Dl

Fa

b

c

a

ARRAA

KFIKKT

1

fifCMobfi

sef(oAi(2

r

0h

Research Policy 42 (2013) 231– 244

Contents lists available at SciVerse ScienceDirect

Research Policy

jou rn al h om epage: www.elsev ier .com/ locate / respol

oes inward foreign direct investment improve the innovative performance ofocal firms?

rancisco Garcíaa,1, Byungchae Jinb,2, Robert Salomonc,∗

Facultad de Economía y Empresa, Universidad de Oviedo, Avda. del Cristo S/N, 33006 Oviedo (Asturias), SpainBeedie School of Business, Simon Fraser University, 250-13450 102 Avenue, Surrey, BC, Canada V3T 0A3Stern School of Business, New York University, 44 West 4th Street, Suite KMC 7-59, New York, NY 10012, United States

r t i c l e i n f o

rticle history:eceived 13 August 2010eceived in revised form 8 June 2012ccepted 11 June 2012vailable online 10 July 2012

a b s t r a c t

Over the past several decades, research in the fields of international business and strategy has devotedincreasing attention to outward foreign direct investment (FDI). Despite extensive scrutiny of the firm-specific motivations for, and consequences of, outward FDI; we know relatively little about inward FDI,the impact of inward FDI on host country firms, and especially, how inward FDI affects the innovativenessof those firms. Extant theoretical arguments predict contrasting effects. One line of research highlights

eywords:oreign direct investmentnward FDInowledge spilloversnowledge transferechnological innovation

the benefits to host country firms. Another line of research highlights the deleterious consequences tohost country firms. Utilizing data from 1799 Spanish manufacturing firms from 1990 to 2002, we investi-gate the relationships between industry-level and firm-level inward FDI and the innovative performanceof host country firms. We find that FDI inflows into Spain are negatively associated with the ex post inno-vation of local firms. We contrast these findings with those using conventional measures of productivity.

. Introduction

Scholars from a variety of disciplines have long examined therm-specific motivations for, and consequences of, engaging in

oreign direct investment (FDI) (e.g., Buckley and Casson, 1976;antwell, 1989; Kogut and Chang, 1991; Martin and Salomon, 2003;orck and Yeung, 1991). This type of investment is referred to as

utward FDI, and it is characterized by individual firms expandingeyond domestic borders to invest in foreign countries via green-eld, alliance, or acquisition (for a review see Caves, 1996).

Despite the extensive theoretical and empirical study of FDI,urprisingly, research in international strategy has centered almostxclusively on the antecedents and consequences of outward FDIor the firms making the investments. Relatively little researchwith a few notable exceptions) has been directed at the impactf inward FDI on firms in the country receiving those investments.nd what little we do know generally addresses the impact of

nward FDI on the total factor, or labor, productivity of local firmsAitken and Harrison, 1999; Blalock and Simon, 2009; Chung et al.,003; Haddad and Harrison, 1993; Javorcik, 2004; Konings, 2001).

∗ Corresponding author. Tel.: +1 212 998 0223.E-mail addresses: fgarciap@uniovi.es (F. García), bjin@sfu.ca (B. Jin),

salomon@stern.nyu.edu (R. Salomon).1 Tel.: +34 985 10 62 19.2 Tel.: +1 778 782 7833.

048-7333/$ – see front matter © 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.respol.2012.06.005

© 2012 Elsevier B.V. All rights reserved.

We understand relatively little about how inward FDI affects inno-vativeness, a measure of learning that scholars suggest is vital tounderstanding the growth of economies (Grossman and Helpman,1994; Salomon and Shaver, 2005). Because FDI is often viewedas a catalyst for economic development, and because countriesincreasingly compete to attract FDI, it is important for managersand policymakers to understand the impact of inward FDI not juston productivity, but also on innovation. To our knowledge, no studyhas examined the impact of inward FDI on the innovativeness ofindigenous firms.

When it comes to the broader impact of inward FDI on theinnovation of local firms, extant theory offers two plausible,yet competing, explanations. One line of reasoning suggests thatinward FDI ought to lead to beneficial outcomes for local firms.The purported mechanism is through knowledge spillovers fromforeign entrants to local firms, and/or through heightened incen-tives to innovate to compete with better-endowed foreign entrants.Another line of research casts doubt on the positive impact ofinward FDI, suggesting instead that inward FDI might adverselyaffect local firm innovation. This research emphasizes how theincreased competition that comes with foreign entry relegatesdomestic firms to less innovative market niches and/or crowdsindigenous competitors out of the market.

The aim of this study is to address this debate by examininghow inward FDI affects the innovative performance of local firms.Utilizing data from 1799 Spanish manufacturing firms from 1990to 2002, we examine how their innovative output is influenced

2 ch Poli

boctStie

miitioiiTi

lstd

2

fwtfiehcFoYiH

cilsetSptmwtMiswpiMrusod

32 F. García et al. / Resear

y industry- and firm-level FDI inflows. With respect to measuresf innovation, we use patent application and product innovationounts. In both cases, we find that inward FDI is negatively relatedo the ex post innovative performance of Spanish firms. That is,panish firms, after having been acquired by foreign entrants, tendo apply for fewer patents. Additionally, indigenous firms operat-ng in industries that receive greater FDI inflows introduce fewerx post product innovations.

In a post hoc analysis, we contrast our findings with those usingore conventional labor productivity and total factor productiv-

ty measures. In contrast with the innovation results, we find thatnward FDI is positively related to ex post labor productivity andotal factor productivity. We interpret this to suggest that whilenward FDI facilitates efficient resource allocation in the local econ-my – helping inefficient firms (relative to the foreign entrants)mprove – it is detrimental to the technological development ofndigenous firms which is critical for long-term economic growth.hese combined findings hold important implications for both pol-cy and practice.

The paper proceeds as follows. In the next section we review theiterature on the impact of inward FDI on host country firms. Weubsequently generate competing hypotheses. We then describehe data, the methodology, and present results. The final sectioniscusses the findings and concludes.

. Theory and hypotheses

Although much of the research in international strategy hasocused on the firm-specific antecedents and consequences of out-ard FDI, there is a literature in economics that has examined

he impact of inward FDI on the host country economy, and itsrms. Some scholars recognize the potential for beneficial knowl-dge spillovers from foreign entrants to host country firms. Othersighlight how foreign entry increases local competition. Reflectingompeting explanations, extant empirical results have been mixed.indings have demonstrated both a positive effect of inward FDIn the productivity of local firms (Haskel et al., 2007; Keller andeaple, 2009) and a negative effect of inward FDI on the productiv-

ty of host country firms (Aitken and Harrison, 1999; Haddad andarrison, 1993; Konings, 2001).

Regardless of the effect, the one thing these studies share inommon is that they have centered almost exclusively on thempact of inward FDI on either the total factor productivity orabor productivity of local firms. Recent research in internationaltrategy, however, suggests that the mixed empirical findings fromconomic studies might be influenced by the use of such produc-ivity measures (e.g., MacGarvie, 2006; Salomon and Jin, 2008;alomon and Shaver, 2005). In fact, Bloom and Van Reenen (2010:. 204), channeling Abramovitz (1956), go so far as to suggesthat, “Productivity differences at the firm level have long been a

easure of our ignorance. . .” This is due to not just the way inhich productivity is estimated, but also because a variety of fac-

ors influence productivity and it is difficult to isolate them. Instead,acGarvie (2006) and Salomon and Shaver (2005) suggest that

nnovation represents a fruitful alternative to the standard mea-ures of productivity. Although the expectation is that innovationill ultimately manifest as increased productivity, improved laborroductivity and/or total factor productivity are not necessarily

ndicative of innovation, nor do they ultimately yield innovations.oreover, because technological innovation is central to theo-

ies of economic growth (e.g., Grossman and Helpman, 1994),

nderstanding the impact of inward FDI on innovation, and notimply labor or total factor productivity, aids our understandingf whether, and if so, how, inward FDI can act as a catalyst forevelopment.

cy 42 (2013) 231– 244

2.1. Inward FDI as a catalyst for innovation

Scholars have long recognized the potential for positive exter-nalities from inward FDI in host economies (e.g., Caves, 1974;Kearns and Ruane, 2001). According to theory, such externali-ties should be driven by the following underlying mechanisms:increased competitive pressure that provides incentives for localfirms to improve; an increase in the demand for upstream sup-ply allowing for increased scale economies that reduce costs forall firms; and/or, the opportunity for local firms to benefit fromknowledge transfer – learning state-of-the-art technologies frombetter-endowed foreign entrants.

As conventional industrial organization economics suggests,the entry of foreign rivals stimulates competition in host-countrymarkets. Enhanced competition inhibits local firms from gainingmonopolistic or oligopolistic profits, which induces local firmsto take action to defend their markets and retain market share(Chung, 2001). As a response to foreign entry therefore, local firmsattempt to improve their productivity in order to remain com-petitive. According to this reasoning, competition, via the marketmechanism, improves the allocative and technical efficiency offirms (Blomström and Kokko, 1998).

Consistent with such an interpretation, Chung et al. (2003)found a positive relationship between the entry of Japanese auto-mobile assemblers into the U.S. and productivity among U.S. autocomponent manufacturers. Although they found evidence that U.S.auto component manufacturers benefited broadly from the entryof Japanese auto assemblers, they failed to find any productivityimprovements for local U.S. suppliers that provided automobilecomponents directly to the Japanese assemblers. This led theauthors to conclude that the productivity improvements were aresult of enhanced competitive pressure rather than any directknowledge transfer.

Although similar in many respects to the first mechanism, thesecond argument highlights the demand-side benefits of foreignentry in local factor markets (Rodríguez-Clare, 1996). The under-lying rationale is that the presence of foreign entrants increasesthe demand for upstream goods. This allows local suppliers to reapthe benefit of scale economies, and for local competitors to bene-fit from decreased input costs. In the aggregate, this increases theproductivity of local firms.

As an example, Hobday (1995) discovered that multinationalentry into various industries in Taiwan resulted in increaseddemand for intermediate goods, resulting in productivity increasesamong local firms. He attributed these productivity improvementsto economies of scale, and quality standards imposed by foreignentrants.

The third mechanism through which foreign entry can improvethe operations of domestic firms is through knowledge transfer –either directly or indirectly. Direct knowledge transfer from inwardFDI can occur through the intraorganizational provisions of intan-gible assets from a parent firm to its foreign affiliates (Hobdayand Rush, 2007) or through contract between foreign entrants andtheir local suppliers (Blalock and Gertler, 2008; Haskel et al., 2007;Keller and Yeaple, 2009). Indirect knowledge transfer can occurthrough unintended knowledge spillovers (Almeida and Kogut,1999). Local firms can learn by observing and imitating foreignentrants, through formal and informal interactions with those com-petitors, and through intelligence gathering from third-parties thatinteract regularly with the foreign entrants. For example, domesticcompetitors can use foreign entrants as an operational benchmark,and even reverse-engineer their products to gain technological

insights (Salomon, 2006).

There is empirical evidence that knowledge flows from foreignentrants to domestic firms. For example, the international strat-egy literature has long demonstrated that foreign parents transfer

ch Poli

iC(ii(LoBwi

iltlfSn(Jpt

Ht

2

ndsA2frtsn

gifipm

rsoiIwp

tc2hqsumfi

m

F. García et al. / Resear

ntangible assets and capabilities to their foreign affiliates (seeaves, 1996). With respect to non-affiliated firms, Keller and Yeaple2009) discovered that knowledge spillovers from foreign directnvestment into the U.S. accounted for about 10% of the productiv-ty gains of U.S. manufacturing firms from 1987 to 1996. Javorcik2004) found that knowledge transferred from foreign entrants inithuania to their upstream suppliers increased the productivityf those suppliers, especially for joint venture entries. Similarly,lalock and Gertler (2008) found that foreign entrants to Indonesiaillingly transferred technology to upstream suppliers resulting in

mprovements in productivity for local firms.Although the studies reviewed in this section examine the

mpact of inward FDI on the productivity (labor or total factor) ofocal firms, there is reason to believe that similar effects are likelyo hold for innovation. In fact, work in the innovation literature hasong documented the role of competition in increasing incentivesor firms to innovate (e.g., Aghion et al., 2001; Schumpeter, 1942).imilarly, scholars have demonstrated a connection between inter-ational business activity and spillovers in the form of innovationMacGarvie, 2006; Penner-Hahn and Shaver, 2005; Salomon andin, 2008, 2010; Salomon and Shaver, 2005). Therefore, we mightlausibly expect inward FDI to be positively related to the innova-ive output of host country firms:

ypothesis 1a. All else equal, inward FDI will be positively relatedo the innovative output of host country firms.

.2. Inward FDI as a hindrance to innovation

Although one set of studies indicate that there are positive exter-alities for domestic firms from inward FDI, a contrasting view castsoubt on the existence of such externalities, even demonstrating, inome cases, negative externalities associated with inward FDI (e.g.,itken and Harrison, 1999; Haddad and Harrison, 1993; Konings,001). For example, in an empirical study of Moroccan manu-acturing plants, Haddad and Harrison (1993) found a negativeelationship between industry-level inward FDI and the produc-ivity of local plants. Similarly, Aitken and Harrison (1999), using aample of Venezuelan firms, demonstrated that inward FDI had aegative effect on the productivity of local domestic firms.

The main mechanism by which inward FDI is purported toenerate negative externalities is through competition. Althoughncreased competitive pressure generated by the entry of foreignrms into the local market can motivate local firms to become moreroductive to stave off competition, increased competition in localarkets may have negative consequences for local firms.Foreign entry increases competition for market share, and may

esult in “market stealing,” whereby better-endowed entrants stealhare from local players. This forces the domestic players to reduceutput, raising their average costs, and leaving less capital to investn new technologies (Aitken and Harrison, 1999; Konings, 2001).n addition, increased supply/capacity in the market puts down-

ard pressure on prices for competing local firms, leading to lowerrofitability (Hanson, 2001).

Not only do foreign entrants steal market share from domes-ic competitors and put downward pressure on prices, but theyan also raise labor and resource costs for local firms (Hanson,001; Spencer, 2008). For example, since multinationals often payigher wages than local firms, the latter may find it difficult to hireualified employees (Spencer, 2008). Similarly, although upstreamuppliers may benefit from economies of scale, to the extent thatpstream supply is constrained in the short run, factor input prices

ay increase for local firms, again impacting the profits of domestic

rms.Taken together, these mechanisms suggest that foreign entry

ay crowd local firms out of the market, especially the least

cy 42 (2013) 231– 244 233

efficient ones (Blomström and Kokko, 1998). Furthermore, evenif local firms are not forced out of the market entirely, foreignentrants often displace them to less-profitable, less-innovative seg-ments of the market (Hanson, 2001). For example, Cantwell (1989)demonstrated that the entry of U.S. multinationals in Europeanmarkets relegated local firms in some industries to less desirable,less innovative market segments in which the multinationals werenot active.

Some scholars argue that increased competition from inwardFDI may also have deleterious consequences for domesticentrepreneurship (De Backer and Sleuwaegen, 2003). That is,inward FDI may not only hasten the exit of local firms, but alsoimpede the entry of domestic entrepreneurs. This occurs becauseforeign competition reduces the expected returns to entrepreneur-ship, creating a situation in which the best entrepreneurs prefer totake employment with foreign entrants instead of founding newenterprises.

As with studies that demonstrate positive externalities ofinward FDI, the aforementioned studies solely examine the impactof inward FDI on the productivity of local firms. Similar effects arelikely to hold for innovation. In fact, some scholars have hinted thatinward FDI might negatively impact the innovative output of localfirms (Cheung and Lin, 2004). Reduced profitability decreases thecash available to invest in innovative effort; foreign entry pusheslocal firms to less desirable, less innovative segments of the mar-ket; inward FDI can bid up the price for the talented human capitalnecessary to achieve innovation; and for local affiliates acquiredby foreign entrants, foreign parents might shift innovative activi-ties previously performed in the local market to the home country.For these reasons, we might plausibly expect inward FDI to be neg-atively related to the innovativeness of local firms. We thereforepropose:

Hypothesis 1b. All else equal, inward FDI will be negativelyrelated to the innovative output of host country firms.

3. Data and methods

3.1. Sample

To test the competing hypotheses, we use data from a yearlysurvey conducted by the Fundación SEPI (“Sociedad Estatal de Par-ticipaciones Industriales” or National Bureau of Industrial Activity)with the support of the Spanish Ministry of Industry. The Fundaciónsurveys the entire population of Spanish manufacturing firms with200 or more employees as well as a random sample of 5% of the pop-ulation of firms with fewer than 200 employees in order to generatea representative picture of the Spanish manufacturing economy atlarge. The first survey was conducted in 1990, and we were able togain access to data gathered through 2002.

The initial survey year (1990) included information on 2188firms from 20 industries. When a firm drops from the sample inany given year, the Fundación replaces it with another of similarsize from the same industry. This led to an initial sample consistingof an unbalanced panel of 3462 firms from 1990 to 2002.

We complemented the Fundación data with industry-level dataon inward FDI flows from the OECD Foreign Direct InvestmentStatistics (FDIS) database. The FDIS publishes yearly FDI inflowsby industry for the 30 OECD member countries. We combined theOECD data for Spain with the Fundación data, matching them byindustry. In the process we removed all 78 firms in the Fundación

data that were classified as “Miscellaneous manufacturing” to pre-serve consistency across data sources. In addition, we were unableto determine an exact industry match with the OECD data for 236of the firms in the Fundación data (those firms classified into the

2 ch Poli

“fi

fiiobfisqoi

3

puvioi

paoyt

gtCa(aMniotaScwta

rofvtpdtlgf

c

isi

34 F. García et al. / Resear

Non-metallic products” industry). This reduced the sample to 3148rms from 18 industries.

To explore the temporal relationship between inward FDI andrm innovation, and to control for some forms of endogene-

ty and unobserved heterogeneity, we incorporate dynamics intour empirical specification (we describe our method in detailelow). The dynamics require that there be no gaps in the within-rm time series. This restriction, coupled with missing data forome variables, sacrifices 1349 firms from the sample.3 Conse-uently, our final sample reduces to 1799 firms and 7506 firm-yearbservations.4 Table 1 presents the breakdown of our sample byndustry with descriptives.

.2. Dependent variables

In this study we examine innovation output rather than laborroductivity or total factor productivity. As we mentioned earlier,nderstanding the relationship between inward FDI and the inno-ativeness of local firms is important to gain greater insight on thempact of inward FDI on the host country. We proxy for innovationutput using two measures: patent application counts and productnnovation counts.

The Fundación SEPI collects information on the number ofatents for which a focal firm applied in a given year. The vari-ble we label PATENT APPLICATIONS is expressed as the numberf patent applications a firm filed for protection in Spain in a givenear, via either the Spanish Industrial Property Registry (SIPR) orhe European Patent Office (EPO).

A firm seeking patent protection must apply to the agency thatoverns patenting in the country or region where it seeks pro-ection. The EPO, established as a result of the European Patentonvention (EPC) of Munich in October 1973, currently overseesnd governs patent applications/grants in 38 European countriesEPO, 2009). Spain formally became a member of the EPC andligned its national patent laws with prevailing European law onarch 20, 1986 (Ulloa and Salas, 1993). However, it still maintains a

ational patent office. Thus, any firm seeking to patent technologyn Spain has two options: to apply to the EPO, designating Spain asne of the countries where it seeks protection, or to apply directly tohe SIPR. The two offices use identical criteria for granting patentsnd offer the same protection to patent holders in Spain (Ulloa andalas, 1993). It is more expensive to file with the EPO, and the pro-ess takes longer (an average of 18 months for the EPO, versus 12ith the SIPR). However, if a firm is applying for protection in more

han one EPC country, applying through the EPO reduces paperworknd administrative costs.

The variable we employ departs somewhat from existingesearch in that we measure total patent applications (regardlessf whether they were granted) rather than only those applicationsor patents that were later granted. As a result, it is likely that ourariable captures a higher approximation of the number of patentshat a firm ultimately receives. The EPO (2009) notes that Europeanatent submissions have a 50% success rate, but since the Fundaciónoes not identify the firms or patents by name, it is not possible

o identify which applications were successful. A potential prob-em in measuring total patent applications instead of only patentsranted is that it may capture spurious applications filed by theocal firm. However, because the application process incurs costs,

3 The dynamic specification that we use necessarily drops any firm without threeontinuous years worth of observations.

4 Results were consistent with those presented herein when we exclude dynam-cs. Therefore, our specification, although it sacrifices some of the power of theample, represents a more conservative test of the phenomenon. We return to thesessues, and others related to sample selectivity and survivor bias, in Section 4.5.

cy 42 (2013) 231– 244

we expect that a firm would file only if it believed it has devised alegitimate innovation. Therefore, a benefit of this measure is thatit captures the number of innovations for which the firm believespatent protection is merited, whereas counts exclusively of applica-tions that were ultimately granted may underestimate the numberof innovations the firm achieved.5

In addition to patent counts, the Fundación SEPI collectsinformation on product innovation. The variable PRODUCT INNO-VATIONS captures the number of new and/or modified productsintroduced by a focal firm in a given year.

Myriad studies use innovation and patent counts as a proxyfor innovative productivity (e.g., Acs and Audretsch, 1990; Pavitt,1984). Moreover, the strengths and weaknesses have been welldocumented (e.g., Cohen and Levin, 1989). Although patents serveas a particularly good indicator of innovation in certain industries(Levin et al., 1987), product innovations provide a nice complement(Acs and Audretsch, 1990; Salomon and Shaver, 2005). Ultimately,the use of two related, but distinct, measures of innovation allowsus to explore the variance unique to each, while mitigating the defi-ciencies inherent in selecting one measure to the exclusion of theother.

3.3. Independent variables

We use two measures of inward FDI in this study: industry-level inward FDI and firm-level inward FDI. The latter captures theFDI inflows directly into the Spanish firms in this sample. The Fun-dación SEPI collects information on the foreign ownership stakestaken in the Spanish companies in its survey. The measure, whichwe label FDI INTO FIRM, captures the percentage ownership stakethat a foreign parent takes in the focal Spanish firm in a givenyear. This measure allows us to assess the direct impact of foreignownership on the innovativeness of Spanish affiliates.

We complement the firm-specific measure of inward FDI withone that captures sector-specific effects. That is, we include a mea-sure of FDI inflows into the focal firm’s industry. We obtained thedata on industry-specific FDI inflows in Spain from the OECD FDISdatabase. We then defined the variable FDI INTO INDUSTRY as thetotal amount of FDI inflows to a given industry in a given year,expressed in millions of Euros. This measure captures how for-eign investment, broadly, impacts the innovative productivity ofdomestic firms.

Because it takes time for the effects of inward FDI to be felt bylocal firms, we lag both of our inward FDI variables. This also allowsus to better explore the temporal relationship between inward FDIand innovation. Based on the size of our panel and previous studies,we use lags of one, two, and three years of the industry- and firm-level inward FDI variables (Bernard and Jensen, 1999; Salomon andShaver, 2005). Table 2 presents data on the firm- and industry-levelinward FDI over the sample period.

3.4. Control variables

Because our dependent variables capture the innovative per-formance of a firm, we control for other factors that couldsystematically affect innovation. We therefore include variablespreviously identified as likely to influence the innovativeness offirms, while controlling for unobservables using a combination of

dynamics and fixed year effects.

Researchers have long suggested that firm size influences inno-vative output (e.g., Schumpeter, 1942). Larger firms tend to possess

5 The relationship between patent applications and R&D in these data follows asimilar pattern to that demonstrated between granted patents and R&D found byHausman et al. (1984).

F. García et al. / Research Policy 42 (2013) 231– 244 235

Table 1Industry breakdown of the sample (1990–2002).

Industry Total number of firms Avg. employees Avg. patents Avg. product innovations

1. Meat products 56 228.59 0.05 1.182. Food and tobacco 209 205.70 0.15 0.963. Beverages 44 353.82 0.25 0.834. Textiles 216 159.87 0.08 4.395. Leather and footwear 70 38.32 0.08 8.586. Wood 55 68.37 0.01 0.667. Paper 59 216.28 0.12 1.048. Editing and graphic arts 100 116.58 0.19 3.109. Chemical products 154 296.20 0.29 2.1810. Rubber and plastic products 113 192.37 0.04 3.1711. Metallurgy 69 477.48 0.13 3.6912. Metallic products 213 118.10 0.14 1.8913. Machinery and mechanical 171 262.85 0.26 2.6414. Office products, data, and optical 38 352.06 0.18 2.5415. Electronic and electrical machinery 153 236.02 0.14 2.2916. Automobiles and motors 100 968.36 0.16 1.7817. Other transport material 47 612.96 0.10 0.6718. Furniture 116 68.39 0.36 4.77

Full sample 1983 276.24 0.15 2.57

simfilta

ciiRiI

girb

TI

ufficient resources to develop new technologies and are thereforen a better position to appropriate returns from their R&D invest-

ents. However, innovation on the part of small, entrepreneurialrms underpins economic growth. Regardless of whether small or

arge firms are better suited to innovation, we control for any poten-ial effect of size on innovation. We define the variable FIRM SIZEs the natural log of total employees of a focal firm in a given year.

Scholars have similarly explored the influence of intangibleapabilities such as R&D and marketing on innovative productiv-ty (see Cohen and Levin, 1989). We control for the effect of R&Dnvestments by including an R&D intensity measure. We define&D INTENSITY as R&D expenditures divided by total sales. We also

nclude a measure of advertising intensity. We define ADVERTISINGNTENSITY as advertising expenditures divided by sales.

Literature in industrial organization economics has long sug-ested that market structure – i.e., industry concentration –

s related to firm innovation (see Van Cayseele, 1998 for aeview). Industry structure not only impacts the competitiveehavior of firms, but also their incentives to innovate. There

able 2ndustry and firm FDI inflows (1990–2002).

Industry Industry FDI inflows(millions D)

Firm FDIinflows (%)

1. Meat products 722.02 8.042. Food and tobacco 2551.16 16.533. Beverages 522.43 16.774. Textiles 282.72 4.475. Leather and footwear 84.00 0.856. Wood 527.16 4.917. Paper 570.32 22.448. Editing and graphic arts 1033.17 5.939. Chemical products 7658.49 40.1910. Rubber and plastic products 913.72 28.3711. Metallurgy 780.17 19.8712. Metallic products 2148.50 12.1513. Machinery and mechanical 415.95 22.8114. Office products, data, and optical 420.00 34.1015. Electronic and electrical machinery 367.00 31.2916. Automobiles and motors 2687.60 45.1517. Other transport material 271.91 14.6918. Furniture 987.48 4.46

Full sample (averages) 1274.62 18.50

remains, however, considerable debate about which market struc-ture is more conducive to innovation. Although some argue thatoligopolistic/monopolistic market structures foster innovation,others contend that competitive markets provide greater incen-tives to innovate (e.g., Aghion et al., 2001; Schumpeter, 1942).Regardless, we control for INDUSTRY CONCENTRATION to accountfor the effect of market structure. Following prior literature, wedefine INDUSTRY CONCENTRATION as the four-firm concentrationratio in the focal firm’s primary market.

Finally, for reasons unobserved to these researchers, innova-tiveness might vary over time, with macro-level changes in theeconomic environment. We therefore include fixed year effects(year dummies) into each specification.

3.5. Statistical method

Our two dependent variables — PATENT APPLICATIONS andPRODUCT INNOVATIONS — are count measures that take non-negative integer values. Moreover, many of the observations arebunched close to zero. Scholars suggest a Poisson regression modelto deal with dependent count variables of this sort (e.g., Greene,2003). The Poisson regression, however, is sensitive to distribu-tional assumptions. For example, should the mean and variancefor the observed sample not equal lambda, the likelihood functionwould be misspecified, generating erroneous results. We found evi-dence of overdispersion in these data; therefore, we turn to thenegative binomial (Cameron and Trivedi, 1998; Hausman et al.,1984). Eq. (1) formally expresses the specification:

�it = exp(ˇ1ni,t−p + ˇ2mi,t−p + ˇ3Wit + εit), p = 1, 2, 3 (1)

where �it represents the expected number of patent applications(or product innovations) for firm i at time t, ni,t−p captures theindustry-level inward FDI for firm i at time t − p, mi,t−p is the firm-level inward FDI for firm i at time t − p, Wit is a vector of controlvariables, and εit is the individual unobserved error term. In thenegative binomial model, εit is assumed to have a standard gammadistribution. The betas are parameter estimates.

Given the panel structure of our data with several observa-

tions per firm, the possibility arises that a systematic componentmay be embedded in εit, leading to serial correlation of residualsacross observations within firms and spurious regression results.Hausman et al. (1984) introduce a fixed-effect to control for serial

2 ch Poli

cduaId

iv

�

wacr

grmticden(

tFtdltCfioa

4

4

aICsFwIaai

sm(a(ct

36 F. García et al. / Resear

orrelation of this sort; however, there has been considerableebate about whether this method effectively controls for individ-al effects (Allison and Waterman, 2002; Guimarães, 2008). As anlternative, we turn to a dynamic longitudinal model. We specify anNAR autoregressive process that incorporates lagged values of theependent variable as regressors (see Alzaid and Al-Osh, 1990).6

Because we use three-year lags of the independent variables ofnterest (inward FDI), we incorporate three lags of the dependentariable into every specification. We estimate the following model:

it = exp(�1yi,t−1 + �2yi,t−2 + �3yi,t−3 + ˇ1ni,t−p + ˇ2mi,t−p

+ ˇ3Wit + uit), p = 1, 2, 3 (2)

here yi,t−1∼3 are the lags of the dependent variable for firm i,nd uit is an unobserved disturbance term that we can now moreonfidently assume is free of serial correlation. The rhos and betasepresent coefficient estimates.

Including firm dynamics (i.e., the yi,t−1∼3 terms) provides twoeneral benefits. First, dynamics reduce the potential for serial cor-elation by controlling for firm-specific unobservables in a flexibleanner. They control for firm-specific unobservables by allowing

he current dependent variable to temporally co-vary with pastnstantiations of the dependent variable. Fixed effects, by contrast,ontrol for unobservables in a static fashion, by de-meaning theependent and independent variables by firm. Both approachesffectively ensure that the independent variables of interest doot simply reflect persistent innovative differences across firmsAllison and Waterman, 2002).

Second, and especially important for identification in our set-ing, dynamics allow us to control for the endogeneity of inwardDI to the extent that inward FDI is influenced by past innova-ion. This is important because research in international strategyemonstrates that foreign firms not only tend to enter growing

ocal industries, but that when entering through acquisition, theyend to acquire more productive, more innovative, local firms (e.g.,aves, 1996; Kogut and Chang, 1991). Including dynamics allows

or inward FDI (both at the industry level and the firm level) andnnovation to freely co-vary, thereby controlling for the influencef past innovation on current inward foreign investment (Cameronnd Trivedi, 1998).

. Results

.1. Correlations and descriptives

Table 3 presents descriptive statistics for, and correlationsmong, the variables used in this study. We find that R&DNTENSITY, ADVERTISING INTENSITY, FIRM SIZE, and INDUSTRYONCENTRATION are positively related to both innovation mea-ures: PATENT APPLICATIONS and PRODUCT INNOVATIONS. BothDI INTO FIRM and FDI INTO INDUSTRY are positively correlatedith R&D INTENSITY, ADVERTISING INTENSITY, FIRM SIZE, and

NDUSTRY CONCENTRATION. This indicates that Spanish affili-

tes of foreign firms are generally larger, invest more in R&D,nd advertise more than purely domestic firms. Moreover, foreignnvestment is drawn to more concentrated markets.

6 For continuous dependent variables, researchers have proposed the autoregres-ive model that includes exogenous regressors and lagged dependent variables as aethod of controlling for firm-specific effects (see Greene, 2003). Al-Osh and Alzaid

1987) argue that the traditional AR(1) model can be extended to the integer-valuedutoregressive (INAR(1)) model applied to count data. Moreover, Alzaid and Al-Osh1990) refine an integer-valued pth-order autoregressive structure (INAR(p)) pro-ess and address differences between the INAR(p) and AR(p) processes. We applyhe INAR(p) process to our negative binomial model.

cy 42 (2013) 231– 244

With respect to the impact of inward FDI on innovative produc-tivity, we find somewhat mixed results. Although FDI INTO FIRM ispositively related to both innovation dependent variables, hintingat some potential positive externalities of inward FDI, the correla-tions between FDI INTO INDUSTRY and innovation are less clear.Specifically, FDI INTO INDUSTRY is positively related to PATENTAPPLICATIONS, yet negatively correlated with PRODUCT INNOVA-TIONS.

We must exercise caution in interpreting these effects. First, thecorrelations are moderate in both statistical and economic magni-tude. Second, the effects are contemporaneous. They cannot informdirectionality, or temporal relations among variables. Finally, pair-wise correlations do not control for intervening effects that weinclude in the multivariate analyses. Therefore, to better under-stand the nature of the relationships, we turn to the multivariateregression analyses.

4.2. Regression results

Table 4 presents the results of the negative binomial regres-sion with patent applications as the dependent variable. Again, tominimize endogeneity and unobserved heterogeneity concerns, weinclude INAR (3) dynamics into each model. Although not reported,year dummies are also included to control for time-dependenteffects.

Column 1 presents the base model with control variables only.The results show persistence in patent behavior. Firms that appliedfor patents in the prior year are more likely to apply for patents inthe current year. As expected, the effects diminish with increas-ing lags. Consistent with prior studies, FIRM SIZE is positive andsignificantly associated with patent applications. Larger firms aremore likely to file for patents. The effects of R&D INTENSITYand ADVERTISING INTENSITY are likewise positive and significant.These results are consistent with previous findings in the innova-tion literature (e.g., Cohen and Levin, 1989; Hausman et al., 1984).The effect of INDUSTRY CONCENTRATION on patent applications isalso positive and significant, indicating that Spanish firms in moreconcentrated markets are more likely to apply for patents.

Turning to our hypotheses, we introduce one-, two-, and three-year lags of FDI INTO INDUSTRY and FDI INTO FIRM in columns 2,3, and 4. We fail to find a significant relationship between FDI INTOINDUSTRY and subsequent firm patent activity. However, we dofind a negative and significant relationship between FDI INTO FIRMin all three specifications (p < .05), suggesting that foreign owner-ship can have deleterious consequences for the patent output offoreign-owned Spanish firms.

Taken together, this suggests that foreign entrants into Spainconcentrate their innovative efforts in their home country and shiftthe locus of innovative activity outside Spain. The foreign par-ent substitutes home country for host country innovation, and theSpanish affiliate becomes an implementer of strategy developed inthe home country.

We consider the same specifications using PRODUCT INNOVA-TIONS as an alternative to PATENT APPLICATIONS in Table 5. Theeffects for the control variables from column 1 are largely consistentwith those from Table 4, with the exception INDUSTRY CONCEN-TRATION.

Interestingly, though directionally consistent, we find contrast-ing results for the inward FDI variables of interest. That is, thereis no statistically significant relationship between FDI INTO FIRMand PRODUCT INNOVATIONS. However, we do find a negativeand significant effect of the one-, two- and three-year lags of FDI

INTO INDUSTRY on PRODUCT INNOVATIONS (p < .10; p < .01; p < .01respectively), and the magnitude of the effect becomes larger as thelags increase. This latter finding suggests that foreign investmentinto Spain crowds out domestic innovation.

F. García et al. / Research Policy 42 (2013) 231– 244 237

Table 3Descriptive statistics and correlations.

Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

1. Patent applications(t) 12. Patent applications(t−1) 0.44 13. Patent applications(t−2) 0.28 0.35 14. Patent applications(t−3) 0.24 0.30 0.35 15. Product innovations(t) 0.07 0.05 0.03 0.03 16. Product innovations(t−1) 0.04 0.03 0.06 0.02 0.51 17. Product innovations(t−2) 0.05 0.02 0.02 0.04 0.35 0.55 18. Product innovations(t−3) 0.03 0.04 0.02 0.01 0.36 0.39 0.52 19. FDI into industry (t−1) 0.03 0.03 0.01 0.01 −0.02 −0.02 −0.02 −0.01 110. FDI into industry (t−2) 0.04 0.04 0.02 0.01 −0.03 −0.03 −0.03 −0.02 0.46 111. FDI into industry (t−3) 0.02 0.03 0.01 0.01 −0.04 −0.03 −0.03 −0.03 0.39 0.63 112. FDI into firm (t−1) 0.01 0.01 0.01 0.03 0.03 0.02 0.01 0.01 0.08 0.16 0.13 113. FDI into firm (t−2) 0.01 0.01 0.03 0.02 0.03 0.02 0.01 0.01 0.08 0.16 0.13 0.95 114. FDI into firm (t−3) 0.01 0.01 0.03 0.04 0.03 0.02 0.02 0.01 0.08 0.16 0.13 0.92 0.95 115. R&D intensity(t) 0.16 0.16 0.10 0.10 0.05 0.03 0.03 0.03 0.11 0.15 0.13 0.07 0.07 0.07 116. Advertising intensity(t) 0.08 0.09 0.06 0.06 0.03 0.03 0.01 0.01 0.15 0.17 0.15 0.10 0.11 0.10 0.15 117. Firm size(t) 0.13 0.12 0.09 0.08 0.08 0.06 0.05 0.04 0.07 0.15 0.12 0.47 0.48 0.47 0.25 0.22 118. Industry concentration(t) 0.07 0.07 0.06 0.06 0.02 0.01 0.02 0.01 0.03 0.05 0.05 0.25 0.25 0.25 0.09 0.09 0.34 1

Mean 0.17 0.18 0.20 0.21 2.79 2.92 3.11 3.11 148.39 180.13 189.12 18.30 17.98 17.61 0.01 0.01 4.22 0.32s.d. 1.03 1.15 1.64 1.69 16.31 21.89 25.55 27.09 271.93 229.60 251.14 36.92 36.54 36.20 0.03 0.03 1.55 0.35Min 0 0 0 0 0 0 0 0 −1176 0 0 0 0 0 0 0 0 0

50

ith

TN

t

Max 25 32 92 92 481 950 950 9

Taken together, the results from Tables 4 and 5 indicate that

ndustry- and firm-level inward FDI inflows are negatively relatedo the ex post innovation of local firms. With respect to ourypotheses, the findings favor hypothesis 1b over hypothesis 1a.

able 4egative binomial regressions (Dependent variable = Patent applications(t)).

Variable Negative binomial regression

1 2

FDI into industry(t−1) −0(−0

FDI into firm(t−1) −0(−2

FDI into industry(t−2)

FDI into firm(t−2)

FDI into industry(t−3)

FDI into firm(t−3)

Patent applications(t−1) 0.66180*** 0.6(8.53) (8.3

Patent applications(t−2) 0.23930*** 0.2(4.10) (4.2

Patent applications(t−3) 0.07176 0.0(1.25) (1.2

R&D intensity(t) 10.05021*** 9.9(3.55) (3.4

Advertising intensity(t) 3.53122** 3.2(1.81) (1.6

Firm size (ln)(t) 0.21250*** 0.2(5.32) (5.7

Industry concentration(t) 0.63237*** 0.6(3.64) (3.7

Constant −4.05161*** −4(−17.07) (−1

N 7506 750Log likelihood −2177.0274 −2−2�L 5.3

-Statistics appear in parentheses.* p < .10 (one-tailed tests).

** p < .05 (one-tailed tests).*** p < .01 (one-tailed tests).

1176 1176 1372.58 100 100 100 0.40 0.53 9.62 1

4.3. Additional evidence

Despite the many benefits of the INAR dynamic negative bino-mial, including past innovation in the specifications may not

3 4

.00002.08)

.00392**

.28)−0.00017(−0.62)−0.00373**

(−2.14)−0.00013(−0.43)−0.00375**

(−2.11)4559*** 0.64563*** 0.64313***

6) (8.36) (8.35)4521*** 0.24691*** 0.24539***

7) (4.28) (4.27)7271 0.06437 0.063077) (1.15) (1.14)

2323*** 10.43731*** 10.38246***

1) (3.57) (3.54)0937* 3.43970** 3.37236**

3) (1.74) (1.71)5534*** 0.25548*** 0.25502***

9) (5.78) (5.77)6638*** 0.64969*** 0.64806***

8) (3.72) (3.71).34289*** −3.86813*** −3.86468***

8.17) (−10.63) (−16.30)

6 7506 7506174.3752 −2174.2854 −2174.4284044* 5.4840* 5.1980*

238 F. García et al. / Research Policy 42 (2013) 231– 244

Table 5Negative binomial regressions (Dependent variable = Product innovations(t)).

Variable Negative binomial regression

1 2 3 4

FDI into industry(t−1) −0.00023*

(−1.57)FDI into firm(t−1) −0.00104

(−0.93)FDI into industry(t−2) −0.00037***

(−2.37)FDI into firm(t−2) −0.00124

(−1.14)FDI into industry(t−3) −0.00047***

(−2.74)FDI into firm(t−3) −0.00081

(−0.74)Product innovations(t−1) 0.06167*** 0.06210*** 0.06233*** 0.06176***

(9.64) (9.66) (9.68) (9.61)Product innovations(t−2) 0.03030*** 0.02979*** 0.02971*** 0.02982***

(6.33) (6.24) (6.23) (6.24)Product innovations(t−3) 0.02876*** 0.02878*** 0.02787*** 0.02800***

(6.48) (6.48) (6.37) (6.34)R&D intensity(t) 14.25532*** 14.23502*** 14.14510*** 14.09135***

(5.97) (5.95) (5.90) (5.89)Advertising intensity(t) 4.28827*** 4.47776*** 4.71613*** 4.80048***

(3.05) (3.19) (3.37) (3.41)Firm size (ln)(t) 0.18904*** 0.20360*** 0.20629*** 0.20111***

(7.25) (7.04) (7.34) (7.13)Industry concentration(t) −0.08248 −0.08244 −0.07489 −0.07285

(−0.76) (−0.75) (−0.69) (−0.67)Constant −1.68723*** −1.01627*** −1.20536*** −1.20553***

(−7.17) (−7.14) (−8.07) (−3.98)

N 7506 7506 7506 7506Log likelihood −9080.1428 −9078.2230 −9076.2679 −9075.8246−2�L 3.8396 7.7498** 8.6364**

t-Statistics appear in parentheses.

atnimeoe(si

ttrIais

cF

pgAres

* p < .10 (one-tailed tests).** p < .05 (one-tailed tests).

*** p < .01 (one-tailed tests).

ccount for the full range of firm-specific unobservables that stando influence innovation. Moreover, including past innovation mightot be sufficient to adequately control for the endogeneity of

nward FDI. We therefore turn to a quasi-differenced generalizedethod of moments (GMM) for count data developed by Blundell

t al. (2002). The quasi-differenced moment conditions partialut individual specific effects and allow for the identification ofndogenous regressors using only weakly exogenous instrumentsBlundell et al., 2002; Chamberlain, 1992). In contrast with the INARpecification, this approach allows us to explicitly instrument fornward FDI (see also Salomon and Shaver, 2005).

Table 6 presents quasi-differenced GMM results that mirrorhose from Table 4. We use an additional one-year lag on each ofhe inward FDI measures of interest as an instrument, hence theeduction in sample size.7 For example, in column 4 we use FDINTO INDUSTRY(t−4) as an instrument for FDI INTO INDUSTRY(t−3)nd FDI INTO FIRM(t−4) as an instrument for FDI INTO FIRM(t−3). Wenstrument for the inward FDI variables in columns 2 and 3 in theame fashion.

The results for the independent variables of interest are largelyonsistent with those in Table 4. We find no significant effect ofDI INTO INDUSTRY, but there remains a negative and significant

7 The results were robust to the inclusion of greater lags and alternative inde-endent variables as instruments. Including additional lags resulted in an evenreater reduction in sample size. We therefore do not report those results herein.s an alternative instrument, we ran specifications with Spanish currency exchangeates, which influence the cost of inward FDI but are less likely to have a directffect on innovation. In all cases, the results remained unchanged. These alternativepecifications are available from the authors upon request.

relationship between FDI INTO FIRM and subsequent patent outputin columns 2–4 (p < .01).

We report product innovation results using quasi-differencedGMM specifications in Table 7. We instrument for our inward FDIvariables of interest as in Table 6. As with the product innovationresults from Table 5, we find that FDI INTO INDUSTRY has a neg-ative and significant effect on PRODUCT INNOVATIONS (p < .01).Likewise, we do not find a significant effect for FDI INTO FIRM.

4.4. Productivity results

In addition to the quasi-differenced GMM models, we presentadditional results using more traditional productivity measures –labor productivity and total factor productivity. This allows us tocontrast our findings with those from the economics literature onthe externalities of inward FDI. Following the extant economicsliterature, we use economic value-added per employee in a givenyear as a proxy for LABOR PRODUCTIVITY.8 Because this measureis continuous, we run linear auto-regressive (AR) models.

Table 8 presents results using labor productivity as the depen-

dent variable. The one-, two-, and three-year lags of FDI INTOINDUSTRY and FDI INTO FIRM appear in columns 2–4. Strikingly,and in contrast with the results using innovation as the dependentvariable, we find that the effects of FDI INTO FIRM and FDI INTO

8 As an alternative labor productivity measure, we also use output per employeedefined as a firm’s sales plus the variation in inventory (finished and intermediategoods) and other revenues, divided by the number of employees. The results remainconsistent with those presented herein.

F. García et al. / Research Policy 42 (2013) 231– 244 239

Table 6Quasi-differenced GMM regressions (Dependent variable = Patent applications(t)).

Variable Quasi-differenced GMM regression

1 2 3 4

FDI into industry(t−1) 0.00017(0.86)

FDI into firm(t−1) −0.00575***

(−2.39)FDI into industry(t−2) 0.00023

(0.83)FDI into firm(t−2) −0.00705***

(−2.78)FDI into industry(t−3) −0.00002

(−0.09)FDI into firm(t−3) −0.00818***

(−3.04)Patent applications(t−1) 0.14724*** 0.14708*** 0.14490*** 0.14149***

(8.37) (8.37) (8.34) (8.24)Patent applications(t−2) 0.04458*** 0.04271*** 0.04833*** 0.04811***

(4.76) (4.52) (5.18) (5.26)Patent applications(t−3) 0.04146*** 0.04047*** 0.03954*** 0.04608***

(4.36) (4.16) (4.08) (4.82)R&D intensity(t) 0.23184 0.31752 0.33970 0.32185

(0.84) (1.19) (1.26) (1.20)Advertising intensity(t) 7.91898*** 6.85857*** 6.60364*** 6.90282***

(6.50) (5.05) (4.75) (4.98)Firm size (ln)(t) 4.12375*** 3.96756*** 3.90234*** 4.18201***

(2.57) (2.49) (2.52) (2.62)Industry concentration(t) 0.37088*** 0.41894*** 0.42646*** 0.43742***

(6.44) (7.03) (7.18) (7.27)Constant −4.05046*** −4.20416*** −4.24475*** −4.23612***

(−12.89) (−13.57) (−13.47) (−13.40)

N 7372 7372 7372 7372

t-Statistics appear in parentheses.*p < .10 (one-tailed tests).**p < .05 (one-tailed tests).

*** p < .01 (one-tailed tests).

Table 7Quasi-differenced GMM regressions (Dependent variable = Product innovations(t)).

Variable Quasi-differenced GMM regression

1 2 3 4

FDI into industry(t−1) −0.00049***

(−3.58)FDI into firm(t−1) 0.00038

(0.14)FDI into industry(t−2) −0.00127***

(−2.85)FDI into firm(t−2) 0.00097

(0.34)FDI into industry(t−3) −0.00136***

(−3.63)FDI into firm(t−3) 0.00171

(0.59)Product innovations(t−1) 0.00346*** 0.00342*** 0.00338*** 0.00338***

(2.78) (2.77) (2.84) (2.79)Product innovations(t−2) 0.00190* 0.00187* 0.00176* 0.00178*

(1.51) (1.48) (1.43) (1.44)Product innovations(t−3) 0.00252** 0.00250** 0.00248** 0.00243**

(1.69) (1.68) (1.75) (1.68)R&D intensity(t) 0.18011 0.16400 0.13512 0.13414

(0.72) (0.70) (0.58) (0.58)Advertising intensity(t) 5.018354** 5.50068** 5.99525*** 6.04077***

(2.08) (2.20) (2.49) (2.65)Firm size (ln)(t) 2.26895* 2.95126** 3.62995** 3.70508**

(1.29) (1.68) (1.95) (2.03)Industry concentration(t) 0.23640*** 0.23374*** 0.24358*** 0.23442***

(4.51) (3.59) (3.70) (3.57)Constant −0.31098 −0.2497728 −0.1746156 −0.1320178

(−1.18) (−0.89) (0.62) (−0.47)

N 7372 7372 7372 7372

t-Statistics appear in parentheses.* p < .10 (one-tailed tests).

** p < .05 (one-tailed tests).*** p < .01 (one-tailed tests).

240 F. García et al. / Research Policy 42 (2013) 231– 244

Table 8OLS regressions (Dependent variable = Labor productivity(t)).

Variable 1 2 3 4

FDI into industry(t−1) 0.00070(1.14)

FDI into firm(t−1) 0.02593***

(5.51)FDI into industry(t−2) 0.00207***

(2.99)FDI into firm(t−2) 0.02593***

(5.43)FDI into industry(t−3) 0.00162**

(2.16)FDI into firm(t−3) 0.02654***

(5.50)Labor productivity(t−1) 0.01704* 0.01669* 0.01681* 0.01669*

(1.32) (1.29) (1.30) (1.29)Labor productivity(t−2) 0.53606*** 0.50719*** 0.50206*** 0.50629***

(9.70) (9.15) (9.06) (9.14)Labor productivity(t−3) 0.35826*** 0.34132*** 0.33399*** 0.33750***

(6.38) (6.08) (5.95) (6.02)R&D intensity(t) 0.08752 1.61475 0.09246 0.92960

(0.01) (0.21) (0.01) (0.12)Advertising intensity(t) 0.72516 0.78860 −0.44758 0.04155

(0.15) (0.16) (−0.09) (0.01)Firm size (ln)(t) 0.00397 −0.24413** −0.25641** −0.25478**

(0.03) (−2.00) (−2.11) (−2.09)Industry concentration(t) 0.54610 0.37954 0.39148 0.37785

(1.19) (0.83) (0.86) (0.83)Year effects Included Included Included IncludedConstant 0.75908 1.48896* 1.49820* 0.21362

(0.79) (1.51) (1.53) (0.18)

N 7506 7506 7506 7506F-test 40.85*** 37.91*** 38.56*** 38.27***

t-Statistics appear in parentheses.*

Il

dGPfcppefiaw

icdlli

te

f

1it

p < .10 (one-tailed tests).** p < .05 (one-tailed tests).

*** p < .01 (one-tailed tests).

NDUSTRY are positive and significantly related to our measures ofabor productivity.

We examine the effect of inward FDI on the total factor pro-uctivity of local firms in Table 9. Following prior research (e.g.,riliches, 1986; Hall et al., 1993), we estimate TOTAL FACTORRODUCTIVITY (TFP) using a modified Cobb-Douglas productionunction with production output and inputs, including labor, physi-al capital, and intermediate inputs.9 We deflate production output,hysical capital and intermediate inputs by the Spanish consumerrice index. Following Hall et al. (1993) we estimate firm fixed-ffects regressions with year dummies to control for unobservedrm fixed effects and macro shocks. We capture each firm’s residuals a proxy for TFP (Chung et al., 2003). As with labor productivity,e run linear auto-regressive (AR) models.

Although the sample reduces to 5763 observations due to miss-ng values for physical capital, overall, the results are broadlyonsistent with those from Table 8.10 Consistent with the labor pro-uctivity results, we find positive and significant effects of all three

ags of FDI INTO FIRM on TFP. We also find positive effects on theags of FDI INTO INDUSTRY on TFP, though the effects are weakern magnitude.

The positive effects of inward FDI into the firm on produc-ivity (from Tables 8 and 9) can help inform the non-significantffects of inward FDI into the firm on product innovation (from

9 We also examined R&D stock in lieu of intermediate inputs (see Griliches, 1986or a detailed discussion). The results were similar across specifications.10 Unfortunately, the Fundación SEPI only collects data on physical capital from990 to 1999. To better compare results across tables, we reduced the sample size

n Tables 4–9 equal to 5763 and re-estimated the same specifications presented inhe tables herein. Results do not change.

Tables 5 and 7). As mentioned previously, our product innovationdependent variable, as measured by the Fundación SEPI, cap-tures not only new product introductions but also modified and/orimproved products. To the extent that productivity improvementsbrought by the foreign parent and implemented in the local affiliatemanifest in improved materials, improved components, improveddesign, and/or improved functionalities, this could mask any under-lying new product effect. Otherwise stated, any dampening effect ofinward FDI into the firm on new products is potentially being offsetby a positive effect on product improvement or modification.

More importantly, comparing results from Tables 8 and 9 withthose from Tables 4–7 indicates that while inward FDI (both intothe industry and into the firm) is positively related to the produc-tivity of local firms, it is negatively related to their innovativeness.This suggests that although inward FDI helps firms operating at apoint within the production possibility frontier become more effi-cient, moving them closer to the frontier, it does not help shift thatfrontier outward through innovation, which is critical to the growthof national economies. Fig. 1 depicts the described effects.

Viewed in the context of a macro-level production possibil-ity frontier in a two-good nation, inward FDI might be viewedas helping economies (in our case, firms) move from point Awithin the production possibility frontier towards points B or C,thereby increasing productivity consistent with the results fromTables 8 and 9. However, the results do not support an interpre-tation whereby inward FDI helps shift the production possibilityfrontier outward through innovation toward points D or E, which

potentially provides greater long-run economic benefits. In fact,results from Tables 4 and 5 indicate that inward FDI hinders theinnovative capabilities of domestic firms. In this sense then, inwardFDI might be helpful in the short run, but harmful in the long run.

F. García et al. / Research Policy 42 (2013) 231– 244 241

Table 9OLS regressions (Dependent variable = TFP(t)).

Variable 1 2 3 4

FDI into industry(t−1) 0.00001(0.77)

FDI into firm(t−1) 0.00025***

(4.38)FDI into industry(t−2) 0.00000

(0.25)FDI into firm(t−2) 0.00027***

(4.58)FDI into industry(t−3) 0.00001*

(1.54)FDI into firm(t−3) 0.00025***

(4.27)TFP(t−1) 0.51683*** 0.51332*** 0.51346*** 0.51339***

(41.32) (41.03) (41.05) (41.04)TFP(t−2) 0.20746*** 0.20567*** 0.20559*** 0.20534***

(15.99) (15.87) (15.86) (15.84)TFP(t−3) 0.13584*** 0.13461*** 0.13453*** 0.13466***

(11.87) (11.77) (11.77) (11.78)R&D intensity(t) −0.25663*** −0.25664*** −0.24406*** −0.26247***

(−2.58) (−2.56) (−2.44) (−2.62)Advertising intensity(t) 0.03508 0.03154 0.03536 0.02525

(0.60) (0.53) (0.60) (0.43)Firm size (ln)(t) 0.02625*** 0.02442*** 0.02439*** 0.02452***

(13.71) (12.49) (12.48) (12.55)Industry concentration(t) 0.01318*** 0.01102** 0.01085** 0.01116**

(2.43) (2.03) (2.00) (2.05)Year effects Included Included Included IncludedConstant −0.11641*** −0.11362*** −0.11261*** −0.11392***

(−12.80) (−12.28) (−12.26) (−12.42)

N 5763 5763 5763 5763F-test 2203.11*** 1916.96*** 1917.24*** 1917.48***

t-Statistics appear in parentheses.

4

st

* p < .10 (one-tailed tests).** p < .05 (one-tailed tests).

*** p < .01 (one-tailed tests).

.5. Sensitivity and robustness

We tested variants of the models presented above to assess theensitivity and robustness of our findings. The aim is to evaluate

he consistency of our findings and subject them to stress tests so

Fig. 1. The net effects of inward FDI on domestic firms.

as to ensure that they are free from bias. We discuss the alternativemodels we tested below.11

First, although our dynamic modeling approach (including threeyears worth of lags of the dependent variable) is extremely con-servative, one lingering concern is that we may be systematicallyexcluding inefficient firms from the sample. Since firms must haveat least three years worth of observations to be included in oursample, it might be biased to successful firms. To check whether asurvival bias of this sort is influencing our results, we reduced thelag structure and re-ran our results, loosening the panel restrictionsto include a greater number of firms. The results remain unchanged.We also explored variants of the Olley and Pakes (1996) approachto explicitly account for firm exit.12 Again, the results confirm that,accounting for survival bias via firm exit, inward FDI dampens expost innovation and improves ex post productivity.

Second, to better inform temporal relationships between theindependent and dependent variables, and to ensure that our infer-ences were cogent, we explored models with various leads and lagsin the timing of inward FDI. Specifically, we created dummy vari-ables to capture the specific year in which a Spanish firm receivedforeign investment. We then examined the innovativeness and

productivity of those firms in the years leading up to, and following,the inward investment. We found that subsequent to investment,innovation generally declines while productivity generally

11 Though not presented, all results mentioned herein are available from theauthors upon request.

12 We note that we took the natural logarithm of patents and product innovationsin order to run the Olley and Pakes (1996) models for the innovation dependentvariables. We do so because the Olley and Pakes (1996) model is based on an OLSestimation approach.

2 ch Poli

ii

cHeoSpinaoda(Wp

5

rftpdceptsp

nbimriotbt1

lfiqSgr

adCietgaas

t

42 F. García et al. / Resear

ncreases. This corroborates our findings and strengthens ournference.

Finally, our patent measure captures the number of patent appli-ations filed for protection in Spain via either the SIPR or the EPO.owever, to the extent that there are significant differences in thexamination process between the two patent-granting institutions,ur results could potentially reflect bias. Although the FundaciónEPI does not break down the number patent of applications byatent-granting institution, it does collect data on the number of

nternational patent applications filed by the focal firm. The inter-ational patent application measure captures the number of patentpplications filed with the USPTO, JPO, and/or the EPO (for countriesther than Spain). We therefore ran additional results using thisependent variable instead of our patent application measure. Welso ran results using various permutations of the patent measurese.g., international plus domestic, domestic minus international).

e find a negative and significant effect of inward FDI on ex postatent applications in all cases.

. Discussion and conclusion

In this study we assessed contrasting theoretical predictionsegarding the relationship between inward FDI and innovation. Weocus on how inward FDI into the firm, and/or into the firm’s indus-ry, affects the firm’s innovative performance. Accordingly, we useatent application counts and product innovation counts as depen-ent variables. We use a complementary set of methods – dynamicount models and quasi-differenced GMM models – to evaluate theffects. We subsequently conduct a post hoc analysis using laborroductivity and total factor productivity measures often used inhe economics literature. We then compare our results using mea-ures of innovation to those using measures of labor and total factorroductivity.

Generally, we find that industry- and firm-level FDI inflows areegatively related to the innovativeness of local firms. Inward FDIlunts domestic innovation. Specifically, we find that firm-level FDI

nflows are negatively related to the ex post patent applications ofultinational affiliates. This suggests that foreign entrants do not

ely on their Spanish affiliates for innovation; rather, after investingn Spain, they either substitute innovation from their home country,r shift responsibility for innovation from the Spanish affiliate tohe parent firm (De Faria and Sofka, 2010). This is consistent withroader findings on asset-exploiting FDI, as foreign entrants oftenransfer their superior capabilities to the host country (see Caves,996; Martin and Salomon, 2003).

Similarly, we find a negative relationship between industry-evel FDI inflows and the ex post product innovation of localrms. Consistent with theory highlighting the deleterious conse-uences of inward FDI (Aitken and Harrison, 1999; De Backer andleuwaegen, 2003; Konings, 2001; Spencer, 2008), this finding sug-ests that foreign entry crowds out domestic innovation and/orelegates domestic firms to less profitable niches.

We find some interesting differences when comparing theforementioned innovation findings to those using labor pro-uctivity and total factor productivity as dependent variables.onsistent with a number of studies in economics, we find that

nward FDI (at both the industry and firm levels) improves thefficiency of local firms, through some combination of competi-ion/knowledge spillovers. Increased inward FDI into the industryenerally increases local competition and can spur local firms todopt more efficient methods of production. Alternatively, foreign

cquirers can exploit their superior capabilities and implementuperior production methods in the local target.

Taken together, these findings highlight an interesting distinc-ion between innovation and productivity, and demonstrate how

cy 42 (2013) 231– 244

inward FDI may differentially impact each. Although inward FDImay be beneficial for less productive (relative to foreign entrants)local firms, helping them increase their efficiency and modern-ize production, at the same time it has detrimental consequencesfor the innovative capabilities of local firms. We interpret this asevidence that inward FDI pushes local firms toward the produc-tion possibility frontier. However, inward FDI does not help thosefirms become more innovative, facilitating an outward shift in theproduction possibility frontier. Put differently, a negative compe-tition or crowding out effect dominates the potential for positiveinnovative externalities from inward FDI.

The findings from this study hold important implications forpolicymakers and managers. For policymakers, our findings aremultifaceted. On the one hand, inward FDI positively contributes toresource allocation efficiency in the host country. Foreign invest-ment improves the overall productivity of local firms. Moreover,inward FDI can result in increased local employment and tax rev-enues. These two outcomes benefit the host country and are welfareenhancing in the near term.

On the other hand, however, the results suggest that as a conse-quence of inward FDI, local innovation stagnates. This indicates thatinward foreign investment may not be the best source of knowl-edge spillovers to local firms. Although foreign firms often bringsuperior technological capabilities with them, they may reduce theability of domestic firms to innovate on their own. To the extent thatinnovation is at the root of economic growth as many in the macro-literature suggest, inward FDI might therefore not be beneficial tothe long-term growth prospects of local economies.

Inward FDI might not only negatively impact the innovative-ness of indigenous firms, but many of the aforementioned efficiency(productivity) gains might be privatized by foreign owners. There-fore, from a national policy perspective, overall local welfare mightbe diminished if the benefits that accrue to the owners of foreigncapital exceed those that accrue to agents in the local market. Out-comes such as these might be detrimental to welfare in the longrun, and suggest that policymakers ought to carefully reconsiderpolicies that attempt to attract foreign investment. Overall then,our findings call into question whether inward FDI can serve as along-run catalyst to growth in domestic economies, or whether itsimply offers a short-term fix.

For domestic firm managers, the findings suggest that firmsought to understand the full range of effects of foreign entry on theirbusiness. Although foreign firms often bring advanced knowledgewith them, the exposure to such knowledge may not necessar-ily benefit the domestic firms. They may stand to benefit fromincreased productivity, and should therefore capitalize on oppor-tunities to implement enhanced production techniques. However,it is important that firms recognize that they might not be able toenhance their innovative capabilities via knowledge spillovers orthrough competitive responses to foreign entry. They should plantheir innovation investment accordingly. Domestic managers fac-ing the prospects of foreign entry should also be aware of theirown capabilities and how to use such capabilities to respond to for-eign competition (Lee, 2009). Moreover, domestic firms should beaware of the potential for “brain drain” types of effects whereby tal-ented staff get lured away from the firm to work for higher-payingmultinationals (Spencer, 2008).

This is not to say, however, that all domestic firms will beaffected equally by foreign entry. Some firms might benefit fromforeign entry while others might be negatively impacted. For exam-ple, Blalock and Gertler (2008) and Blalock and Simon (2009) bothsuggest that technologically weaker domestic firms benefit more

from “low-hanging fruit” productivity enhancements followinginward FDI. However, technologically competent domestic firmsmight stand to benefit more from the complex technological knowl-edge that foreign entrants bring with them (Blalock and Gertler,

ch Poli

2t

tfaochpfcremiwdvagtit

ahsiwd

R

A

A

A

A

A

A

A

A

B

B

B

B

B

B

B

C

C

C

C

F. García et al. / Resear

008). Future research would be well served to examine whetherhose effects hold for innovation.

At this point, we acknowledge several additional limitations ofhis study that can help motivate future research. First, the findingsrom this study are limited to the Spanish manufacturing context,nd so future research would be well served to examine the effectsf inward FDI in other contexts, and potentially, in a cross-countryomparative setting that effectively controls for country-specificeterogeneity. Second, since the Fundación SEPI survey does notrovide home country-specific or firm-specific information on theoreign firms making investments (e.g., Patel and Vega, 1999), weannot fully control for such heterogeneity. We encourage futureesearch to revisit these issues to understand the full range offfects. Finally, we are not able to fully speak to the competitionechanisms associated with inward FDI. For example, if we could

solate data on the pay of local R&D workers, we could better assesshether inward FDI raises local R&D wages, thereby crowding outomestic firms. Moreover, if we had disaggregated data on indi-idual R&D workers, we might be able to trace employee mobilitynd knowledge flows before and after foreign investment. Finer-rained data would help to tell a more complete story with respecto the precise mechanisms through which inward FDI influencesnnovation versus productivity. We encourage future research toackle these issues.

For all the reasons mentioned above, we are cautious to gener-lize our findings. Further corroboratory research would certainlyelp strengthen our conclusions. However, limitations notwith-tanding, this study stands to make contributions to the fields ofnternational business and strategy. We hope that future research

ill refine and extend our contributions by examining improvedata sets with more precise measures.

eferences

cs, Z.J., Audretsch, D.B., 1990. Innovation and Small Firms. MIT Press, Cambridge,MA.

bramovitz, M., 1956. Resource and output trends in the United States since 1870.American Economic Review 46 (2), 5–23.

ghion, P., Harris, C., Howitt, P., Vickers, J., 2001. Competition, imitation and growthwith step-by-step innovation. Review of Economic Studies 68 (3), 467–492.

itken, B.J., Harrison, A.E., 1999. Do domestic firms benefit from direct foreign invest-ment? Evidence from Venezuela. American Economic Review 89 (3), 605–618.

llison, P.D., Waterman, R.P., 2002. Fixed-effects negative binomial regression mod-els. Sociological Methodology 32, 247–265.

lmeida, P., Kogut, B., 1999. Localization of knowledge and the mobility of engineersin regional networks. Management Science 45 (7), 905–917.

l-Osh, M.A., Alzaid, A.A., 1987. First-order integer-valued autoregressive (INAR(1))process. Journal of Time Series Analysis 8 (3), 261–275.

lzaid, A.A., Al-Osh, M.A., 1990. An integer-valued pth-order autoregressive struc-ture (INAR(p)) process. Journal of Applied Probability 27 (2), 314–324.

ernard, A.B., Jensen, J.B., 1999. Exceptional exporter performance: cause, effect, orboth? Journal of International Economics 47 (1), 1–25.

lalock, G., Gertler, P.J., 2008. Welfare gains from foreign direct investment throughtechnology transfer to local suppliers. Journal of International Economics 74 (2),402–421.

lalock, G., Simon, D.H., 2009. Do all firms benefit equally from downstream FDI themoderating effect of local suppliers capabilities on productivity gains. Journalof International Business Studies 40 (7), 1095–1112.

lomström, M., Kokko, A., 1998. Multinational corporations and spillovers. Journalof Economic Surveys 12 (3), 247–277.

loom, N., Van Reenen, J., 2010. Why do management practices differ across firmsand countries? Journal of Economic Perspectives 24 (1), 203–224.

lundell, R., Griffith, R., Windmeijer, F., 2002. Individual effects and dynamics incount data models. Journal of Econometrics 108 (1), 113–131.

uckley, P.J., Casson, M., 1976. The Future of the Multinational Enterprise. Holmes& Meier, London, UK.

ameron, A.C., Trivedi, P.K., 1998. Regression Analysis of Count Data. CambridgeUniversity Press, New York.

antwell, J., 1989. Technological Innovation and Multinational Corporations. Basil

Blackwell, Oxford, UK.

aves, R.E., 1974. Multinational firms, competition, and productivity in host-countrymarkets. Economica 41 (162), 176–193.

aves, R.E., 1996. Multinational Enterprise and Economic Analysis. Cambridge Uni-versity Press, New York.

cy 42 (2013) 231– 244 243

Chamberlain, G., 1992. Comment: sequential moment restrictions in panel data.Journal of Business & Economic Statistics 10 (1), 20–26.

Cheung, K.-Y., Lin, P., 2004. Spillover effects of FDI on innovation in China: evidencefrom the provincial data. China Economic Review 15 (1), 25–44.

Chung, W., 2001. Identifying technology transfer in foreign direct investment: influ-ence of industry conditions and investing firm motives. Journal of InternationalBusiness Studies 32 (2), 211–229.

Chung, W., Mitchell, W., Yeung, B., 2003. Foreign direct investment and host coun-try productivity: the American automotive component industry in the 1980s.Journal of International Business Studies 34 (2), 199–218.

Cohen, W.M., Levin, R.C., 1989. Empirical studies of innovation and market struc-ture. In: Schmalensee, R., Willig, R.D. (Eds.), Handbook of Industrial Organization.North-Holland, New York, pp. 1059–1107.

De Backer, K., Sleuwaegen, L., 2003. Does foreign direct investment crowd outdomestic entrepreneurship? Review of Industrial Organization 22 (1), 67–84.

De Faria, P., Sofka, W., 2010. Knowledge protection strategies of multinational firms:a cross-country comparison. Research Policy 39 (7), 956–968.

EPO, 2009. Facts and Figures. European Patent Office, Munich, Germany.Greene, W.H., 2003. Econometric Analysis, 5th ed. Prentice Hall, Upper Saddle River,

NJ.Griliches, Z., 1986. Productivity R and D, and basic research at the firm level in the

1970s. American Economic Review 76 (1), 141–154.Grossman, G.M., Helpman, E., 1994. Endogenous innovation in the theory of growth.

Journal of Economic Perspectives 8 (1), 23–44.Guimarães, P., 2008. The fixed effects negative binomial model revisited. Economics

Letters 99 (1), 63–66.Haddad, M., Harrison, A.E., 1993. Are there positive spillovers from direct foreign

investment? Evidence from panel data for Morocco. Journal of DevelopmentEconomics 42 (1), 51–74.

Hall, B.H., Mansfield, E., Jaffe, A.B., 1993. Industrial research during the 1980s: didthe rate of return fall? Brookings Papers on Economic Activity. Microeconomics1993 (2), 289–343.

Hanson, G.H., 2001. Should countries promote foreign direct investment? In: G-24Discussion Paper Series #9, United Nations.

Haskel, J.E., Pereira, S.C., Slaughter, M.J., 2007. Does inward foreign direct investmentboost the productivity of domestic firms? Review of Economics and Statistics 89(3), 482–496.

Hausman, J., Hall, B.H., Griliches, Z., 1984. Econometric models for count data withan application to the patents-R & D relationship. Econometrica 52 (4), 909–938.

Hobday, M., 1995. Innovation in East Asia: The Challenge to Japan. Aldershot, London,UK.

Hobday, M., Rush, H., 2007. Upgrading the technological capabilities of foreigntransnational subsidiaries in developing countries: the case of electronics inThailand. Research Policy 36 (9), 1335–1356.

Javorcik, B.S., 2004. Does foreign direct investment increase the productivity ofdomestic firms? In search of spillovers through backward linkages. AmericanEconomic Review 94 (3), 605–627.

Kearns, A., Ruane, F., 2001. The tangible contribution of R&D-spending foreign-owned plants to a host region: a plant level study of the Irish manufacturingsector (1980–1996). Research Policy 30 (2), 227–244.

Keller, W., Yeaple, S.R., 2009. Multinational enterprises, international trade, andproductivity growth: firm-level evidence from the United States. Review ofEconomics and Statistics 91 (4), 821–831.

Kogut, B., Chang, S.J., 1991. Technological capabilities and Japanese foreign directinvestment in the United States. Review of Economics and Statistics 73 (3),401–413.

Konings, J., 2001. The effects of foreign direct investment on domestic firms:evidence from firm-level panel data in emerging economies. Economics of Tran-sition 9 (3), 619.

Lee, C.-Y., 2009. Competition favors the prepared firm: firms’ R&Dresponses to competitive market pressure. Research Policy 38 (5),861–870.

Levin, R.C., Klevorick, A.K., Nelson, R.R., Winter, S.G., Gilbert, R., Griliches, Z., 1987.Appropriating the returns from industrial research and development. BrookingsPapers on Economic Activity 1987 (3), 783–831.

MacGarvie, M., 2006. Do firms learn from international trade? Review of Economicsand Statistics 88 (1), 46–60.

Martin, X., Salomon, R., 2003. Knowledge transfer capacity and its implications forthe theory of the multinational corporation. Journal of International BusinessStudies 34 (4), 356–373.

Morck, R., Yeung, B., 1991. Why investors value multinationality. Journal of Business64 (2), 165–187.

Olley, G.S., Pakes, A., 1996. The dynamics of productivity in the telecommunicationsequipment industry. Econometrica 64 (6), 1263–1297.

Patel, P., Vega, M., 1999. Patterns of internationalisation of corporate technology:location vs. home country advantages. Research Policy 28 (2–3), 145–155.

Pavitt, K., 1984. Sectoral patterns of technical change: towards a taxonomy and atheory. Research Policy 13 (6), 343–373.

Penner-Hahn, J., Shaver, J.M., 2005. Does international research and developmentincrease patent output? An analysis of Japanese pharmaceutical firms. StrategicManagement Journal 26 (2), 121–140.

Rodríguez-Clare, A., 1996. Multinationals, linkages, and economic development.American Economic Review 86 (4), 852–873.

Salomon, R.M., 2006. Spillovers to foreign market participants: assessing the impactof export strategies on innovative productivity. Strategic Organization 4 (2),135–164.

2 ch Poli

S

S

S

S

Ulloa, G., Salas, E., 1993. The basic features of Spain’s patents and models, Patent

44 F. García et al. / Resear

alomon, R.M., Jin, B., 2008. Does knowledge spill to leaders or laggards? Explor-ing industry heterogeneity in learning by exporting. Journal of InternationalBusiness Studies 39 (1), 132–150.

alomon, R.M., Jin, B., 2010. Do leading or lagging firms learn more from exporting?Strategic Management Journal 31 (10), 1088–1113.

alomon, R.M., Shaver, J.M., 2005. Learning by exporting: new insights from exam-ining firm innovation. Journal of Economics & Management Strategy 14 (2),431–460.

chumpeter, J.A., 1942. Capitalism Socialism and Democracy. Harper and Brothers,New York.

cy 42 (2013) 231– 244

Spencer, J.W., 2008. The impact of multinational enterprise strategyon indigenous enterprises: horizontal spillovers and crowding outin developing countries. Academy of Management Review 33 (2),341–361.

Yearbook 1993. In: Phillips, J. (Ed.), Managing Intellectual Property. EuromoneyPublications, plc., London, UK, pp. 66–75.

Van Cayseele, P.J.G., 1998. Market structure and innovation: a survey of the lasttwenty years. De Economist 146 (3), 391–417.