the impact of venture capital investments on public firm stock performance

15
This article was downloaded by: [Washburn University] On: 31 October 2014, At: 10:37 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Journal of Behavioral Finance Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/hbhf20 The Impact of Venture Capital Investments on Public Firm Stock Performance Tim Loughran a & Sophie Shive a a University of Notre Dame Published online: 02 Dec 2011. To cite this article: Tim Loughran & Sophie Shive (2011) The Impact of Venture Capital Investments on Public Firm Stock Performance, Journal of Behavioral Finance, 12:4, 233-246, DOI: 10.1080/15427560.2011.620723 To link to this article: http://dx.doi.org/10.1080/15427560.2011.620723 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions

Upload: sophie

Post on 07-Mar-2017

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The Impact of Venture Capital Investments on Public Firm Stock Performance

This article was downloaded by: [Washburn University]On: 31 October 2014, At: 10:37Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Journal of Behavioral FinancePublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/hbhf20

The Impact of Venture Capital Investments on PublicFirm Stock PerformanceTim Loughran a & Sophie Shive aa University of Notre DamePublished online: 02 Dec 2011.

To cite this article: Tim Loughran & Sophie Shive (2011) The Impact of Venture Capital Investments on Public Firm StockPerformance, Journal of Behavioral Finance, 12:4, 233-246, DOI: 10.1080/15427560.2011.620723

To link to this article: http://dx.doi.org/10.1080/15427560.2011.620723

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Page 2: The Impact of Venture Capital Investments on Public Firm Stock Performance

THE JOURNAL OF BEHAVIORAL FINANCE, 12: 233–246, 2011Copyright C© The Institute of Behavioral FinanceISSN: 1542-7560 print / 1542-7579 onlineDOI: 10.1080/15427560.2011.620723

The Impact of Venture Capital Investments on PublicFirm Stock Performance

Tim Loughran and Sophie ShiveUniversity of Notre Dame

The aggregate amount of venture capital investments in nonpublicly traded firms since 1980 ismore than $390 billion. We test two economic hypotheses on the connection between venturecapital investment and subsequent firm performance. We find that lagged VC investments scaledby industry assets are negatively related to subsequent firm stock returns after adjusting forother factors. However, not all firms are equally impacted. We find that financially constrainedfirms suffer the most when new VC money pours into an industry. Firms receiving VC moneyare active in patent creation which appears to increase innovation pressures on establishedcompanies. It appears that the market is slow to incorporate the information contained in theventure capital investments.

Keywords: Venture capital, Financial constraints, Patents, Innovations

From 1980 to 2005, the aggregate amount of equity invest-ments made by venture capitalists into U.S. firms, accordingto VentureXpert, totaled $394.6 billion. Venture capitalistsplay an increasingly important role in the U.S. economy.Highly successful firms such as Apple Computer, Sun Mi-crosystems, Yahoo, eBay, and Google all had venture capitalfunding before going public. The venture capitalist not onlyprovides cash investment to young firms but also often serveson boards of directors and can provide critical support to al-low entrepreneurs to transform small, private start-ups intolarge publicly-traded firms.

Prior work has shown that venture capitalists have skillat responding to valuation signals provided by the market.For example, Lerner [1994a] suggests venture capitalists arequite good at selecting times when public market valuationsare especially high to take young firms public. Brav andGompers [1997] provide evidence that venture capital (VC)-backed initial public offerings have substantially higher long-run returns after an initial public offering (IPO) than non-VC backed IPOs. Gompers, Kovner, Lerner, and Scharfstein[2008] report that the most experienced VCs make the mostmoney and are most responsive to investment opportunitiessignaled by financial markets.

Address correspondence to Tim Loughran, Mendoza College of Busi-ness, University of Notre Dame, Notre Dame, IN 46556-5646. E-mail:[email protected]

Little attention, however, has been paid to how the pub-lic equity markets respond to private VC investments. It ispossible that VC investments into private firms serve as anendorsement of an industry and increase investor optimismabout the industry’s prospects. This would increase the valueof established public companies. In this case, we would ex-pect VC investment to forecast higher firm returns in a par-ticular industry, at least in the short term. This appears to bea commonly held view. A December 27, 2006, article in theWall Street Journal states: “Venture capitalists, who fueledthe previous Internet bubble, are pumping money into thenew crop of Web startups.”1

Alternatively, it might be that new VC money signal trou-ble for existing firms in the form of increased pressure. VCmoney could lead to innovations by nonpublic firms in theform of new technologies or patents which render existinginfrastructure obsolete. As new private investments pour intoilliquid and nonpublicly traded firms bent on achieving inno-vative techniques and patents, the returns of companies thatare already public could suffer as a result.

There is evidence in the prior empirical research in relatedareas of finance suggests that venture capital investment in anindustry might be related to public firm performance. Cooper,Gulen, and Schill [2008] find that the higher is an exchangelisted firm’s growth in total assets; the lower are the firm’ssubsequent returns. The three authors state that their assetgrowth effect is most consistent with over extrapolation ofpast gains to growth by financial market participants.

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 3: The Impact of Venture Capital Investments on Public Firm Stock Performance

234 LOUGHRAN AND SHIVE

Hoberg and Phillips [2010] show that a strong negativeeffect of investment, valuation, and new financing on returnsduring boom and bust cycles exists, though it is concentratedin competitive product markets. Their work provides evi-dence that investors and firms in competitive industries sys-tematically underestimate the negative externality that theirinvestment creates. Hoberg and Phillips [2010] show that thiscan be a rational response because in competitive industries,firm-specific information is hard to gather. Thus, firms aremore likely to rely on industry-wide signals and to herd intheir investment decisions.

One should expect VC funding to have a stronger per dol-lar effect on subsequent returns than total investment or assetgrowth in public firms. For example, Kortum and Lerner[2000] find that a dollar of venture capital has three timesthe impact in stimulating patent creation than a dollar ofregular corporate research and development. As individualfirms issue equity or debt to build new factories, the stock re-turns of firms in that industry might suffer from the increasedcompetition from the new investment.

However, venture capital money usually focuses on newinnovations which have the potential to destroy or severelydamage the existing firms in the industry (think of the creativedestruction explanation of Schumpeter [1942]). As Gompersand Lerner [2001], p. 78) state: “Unless a venture firm seesthe potential for patents or some other form of protectedintellectual property, it isn’t likely to invest.” This helps toexamine the wide variation in VC funding across industries.

Gompers and Lerner [2001] find that industries with alower propensity for major innovations see a paucity of pub-licly traded VC-backed companies. They believe that theventure capitalist mission “is to capitalize on revolutionarychanges in an industry” (p. 70). Thus, given the focus to-wards innovations, it is conceivable for millions of venturecapital dollars invested to lower the values of existing firmsby billions of dollars.

We test two economic hypotheses on the market’s reac-tion to VC investments. Our first hypothesis relates to howthe stock market eventually reacts to VC flows into an in-dustry. Are the flows a positive, negative, or neutral signalabout subsequent stock returns? The second hypothesis dealswith whether financially constrained firms are more or lessadversely affected by VC investments than nonconstrainedfirms. It is possible that the more experienced, more glam-orous firms in an industry are most damaged by the newcapital investments. These firms with high growth prospectsmight be damaged most with the new flows of innovativecapital. Or it might be that younger, more financially con-strained public companies are the ones that bear the brunt ofthe effect of VC capital flows.

Our paper makes several contributions. First, in a samplefrom VentureXpert of more than 73,400 rounds of venturecapital investments during 1980–2005, we show that VCinvestment has no perceptible endorsement effect and fore-casts sharply lower returns for the existing public firms in

the same industry. We find evidence that higher levels ofquarterly venture capital investments scaled by Fama-Frenchindustry assets are associated with significantly lower subse-quent quarterly firm returns in a particular industry. That is, ifa high amount of VC dollars are invested in the telecommu-nications industry in a particular quarter, telecommunicationfirm stock returns will be lower in the next quarter. Thisis even after adjusting for factors such as industry-wide in-vestment growth, prior firm year return, asset growth of thefirm, and book-to-market. Our results are significant, both in-cluding and excluding the Internet bubble period, for small,medium, and large firms and for early, middle, and late roundsof VC financing.

Our results are also economically significant. A one stan-dard deviation increase in change in VC dollars/assets isassociated with a 4% decline in stock returns on an annual-ized basis. This is after controlling for other factors knownto explain stock return patterns.

We hypothesize that firms saddled by financial constraintsshould be less able to adapt to change in their industry thanfinancially healthy firms. More financially constrained com-panies might not have the funds or the experience to invest innew technologies or infrastructure to properly compete. Wespecifically use the recent Hadlock and Pierce [2010] sug-gested measure for our financial constraint definition. Oursecond contribution is to show that firms that are more finan-cially constrained suffer worse performance when venturecapital flows into their industry. Specifically, less experi-enced firms with lower payout yield are hurt more by VCinvestment than are financially healthy firms. We believe thatVC money leads to greater innovation and hence greaterpressures on publicly traded firms with less knowhow andexperience.

Consistent with our assertion that venture capital leads toincreased innovation, we find enormous eventual patent is-suance for the sample of private firms receiving VC money.More than 111,000 patents were issued to our VC-backedsample within 10 years after their first VC investment. VC-backed firms in our sample, such as Micron Technology, Mi-crosoft, Sun Microsystems, and Genentech, each eventuallyhad more than 1,500 U.S. patents. We find that the averagenumber of issued U.S. patents in the 10 years following afirm’s first VC investment was 10.6. The median value isfour patents for the universe of VC-backed private firms withat least one patent.

Along the lines of Gompers and Lerner [2001], we findthe Fama-French industries with the highest aggregate VCinvestments are the same industries with the largest patentissuance. Patent creation, venture capital investments, andinnovations within an industry seem to go hand-in-hand.Though it is difficult to perfectly document the direct linkbetween patent issuance by VC-backed firms and innovationwithin an industry, we have some indirect evidence.

Lerner [1994b] lists the 13 most important biotech-nology patents. Lerner created the list by interviewing a

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 4: The Impact of Venture Capital Investments on Public Firm Stock Performance

IMPACT OF VENTURE CAPITAL INVESTMENTS 235

number of intellectual property attorneys for their sugges-tions of the most economically important patents. The listexcludes patents awarded to universities and patents issuedafter September 1992. Interestingly, all 13 of the patents onhis list were issued to firms in our VentureXpert sample.

We find that the effect of VC dollars on returns lasts abouta quarter and is permanent. Why do the financial marketstake this long to incorporate the news of the venture capitalinvestments into industry stock prices? Even though the ven-ture capitalists have disbursed the money, uncertainty stillexists in exactly how the managers will execute the projects.As noted in Hong, Torous, and Valkanov [2007], informationmay flow slowly across markets possibly due to the limitedinformation-processing capacity of investors. Thus, it is notsurprising that information of VC investments into privatelyheld firms might slowly seep into the values of publicly tradedindustry companies.

VC investments often help create innovations within anindustry. The industries with the most capital flows are the in-dustries with the highest patent issuance by VC-backed firms.Our paper sheds light on the stock market’s delayed negativereaction to flows of venture capital investments within anindustry.

The paper proceeds as follows. The first section presentsthe sample and the second section presents the estimationprocedure. The third section reports the empirical results andthe fourth section concludes.

SAMPLE

Our source for the venture capital data is VentureXpert(Thomson Financial Economics). We use data on all ven-ture capital investments between 1980 and 2005.2 This is theactual equity investment made by the venture capitalist intoeach company. The VentureXpert database classifies indus-tries in its own way, so we re-code them according to theFama and French [1997] 48 industry classifications by com-paring individual VentureXpert industry codes and StandardIndustrial Classification (SIC) Codes. Appendix A reportshow the VentureXpert classifications were categorized intoFama and French industries.

VC Funding Classifications and Patents

The main classifications of funding available in VentureXpertare early stage (seed, startup, and other early stage), expan-sion, later stage, and other final stage (acquisition, specialsituation, and VC partnership). We exclude the other finalstage category because it is not clear that funding of suchenterprises represents the creation/expansion of new enter-prises by the venture capitalist. Including this other stagecategory, however, does not change the results. In the paper,we include all VC funding directed at the following stages:seed, startup, other early stage, expansion, and later stage.3

One obvious source of innovation is the number of issuedpatents by VC-backed firms. We programmatically accessthe U.S. patent office to obtain the number of issued patentsfor each firm receiving VC cash.4 Each of our sample Ventur-eXpert firms which had at least one round of VC funding aresearched for both lifetime issued patents and issued patentsin the first 10 years following the initial VC round. One majoradvantage in using the patent office to gather patent issuancedata is that firms are not required to be on Center for Re-search in Security Prices (CRSP) or Compustat to enter ourpatent universe. There are many sample nonexchange listedfirms which receive VC money, issue a few patents, and theneither get acquired by an outside firm or go bankrupt.

Time Series of Venture Capital Investments

Figure 1 presents the time series of quarterly venture capi-tal investments compared to the quarterly level of Nasdaq.Figure 2 reports the relation between Nasdaq levels and VCinvestments scaled by industry total assets. Scaled VC invest-ments experienced a decrease from 1983–1994, but then boththey and Nasdaq saw a tremendous spike in the Internet bub-ble period of 1998–2000. These two figures show aggregateventure capital funding patterns.

Table 1 provides descriptive statistics of VC investmentrounds and aggregate number of issued patents by Famaand French [1997] industry classifications. Over the sampleperiod, there was an aggregate $394.6 billion in investmentsby venture capitalists made in 73,401 separate rounds. Thus,there were more than 73,000 separate investments by venturecapitalists in nonpublicly traded firms during the period. Wefocus on the dollar amount invested scaled by industry assets,not on the number of rounds that venture capitalists engagedin. Thus, in a quarter for a particular industry, five roundsof $1 million each will count the same as one round of $5million.

Also, we focus on the level of investment, not the changein investment, as the critical determinant of VC impact. Thelevel of investment in a quarter should be a good measure ofthe overall impact that the VCs will have on an industry. Ifa billion dollars flows into the private firms in an industry,there could be substantial innovation in the industry.

Not all industries are equally likely to attract financial in-terest from venture capitalists. More than 83% of the totalventure capital investments went to only five industries: busi-ness services (37.2%), telecommunications (24.4%), phar-maceutical products (9.2%), computers (6.7%), and chipsand electronic equipment (5.6%).

The last column in Table 1 reports the aggregate numberof patents issued by firms receiving VC money categorizedby Fama and French industries. We count only the number ofpatents issued within the first 10 years after the initial roundof VC investment. Not surprisingly, the industries with thehighest flows of VC money are the same industries where thegreatest patent creation is centered. This is an important link:

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 5: The Impact of Venture Capital Investments on Public Firm Stock Performance

236 LOUGHRAN AND SHIVE

FIGURE 1 Quarterly level of Nasdaq and venture capital investments, 1980–2005.

the higher the venture capital dollars flowing into an industry,the higher subsequent patent production. This provides abasis for our hypothesis that higher investment by venturecapital leads to increased subsequent innovation within anindustry. In the aggregate, firms in the VentureXpert databasecreated 111,500 U.S. patents within 10 years of receiving thefirst VC investment.

Figure 3 shows the time series trend in venture capitalinvestments as a percent of total VC funding. In the early1980s, the computer industry attracted the most funding, ac-counting for about 30% of all investments in the first quarterof 1983. In the late 1980s, business services surpassed com-puters as the most heavily funded industry. During the bubbleperiod, business services accounted for more than half of VC

FIGURE 2 Quarterly level of Nasdaq and venture capital investments scaled by industry total assets, 1980–2005.

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 6: The Impact of Venture Capital Investments on Public Firm Stock Performance

IMPACT OF VENTURE CAPITAL INVESTMENTS 237

TABLE 1Aggregate Venture Capital Investments and Patents

Categorized by Fama and French Industries,1980–2005

FF IndustryTotal in $Billions % Total $ Total Rounds

Number ofPatents

1 Agric 1.7 0.4% 699 7132 Food 2.2 0.6% 674 2303 Soda 0.1 0.0% 63 204 Beer 0.2 0.0% 69 97 Fun 3.3 0.8% 818 1,3138 Books 0.4 0.1% 233 4939 Hshld 1.7 0.4% 636 841

10 Clths 0.4 0.1% 259 11211 Hlth 6.2 1.6% 1,600 1,29012 MedEq 14.9 3.8% 3,553 8,92113 Drugs 36.2 9.2% 6,255 10,38014 Chems 1.3 0.3% 477 1,39117 BldMt 2.0 0.5% 812 1,56818 Cnstr 0.8 0.2% 360 45321 Mach 3.7 1.0% 1,969 3,75922 ElcEq 0.8 0.2% 195 58328 Mines 0.5 0.1% 31 2029 Coal 0.1 0.0% 28 5530 Oil 1.6 0.4% 334 51031 Util 1.5 0.4% 398 1,00732 Telcm 96.2 24.4% 11,926 19,29033 PerSv 1.2 0.3% 201 27634 BusSv 146.9 37.2% 26,993 26,13435 Comps 26.3 6.7% 5,562 10,46836 Chips 22.3 5.6% 4,396 14,69437 LabEq 5.5 1.4% 1,258 3,67238 Paper 0.6 0.2% 209 50039 Boxes 3.8 1.0% 888 94242 Rtail 3.2 0.8% 840 46843 Meals 2.0 0.5% 525 16544 Banks 3.1 0.8% 581 60845 Insur 1.0 0.2% 148 14446 RlEst 0.6 0.2% 183 24047 Fin 2.3 0.6% 228 281Total 394.6 100.0% 73,401 111,550

Note. Our data source for the venture capital data is VentureXpert fromThomson Financial Economics. All firms are classified into Fama and French[1997] industries. All investments are in one of five stages: seed, startup,other early stage, expansion, or later stage. Thirty-four different industrieshave at least one round of VC investments. The last column lists the aggregatenumber of issued U.S. patents to VentureXpert firms in the first 10 yearsafter the first round of VC funding.

dollars, while telecommunications received almost 40% ofall dollars.

Summary Statistics

Table 2 reports the summary statistics for our sample. Al-though only 34 Fama-French industries have VC investmentat some point in the sample period, we use firms in all 48 in-dustries in our analysis. Since we are using lagged quarterlyVC investments, the analysis starts at the end of first-quarter

TABLE 2Sample Summary Statistics, 1980–2005

Mean Median St. Deviation Min Max

Quarterly VCDollars/Assets

0.09% 0.01% 0.27% 0.00% 5.30%

Quarterly return 4.62% 2.37% 27.75% −98.86% 1, 411.11%Payout Yield 2.85% 1.24% 4.25% 0.00% 24.67%Asset Growth 15.37% 8.00% 35.02% −40.71% 214.06%Firm New

Finance5.47% 0.11% 21.13% −24.53% 135.77%

Industry NewFinance

1.61% 1.08% 3.42% −19.86% 42.10%

IndustryInvestment

1.79% 1.51% 4.74% −36.50% 53.63%

Herfindahl Index 0.26 0.20 0.21 0.02 1Log (Book/

Market)−0.58 −0.49 0.88 −6.91 3.36

Log (MV) 5.13 5.00 2.08 0.98 10.33Prior Year

Return22.92% 10.00% 79.77% −98.03% 5, 868.00%

Age 18.11 13.42 14.74 3.50 70.50Log (Assets) 5.30 5.15 2.07 1.21 10.50

Note. VC dollars are the lagged aggregate quarterly VC investments,in millions of dollars, within a particular Fama-French industry. VC Dol-lars/Assets are the lagged aggregate quarterly VC investments, divided bytotal assets (Compustat item 6) within a particular Fama-French industry.Payout Yield is dividends plus repurchases divided by share price, (Com-pustat items [26 + (115/54)] / Price. Asset Growth is the lagged annualchange in assets (Compustat annual item 6). Firm New Finance is Com-pustat items (108-115+111-114)/6. Industry New Finance is the industrymean of Compustat items (108-115+111-114)/6. Industry Investment is thelagged change in industry-level capital expenditure scaled by property, plantand equipment. The Herfindahl index is computed by dividing the sum of thesquares of the sales (Compustat item 12) in each 3-digit SIC code industryby the squared total sales of that industry. Log (Book/Market) is the laggedfirm-level natural log of the book-to-market ratio. Log (MV) is the laggedfirm-level natural log of market equity. Prior Year Return is the last 12 monthreturn of the firm. Age is the number of years the firm has been listed onCRSP. Log (Assets) is the natural log of firm total assets. Payout yield, assetgrowth, firm new finance, log(Book/Market), log(MV), Age, and log(Assets)are winsorized at the 1/99 percentile level. There are 103 quarters in theanalysis.

1980. When VC dollars are scaled by industry total assets,the mean value over the 103 quarters is 0.09%.

We compute quarterly returns using the monthly returnsprovided by CRSP. Only operating companies as defined byCRSP enter our sample. We chose quarterly time intervals forour stock return analysis to allow a long enough time periodfor the market to incorporate the news of the VC investment.The average quarterly stock return including distributions is4.62%, with a median of 2.37%.

Notice that there is substantial skewness in the quarterlystock returns. The maximum firm three month return is over1,400%. Because of the stock return skewness, in subsequentregressions we will use the natural log of one plus the quar-terly return as the dependent variable.

There is the possibility that venture capital investment iscorrelated with other factors or changes that are present in

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 7: The Impact of Venture Capital Investments on Public Firm Stock Performance

238 LOUGHRAN AND SHIVE

0.1

.2.3

.4.5

Pro

port

ion

of T

otal

Fun

ding

1980 1985 1990 1995 2000 2005Year

BusSv TelcmDrugs CompsChips

FIGURE 3 Venture capital investments as a percent of total funding for the five most-funded industries, 1980–2005. (Color figure available online).

the industry. That is, growth firms (low book-to-market ratio)may be a common characteristic of industry firms wherethe VC dollars flow. Fama and French [1992] have shownthat growth firms historically have experienced lower stockreturns than value firms.

Also, perhaps overall investment is increasing in an in-dustry and our documented effect is merely a reflection ofexpansion or growth among publicly traded firms in the in-dustry. Thus, in the regressions, we will control for firmcharacteristics in addition to investment or growth amongthe publicly traded firms to isolate a unique effect of venturecapital investment.

In all cases, we only use data available before the startof each quarter. All Compustat items use the most recentlyavailable data for each firm. We eliminate firms with stockprices below $1, and we do not use the first two years of afirm’s existence on Compustat to avoid potential backfillingbiases (Fama and French [1993]).

We also include payout yield, firm asset growth, firmnew finance, industry new finance, industry investment, theHerfindahl index of the 3-digit SIC code industry of the firm,natural log of the book-to-market ratio of the firm, natural logof market value, and the prior year return of the firm, as wellas quarter and industry dummies. Our results are strongerwithout the dummies being included in the regressions. Pay-out yield, asset growth, firm new finance, log(book/market),and log(market value) are all winsorized at the 1/99 percentilelevel.

The first of our controls is payout yield. Boudoukh,Michaely, Richardson, and Roberts [2007] show that payoutyield is a strong predictor of returns. We compute total payout

yield as the total dollars spent on dividends and repurchasesduring the quarter divided by share price. Specifically, pay-out yield is Compustat annual items [26 + (115 / 54)]/Price,where item 26 is dividends per share; item 115 is purchase ofcommon and preferred stock; item 54 is number of commonshares; and Price is the end-of-prior-quarter share price fromCRSP. The median payout yield in our sample is 1.24%.

Following Cooper, Gulen, and Schill [2008], we controlfor the asset growth of each stock. The three authors doc-ument a strong negative relationship between a firm’s assetgrowth and its subsequent stock returns. For each firm, wecompute the annual percentage change in assets of the latestfiscal year from the one before. The median asset growth forfirms in our sample is 8.00%.

Following Hoberg and Phillips [2010], we include firmnew finance and industry investment to control for the levelof public investment into an industry while we examine theimpact of VC investments. For each firm, firm new financeis the sum of net equity and debt issuance divided by thevalue of assets (Compustat items (108–115 + 111–114)/6).Industry new finance is defined as the summed total newfinancing by all firms in an industry divided by the aggregatetotal industry assets.

The Herfindahl index is computed by dividing the sumof the squares of the sales (Compustat annual item 12) in a3-digit SIC code industry by the squared total sales of thatindustry. A high value for the Herfindahl index indicates a lowlevel of competition in an industry. Hou and Robinson [2006]show that concentrated or high Herfindahl industries havelower returns. The mean Herfindahl index is 0.26 comparedto a median value of 0.20.

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 8: The Impact of Venture Capital Investments on Public Firm Stock Performance

IMPACT OF VENTURE CAPITAL INVESTMENTS 239

It might be that VC dollars are attracted to high growthindustries. Prior evidence has shown that firms with lowbook-to-market ratios (i.e., growth firms) experience lowersubsequent returns than value firms. Previous studies havealso shown that the market value of equity helps explainrealized stock returns. Thus, we include both the natural logof the book-to-market ratio and the market value of equityof the firm at the time of the VC investment to show that ourfindings are not driven by large capitalization, growth firmsexperiencing low subsequent stock returns. Stock prices andnumber of shares are from CRSP and are from the end ofeach firm’s fiscal year.5

Issued U.S. Patents

To provide a link between the VC investments and innova-tion, Table 3 lists the top 15 firms in terms of lifetime issuedU.S. patents. The list includes some of the most importantand innovative U.S. firms of the last 30 years. Also reportedin the table are the first venture capital investment into eachfirm according to VentureXpert and its IPO date. For exam-ple, Microsoft had its first round of VC money on 19810901,about 4.5 years before its initial public offering.

Microsoft, like the other firms listed in the table, was ini-tially slow at obtaining patent issuance. The last column ofTable 3 reports that Microsoft had only 10 issued patents inthe first 10 years after getting its first VC money. One ofthe more successful firms at issuing patents in a tradition-ally nontechnical industry (Entertainment) is Callaway Golf.Most of Callaway’s patents involve putter head, club, or golfball designs. This table provides evidence of eventual inno-vation (as proxied by patent issuance) by firms backed by VCmoney. Given the high production of patents by VC-backed

firms, innovative pressures must have increased for the estab-lished publicly traded companies in the same industry afterthe venture capital investments were made.

We do not use the patent information in the subsequentstock return regressions due to a clear look-ahead bias. Theinformation of eventual patent issuance was not availableto investors at the time of the initial venture capital invest-ment. Yes, today we know that Micron Technology has hadenormous patent productivity. However, in 1983, there wasstill some uncertainty as to the firm’s eventual success. Thepurpose of Table 3 is to illustrate that some of the sampleof firms receiving venture capital investments went on tobecome exceptionally innovative forces in their industries.

ESTIMATION

We estimate our models at the firm level using panel re-gressions with both quarterly and industry dummies. We usefirm-level data in order to test hypotheses about the causeof the relation between VC funding and returns, but the re-sults are strong if we use industry-level regressions witheither value-weighted or equal weighted returns along withexplanatory variables. Fixed-effects regressions have the ad-vantage of allowing industry effects.As many of the industryfixed effects are statistically significant, we think they shouldbe controlled for. Our results, however, do not depend on theirinclusion.

We also cluster the standard errors by both quarter and firmusing the methodology proposed by Thompson [2009] andCameron, Gelbach, and Miller [2006]. As Thompson [2009]explains, clustering standard errors by just one variable or not

TABLE 3Top 15 Firms in Terms of Lifetime Issued U.S. Patents

Rank Firm Name First VC Round Date IPO Date Lifetime Patents Number of Patents Within First 10 Years

1 Micron Technology 19830301 19840601 17,516 2492 Microsoft Corporation 19810901 19860313 11,670 103 Sun Microsystems 19820401 19860304 7,153 684 LSI Logic Corporation 19810101 19830513 3,906 225 Cisco Systems 19880101 19900216 3,659 116 Xilinx 19840501 19900612 2,077 287 Compaq Computer Corporation 19820301 19831209 2,002 648 Genentech 19800101 19801014 1,655 709 Altera Corporation 19830601 19880330 1,654 26

10 3Com Corporation 19810201 19840321 1,395 411 Lam Research Corporation 19801001 19840504 1,092 712 VLSI Technology 19801201 19830224 1,088 1813 Cirrus Logic, Inc. 19840501 19890608 1,041 1814 Callaway Golf Company 19890101 19920227 909 8115 Sandisk Corporation 19880901 19951107 904 32

Note. Using data on issued patents from the U.S. Patent Office, the table lists the top 15 patent producing firms which received venture capital fundingaccording to VentureXpert. The third column lists the date of the first venture capital financing round. The last column reports the number of issued patents inthe first 10 years following the initial venture capital round of financing. For the firms with patents, the average lifetime number of patents is 23.7 while themedian number is 5. For the patents issued within first 10 years, the average number of patents per firm is 10.6 while the median is 4.

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 9: The Impact of Venture Capital Investments on Public Firm Stock Performance

240 LOUGHRAN AND SHIVE

clustering them at all can bias the standard errors when bothcross-sectional and time series effects are present. Clusteredstandard errors are always heteroskedasticity-consistent.

EMPIRICAL RESULTS

Main Results

We want to test whether VC funding in a particular indus-try is related to industry firm returns in subsequent quarters,controlling for variables that have been shown to have someexplanatory power for quarterly returns. The main resultsare presented in Table 4. The control variables are the firm’spayout yield, the asset growth of the stock, the net issuanceof debt and equity by public firms, industry new finance, in-dustry investment, the Herfindahl index, the book-to-marketratio, market value of equity, and firm’s prior year stock re-turn. In the fixed-effect regressions, we also include industryand quarter dummies. In all the Table 4 regressions, the de-pendent variable for each firm is the natural log of one plusthe quarterly stock return.

Table 4 uses venture capital dollars scaled by total industryassets as the independent variable of interest. We presentdifferent specifications ranging from this variable alone to apurely predictive model using control variables. Scaling theVC investments may provide a better gauge of the effect ofthe funding amount. That is, $1 billion of investments in avery large industry would likely not have the same impact asa $1 billion investment in a relatively small industry.6

The regression specification is as follows:

Log(1 + Return)i,t = a0 + a1VC Dollars/Assetsi,t−1

+ a2Payout Yieldi,t−1 + a3Asset Growthi,t−1

+ a4Firm New Financei,t−1

+ a5Industry New Financei,t−1

+ a6Industry Investmenti,t−1 + a7Herf indahli,t−1

+ a8Log(Book/Market)i,t−1 + a9Log(MV )i,t−1

+ a10Prior Y ear Returni,t−1

+ Industry Dummies + Quarter Dummies + ei,t

The t-statistics generated from the double clustered errors(i.e., both firm and quarter) are in parentheses under thecoefficients. In the first four regressions, the sample size is269,257 firm-quarter observations.

The coefficient on lagged VC dollars divided by industryassets is consistently negative. The t-statistics for VC dol-lars/assets in the first four columns of Table 4 range from−2.27 and −2.43. The coefficients are economically signifi-cant as well. If we use the full model in column 4 of Table 4,a one standard deviation change in VC dollars/assets is as-sociated with about a 1.00% decrease in quarterly return. Incomparison, a one standard deviation change in asset growth

TABLE 4Quarterly Stock Return Regressions, 1980–2005

(1) (2) (3) (4)

VC Dollars/Assets

−5.78 −4.11 −3.72(−2.31) (−2.43) (−2.27)

Payout Yield 0.10 0.10(4.15) (4.15)

Asset Growth −0.03 −0.03(−6.60) (−6.55)

Firm NewFinance

0.02 0.02(1.56) (1.59)

Industry NewFinance

−0.03 −0.02(−0.73) (−0.58)

IndustryInvestment

−0.07 −0.04(−2.05) (−1.56)

Herfindahl 0.00 0.00(0.54) (0.36)

Log(Book/Market) 0.02 0.02(6.15) (6.19)

Log(MV) 0.37 0.38(2.65) (2.66)

Prior YearReturn

0.00 0.00(0.76) (0.81)

N (firm-quarters)

269,257 269,257 269,257 269,257

Number of firms 8,191 8,191 8,191 8,191Quarter and

IndustryDummies

No Yes Yes Yes

R2 0.00 0.18 0.18 0.19

Note. The dependent variable is the natural log of one plus quarterlyfirm return. VC Dollars/Assets are the lagged aggregate quarterly VC in-vestments, divided by total assets within a particular Fama-French industry.See Appendix B for other variable definitions. All Compustat items use themost recently available data for each firm. The t-statistics are in parentheses.Errors are clustered by firm and by quarter.

Log(1 + Return)i,t = a0 + a1VC Dollars/Assetsi,t-1 + a2PayoutYieldi,t-1 + a3Asset Growthi,t-1 + a4Firm New Financei,t-1 + a5IndustryNew Financei,t-1 + a6Industry Investmenti,t-1 +a7Herfindahli,t-1 +a8Log(Book/Market)i,t-1 + a9Log(MV)i,t-1 + a10 Prior Year Returni,t-1 +Industry Dummies + Quarter Dummies + ei,t

is associated with a 1.05% decrease in quarterly return. Onan annual basis, both VC dollars/assets and the change inasset growth have an impact of about a 4% decline in firmstock returns given a one standard deviation change in thevariables.

The negative coefficients on VC dollars/industry assetsare consistent with a hypothesis that there is slow diffusionin the stock market of the information content of the venturecapital investment. Our result is similar to the results in Hong,Torous, and Valkanov [2007], who find that the stock marketreacts with a delay to new information embedded in industryreturns. Information on fundamentals appears to be diffusedonly gradually across markets.

We do not investigate lags smaller than a quarter be-cause of the imprecise nature of the dates available fromVentureXpert. Kaplan, Sensoy, and Stromberg [2002] reportthat VentureXpert’s funding date is within one month of the

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 10: The Impact of Venture Capital Investments on Public Firm Stock Performance

IMPACT OF VENTURE CAPITAL INVESTMENTS 241

actual funding date in only 107 of 124 financings. Thus,while we are fairly sure about the quarter that the fundingtakes place in, we do not want to rely on monthly or dailydata to provide an event study on the announcement dates.There are no significant additional lags when we add addi-tional lags of VC dollars divided by industry assets to thecolumn 4 regression.

Other independent variables that are often significant inthe Table 4 regressions include payout yield, asset growth,log(book/market) and log(market value). The sign and sig-nificance of the coefficient on payout yield is consistent withthe empirical evidence of Boudoukh, Michaely, Richard-son, and Roberts [2007]. The negative values on the as-set growth variable are as would be predicted by Cooper,Gulen, and Schill [2008]. The positive and significant co-efficient on log(book/market) for stock returns implies thatwhen the book-to-market ratio is high (i.e., tilted towardvalue), subsequent firm returns are higher, all else beingequal.

The industry Herfindahl index is never a significant pre-dictor of quarterly firm returns. This differs from the findingsin Hou and Robinson [2006], but they use a different time

series, from 1963–2001. They also average the Herfindahlindex over the past three years.

Subsample Robustness Tests

This subsection reports the results by various partitions ofthe data. The three groupings are on the basis of bubble/non-bubble time periods, asset size of the firm, and stage of VCfinancing. Might the observations during the bubble periodof 1998–2000 be driving the relation between VC dollars andstock returns? We saw in Figures 1 and 2 a spike in both thelevel of Nasdaq and the level of venture capital investmentsduring the Internet bubble period. To test this hypothesis,we divide the sample into bubble and nonbubble periodsand re-run the tests in columns 1 and 2 of Table 5 for eachsubsample.

VC dollars scaled by industry assets are again the explana-tory variable of interest. Both industry and quarter dummiesare included in all eight regressions. The t-statistic is slightlylower and the coefficient slightly smaller in the bubble pe-riod than in the nonbubble period. The VC investment coef-ficient is statistically significant in the first two regressions in

TABLE 5Robustness – Subsets of the Data

Bubble Non-Bubble Small Medium Large Early Stage Expansion Late Stage(1) (2) (3) (4) (5) (6) (7) (8)

VC Dollars/Assets −4.86 −5.51 −3.81 −3.44 −2.53 −11.50 −5.85 −15.80(−2.32) (−3.04) (−2.77) (−2.87) (−2.03) (−2.23) (−2.05) (−2.24)

Payout Yield 0.10 0.10 0.15 0.13 0.06 0.10 0.10 0.10(1.40) (3.91) (4.09) (4.50) (2.19) (4.13) (4.15) (4.18)

Asset Growth −0.05 −0.03 −0.03 −0.04 −0.03 −0.03 −0.03 −0.03(−5.20) (−5.14) (−6.47) (−5.99) (−4.23) (−6.54) (−6.56) (−6.59)

Firm New Finance 0.02 0.02 0.03 0.03 −0.00 0.02 0.02 0.02(0.61) (2.17) (2.97) (2.88) (−0.28) (1.57) (1.60) (1.60)

Industry New Finance −0.43 −0.00 −0.02 −0.01 −0.01 −0.02 −0.02 −0.02(−2.34) (−0.10) (−0.43) (−0.31) (−0.24) (−0.48) (−0.66) (−0.63)

Industry Investment −0.55 −0.03 −0.06 −0.04 −0.03 −0.05 −0.05 −0.06(−4.32) (−1.01) (−1.82) (−1.35) (−0.83) (−1.59) (−1.68) (−1.88)

Herfindahl 0.00 0.00 −0.00 0.00 0.01 0.00 0.00 0.00(0.05) (0.32) (−0.15) (0.41) (0.99) (0.49) (0.39) (0.26)

Log(Book/Market) 0.02 0.02 0.02 0.01 0.01 0.02 0.02 0.02(1.18) (6.98) (7.67) (3.33) (1.87) (6.19) (6.19) (6.17)

Log(MV) 1.05 0.27 −0.11 0.29 0.06 0.38 0.38 0.37(1.36) (2.18) (−0.33) (1.03) (0.41) (2.66) (2.66) (2.66)

Prior Year Return −0.00 0.00 −0.00 0.01 0.02 0.00 0.00 0.00(−0.54) (1.43) (−0.10) (2.21) (2.41) (0.78) (0.82) (0.77)

N (firm-quarters) 35,439 233,818 82,599 94,218 92,440 269,257 269,257 269,257Number of firms 3,919 7,958 4,089 4,119 2,524 8,191 8,191 8,191R2 0.15 0.20 0.19 0.21 0.23 0.19 0.19 0.19

Note. The dependent variable is the natural log of one plus quarterly firm return. VC Dollars/Assets are the lagged aggregate quarterly VC investments,divided by total assets (Compustat item 6) within a particular Fama-French industry. See Appendix B for other variable definitions. All Compustat itemsuse the most recently available data for each firm. The t-statistics are in parentheses. Errors are clustered by firm and by quarter. The bubble period is years1998–2000. The non-bubble period is years 1980–1997 and 2001–2005. On a yearly basis, sample firms are placed into total asset size terciles. All eightregressions include quarter and industry dummy variables.

Log(1 + Return)i,t = a0 + a1 VC Dollars/Assetsi,t-1 + a2 Payout Yieldi,t-1 + a3 Asset Growthi,t-1 + a4 Firm New Financei,t-1 + a5 Industry New Financei,t-1

+ a6 Industry Investmenti,t-1 +a7 Herfindahli,t-1 + a8 Log(Book/Market)i,t-1 + a9 Log(MV)i,t-1 + a10 Prior Year Returni,t-1 + Industry Dummies + QuarterDummies + ei,t

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 11: The Impact of Venture Capital Investments on Public Firm Stock Performance

242 LOUGHRAN AND SHIVE

Table 5. Our empirical results are thus not being driven by theunusual return patterns of the Internet bubble time period. Infact, our results are slightly stronger in the nonbubble period.

It could also be that our results are driven by only firmswith low levels of total assets. We replicate the analysis incolumns 3, 4, and 5 of Table 5 for small, medium, and largefirms in terms of total assets (Compustat item 6). On a yearlybasis, sample firms are placed into total assets size terciles.The results show that the VC dollars/assets coefficient dif-fers very little across firm size groupings It is slightly largerfor medium and small firms, and the t-statistic is slightlylarger as well, but the effect remains even for the largestfirms.

Finally, are the results present across different stages offunding? This study uses total VC funding dollars, but Ven-tureXpert allows us to break funding down by stage. Acrossthe three stages of venture capital financing (early stage,expansion, and late stage), the coefficients on lagged VCdollars/assets are negative and statistically significant (witht-statistics between −2.05 and −2.24).

To ensure that one industry or an omitted industry-levelvariable is not driving the results, in Table 6 we test therelation using individual industry-level ordinary least squarestests on the five largest industries in terms of venture capitalfunding. These five industries make up 83.1% of the VC

funding in our sample period. These industries are BusinessServices, Telecom, Drugs, Computers, and Chips.

Four of the five industries have negative and significantcoefficients on VC funding/industry assets.7 The BusinessServices industry has a negative coefficient that is significantat only the 9% level. It should also be noted that BusinessServices is the only industry in which firm new finance hasan insignificant coefficient value.

Financial Constraints and the Ability ofCompete/Innovate

We have hypothesized that the effect of VC dollars/assets isdue to the increased pressure new innovative capital bringsto firms in an industry. To test some implications of thisstory, we need measures of a publicly traded firm’s ability tocompete against new entrants. The financial constraints facedby the firm should be a good measure of a firm’s flexibilityand ability to compete. We use a firm’s payout yield, age, andtotal assets as measures of the ability to withstand increasedinnovation brought on by the venture capital dollars. Ageis the number of years a firm has been listed on CRSP justbefore the particular calendar quarter while log(assets) is thelog of total assets (Compustat item 6).

TABLE 6OLS Regression Results by Top Fama-French Industries

BusSv Telcm Drugs Comps Chips

VC Dollars/Assets −5.00 −24.32 −74.99 −31.79 −57.91(−1.69) (−3.15) (−4.02) (−2.02) (−2.74)

Payout Yield 0.23 0.16 0.39 0.16 0.05(4.39) (1.78) (1.86) (1.53) (0.45)

Asset Growth −0.05 −0.05 −0.04 −0.02 −0.04(−3.38) (−3.44) (−3.12) (−1.37) (−2.19)

Firm New Finance 0.02 0.03 0.07 0.04 0.05(1.29) (2.41) (3.43) (2.67) (3.33)

Industry New Finance 0.14 0.74 0.11 0.05 0.08(0.23) (1.85) (0.16) (0.08) (0.13)

Industry Investment −0.12 −0.17 0.82 −0.25 −0.24(−0.26) (−1.18) (1.20) (−0.87) (−0.75)

Herfindahl 0.02 −0.04 2.26 −0.15 −0.02(0.83) (−1.42) (2.18) (−0.61) (−0.62)

Log(Book/Market) 0.01 0.01 0.03 0.01 0.02(2.63) (2.20) (3.32) (2.16) (2.65)

Log(MV) 0.20 0.42 0.78 0.23 0.54(0.54) (1.71) (2.04) (0.87) (1.69)

Prior Year Return −0.00 0.00 −0.00 −0.01 −0.01(−0.59) (0.15) (−0.40) (−1.05) (−0.73)

N (firm-quarters) 24,519 4,577 10,107 9,402 15,721Number of firms 1,207 252 397 379 533R2 0.03 0.05 0.06 0.02 0.03

Note. The dependent variable is the natural log of one plus quarterly firm return. VC Dollars/Assets are the lagged aggregate quarterly VC investments,divided by total assets (Compustat annual item 6) within a particular Fama-French industry. See Appendix B for other variable definitions. All Compustatitems use the most recently available data for each firm. The t-statistics are in parentheses. Errors are clustered by firm and by quarter.

Log(1 + Return)i,t = a0 + a1 VC Dollars/Assetsi,t-1 + a2 Payout Yieldi,t-1 + a3 Asset Growthi,t-1 + a4 Firm New Financei,t-1 + a5 Industry New Financei,t-1

+ a6 Industry Investmenti,t-1 +a7Herfindahli,t-1 + a8 Log(Book/Market)i,t-1 + a9Log(MV)i,t-1 + a10 Prior Year Returni,t-1 + Quarter Dummies + ei,t

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 12: The Impact of Venture Capital Investments on Public Firm Stock Performance

IMPACT OF VENTURE CAPITAL INVESTMENTS 243

The empirical evidence of Hadlock and Pierce [2010] mo-tivates our use of age and total assets. The two authors arguefor the use of exogenous measures of financial constraint.They suggest that as a firm gets older and larger, its level offinancial constraint falls. Eventually, however, the relationswill flatten out.

Interestingly, papers such as Hadlock and Pierce [2010],Opler, Pinkowitz, Stulz, and Williamson [1999], and Whitedand Wu [2006] find that firms with higher levels of cashholdings are actually more financially constrained. Clearly,some companies will increase their cash holdings as a precau-tionary action against the contemplation of being financiallyrestricted in the future. Hadlock and Pierce argue that sinceleverage, cash flows, and cash holding are endogenous, theyall make poor measures of a firm’s financial constraint.

In theory, an older, mature firm with substantial financialslack will be able to invest in new technologies or survivean economic downturn against the new entrants. Thus, ifour hypothesis is correct, the interaction terms between VCdollars/assets and payout yield, age, and log(assets) will bepositive. This means that older, larger, higher payout firmsshould see less of a stock price decrease due to new VCinvestments.

Yet the marginal effect of both firm age and size will even-tually weaken. To address this issue, we will also includequadratic terms and their interaction with VC dollars/assets.Following the logic on Hadlock and Pierce [2010], the inter-actions of age2 and log(assets)2 with VC dollars/assets shouldbe negative and significant.

The results of regressions including some measures offinancial constraints and the interaction terms with VC dol-lars/assets appear in Table 7. Column 1 is the original regres-sion from column 4 of Table 4, with all of the prior controlvariables. Column 2 adds age and age2 along with both its in-teraction terms with VC dollars/assets. The interaction termbetween age and VC dollars/assets has a positive and signif-icant coefficient. Hence, older, more mature firms have lessof a negative response to VC funding, as our hypothesis pre-dicts. The interaction between age2 and VC dollars/assets isas predicted negative and significant (t-statistic of -5.39).

The specification in column 3 adds the interaction termwith payout yield. As predicted, this interaction term is pos-itive and significant, meaning that firms with higher payouts(i.e., dividends and share repurchases) have higher quarterlyreturns to VC dollars/assets.

In untabulated results, we find that log(assets)∗ VC dol-lars/assets and log(assets)2∗

VC dollars/assets have the pre-dicted sign, but are not significant at conventional levels.The coefficient on interaction term of log(assets) is 1.68 (t-statistic of 0.81) while log(assets)2 has a negative coefficientwith a t-statistic of -0.63.

The results from Table 7 show that not all firms are equallyaffected by the VC money. Financially constrained firms asproxied by payout yield and firm age have lower returns thehigher is the VC investments in their industry.

TABLE 7Effect of Financial Constraints on Stock Returns

(1) (2) (3)

VC Dollars/Assets −3.72 −7.00 −4.27(−2.27) (−2.73) (−2.29)

Age 0.00(1.58)

Age∗VC Dollars/Assets 0.35(3.94)

Age2 −0.00(−2.25)

Age2∗VC Dollars/Assets −0.40(−5.39)

Payout Yield 0.10 0.09 0.08(4.15) (4.08) (3.10)

Payout Yield∗VC Dollars/Assets 25.14(2.10)

Asset Growth −0.03 −0.03 −0.03(−6.55) (−6.60) (−6.52)

Firm New Finance 0.02 0.02 0.02(1.59) (1.72) (1.60)

Industry New Finance −0.02 −0.02 −0.02(−0.58) (−0.65) (−0.57)

Industry Investment −0.04 −0.04 −0.04(−1.56) (−1.53) (−1.57)

Herfindahl 0.00 0.00 0.00(0.36) (0.12) (0.32)

Log(Book/Market) 0.02 0.02 0.02(6.19) (6.24) (6.26)

Log(MV) 0.38 0.36 0.38(2.66) (2.47) (2.71)

Prior Year Return 0.00 0.00 0.00(0.81) (0.82) (0.82)

N (firm-quarters) 269,257 269,257 269,257Number of firms 8,191 8,191 8,191Quarter and Industry Dummies Yes Yes YesR2 0.19 0.19 0.19

Note.The dependent variable is the natural log of one plus quarterlyfirm return. VC Dollars/Assets are the lagged aggregate quarterly VC in-vestments, divided by total assets (Compustat item 6) within a particularFama-French industry. Payout Yield is dividends plus repurchases dividedby share price, (Compustat items [26 + (115/54)] / Price). Age is the numberof years the firm has been listed on CRSP. See Appendix B for other variabledefinitions. The t-statistics are in parentheses. Errors are clustered by firmand by quarter.

Plausibility of the Venture Capital Effect

We have shown that venture capital backing divided by thetotal industry assets is a relatively small number. The meanvalue is only 0.09%. Yet we report that this relatively smallinvestment of $394.6 billion spread over several decades canhave a substantial and economically significant effect on in-dustry stock returns. Is this documented venture capital effecton industry firm performance plausible?

First, to provide some perspective, the total amount of IPOproceeds (including banks, savings and loans, and non-CRSPlisted IPOs) over the 1980–2005 time period is $546.7 billion,according to Jay Ritter’s website.8 Thus, the aggregate VCinvestment is 72% of aggregate IPO gross proceeds over

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 13: The Impact of Venture Capital Investments on Public Firm Stock Performance

244 LOUGHRAN AND SHIVE

the same time period. Hence, the total amount of venturecapital investments is only about 25% less than the totalamount of capital raised from all IPOs over an identical timeperiod.

Second, Kortum and Lerner [2000] estimate that by 1998venture capital funding accounted for about 14% of all in-novative activity in the United States. Innovation could beexpected to harm publicly traded firms with less cutting edgetechnology. Third, our paper presents evidence that within 10years after receiving the VC investments, the average patentissuing firm has 10.6 patents. In the aggregate, firms receiv-ing money from venture capitalists in our sample producemore than 111,000 U.S. patents within the first 10 years aftergetting their VC money.

As noted earlier, Lerner [1994b] reports a list of themost important biotechnology patents by surveying promi-nent patent attorneys. All 13 of the listed patents were byfirms in our VentureXpert sample. Clearly, the venture capi-tal dollars created an eventual downpour of innovation in theyears following the initial investment.

CONCLUSION

Venture capitalists play an important role in the U.S. econ-omy. During 1980–2005, venture capital investments totaled$394.6 billion in nonpublic companies over 73,401 uniquerounds. Quite a number of these investments enabled youngfirms to issue public equity at relatively high valuations.Many of the VC-backed firms were issued patents and helpedinnovate entire industries.

Yet what is the effect on publicly traded established firmswhen VCs fund investments in a particular industry? Wefind that the effect on subsequent returns is negative, whichis consistent with the hypothesis that VC investment bringssignificant innovation pressures to the industry.

Similar to Cooper, Gulen, and Schill [2008], who findthat mispricings of asset investment are economically mean-ingful, we find that there is a negative relation between VCfunding scaled by industry assets and subsequent quarterlyfirm stock returns. Our results are also consistent with the in-dustry boom and bust analysis of Hoberg and Phillips [2010],which focuses on the role of capital expenditures by an indus-try’s publicly traded firms. Our new twist is in highlightingthe importance of private capital on subsequent industry firmperformance.

Not all firms are equally affected by the increased pres-sures brought on by the VC dollars. Using two proxies forfinancial constraint (payout yield and firm age) we find that fi-nancially constrained firms generally have lower subsequentstock returns than nonfinancially constrained firms in thesame industry. Mature firms with financial resources appearto better withstand the pressure brought on by venture capitaldollars.

As money goes to companies that are not public yet, thestock performance in an industry suffers. VC-backed firmsare found to be proficient in being awarded U.S. patents. Thisinnovation appears to apply downward pressure on the futurestock performance of an industry. On the other hand, billionsof dollars in wealth have been created by the innovations ofventure capitalist-funded firms. A potential avenue of futureresearch would investigate whether the wealth created bythe venture capital backing outweighs the losses incurred byindustry rivals.

ACKNOWLEDGMENTS

We would like to thank Robert Battalio, Harry DeAngelo,Joshua Lerner, Bill McDonald, Jay Ritter, Paul Schultz, ScottSmart, Laura Starks, and seminar participants at the BrighamYoung University, Florida State University, IU-Notre Dame-Purdue Finance Symposium, Singapore Management Uni-versity, University of Utah, and University of Notre Damefor helpful comments. We are grateful to Hang Li and JohnSaulitis for research assistance.

NOTES

1. Is ‘Web 2.0’ Another Bubble?, Wall Street Journal,December 27, 2006.

2. Although venture capital investment started in theUnited States shortly after the end of World War II,it really took off after 1979 when a “prudent man”ruling allowed pension plans to invest in VCs. Hence,we use a starting point of 1980 for our sample. Oursample also includes corporate venture capital (CVC)investments by firms like Microsoft, Intel, and CiscoSystems.

3. Across the five stages, on average, more than 57% oftotal venture capital money is in the expansion stage.The lowest amount of VC dollars (only 1.87% of allVC dollars) is directed towards seed stage financing.

4. The web site is http://patft.uspto.gov/netahtml/PTO/search-adv.htm. We use the assignee name field to paireach of our VC-backed firms.

5. Other variables that we used in untabulated results aredividend yield, payout ratio (dividends and repurchasesscaled by net income), IPO first-day returns, and lognumber of quarterly IPOs within an industry. Thesevariables do not affect the results and we do not includethem to avoid using too many potentially correlatedvariables. In addition, if we control for the contem-poraneous Fama-French factors (size, book-to-market,and momentum), we see only minor value changes inthe VC dollars/assets regression coefficients.

6. When we use VC dollars instead of scaled VC dol-lars by industry assets as the explanatory variable of

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 14: The Impact of Venture Capital Investments on Public Firm Stock Performance

IMPACT OF VENTURE CAPITAL INVESTMENTS 245

interest, our results are similar. If unscaled VC dollarsare used as an independent variable, the coefficientsare always negative and statistically significant in theregressions or included with the other explanatory vari-ables.

7. For robustness, we also examine only technology firmsaccording to the Loughran and Ritter [2004] tech clas-sifications and obtain significant coefficients on VCdollars scaled by assets.

8. The data from Ritter’s table 8 report grossproceeds by calendar year (http://bear.cba.ufl.edu/ritter/IPOs2008sorts.pdf).

REFERENCES

Boudoukh, J., R. Michaely, M. Richardson, and M. Roberts. “On the Impor-tance of Measuring Payout Yield: Implications for Asset Pricing.” Journalof Finance, 62, (2007), pp. 877–915.

Brav, A. and P. Gompers. “Myth or Reality? The Long-Run Underperfor-mance of Initial Public Offerings: Evidence from Venture and Nonven-ture Capital-Backed Companies.” Journal of Finance, 52, (1997), pp.1791–1821.

Cameron, C., J. Gelbach, and D. Miller. “Robust Inference With Multi-wayClustering.” Working Paper (2006), University of California-Davis.

Cooper, M., H. Gulen, and M. Schill. “Asset Growth and the Cross-Section of Stock Returns.” Journal of Finance, 63, (2008), pp. 1609–1651.

Fama, E. and K. French. “The Cross-section of Expected Stock Returns.”Journal of Finance, 47, (1992), pp. 427–465.

Fama, E. and K. French. “Common Risk Factors in the Returns on Stocksand Bonds.” Journal of Financial Economics, 33, (1993), pp. 3–56.

Fama, E. and K. French. “Industry Costs of Equity.” Journal of FinancialEconomics, 43, (1997), pp. 153–193.

Gompers, P., A. Kovner, J. Lerner, and D. Scharfstein. “Venture CapitalInvestment Cycles: The Impact of Public Markets.” Journal of FinancialEconomics, 87, (2008), pp. 1–23.

Gompers, P. and J. Lerner. The Money of Invention. Boston, MA: HarvardBusiness School Press, 2001.

Hadlock, C. and J. Pierce. “New Evidence on Measuring Financial Con-straints: Moving Beyond the KZ Index.” Review of Financial Studies, 23,(2010), pp. 1909–1940.

Hoberg, G. and G. Phillips. “Real and Financial Industry Booms and Busts.”Journal of Finance, 65, (2010), pp. 45–86.

Hong, H., W. Torous, and R. Valkanov. “Do Industries Lead Stock Markets?”Journal of Financial Economics, 83, (2007), pp. 367–396.

Hou, K. and D. Robinson. “Industry Concentration and Average Stock Re-turns.” Journal of Finance, 61, (2006), pp. 1927–1956.

Kaplan, S., B. Sensoy, and P. Stromberg. “How Well Do Venture Capital DataBases Reflect Actual Investments?” Working Paper (2002), University ofChicago.

Kortum, S. and J. Lerner “Assessing the Contribution of Venture Capital toInnovation.” Rand Journal of Economics, 31, (2000), pp. 674–692.

Lerner, J. “Venture Capitalists and the Decision to Go Public.” Journal ofFinancial Economics, 35, (1994), pp. 293–316.

Lerner, J. “The Importance of Patent Scope: An Empirical Analysis.” RANDJournal of Economics, 25, (1994), pp. 319–333.

Loughran, T. and J. Ritter. “Why Has IPO Underpricing Changed OverTime?” Financial Management, 33, (2004), pp. 5–37.

Opler, T., L. Pinkowitz, R. Stulz, and R. Williamson. “The Determinantsand Implications of Corporate Cash Holdings.” Journal of Financial Eco-nomics, 52, (1999), pp. 3–46.

Schumpeter, J. Capitalism, Socialism and Democracy. New York, NY:Harper and Row, 1942.

Thompson, S. “Simple Formulas for Standard Errors That Cluster by BothFirm and Time.” Working Paper (2009), Harvard University.

Whited, T. and G. Wu. “Financial Constraints Risk.” Review of FinancialStudies, 19, (2006), pp. 531–559.

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014

Page 15: The Impact of Venture Capital Investments on Public Firm Stock Performance

246 LOUGHRAN AND SHIVE

APPENDIX A:Industry Linkage between Fama and French [1997] and VentureXpert

Fama-French Fama-FrenchIndustry Number Industry Name VentureXpert codes

1 Agric 9500–9699, 4200–4299, 95402 Food 7320, 7340–7359, 7399, 73003 Soda 73304 Beer 73107 Fun 7100–71998 Books 9450, 94709 Hshld 7000, 7400, 7420–7499, 7999

10 Clths 741011 Hlth 5400–5499, 521012 MedEq 4400–4499, 5000, 5200–5209, 5220–539913 Drugs 4000, 4100–4139, 4900, 5100–5149, 5500–5599,14 Chems 4300–4399, 8150–819917 BldMt 7450–7459, 8100–8149, 9520–9530, 9440–944918 Cnstr 9700–979921 Mach 8000, 8200–8399, 8500–869922 ElcEq 3200–339928 Mines 9600–969929 Coal 6700–679930 Oil 6100–649931 Util 6000, 6500–6699, 6800–6799, 6900, 9800–989932 Telcm 1000–189933 PerSv 7540–755934 BusSv 2600–2899, 4600–4699, 8700–8799, 9300–9399, 9470–947935 Comps 2000–2149, 2200–2599, 2900–2999, 3600–3699, 941536 Chips 3000–3179, 3400–3599, 3800–389937 LabEq 3500–3599, 3700–3799, 3900–3999, 4500–459938 Paper 9410–9419, 9430–943939 Boxes 7560–7569, 9100–9199, 9460–946942 Rtail 7200–729943 Meals 7500–7529, 759944 Banks 9200, 9230–9239, 929945 Insur 9210–921946 RlEst 9220–922947 Fin 9240–9259, 9250, 9254, 9255

APPENDIX B:Definitions of the Variables Used in the Paper

VC Dollars/Assets This variable is the lagged aggregate quarterly VC investments from VentureXpert, divided by total assets (Compustat item 6) within aparticular Fama-French industry.

Prior Year Return This firm-level variable is the last 12 month buy-and-hold return of the firm.Payout Yield We compute total payout yield as the total dollars spent on dividends and repurchases during the quarter divided by share price. This

firm-level variable is dividends plus repurchases divided by share price, (Compustat items [26 + (115/54)] / share price).Asset Growth For each firm, we compute the annual percentage change in assets (Compustat annual item 6) of the latest fiscal year from the one

before.Firm New Finance For each firm, firm new finance is the sum of net equity and debt issuance divided by the value of assets (Compustat items

(108–115+111–114)/6).Industry New

FinanceThis is the particular Fama-French industry mean of the sum of net equity and debt issuance divided by the value of assets (Compustat

items (108–115+111–114)/6).Industry Investment This is the lagged change in industry-level capital expenditure (Compustat item 30) scaled by property, plant and equipment

(Compustat item 8).Herfindahl Index This index is computed by dividing the sum of the squares of the sales (Compustat item 12) in each 3-digit SIC code industry by the

squared total sales of that industry.Log (Book/Market) This firm-level variable is the lagged firm-level natural log of the book-to-market ratio.Log (MV) This variable is the lagged natural log of the firm’s market value (shares outstanding multiplied by stock price) on the day prior to the

beginning of the quarter.Age Age is the number of years the firm has been listed on CRSP.Log (Assets) This firm variable is the natural log of total assets (Compustat item 6).

Dow

nloa

ded

by [

Was

hbur

n U

nive

rsity

] at

10:

37 3

1 O

ctob

er 2

014