that an exogenous increase in liquidity (decimalization) leads to a higher level of institutional ownership by transient and quasi-indexers which reduces innovation Increases in dedicated institutional ownership are not correlated with an exogenous increase in liquidity but are positively associated with innovation Our findings suggest that institutional investors who gather private information enhance innovation while transient and quasi-indexer institutional investors impede innovation

JEL classifications: G12; G19; G34; G38; O31

Keywords: Stock Liquidity; Innovation; Managerial Myopia; Institutional Ownership; Corporate

Governance

Porter (1992) and Bhide (1993) stress that U.S regulators have developed securities laws and regulations that promote market liquidity at the expense of firm governance.1 Their concern

is that high stock liquidity stifles innovation by making it easier for dissatisfied blockholders or institutional investors to exit This may pressure managers to be too short-sighted, or myopic, when selecting investment projects Investment in long-term projects tends to reduce short-term earnings as the initial outlays are often treated as expenses under U.S GAAP and the investment may not generate profits in the foreseeable future Low profits increase the need for information gathering and monitoring by investors especially since long-term projects may require a significant investment in intangible assets which are difficult to observe If stock liquidity is high, low earnings may cause impatient investors to exit resulting in a drop in stock price If managers have an incentive to keep stock price high due to career concerns or takeover fears, and if stockholders have imperfect information, Stein (1988) predicts that managers will feel pressured to cut investment in long-term projects to boost current profits There may also be pressures for managers to invest in short-term projects at the expense of long-term projects to

      

rules, rules to eliminate price manipulation, reduction of minimum tick size, and deregulation of stock commissions

reduce the mispricing associated with long-term assets as the uncertainty of these assets may take

a long time to resolve (Shleifer and Vishny (1990)) In either case, firms with higher stock liquidity will invest less in long-term projects and will generate less innovation than is optimal

In contrast, Edmans (2009) argues that high stock liquidity mitigates managerial myopia

by making it easier for blockholders to exit Edmans and Manso (2010) suggest that this effect is especially pronounced for firms with multiple blockholders Both papers show that blockholders collect costly information but they provide governance without intervention If blockholders remain as investors after collecting costly information, they send the signal of good prospects and the stock price remains high If they sell shares they send the signal of poor prospects and the stock price drops This behavior by blockholders will result in more efficient stock prices If stock prices are efficient, fundamental information linked to long-term investments will be reflected in the stock price This reduces the pressure on managers to boost short-term profits and instead encourages them to invest in long-term projects that maximize firm value and generate innovation even if these projects reduce short-term profits According to the models in Edmans (2009) and Edmans and Manso (2010), firms with high stock liquidity will invest more

in long-term projects and generate more innovation than firms with low stock liquidity

We enter the debate by examining the effect of stock liquidity on firm-level innovation

We surmount several empirical challenges in this study First, as pointed out by Stein (2003),

“Managerial myopia is difficult to test because it results in underinvestment in activities that are difficult to observe.”2 Our use of an observable investment output (innovation productivity) helps us assess the success of long-term investment and investment in intangible assets which have traditionally been difficult to observe Second, stock liquidity may be endogenous as it could be correlated with unobservables that are correlated with innovation There could also be reverse causality where the level of expected future innovation predicts the level of stock liquidity We use an exogenous shock to liquidity to mitigate endogeneity and reverse causality concerns More specifically, we examine whether an exogenous shock to liquidity (decimalization) increases or decreases innovation productivity several years into the future

      

2 Due to the difficulty in measuring whether there is underinvestment in unobservable activities, Stein (2003) points out that most studies in managerial myopia have tended to examine firm operating performance around equity issues This is because managers face short-term pressure to increase earnings to boost the current stock price prior

to equity issues

Our baseline tests show a negative relationship between stock liquidity (measured by the inverse of relative effective spread) and innovation output (patents and citations-per-patent) one, two and three years in the future Provided that higher levels of innovation are value enhancing, our findings are consistent with Porter (1992) and Bhide (1993) as they predict that higher stock liquidity magnifies managerial myopia and reduces innovation.3 The assumption that higher innovation is value enhancing seems reasonable since Hall, Jaffe and Trajtenberg (2005) and Atanassov, Nanda, and Seru (2007) find that higher innovation has a positive effect on firm Tobin’s Q

To establish causality, we show that the negative relationship between stock liquidity and future innovation holds following an exogenous shock to liquidity (i.e., decimalization of the minimum tick size) Decimalization provides a natural experiment to test the effect of stock liquidity on innovation productivity First, decimalization directly affected stock liquidity (liquidity rose on average) and the change in liquidity surrounding decimalization exhibited variation in the cross-section of stocks (Bessembinder (2003) and Furfine (2003)) Second, it seems unlikely that decimalization directly affected innovation Third, it also seems unlikely that a change in expected future innovation influenced the cross-sectional change in liquidity brought about by decimalization

We further identify the causal relationship from stock liquidity to innovation productivity

by relying on cross-sectional variations in “pressures” affecting managers If high transaction costs serve as a mechanism that “insulates” managers from pressures arising from short-term investors, the pressure to invest in short-term projects should be lower when managers are more entrenched (the G-index is higher or the CEO is also the chairman of the board of directors) We confirm that the negative effect of stock liquidity on innovation productivity is weaker when managers are more entrenched Similarly, since long-term investment reduces current earnings which may depress stock price, the pressure to reduce investment in innovation is most likely the highest when current profitability is low We confirm that the negative effect of stock liquidity

on innovation productivity is magnified in these situations Together, these results are consistent

      

3 An alternative interpretation of the results is that illiquidity rather than liquidity leads to myopic investment decisions If patents generated over the next three years are actually short-term rather than long-term investment outputs, the results would be consistent with illiquidity pressuring managers to focus on short-term innovation (which generates patents within 3-years) at the expense of long-term innovation This alternative interpretation is consistent with Edmans (2009) and the finding in Fang, Noe, and Tice (2009) that higher stock liquidity causes higher firm value

with high transaction costs serving as a mechanism that “insulates” managers from pressures arising from short-term investors

We next identify a specific channel through which liquidity affects innovation, i.e., institutional investors There are theoretical reasons to believe that stock liquidity affects the level and type of institutional ownership and that the level and type of institutional ownership affect innovation Maug (1998) argues that high liquidity will facilitate the entry of blockholders who actively monitor which presumably would help mitigate managerial myopia and improve performance Similarly, Edmans (2009) argues that high liquidity makes the threat of exit by blockholders more credible and a credible threat serves as a governance mechanism that reduces managerial myopia However, Porter (1992) and Bhide (1993) worry that high liquidity makes it easier for dissatisfied institutional investors to exit The ease with which dissatisfied institutional investors can exit discourages them from monitoring and puts short-term performance pressures

on managers which may lead to a reduction in long-term investment in innovation and a

reduction in innovation efficiency

Testing between the aforementioned hypotheses is challenging for two reasons First, the models we are testing predict that liquidity is a determinant of institutional ownership and institutional ownership affects innovation Second, liquidity may be endogenous as unobservables that affect liquidity may be correlated with innovation To address these challenges we identify the change in institutional ownership caused by an exogenous shock to liquidity (decimalization) then examine the effect of the change in institutional ownership caused

by the exogenous shock to liquidity on innovation output one, two, and, three years in the future

Using the Bushee (1998, 2001) definitions of institutional investor types, we find that an exogenous increase in liquidity, caused by decimalization, leads to a higher level of ownership

by transient and indexer institutional investors Furthermore, for transient and indexers, the increase in institutional ownership caused by an exogenous increase in liquidity reduces innovation productivity The results seem to suggest that the higher liquidity following decimalization facilitates the entry of fickle transient and quasi-indexer institutional investors who do not gather private information but instead put short-term pressures on managers

quasi-Interestingly, the results are different when we examine dedicated institutional investors

An exogenous increase in liquidity, caused by decimalization, does not lead to a higher level of institutional ownership by dedicated investors Furthermore, we find that changes in the holdings

of dedicated institutional investors (which are not caused by changes in liquidity) are positively correlated with innovation We interpret this finding as consistent with monitoring or a reduction

in short-term pressures by institutional investors who trade for reasons unrelated to liquidity Once we control for the fact that institutional investors are not homogenous, we find that transient and quasi indexer institutional investors impose the short-term pressures predicted by Porter (1992) and Bhide (1993) while dedicated institutional investors who gather private information mitigate managerial myopia. 4

Our finding of higher levels of innovation for illiquid stocks complements the findings of recent papers on innovation Aghion, Van Reenen, and Zingales (2009) show that stocks with higher institutional ownership have more innovation as higher institutional ownership lowers manager career concerns that arise with riskier innovation We provide insight into their paper

by showing that dedicated institutional investors who trade for reasons other than liquidity could

be driving their findings Edmans (2010) contains a theoretical model which predicts that LBO firms will have less managerial myopia due to more institutional investor monitoring Lerner, Sorensen, and Stromberg (2010) examine this issue empirically and find that LBO firms generate more important patents after the LBO transaction Hence, they find an increase in innovation productivity in the extreme case where the liquidity of a firm’s stock is gone (since the firm is delisted and its stock is no longer traded) Similarly, the model in Ferreira, Manso and Silva (2010) shows it is optimal for firms to be private (in essence have no stock liquidity) when they wish to innovate Lastly, Chemmanur and Tian (2010) find that the adoption of anti-takeover defenses increases innovation by insulating managers from short-term stock market pressures

We find that having an illiquid stock also shields managers from short-term market pressures  

Our paper also adds to the small but growing literature linking stock liquidity to firm performance Fang, Noe, and Tice (2009) find that an exogenous shock to liquidity leads to an increase in firm performance (higher firm Q) by creating a more efficient feedback mechanism from investors to managers via stock prices or by enhancing manager pay-for-performance sensitivity Bharath, Jayaraman, and Nagar (2010) add to their work and find that exogenous shocks to liquidity lead to greater increases in firm value for stocks with a higher level of blockholders We contribute to this literature by studying the effect of liquidity on firm       

4 One caveat here is that Edmans (2009) and Edmans and Manso (2010) are built on the assumption that there are

blockholders who own a significant stake and gather private information Dedicated institutional investors are not necessarily blockholders Therefore, we do not directly test Edmans (2009) or Edmans and Manso (2010)

innovation productivity Though Fang, Noe, and Tice (2009) show that higher stock liquidity results in a higher firm Q on average, we show that higher stock liquidity results in less future innovation on average If lower innovation results in a lower firm Q, we identify one channel through which higher liquidity may result in a lower firm Q

Our paper also builds on the empirical literature examining managerial myopia The papers in this literature tend to find evidence of managerial myopia for publicly listed firms For example, Bhojraj and Libby (2005) find that managers choose projects that will maximize short- term earnings as opposed to total cash flows in response to a pending stock issue; Holden and Lundstrum (2009) find that managerial myopia is reduced when Long-Term Equity Appreciation Securities (LEAPS) are introduced due to a reduction in the cost of trading on long-term investment information; and Asker, Farre-Mensa and Ljungqvist (2010) find that listed firms exhibit myopia as they invest less and their investment levels are less responsive to changes in investment opportunities when compared to unlisted firms Others like Wahal and McConnell (2000) and Bushee (1998) find evidence that institutional investors mitigate managerial myopia Wahal and McConnell (2000) find a positive relationship between R&D expenditures and institutional holdings while Bushee (1998) finds that managers are less likely to cut R&D in response to an earnings decline if institutional holdings are high.5 Our paper complements these papers as our focus is on the effect of stock liquidity on managerial myopia We find that an increase of transient or quasi-indexer institutional investors due to an increase in liquidity leads

to myopic managerial behavior However, an increase in dedicated institutional investors is not associated with a change in liquidity and their presence reduces myopic managerial behavior

We also use innovation productivity (or successful innovation output) rather than R&D expenditures or investment growth rates to evaluate long-term investments

The rest of the paper is organized as follows The next section describes the sample, measurement of variables, and descriptive statistics In Section III, we present our baseline empirical tests and robustness checks In Section IV, we examine a channel through which liquidity affects innovation, institutional investors Section V concludes

II Sample Selection, Variable Measurement, and Descriptive Statistics

      

institutions that have high portfolio turnover and engage in momentum trading, this increases the probability that managers reduce R&D to reverse an earnings decline Otherwise, institutional investors lower myopic behavior

2.1 Sample Selection

Firm-year patent and citation information is retrieved from the latest version of the National Bureau of Economics Research (NBER) patent database initially created by Hall, Jaffe, and Trajtenberg (2001) Intraday trades and quotes are obtained from the Trade and Quote (TAQ) database to construct stock liquidity measures To calculate the control variables, we collect financial statement items from Compustat Industrial Annual Files, institutional holdings data from Thomson's CDA/ Spectrum database (form 13F), institutional investor classification data from Brian Bushee’s website (http://acct3.wharton.upenn.edu/faculty/bushee), and CEO duality status and the governance index (G-index) from the RiskMetrics (formerly Investor Responsibility Research Center or IRRC) database

As in Fang, Noe and Tice (2009), we require a firm to be traded on the NYSE, the AMEX, or the NASDAQ continuously for at least six months in a fiscal year to be included in the sample The final sample used to investigate the relation between stock liquidity and one-year-ahead number of patents consists of 39,469 firm-year observations between 1994 and

2005.6

2.2 Variable Measurement

2.2.1 Measuring Innovation

The latest version of the NBER patent database provides annual information from 1976

to 2006 on patent assignee names, the number of patents, the number of citations received by each patent, a patent’s application year, and a patent’s grant year, etc Based on the information retrieved from the NBER patent database, we construct two measures for a firm’s innovation productivity The first measure is a firm’s number of patent applications filed in a year that are eventually granted We use a patent’s application year instead of its grant year as the application year is argued to better capture the actual time of innovation (Griliches, Pakes, and Hall (1988)) Although straightforward and easy to compute, this measure cannot distinguish groundbreaking innovations from incremental technological discoveries To further assess a patent’s influence,

we construct the second measure of corporate innovation productivity by counting the number of non-self citations each patent receives in subsequent years Controlling for firm size, the number

      

6 TAQ database dates back to 1993 and patent and citation information is only available through 2006 Therefore,

we limit our sample of the one-year ahead patents and citations between 1994 and 2005

of patents captures innovation productivity while citations-per-patent captures the importance of innovation output To reflect the long-term nature of investment in innovation, both measures

of innovation productivity are measured one, two and three years in the future

Following the existing innovation literature, we adjust the two measures of corporate innovation to address the truncation problems associated with the NBER patent database The first truncation problem arises as the patents appear in the database only after they are granted

In fact, we observe a gradual decrease in the number of patent applications that are eventually granted as we approach the last few years in the sample period (e.g., 2004 and 2005) This is because the lag between patent’s application year and patent’s grant year is significant (about two years on average) and many patent applications filed during these years were still under review and had not been granted by 2006 Following Hall, Jaffe, and Trajtenberg (2001, 2005),

we correct for this truncation bias in patent counts using the “weight factors” computed from the application-grant empirical distribution The second type of truncation problem is regarding the citation counts, as a patent can keep receiving citations over a long period of time, but we observe at best the citations received up to 2006 Following Hall, Jaffe, and Trajtenberg (2001, 2005), the truncation in citation counts is corrected by estimating the shape of the citation-lag distribution

The distribution of patent grants in the pooling sample is right skewed, with the 75thpercentile of the distribution at zero.7 Due to the right-skewed distributions of patent counts and citations per patent, we then use the natural logarithm of the weight-factor adjusted patent counts

and the natural logarithm of the citation-lag adjusted citations per patent, LN_PATENTS and LN_CITEPAT, as the main innovation measures in our analysis To avoid losing firm-year

observations with zero patents or citations per patent, we add one to the actual values when calculating the natural logarithm

2.2.2 Measuring Stock Liquidity

We use the relative effective spread during fiscal year t as our proxy for stock liquidity

(higher relative effective spread means lower liquidity), where relative effective spread is

      

reported in Atanassov, Nanda, and Seru (2007), i.e., 84%, and in Tian and Wang (2010), i.e., 73% Their samples include the universe of Compustat firms between 1974 and 2000 and VC-backed IPO firms between 1985 and 2006, respectively

defined as the absolute value of the difference between the execution price and the mid-point of the prevailing bid-ask quote (effective spread) standardized by the mid-point of the prevailing bid-ask quote While the market microstructure literature has proposed a handful of stock liquidity measures, the effective spread is generally considered to be the best proxy for stock liquidity as it is based on realized high-frequency trading data In fact, it often serves as a benchmark to evaluate the effectiveness of other liquidity measures computed using low-frequency price and volume data (see, e.g., Hasbrouck (2009) and Goyenko, Holden, and Trzcinka (2009))

To construct relative effective spreads, we follow the procedures detailed in Chordia, Roll, and Subrahmanyam (2001) and Fang, Noe, and Tice (2009) Specifically, for each stock in our sample, we first calculate the relative effective spread for each matched quote/trade during a trading day To do so, we match any trade from 1993 – 1998 to the first quote at least five seconds before the trade and any trade after 1998 to the first quote prior to the trade.8 Trades out of sequence, trades recorded before the open or after the close, and/or trades with special settlement conditions are dropped To minimize matching errors, trades that have a quoted spread larger than five dollars, a ratio of effective spread to quoted spread larger than four, and/or a ratio of quoted spread to execution price larger than 0.4 are further deleted from the sample.9

Next, the arithmetic mean of the relative effective spreads for each matched quote/trade over a trading day for a stock is defined as its daily relative effective spread Each daily relative

effective spread within a month is then weighted equally to calculate the monthly relative

effective spread Finally, the annual relative effective spread is defined as the arithmetic mean of the monthly relative effective spreads over a stock’s fiscal year Due to the non-normality of

effective spreads, the natural logarithm of the annual relative effective spread LN_RESPRD is

used in all cross-sectional regression analyses

In supplementary analyses, we calculate the annual Amihud illiquidity measure following Amihud (2002) Specifically, it is calculated as the daily price response associated with one

dollar of trading volume and averaged over fiscal year t for firm i To build the sample using the

      

quotes However, this observed delay has dissipated in recent years Chordia, Roll, and Subrahmanyam (2001) suggested matching any trade to the first quote prior to the trade after 1998

9 Quoted spread is defined as the quoted bid-ask spread of the transaction

Amihud illiquidity measure, we require a stock to be listed at the end of its fiscal year t, to have

at least 200 days of return and volume data available in the CRSP daily files during fiscal year t, and to have a price of $5 or more at the end of fiscal year t The empirical results using Amihud

illiquidity measure are not tabulated but are discussed in the text

2.2.3 Measuring Institutional Ownership

We obtain quarterly institutional investors’ holdings of U.S securities from Thomson's

CDA/ Spectrum institutional 13(f) database For each firm i over its fiscal year t, four quarterly

institutional holdings observations are then weighted equally to compile an annual measure of

the institutional holdings (INSTIPCT).10

As suggested by Porter (1992), the effect of institutional investors on managerial myopia

or the degree of monitoring can be heterogeneous Indeed, Bushee (1998) finds that although institutional holdings as a whole reduce managerial myopia, the presence of a large proportion of transient investors induces myopic investment behavior Since the effect of stock liquidity on innovation may differ for different groups of institutional investors, we disaggregate the annual

institutional holdings (INSTIPCT) into the holdings owned by transient investors (TRAPCT), quasi-indexers (QUAPCT), and dedicated investors (DEDPCT) following the classification

method created by Bushee (1998, 2001) Transient investors represent short-term institutional holders who trade frequently to chase after current profits In contrast, dedicated investors are long-term institutional holders who concentrate in a few firms and exhibit low portfolio turnover Quasi-indexers follow indexing or other passive investment strategies These passive institutional investors trade infrequently to rebalance portfolios and maintain a high degree of diversification Bushee (1998) argues that this classification scheme is superior to other classifications based on type, investment style, or institutional size in the context of examining managerial myopia

2.2.4 Measuring Control Variables

Following the innovation literature, we control for a vector of firm and industry characteristics that may affect a firm’s future innovation productivity All variables are computed       

10 As a robustness check, we replace the annual measure of institutional ownership calculated as the arithmetic mean

of the four quarterly institutional ownership (INSTIPCT) with the institutional ownership reported in the last quarter

of fiscal year t (LASTPCT) The results using LASTPCT are qualitatively and quantitatively similar

for firm i over its fiscal year t In the baseline regressions, our control variables include firm size, LN_MV, measured by the natural logarithm of firm market capitalization; profitability, ROA, measured by return-on-assets ratio; investments in innovation, RDTA, measured by R&D expenditure over total assets; asset tangibility, PPETA, measured by net property, plants and equipment scaled by total assets; leverage, LEV, measured by total debt to total assets ratio; capital expenditure scaled by total assets, CAPEXTA; product market competition, HINDEX, measured by the Herfindahl index based on annual sales; growth opportunities, Q, measured by Tobin’s Q; financial constraints, KZINDEX, measured by the Kaplan and Zingales (1997) five- variable KZ index; and firm age, LN_AGE, measured by the natural logarithm of one plus the

number of years the firm is listed on Compustat To mitigate non-linear effects of product market competition, we also include the squared Herfindahl index in our baseline regressions Detailed variable definitions are described in Panel A of Table 1

2.3 Descriptive Statistics

To minimize the effect of outliers, we winsorize all variables at the top and bottom 1% of each variable’s fundamental distribution.11 Panel B of Table 1 provides summary statistics of the variables used in this study On average, a firm in our final sample has 6.5 granted patents per

year and each patent receives 3.4 non-self citations The stock illiquidity measure LN_RESPRD

has a mean value of -4.482 and a median value of -4.377 (the mean relative effective spread for the sample is 0.020 and median relative effective spread is 0.011), which is comparable to previous studies (e.g., Fang, Noe, and Tice (2009)) Panel B also reports the descriptive statistics

of the control variables In our sample, an average firm has market capitalization of $2.21

billion, ROA of 7.8%, PPE ratio of 28.5%, leverage of 20.9%, Tobin’s Q of 2.1, and is 9.9 years

old since its IPO date

One concern is that patents exist in only a couple of industries which would make our findings relevant for only a couple of industries However, in our sample firms with patents are spread broadly across industries This can be seen in Panel C of Table 1 which shows the number and fraction of firms with and without patents in each industry Using the 49 industry groups as defined by Fama and French (1997), the fraction of firms with zero patents ranges from a low of

      

11 We winsorize all of the variables except for three discrete variables: firm age, governance index, and CEO duality

23% to a high of 100% However, all but one industry (coal) has some firms with patents during our sample period

III Empirical Results

3 1 Baseline Specification

To assess whether stock liquidity enhances or impedes corporate innovation, we estimate the following model:

LN_PATENTS i,t+n (LN_CITEPAT i,t+n ) = a + bLN_RESPRD i,t + cLN_MV i,t + dRDTA i,t + eROA i,t

+ fPPETA i,t + gLEV i,t + hCAPEXTA i,t + iHINDEX i,t +jHINDEX 2 i,t + kQ i,t + lKZINDEX i,t +

mLN_AGE i,t +YR t +FIRM i + error i,t (1)

where i indexes firm, t indexes time, and n equals one, two, or three The dependent variables

capture corporate innovation: the natural logarithm of one plus the number of patents filed (and

eventually granted) (LN_PATENTS) and the natural logarithm of one plus the number of non-self citations per patent (LN_CITEPAT) Since the innovation process generally takes longer than one

year, we examine the effect of a firm’s stock liquidity on its patenting in subsequent years The

liquidity measure, relative effective spread (LN_RESPRD), is measured for firm i over its fiscal year t Since both innovation and stock liquidity are in logarithm form, the regression coefficient estimate on LN_RESPRD gives us the elasticity of innovation productivity to stock liquidity

As most empirical studies involve an endogeneity concern, it is possible that an unobservable variable omitted from our empirical model affects both stock liquidity and corporate innovation, rendering our findings spurious For example, high quality managers may tend to manage companies with more liquid stocks, while high quality managers may also actively engage in long-term innovative projects which result in higher innovation productivity

In this case, management quality is unobservable and correlated with both stock liquidity and innovation, which could bias our coefficient estimates of the stock liquidity measure To address this issue we include firm fixed effects This alleviates the endogeneity concern if the omitted firm characteristics that are correlated with stock liquidity and innovation are constant over time

Table 2 Panel A reports the OLS regression results estimating Eq (1) with LN_PATENTS

as the dependent variable.12 In column (1), we examine the effect of a firm’s stock liquidity on

      

12 In addition to the pooled OLS regressions results reported throughout the paper, we use a Tobit model that takes into consideration the non-negative nature of patent and citation data The results remain robust to using Tobit regression analyses

its number of patents filed (and eventually granted) in one year The coefficient estimate of

LN_RESPRD is positive and significant at the 1% level, suggesting that firms with lower stock

liquidity (higher relative effective spreads) generate a larger number of patents in one year The

magnitude of LN_RESPRD from column (1) suggests that decreasing stock liquidity (increasing LN_RESPRD) by 10% increases the number of patents filed in one year by 1.4% In columns (2)

and (3), we replace the dependent variable with the natural logarithm of the number of patents

filed in two and three years, respectively The coefficient estimates of LN_RESPRD continue to

be positive and significant at the 1% level

We control for a comprehensive set of firm characteristics that may affect future innovation productivity As shown, larger firms (firms with larger market capitalization), more profitable firms (firms with higher ROA), older firms, firms with more tangible assets, and firms with lower leverage are more innovative The baseline results are robust to replacing the firm size proxy (the market capitalization of equity) with either the book value of total assets or firm

sales A larger innovation input measured by a higher R&D-to-assets ratio in year t leads to more innovation output in year t+1, year t+2 and year t+3.13 The regression results are similar if the lagged R&D-to-assets ratio is excluded from the regression Firm product market competition

measured by the Herfindahl index, growth opportunities measured by Tobin’s Q, and financial

constraints measured by the KZ index, do not significantly affect firm innovation productivity

Panel B of Table 2 reports the regression results estimating Eq (1) with the dependent

variable replaced by LN_CITEPAT The coefficient estimates of LN_RESPRD are positive and

significant at the 1% level in all three columns For example, column (1) suggests that a 10%

decrease in firm’s stock liquidity (an increase in LN_RESPRD) increases the number of citations

received by each patent in one year by 1.04% Once again, more profitable firms, older firms, firms with larger innovation input, and larger investment (i.e., higher capital expenditure) are more likely to generate patents with a large impact

      

13 Myopic managers may cut investment in a project too early which will reduce innovation productivity in fiscal

years t+1, t+2, and t+3 controlling for the level of lagged R&D expense in fiscal year t Similarly, myopic

managers may select projects with a faster payback even though they may create less value and less total innovation productivity for the firm The use of innovation productivity (output) as a dependent variable should give us a better idea of the level of myopia than the use of R&D expenses as a dependent variable

 

To address the concern that our results may be driven by the large number of firm-year observations with zero patents and citations per patent, we focus on a subsample of firms that have at least one patent in the pooling sample In an untabulated analysis, we continue to observe positive and significant coefficient estimates on the stock illiquidity measure For

example, the coefficient estimate on LN_RESPRD is 0.232 (p-value<0.001) in model (1) of Table 2 when one-year ahead LN_PATENTS is the dependent variable, and 0.129 (p- value<0.001) in model (1) of Table 2 when one-year ahead LN_CITEPAT is the dependent variable The magnitudes of the coefficient estimates on LN_RESPRD are much larger than

those estimated from the full sample because innovation is more relevant in this subsample

In the baseline model, we include firm fixed effects in the regressions and mainly rely on the time-series variation within a firm to study the effect of stock liquidity on a firm’s innovation productivity As a robustness check, we explore the cross-sectional effect of stock liquidity on corporate innovation in an untabulated analysis We replace firm fixed effects with the Fama-French 49 industry fixed effects (Fama and French (1997)) in Eq (1) The coefficient estimates

on LN_RESPRD remain positive and significant at the 1% level For example, the coefficient estimate on LN_RESPRD is 0.236 (p-value<0.001) in model (1) of Table 2 Panel A when one- year ahead LN_PATENTS is the dependent variable, and 0.060 (p-value<0.001) in model (1) of Table 2 Panel B when one-year ahead LN_CITEPAT is the dependent variable

Finally, we replace the illiquidity measure, relative effective spread (LN_RESPRD), with

the annual Amihud illiquidity measure and repeat the baseline regressions in Table 2 Although the results are not shown for brevity, the coefficient estimates on the Amihud illiquidity measure are significantly positive.14 For example, the coefficient estimate on the annual Amihud illiquidity measure is 0.193 (p-value<0.001) in model (1) of Table 2 Panel A when one-year

ahead LN_PATENTS is the dependent variable

In summary, we document a negative relationship between stock liquidity and firm innovation controlling for the other factors that have been identified to affect innovation The results support the hypothesis that stock illiquidity enhances firm innovation and that transaction costs serve as a mechanism that “insulates” managers from pressures arising from short-term investors

      

the annual Amihud illiquidity measure is positive but not statistically significant

3.2 Identification

As discussed earlier, there is an endogeneity concern that an omitted variable correlated with both stock liquidity and corporate innovation may bias the results towards our findings in section 3.1 While including firm fixed effects alleviates the concern of an omitted variable that remains constant over time, it cannot fully solve the issue if the omitted variable is time-varying

In addition, there is a potential reverse causality problem where expected firm innovation may affect a firm’s stock liquidity

To further address the endogeneity concerns, we identify an exogenous shock to stock liquidity (decimalization of the minimum tick size) during our sample period Prior to 2001, the minimum tick size for quotes and trades on the three major U.S exchanges was $1/16 On January 29, 2001, the New York Stock Exchange (NYSE) and the American Stock Exchange (AMEX) reduced the minimum tick size to pennies and terminated the system of fractional pricing.15 The NASDAQ Stock Exchange decimalized shortly thereafter, over the interval of March 12, 2001 – April 9, 2001 Prior studies have shown significant increases in stock liquidity

as a result of decimalization, especially among actively traded stocks (Bessembinder (2003) and Furfine (2003)) 16

Decimalization appears to be a good candidate to generate exogenous variation in liquidity since decimalization directly affected stock liquidity, decimalization was unlikely to directly affect innovation, and the change in liquidity surrounding decimalization exhibited variation in the cross-section of stocks Regarding reverse causality concerns, we would not expect the change in future innovation to affect the change in stock liquidity brought about by the decimalization change Hence, an examination of the change in innovation productivity following the change in liquidity due to decimalization provides a natural experiment for our

      

stocks in decimals The number was expanded to 57 companies in September 2000 By February 2001, the stocks

of over 3,000 companies listed on the NYSE were traded in decimals, or pennies The U.S Securities and Exchange Commission ordered all U.S stock markets to convert to decimals by April 9, 2001

and Subrahmanyam (2008), the drop in the value-weighted daily average effective spread due to the move from the eighths regime to the sixteenths regime was not as large as the drop in value-weighted daily average effective spread due to the move from sixteenths to decimal quotes Also, there were several spikes in spreads (due to the Asian financial crisis and the Russian financial crisis) in the sixteenths regime which resulted in spreads higher than those

in the eighths regime Hence, the move from sixteenths to decimals is a larger and cleaner shock for our tests A similar graph can be produced using our sample which is available from the authors upon request

tests One remaining concern is that there could be a change in an omitted variable around the

move to decimalization that is correlated with the change in innovation and the change in

liquidity surrounding decimalization However, an omitted variable bias is clearly less likely in a

specification that uses changes surrounding an exogenous shock than in a specification that uses

the levels of variables unconditionally A valid concern of our identification strategy is that the

timing of decimalization coincides with the burst of the dot-com bubble, which led to a large

drop in investment in innovation in general We run robustness tests to control for the possibility

that internet stocks had a large change in liquidity surrounding decimalization (due to the burst

of the dot-com bubble) that was also correlated with the change in innovation surrounding

decimalization

Following the previous literature (e.g., Fang, Noe, and Tice 2009), we measure the

change in the average annual relative effective spread from the pre-decimalization year (year t-1)

to the post-decimalization year (year t+1), where t is the fiscal year during which decimalization

occurred for the firm Due to the long-term nature of innovation projects, we expect innovation

productivity to respond with a lag to changes in stock liquidity To capture this lag, we measure

the change in firm innovation productivity from fiscal year t (the decimalization year) to fiscal

year t+2 The change in innovation productivity out two years after decimalization is regressed

on the change in stock liquidity surrounding decimalization The specification is as follows:

ΔLN_PATENTS i, t to t+2 ( ΔLN_CITEPAT i, t to t+2 ) = a + bΔLN_RESPRD i, t-1 to t+1 + cΔLN_MV i, t-1 to t+1

+ dΔRDTA i, t-1 to t+ 1 + eΔROA i, t-1 to t+1 + fΔPPETA i, t-1 to t+1 + gΔLEV i, t-1 to t+1 + hΔCAPEXTA i, t-1 to t+1

+ iΔHINDEX i, t-1 to t+1 + jΔHINDEX 2

i, t-1 to t+1 + kΔQ i, t-1 to t+1 + lΔKZINDEX i, t-1 to t+1 + IND j + error i,t (2)

where t is the fiscal year during which decimalization occurred for firm i Since we only include

the sub-sample of observations surrounding the shock in this test we control for Fama-French 49

industry fixed effects instead of firm fixed effects Column 1 in Table 3 Panel A reports the

regression results estimating Eq (2) with ΔLN_PATENTS i, t to t+2 as the dependent variable The

dependent variable is replaced with ΔLN_PATENTS i, t+1 to t+3 and ΔLN_PATENTS i, t+2 to t+4 in

columns (2) and (3) respectively to capture the idea that changes in innovation output most likely

respond slowly to changes in liquidity As shown in Table 3 Panel A, an increase in liquidity (a

decrease in LN_RESPRD) surrounding decimalization results in a decrease in the number of

patents filed in subsequent years We repeat the analysis with ΔLN_CITEPAT i, t to t+2,

ΔLN_CITEPAT i, t+1 to t+3 , and ΔLN_CITEPAT i, t+2 to t+4 as the dependent variables in Table 3 Panel

B and show that an increase in liquidity surrounding decimalization also leads to a decrease in the number of non-self citations each patent receives in subsequent years

A valid concern of our identification strategy is that the timing of decimalization coincides with the burst of the dot-com bubble, which led to a large drop in investment in innovation in general Therefore, it is possible that our decimalization test results are mainly driven by the slump in innovation of these high-tech firms To address this concern, we exclude high-tech firms from our decimalization test sample and repeat the analysis Panel C of Table 3

reports the results using the change in firm innovation productivity from fiscal year t (the decimalization year) to fiscal year t+2 as the dependent variable First, we follow the Ljungqvist

and Wilhelm (2003) classification to identify high-tech firms and exclude them from the regressions We report the results in columns (1) and (2) The coefficient estimates of

LN_RESPRD are both positive and significant at the 1% level Next, following Lamont and Stein

(2004), we exclude firms traded on NASDAQ from the sample, and report the regression results

in columns (3) and (4) The coefficient estimates of LN_RESPRD are still positive and

significant at the 5% level In an un-tabulated analysis, we repeat the regressions with the

dependent variables replaced with the two-year-ahead and three-year-ahead ΔLN_PATENTS (ΔLN_CITEPAT), and find both qualitatively and quantitatively similar results We conclude that

our decimalization test results are not driven by the drop in innovation productivity of high-tech firms due to the burst of the dot-com bubble

In summary, in this section we show a causal relationship from stock liquidity to future innovation by utilizing an exogenous shock to liquidity (i.e., decimalization) We find that an increase in liquidity surrounding decimalization significantly reduces firm innovation in subsequent years Overall, the evidence supports the hypothesis that stock illiquidity enhances corporate innovation

3.3 Cross-sectional Analysis

In this section, we aim to further establish the causal relationship from stock liquidity to corporate innovation by relying on cross-sectional variations in “pressures” that induce managerial myopia If high transaction costs serve as a mechanism that “insulates” managers

from stock price pressures imposed by short-term investors, we expect the positive effect of illiquidity on innovation to be more pronounced in situations where firm managers are vulnerable to short-term pressures Along this line of thinking, we examine how variations in two managerial entrenchment measures, CEO duality and shareholder rights, alter the marginal effect of stock liquidity on innovation The innovation productivity of firms with lower levels of manager entrenchment should be more sensitive to changes in liquidity since they will be more vulnerable to short-term price pressures Similarly, since firm managers may be under more pressure when profits are low (as the stock price may drop), the negative effect of stock liquidity

on innovation productivity (more short-term price pressure) should be more pronounced when firm profits are lower

3.3.1 Management Entrenchment

In this section, we examine how the degree of management entrenchment affects the marginal effect of stock liquidity on firm innovation To measure the degree of management

entrenchment, we first use an indicator (DUALCEO) that equals one if firm i’s CEO is also the

chairman of the board (COB) Prior studies show that CEO duality increases CEO power, reduces board independence, and insulates the CEO from market pressures (e.g., Fama and Jensen (1983) and Jensen (1993)) To test this we estimate the following model:

LN_PATENTS i,t+n = a + bLN_RESPRD i,t + cDUALCEO i,t + d Mean Adj. LN_RESPRD i,t × DUALCEO i,t + eLN_MV i,t + fRDTA i,t + gROA i,t + hPPETA i,t + iLEV i,t + jCAPEXTA i,t

+kHINDEX i,t +lHINDEX 2

i,t + mQ i,t + nKZINDEX i,t + oLN_AGE i,t +YR t + IND j + error i,t (3)

where i indexes firm, t indexes time, j indexes industry, and n equals one, two, or three.17 Since

the DUALCEO variable changes slowly over time for a given firm, we control for Fama-French

49 industry fixed effects instead of firm fixed effects in the regressions

Table 4 reports the regression estimates for Eq (3) As shown, the coefficient estimates

of the interaction term, de-meaned LN_RESPRD × DUALCEO, are negative in all three columns

and statistically significant at the 1% level in all three columns The sign on the coefficient estimate of the interaction term suggests that the negative effect of stock liquidity on firm innovation (short-term price pressure) is mitigated when a firm’s CEO is also COB The       

17 For brevity, we only report the regression results with LN_PATENTS as the dependent variable hereinafter, but all results hold if LN_CITEPAT is used as the dependent variable instead

coefficient estimates in column (1) imply that if a firm’s DUALCEO is zero (i.e., the CEO is not the COB), a 10% decrease in stock liquidity (an increase in LN_RESPRD) increases the number

of patents filed in one year by 2.3%; however, if the CEO is the COB, i.e., DUALCEO equals

one, a 10% decrease in stock liquidity increases the number of patents filed in one year by only 0.91% (=2.3% +10%*(-0.139)), a 60.4% (=(2.3%-0.91%)/2.3%) reduction in innovation productivity

Our second measure for management entrenchment is shareholder rights We make use

of a commonly cited anti-takeover index GINDEX introduced in Gompers, Ishii, and Metrick (2003) GINDEX is constructed by identifying 24 distinct anti-takeover provisions (ATPs) and

counting the number of the ATPs a firm has adopted DeAngelo and Rice (1983) show that managers are more entrenched in firms that have adopted antitakeover charter amendments

Thus, a higher value of GINDEX corresponds to a more entrenched management and a lower degree of manager exposure to market pressures We examine the effect of GINDEX by

estimating the following model:

LN_PATENTS i,t+n = a + bLN_RESPRD i,t + cGINDEX i,t + d Mean Adj LN_RESPRD i,t × Mean Adj

GINDEX i,t + eLN_MV i,t + fRDTA i,t + gROA i,t + hPPETA i,t + iLEV i,t + jCAPEXTA i,t +kHINDEX i,t

+lHINDEX 2 i,t + mQ i,t + nKZINDEX i,t + oLN_AGE i,t +YR t + IND j + error i,t (4)

where i indexes firm, t indexes time, j indexes industry, and n equals one, two, or three The variable of interest is the interaction between de-meaned LN_RESPRD and de-meaned GINDEX.18 Since the GINDEX changes slowly over time for a given firm, we control for Fama-

French 49 industry fixed effects instead of firm fixed effects in the regressions

Table 5 reports the regression results estimating Eq (4) The coefficient estimates of the

interaction term, de-meaned LN_RESPRD × de-meaned GINDEX, are negative and significant at

the 5% level in column (1) and at the 10% level in columns (2) and (3), which suggests that the negative effect of stock liquidity on corporate innovation is mitigated by a higher value of

GINDEX, i.e., when the management is more entrenched According to the coefficient estimates reported in column (1), if the firm’s GINDEX is at the sample’s mean value (i.e., the number of ATPs is 9.034), a decrease in stock liquidity (an increase in LN_RESPRD) by 10% increases the number of patents filed in one year by 1.68% If the firm’s GINDEX is increased by one

      

coefficient easier to interpret Each variable is de-meaned using observations pooled across time and across firms

standard deviation (2.769) from the mean value, a decrease in stock liquidity by 10% increases

the number of patents filed in one year by only 0.99% (=1.68%+10%*(-0.025)*2.769), a 41.1%

reduction in innovation productivity

In summary, in this section, we rely on heterogeneity in corporate governance to explore

the marginal effect of stock liquidity on corporate innovation We find that the negative effect of

liquidity on innovation is largely mitigated if the manager is more entrenched Entrenchment

measures and stock illiquidity both insulate managers from the pressures imposed by short-term

investors and consequently help enhance innovation productivity

3.3.2 Profitability

Financial reporting conveys substantial information to outsiders about firm performance

and significantly influences market expectations and stock prices (Kothari (2001)) It is

reasonable to infer that firms whose earnings reflect poor profitability are subject to greater

market pressure imposed by short-term investors as investors may exit and the stock price may

drop They are also subject to greater monitoring by institutional investors If it is true that

illiquid stocks provide a harbor for corporate innovation (which may depress short-term

profitability), we conjecture that the negative effect of stock liquidity on innovation is magnified

for firms with lower profitability as this is when the insulation provided by high transaction costs

is most valuable

Under U.S GAAP, firms are required to charge R&D expenditures to expenses when

they incur Thus, there is a mechanical relation between current profitability and innovation, i.e.,

investment in innovation will lower current profitability To make sure that the results are not

driven by this mechanical relation, we follow the procedures detailed in Chan, Lakonishok, and

Sougiannis (2001) and adjust ROA as if R&D expenditures were capitalized and amortized over

a 5-year period (ADJROA) We estimate the following model:

LN_PATENTS i,t+n = a + bLN_RESPRD i,t + cADJROA i,t + d Mean Adj LN_RESPRD i,t × Mean Adj

ADJROA i,t + eLN_MV i,t + fRDTA i,t + gPPETA i,t + hLEV i,t + iCAPEXTA i,t + jHINDEX i,t

+kHINDEX 2

i,t + lQ i ,t + mKZINDEX i ,t + nLN_AGE i,t + YR t + FIRM i + error i,t (5)

where i indexes firm, t indexes time, and n equals one, two, or three Compared to Eq (1), we

add an interaction term composed of de-meaned LN_RESPRD and de-meaned ADJROA to

capture how firm profitability alters the marginal effect of stock liquidity on innovation Since

firm profitability can vary significantly over time for a given firm, we control for firm fixed effects in the regressions as in our baseline regressions

Table 6 reports the regression results estimating Eq (5) The variable of interest is the

interaction between de-meaned LN_RESPRD and de-meaned ADJROA The coefficient

estimates on the interaction term are negative and significant at the 1% level in all three columns, suggesting that innovation productivity is more sensitive to stock liquidity when firm current profitability is lower The economic effect is significant as well In fact, the magnitudes of

coefficient estimates in column (1) suggest that if the firm’s current ADJROA is at the sample’s mean value, a 10% decrease in stock liquidity (i.e., increasing LN_RESPRD) results in an

increase in the number of patents filed in one year by 1.34% However, if the firm’s current

ADJROA is one standard deviation (0.118) below the mean value, a 10% decrease in stock

liquidity results in an increase in the number of patents filed in one year by 1.68% (=1.34%+10%*(-0.285)*(-0.118)), a 25% (=(1.68%-1.34%)/1.34%) increase in innovation productivity

In summary, the empirical evidence presented in this section suggests that when firms have exhibited poor prior profitability (i.e., managers are under greater pressure to generate higher profits in the near future), the negative effect of stock liquidity on innovation is more pronounced

IV Institutional Investors

The reported results are consistent with the hypothesis that high transaction costs serve as

a mechanism that insulates managers from pressures arising from short-term investors In this section, we identify a specific channel through which this happens: institutional ownership Studies in the literature (see, e.g., Coffee (1991), Porter (1992), Bhide (1993), Maug (1998), and Edmans (2009)) all agree that liquidity affects the ability of institutional investors to exit and enter easily Where they disagree is on the effect of an increased ability to exit and enter on firm governance and managerial myopia We examine this empirically More specifically, we examine how stock liquidity affects the level and type of institutional ownership, and how the level and type of institutional ownership caused by stock liquidity affects corporate innovation

Since stock liquidity is only one possible determinant of institutional ownership, we need

to identify the portion of institutional holdings explained by liquidity before we can examine

how liquidity affects innovation via institutional investors To do this we make use of an exogenous shock to liquidity -decimalization in 2001 We adopt a two-step approach We first identify the change in institutional ownership caused by the change in liquidity due to an exogenous shock to liquidity (decimalization) We then examine the effect of the change in institutional ownership caused by the exogenous shock to liquidity on innovation

In the first step, we regress the change in institutional ownership surrounding decimalization on the change in liquidity surrounding decimalization, and decompose the variation in change in institutional ownership into two parts: one is due to the change in stock liquidity, and the other part is due to factors other than stock liquidity The specification for the first-step regression is as the follows:

ΔINSTIPCT i,t-1 to t+1 = a + bΔLN_RESPRD i,t-1 to t+1 + error i,t (6)

where t is the fiscal year during which decimalization occurred for firm i

The first-step regression estimates for Eq (6) are reported in the column (1) of Table 7

Panel A The coefficient estimate of ΔLN_RESPRD is negative and significant at the 1% level,

consistent with the notion that an exogenous increase in stock liquidity (a lower value of

ΔLN_RESPRD) due to decimalization makes it easier for institutional investors to enter and

therefore increases the level of institutional ownership

We then decompose ΔINSTIPCT by calculating the predicted value of ΔINSTIPCT (ΔINSTIPCT_HAT), which captures the variation in ΔINSTIPCT due to the variation in ΔLN_RESPRD, and the residual value of ΔINSTIPCT (RESIDUAL), which captures the variation

in ΔINSTIPCT not explained by the variation in ΔLN_RESPRD In the second step, we include both ΔINSTIPCT_HAT and RESDUAL and estimate the following model:

ΔLN_PATENTS i,t to t+2 = a + bΔINSTIPCT _HAT i,t-1 to t+1 + cRESIDUAL i,t-1 to t+1 + dΔLN_MV i,t-1

to t+1 + eΔRDTA i,t-1 to t+1 + fΔROA i,t-1 to t+1 + gΔPPETA i,t-1 to t+1 + hΔLEV i,t-1 to t+1 + iΔCAPEXTA

i,t-1 to t+1 + jΔHINDEX i,t-1 to t+1 +kΔHINDEX 2 i,t-1 to t+1 + lΔQ i,t-1 to i,t+1 + mΔKZINDEX i,t-1 to t+1 +IND j