Differences in R&D efficiency between more and less internationalized firms: Evidence from newly industrialized countries

Differences in R&D efficiency between more and less

internationalized firms: Evidence from newly

industrialized countries

Teatske-Anne Bergsma

Supervisor: Dr. B. Qin

Co-assessor: Dr. Ing. N. Brunia

February 2012

Differences in R&D efficiency between more and less

internationalized firms: Evidence from newly

industrialized countries

Teatske-Anne Bergsma

Abstract

According to the efficient market theory all information is reflected in the market value of a firm. First, this paper investigates if expenditures on research and development are absorbed in the market value of firms in newly industrialized countries. The investigation is performed on 123 listed firms in the period 2005-2010. The results in this paper support that R&D efficiency exists in emerging economies, even during a period of crisis. Second, the worldwide trend of geographic expansion brings improved technology and access to international networks. However, it can be hard to reap the benefits of R&D investment in the complexity of international operations. The results in this paper indicate that internationalization has a negative effect on R&D efficiency in newly industrialized countries. This is different from empirical evidence found in developed countries. The results support the resource-based view of internationalization and the three-stage-internationalization theory.

Keywords: Internationalization, abnormal return, research and development expenditure and newly industrialized countries

1

TABLE OF CONTENTS

TABLE OF CONTENTS 3

1 INTRODUCTION 4

2 LITERATURE REVIEW AND HYPOTHESES DEVELOPMENT 6

2.1 R&D Efficiency 6

2.2 Degree of internationalization and R&D efficiency 9 2.3 Country differences in R&D efficiency 11

2.4 Effect crisis on R&D efficiency 12

3 METHODOLOGY 12

3.1 Model 12

3.2 Dependent variable 14

3.3 Independent variable 16

4 DATA AND DESCRIPTIVE STATISTICS 19

4.1 Data 19

4.2 Descriptive statistics 20

5 RESULTS AND FINDINGS 23

5.1 Evidence of R&D efficiency 23

5.2 Internationalization negatively affects R&D efficiency 26 5.3 The existence of R&D efficiency per country 29 5.4 R&D efficiency exists during crisis 33

5.5 Robustness checks 34

6 SUMMARY AND CONCLUSION 35

REFERENCES 38

1 INTRODUCTION

In today‟s world the focus of firms changes from tangible to intangible assets. There is a shift from traditional industry to knowledge and science-based industry. Firms need to get a competitive advantage through knowledge as knowledge is recognized as the driver of productivity and growth (OECD, 1996). This is reflected in the growing interest for literature about R&D efficiency and the valuation of intangibles. If firms are able to spend more efficiently on research and development the effect of R&D investment on firm market value is stronger, also known as R&D efficiency (Duqi & Torluccio, 2010). Another factor which can increase productivity and growth is international expansion of firms. Foreign activities can lead to economies of scale and scope. Moreover, Kotabe et al. (2002) and Morck and Yeung (1991) argue that firms expand internationally to take advantage of their unique resources and capabilities related to R&D investment. However, it can be hard to reap the benefits of R&D investment in the complexity of international operations. The market valuation of R&D expenditures or in a more general context, the valuation of intangibles, is difficult (Edmans, 2011). Investors mechanically accept firms' financial statements at face value, without adjusting for the long-term benefits of R&D, known as: “functional fixation hypothesis” (Florou & Chalevas, 2010). Therefore, the way the market rewards innovation in a world where firms operate internationally is of interest to a firm as it can influences the R&D investment decision.

Therefore this paper focus on the following: “Does the degree of foreign activity within a firm have an

influence on R&D efficiency in newly industrialized countries?

Graph 1.1 R&D investment growth for developed and emerging economies2

This paper aims to explore an under investigated sample, the newly industrialized countries in the period 2005-2010, which includes the crisis. I create a panel dataset of 123 listed firms who reported R&D investments over 2005-2010. First, this paper gives insight into the fast growing emerging markets. Second, light is shed on the influence of the degree of internationalization on R&D efficiency within firms, which is under researched in literature. Third, this study evaluates control variables which prove to have a significant effect on firm value and are widely discussed in the western R&D efficiency literature. By including theses control variables, the study builds upon a more uniform model for measuring R&D efficiency. Fourth, literature signals that R&D efficiency differs per country. To analyze the R&D efficiency per country and the impact of internationalization thereon a subsample is created for China, India, Malaysia and Turkey. Fifth, due to the lack of information and the ability to create a sufficient sample, this study includes the crisis period. According to Wyatt (2008) this is of interest, as further research is needed in new economic benchmarks to test reliability and focus on settings where intangibles are changing due to shocks or events as a crisis. Meyer (2011) found that R&D expenditures stay stable during a crisis in developed countries which is contrary to past experience. This implies that investors value R&D expenditures during the crisis. Finally, to verify the robustness of the results, different dependent variable measures and measures for internationalization are considered.

2

Source: Information about innovation was obtained from the classification that the OECD draws up each year based on the

I find a significant positive relation between abnormal return and R&D expenditure, even during a period of crisis. However, when the sample is regressed per country, not all countries presented R&D efficiency and interesting is that I find R&D inefficiency. This phenomenon can be explained by the fact that countries having R&D efficiency are characterized by well-trained engineers and scientists, and endowment of low-cost which is complemented with a fast growing domestic market, leading to R&D efficiency. Second, past empirical evidence found a positive relation between internationalization and R&D efficiency, only they focused on developed countries. Hence, theory gives different implications as it is complex to reap benefits of intangible assets in international operations. The results in this paper indicate the opposite of evidence found in developed countries. Namely, internationalization has a negative effect on R&D efficiency in newly industrialized countries. Arguments to clarify the findings are that newly industrialized countries are in the early stage of internationalization, have a cost advantage in mature products, and lack institutional support for market base transactions. Overall, it can be concluded that R&D efficiency exists in emerging economies, even during a period of crisis and that internationalization has a negative impact on return of R&D firms in emerging economies.

The remainder of this paper proceeds as follows. Section 2 contains background information on the market valuation of R&D investment and the impact of internationalization thereon. It reviews the vast body of empirical and theoretical literature, about R&D efficiency and from the resource-based view and internationalization theory in the international business literature, and describes the hypotheses to be tested. Section 3, presents the research methodology and variables to be used. The dataset and descriptive statistics are presented in Section 4. Section 5 expresses the empirical results attained and the investigation undertaken with the aim of verifying the robustness of the results. A brief conclusion and final remarks follow in section 6.

2 LITERATURE REVIEW AND HYPOTHESES DEVELOPMENT

In this section I review the literature and empirical evidence of R&D efficiency and the impact of internationalization, country difference and crisis on R&D efficiency. Together with the literature review hypotheses are formulated.

2.1 R&D Efficiency

Investments in knowledge and innovation are seen as long-term investments in intangible assets that contribute to future growth. An indicator for knowledge is R&D expenditure. If firms are able to spend efficiently in research and development: the effect of R&D investment on firm market value is stronger, also known as R&D efficiency (Duqi & Torluccio, 2010).

should be reflected in the stock prices immediately as it changes expectations about the future performance and free cash flow of the firm (Pakes, 1985; Fama, 1970). More to the point, cash flow news influences stock returns as it affects investors‟ expectations about future growth (Fama & French, 1989). A firm is evaluated by the market as a bundle of intangible and tangible assets. Since, R&D creates intangible assets it should be capitalized. However, the market fails to fully incorporate intangibles as innovation because the market lacks information on their value (Edmans, 2011; Florou & Chalevas, 2010). R&D is recorded as expenses in the income statement as the Financial Accounting Standards Board (FASB) requires. All R&D costs should be treated as revenue expenditures and charged to expenses in the period in which they are incurred. The reasoning behind this requirement is that it is too hard to trace specific costs to specific profitable developments. The costs of research and development are continuous and necessary for the success of a business and should be treated as current expenses. To support this conclusion, the FASB cited studies showing that 30 to 90 percent of all new products fail and that 75 percent of new-product expenses go to unsuccessful products. Thus, their costs do not represent future benefit (Power & Needles, 2010; Lev, 2004). Therefore, it is difficult to value R&D investments as it is uninformative and does not say anything about the success or quality.

Another fundamental problem in the relationship between innovation and profitability are the competitors. Innovators want to protect their novel product and process from imitation whereby the average level of investment and knowledge within an industry plays an important role (Tsang et al., 2008). For example, the high-tech industry has an understanding of technology, knows how to integrate R&D results, has a better infrastructure, and platform to share knowledge (Kessler, 2003). Therefore, imitation in the high-tech industry can be more difficult than in the low-tech industry. Arguing that, the less time a competitor need to imitate the innovation has a negative impact on the pay-offs of the R&D investing firm. According to Geroski (1995) it is known as the appropriability problem. Kafouros (2005), Gustavsson et al. (1999) and Pegels and Thirumurthy (1996) find empirical evidence of the fact that R&D investment in the high-tech industry affects firm performance the most. The researchers find a positive and significant result for the medium- and high-tech industry. Moreover, in the low-tech industry the result was insignificant. Therefore, accumulation of knowledge and development of technological strengths affects firm value more in the high-tech than low-tech industry. However, Ballardini et al. (2005) and Chan et al. (2001) have proven that the first costs creating that enterprise value do not out weight expenditures in innovation generating that market value. Thereby arguing that firms; high or low-tech with different business models generate the same return. Besides the fact that it is difficult to value R&D investments as it is uninformative and does not present future benefit, the type of industry and competitive environment can have an influence on the relationship.

Jaruzelski et al. (2005) argue that instead of the R&D investment the quality of the innovation process within the firm is the one that influences firm performance. Koellinger (2008) find that it is not about how a firm innovates but what it innovates. Economic theory suggests that innovations can lead to growth for the firm even if the innovation investment does not lead to profits. It is due to the fact that innovation can create an outward shift of the supply function leading to higher levels of output or lower variable costs which is a productivity increase (Götz, 1999; Reinganum 1981). In the light of the resource-based theory, a firm develops a strategy by using resources of primary importance to get a competitive advantage. The movement whereby physical assets play a less important role than intangible assets in gaining competiveness underlines the importance of innovation in the resource-based view. The way firms prioritize, coordinate and organize resources as R&D is of meaning (Barney, 1991). The linkage between R&D investments and firm value has been widely debated in literature. Moreover, there is an ambiguous debate in literature going on about the way R&D efficiency should be measured. Petkova (2006) find that stock returns significantly correlate with innovation. Moreover, Koellinger (2008) provided evidence that innovative firms are more likely to grow, but not necessarily more profitable. However, recent empirical findings indicate that an R&D investment is positively valued by investors (Appendix I: Table 2.1). Moreover, R&D efficiency indicated positive variations in the scale of the influence which differed per industry and country.

Above and beyond the fact if R&D expenditure has or has not an influence on market value, what should be the direction of the relationship? Namely, the relationship between R&D expenditure and performance is not necessarily unidirectional. Firms that perform well may have easier access to capital to fund future investment (Hubbard & Kashyap, 1992). Furthermore, past investment in innovation may lead to learning-by-doing effect, better absorptive capacity and availability of skilled labor known as complementary resources (Brynjolfsson & Hitt, 2002; Cohen & Levinthal, 1989). There might be a bidirectional relationship between R&D investment and performance. Ho et al. (2005) find that R&D investment contributes positively to the one year market performance. In addition, different from the one year market performance, the three year market performance presented greater R&D efficiency. Confirming the theory of Brynjolfsson and Hitt (2002) and Cohen and Levinthal (1989) that innovation may lead to learning-by-doing effect. It is a complementary resource, whereby investment in R&D has a greater positive impact on return after a longer period of time.

uniform model for measuring R&D efficiency. Concluding, many researchers which did similar research find that there exist a positive relationship between R&D and market value of a firm (Appendix I: Table 2.1). Whereas, to my knowledge there have never been an investigation on the effect of R&D on firm market value for emerging countries. Therefore, the first hypothesis is tested for emerging countries, concerning research and development expenditures and stock return.

H1: Research and development expenditure has a positive effect on stock return.

2.2 Degree of internationalization and R&D efficiency

Degree of internationalization generally refers to the extent to which firms operate beyond their national border. The increased market liberalization contributed to the expansion of firms into foreign markets (Aulakh et al., 2000). Moreover, academic literature and press highlight the necessity of firms to increase multinationality and to organize their value chain around the world for their products and services, to gain scale, location and learning advantages. Morck and Yeung (1991) argue that firms expand internationally to take advantage of their unique resources and capabilities related to R&D investment and increase its value. Criscualo et al. (2005) and Basile et al. (2003) find empirical evidence for firms in the UK and Italy that support this theory, representing that R&D investment is an important determinant of the degree of involvement in international operations. Put differently a higher tendency to innovate in internationalized firms can be appointed to the access of external and internal stock of knowledge. The trade theory argues that firms establish abroad if the costs of maintaining capacity in multiple markets is less than the tariff and transport costs (Castellani & Zanfei, 2007).

Besides the different motives to internationalize, establishing abroad generates benefits. Castellani and Zanfei (2007) argue that foreign direct investment (FDI) brings improved technology and access to international networks, all of which further raise growth and productivity. This can be connected to the resource-based theory. When firms gain access to unique capabilities and resources through internationalization it can contribute in achieving a sustainable differential competitive advantage (Barney, 1991). Furthermore, a multinational can reap economies of scale and scope in marketing, production and R&D (Caves, 1996; Chandler, 1986). According to the efficient market theory all this information should be reflected in the market value of a firm (Duqi & Torlucci, 2010).

geographical dispersion. The debate issue in theory is, does internationalization affects firm performance or does the resource-based theory, access to unique resources as R&D, strengthens the internationalization paradigm. A drawback in the research of Bae et al. (2008) and Lu and Beamish (2004) is that the findings may be country specific. This is proved by Duqi and Torlucci (2010) and Hall and Oriani (2006) who find that results differ per country.

Recent scholars about internationalization from Zahra, Ireland and Hitt (2000), Delios and Henisz (2000) and Barkema and Vermeulen (1998) have drawn attention on exploration benefits in the learning perspective. This learning perspective highlights that a firm‟s subsidiaries in disparate host countries can help to enhance its capabilities, competitiveness and knowledge base through experiential learning. Barlett and Ghoshal (1990) refer in their article to multinationals and the creation of flexible linkages, creating synergies. Linkages within multinationals aids in the innovation process and exploiting country-specific resources resulting in the benefit of worldwide learning and capturing scope economies. Most theories about internationalization argue that internationalization has a positive effect on R&D efficiency due to linkages, networks of knowledge sharing and economies of scale and scope. Morck and Yeung (2003) bring up a dimension which can have a negative influence on the relationship, as processes for innovation are not internalized. It should be taken into account that internationalization is known as stage theory, a theory where firms go through stages of evolution representing a long time span (Bae, et al. 2008; Oviatt & McDougall, 2005; Lu & Beamish, 2004). An important dimension highlighted by Morck and Yeung (2003) is that intangible assets as R&D investments are informative intensive. For efficient exploitation this asset should be internalized, requiring extensive effort and time. Therefore, I expect that in an early stage of internationalization, R&D is less efficient in comparison with firms in a later stage. Kotabe et al. (2002) find evidence for the fact that R&D is less efficient in the early stage. This is in line with the S-curve relation also known as the three-stage theory, whereby in the early stage there is a negative relation between multinationalty and firm performance due to insufficient economies of scope and scale and initial learning costs. In the middle stage, multinationality increases and this leads to a positive relation as the benefits outweigh the cost. Finally, multinationality reaches a turning point whereby coordination costs exceed the benefit, resulting in a decrease in performance (Bae et al. (2008). The finding implies that an increase in R&D investment affects the financial performance of internationalized firms much later than operational performance. Put differently, firms enjoy operational improvement before financial improvement from multinationality and upturn R&D investments. This is in line with literature from Porter (1986) that a firm builds first on positional strength and then it leads to improved financial performance. This confirms the three-stage-theory of internationalization, which postulates an S-curve (Bae, et al. 2008; Oviatt & McDougall, 2005; Lu & Beamish, 2004).

geographical diversification through economies of scope and scale (Castellani & Zanfei, 2007; Zahra et al., 2000; Delios & Henisz , 2000; Barkema & Vermeulen, 1998). Hitt et al. (1994) argue that R&D intensity becomes important when a firm expands into international markets. It allows a firm to achieve more efficiency in its operations. These firms can lower production costs, achieve economies of scale and charge premium prices for its innovative products (Kotabe et al., 2002; Porter, 1986). Bae et al. (2008) and Lu and Beamish (2004) find empirical evidence for the theory only they focused on developed countries. On the other hand, an important dimension highlighted by Morck and Yeung (2003) and Kotabe et al. (2002) is that intangible assets as R&D investments are informative intensive. For efficient exploitation this asset should be internalized, requiring extensive effort and time. Therefore, I expect that in an early stage of internationalization, R&D is less efficient in comparison with firms in a later stage. Hence, theory gives different implications. This makes the direction of the relation uncertain.

H2: Degree of internationalization has an effect on R&D efficiency

2. 3 Country differences in R&D efficiency

can be expected and many researchers have proven that country of origin has on impact on R&D efficiency. To summarize, results for R&D efficiency differ per country according to Duqi and Torlucci (2010), Hall and Oriani (2006) and Hofstede (2010, 2002, 2001, 1988). As most research is performed within developed countries I test if R&D efficiency differs per country in emerging economies. This brings the following hypothesis.

H3: R&D efficiency differs per country

2.4 Effect crisis on R&D efficiency

Since the period of the study covers the recent crisis that impacts the global market, I examine the effect of a crisis on R&D efficiency. According to Meyer (2011) investors take a more sceptical view on innovations in a period of crisis due to uncertainty of future pay-off. Theory about the cyclical view states that implementation of new ideas and innovation is postponed during a period of economic downturn waiting for the next upswing (Francois & Lloyd-Ellis, 2003). However, Meyer (2011), Voigt and Moncada-Paternò-Castello (2009) Lee et al. (2008) and Frankenberger and Graham (2003) find empirical evidence that during and after the crisis R&D expenditures positively affect firm value. An explanation for this is that during times of crisis it is important to invest in technology as every firm in the industry is affected. Thereby, investments in R&D can lead to flexibility, as options for developing products, leads to competiveness (Jagle, 1999; Sanchez, 1993). During previous recession R&D investment always suffered. Controversial is the fact that economic downturn has a positive effect on innovation dynamics. To clarify, firms cut on R&D expenditures during a crisis but the expenses on R&D they make is rewarded by investors. According to Wyatt (2008), research is needed in new economic benchmarks to test reliability and focus on settings where intangibles are changing due to shocks or events as a crisis. Hence, I posit the following.

H4: R&D efficiency exists during a period of crisis

3. METHODOLOGY

In this section I present the methodology that I use to test the hypotheses. First I present the model, thereafter I explain the dependent and independent variables that I use in the estimation.

3.1 Model

(2006), Ho et al. (2005), Stark and Thomas (1998) MacKinlay (1997) and Green et al. (1996). The empirical model presents firm return as a linear function of the firm‟s R&D ratio. The year is defined as the stock prices six month prior to and six months after a firm‟s fiscal year end. Ho et al. (2005), Stark and Thomas (1998) and Green et al. (1996) argue that in the sixth month after the fiscal year all investors had the time to absorb the information of the end of the fiscal year. A longer time period can lead to noise and may not be attributable to the variable under study. Moreover, including a longer-time period may reduce the sample size and induce a logical error, known as the survivorship bias. A shorter time-period may overlook any lag in the response (Fama, 1998). A chance of lag in response is bigger for R&D investments as investors fails to fully incorporate intangibles as innovation since the market lacks information on their value (Edmans, 2011; Florou & Chalevas, 2010). The model estimates the additional value the stock market incorporated into the market value as a result from a cash outflow as R&D investment over the fiscal year (Faulkender & Wang, 2006; Ho et al., 2005). Throughout the analysis, the main focus is on R&D efficiency, captured by the coefficient of the R&D ratio. The model specification takes into account six control variables which affect the return for R&D firms (Duqi & Torlucci, 2010). Some of the control variables are exogenous, while others are endogenous. The additional value is measured by annual buy-and-hold abnormal return (AR). The model, to test H1, is specified with the following equation:

Equation (1)

Equation (2)

Since panel data is used, I control for firm specific and country, industry and time effects to capture common shocks in these countries, industries, and periods that might have an effect on abnormal return. H1 to H4 are tested using fixed effects panel analysis. The fixed effect panel method allows working with data with non-observable effects. Simpler OLS model cannot work with this element because it leads to biased estimators (Verbeek, 2008:355-411). Moreover including time-series observations and cross-sectional observations together allows controlling for heterogeneity both across industry, countries and time leading to lower probability of misspecification bias. Furthermore, the Withe-cross section coefficient covariance matrix is used to correct for the presence of heteroskedasticity and autocorrelation in the residuals.

3.2 Dependent variable

Buy-and-hold abnormal return (AR) is the dependent variable in this model. According to Faulkender and Wang (2006) and Ho et al. (2005) the methodology for estimating the value associated with firm characteristics is an improvement to the market-to book value proposed by Fama and French (1998). An argument in favour of using abnormal return is that it can be corrected for time-varying risk factors. I address this problem using return on the market as benchmark. Another argument is that equity return is easy to measure and interpret. According to Faulkander and Wang (2006) and MacKinlay (1997) a part of the variability in time-series is derived from differences over time in the compensation for risk. Therefore, total shareholders return (TSR) is corrected for market return (Rm). AR i,t is the annual buy-and-hold abnormal return (dividend adjusted) for firm i during six month prior to and six months after a firm‟s fiscal year, adjusted by the market return during the same period, as the dependent variable, which is calculated as follows:

TSR is based on the stock prices six month prior to and six mother after a firm‟s fiscal year end including dividend pair during the period. Thereby Pi,t is the stock price, six months after the firm specific fiscal year end for the particular year and Pi,t-1 is the stock price six month ahead the of the fiscal year end. Ho et al. (2005), Stark and Thomas (1998) and Green et al. (1996) argue that in the sixth month after the fiscal year all investors had the time to absorb the information of the financial statement published at the end of the year. Moreover all the investors had the time to act on this information and therefore in the sixth month after the fiscal year it should be reflected in the stock price. To correct for caveats in measuring AR, firm‟s specific fiscal year end is used as not for all firms the fiscal year ends in December.

The return of the market (Rm) is calculated in the same way as TSR. Rm is based on the return of the respective local market and in that sense emerging markets contrast with developed markets. Evidence finds that return of emerging markets is more influenced by local rather than global information variables (Harvey, 1994). In other words, emerging markets are segmented from world capital markets. Moreover, in developed countries, researchers studying time-varying asset returns it is assumed that risk stays constant. As the industry structure develops in newly industrialized countries, keeping risk constant is a far less reasonable assumption. Hence, changes in industry structure can change the risk exposure and weights of the individual firms. This argument contributes to the fact that risk exposures are influenced by the local market instead of the global market. Therefore, this research uses the market return of the respective local market.

3.3 Independent variables

In this section the independent variables are discussed. First, I explain the main explanatory variables as the research and development ratio and the internationalization variable. Second, I describe the control variables which prove to have a significant effect on return and widely discussed in the R&D efficiency literature.

Research and development ratio

I am mindful that, since, R&D creates intangible assets it should be capitalized. However, the market fails to fully incorporate intangibles as innovation because the market lacks information on their value (Edmans, 2011; Florou & Chalevas, 2010). Therefore, I use a research and development ratio (R&D ratio). The R&D ratio is the research and development expenditure of firm i at time t divided by total assets in the year in progress. Common is that total assets is used as a deflator for R&D expenditure to diminish scale effects (Duqi &Torlucci, 2010; Faulkender & Wang, 2006; Ho et al., 2005). Only R&D expenditures in the current year are incorporated as according to Green et al. (1996) past expenditure has already produced tangible assets and is mirrored in the earnings before interest and tax (ebit). Furthermore, it is common that R&D investment is treated as revenue expenditures and charged to expenses in the period in which they are incurred. At last, Ho et al. (2005) find that R&D investment in the year in progress contributes positively to the one year market performance.

Internationalization

These two variables are highly correlated (r =0.91) and therefore added into one equation, namely, the degree of internationalization. Internationalization is expressed in a range taking values from 0 to 1, with 1 representing the highest level of internationalization.

Control variables

Despite the fact that I am interested in the change in market value associated with R&D expenditures, it is important to control for other factors having an impact on market value. Therefore, the model build in this study includes control variables which prove to have a significant effect on return for R&D firms and are widely discussed in the R&D efficiency literature. The control variables are represented by Zi,j in Equation 1 and 2. This is needed to control for variables mainly driving

return in R&D firms and to build a uniform model for measuring R&D efficiency. Throughout the analysis, the main focus is on R&D efficiency and the interaction term, captured by the coefficient of the R&D ratio and interaction term. The model specification takes into account six control variables which affect AR for R&D firms.

Country, industry dummies are included, to control for industry- and country wide effects, as previous research find that AR for R&D firms differed per country and industry (Duqi & Torlucci, 2010; Hall & Oriani, 2006; Ho et al., 2005; Stark &Thomas, 1998; Green et al., 1996). Two industry dummies are created. First, industry is categorized in low, medium and high tech to check the industry level of innovation expected from each firm and control for industry effects. This classification is made using OCSE (STAN ABERD) which is R&D expenditure divided by sales of the firm industry sector. Second, the other industry dummy variable is based on two-digit SIC code to capture more industry-wide effects.

Next to the type of industry a firm belongs to, size plays a significant role in its valuation by the market as previous studies have highlighted. The innovation in small firms is given higher stock market valuations as it reflects growth potential as these firms have a sort of flexibility (Knight & Cavusgil, 2004). An argument for this is that small firms invest a significant proportion of their turnover in innovation and are riskier (Tutticci et al., 2007). However, in large companies return on sales plays a more important role than innovation (Meyer, 2011; Duqi & Torlucci, 2010; Lewin & Massini, 2003). Nevertheless, as the sample consists of firms which report R&D investments, I expect a negative relation between AR and size. LN_sales represents the size as it is the logarithm of annual firm sales.

balancing costs and benefits. The pecking order theory is related to asymmetric information and is related to the fact that insiders know more about the firm and their risk than the investor therefore debt is preferred over equity (Fama & French, 2002). The financial flexibility theory argues that small firms and research intensive firms maintain low leverage because equity financing allows firms to raise cash without impeding financial flexibility (Powers & Tsyplakov, 2008). The theory implies that firms with R&D investments should have less leverage as high leverage leads to pre-committed future cash flow. This is supported by the financial flexibility theory and trade-off theory. On the other hand, if a firm is highly leveraged it can lead to higher abnormal returns and more institutional oversight known as pecking order theory. In the main, innovative firms have more options to risk capital in a period of economic downturn because future pay-offs are unknown (Opler & Titman, 1994). Szewcxyk et al. (1996) estimates, that abnormal return of firms that invest in R&D are higher when they are highly leveraged. Therefore, a positive relation between leverage and AR exists, whereby leverage is a control for risk.

The control variable, earnings before interest and tax (Ebit), divided through total assets (TA) is positively related to market value. Therefore, Ebit is included in the model as it is an explanatory variable for stock return. Finally, the sample consists of a time span which includes the crisis. In times of crises investors take a more skeptical view of the stock market and stock prices go down (Meyer, 2011). A negative relationship between AR and crisis is expected. Therefore, a dummy for the crisis is added. The explanatory variables and control variables with their expected signs are presented in Table 3.1 below.

Variable Abbreviation Variable description Expected sign

Dependent variable

Return The annual buy-and-hold abnormal return (dividend adjusted) for firm i during six month prior to and six mother after a firm‟s fiscal year, adjusted by the market return during the same period.

Independent variables

RD RD ratio is the research and development expenditure of firm i at time t divided by total assets in the year in progress.

+

Internat. Number of foreign subsidiaries of each firm divided to total foreign subsidiaries of the whole sample plus number of foreign countries of each firm divided by total foreign countries of the whole sample.

At last, the average of both measurements.

Interaction A product of the R&D ratio and

internationalization dimension. ?

LN_Sales Size Logarithm of annual firm sales. -

Leverage (%) Lev Firm “Debt to equity ratio”. +

Ebit Earnings before interest and tax divided by total assets in the year in progress.

+

Dummy crisis Crisis Before (0) and during (1) the crisis.

-Table 3.1 The proposed variables and their expected signs

4 DATA AND DESCRIPTIVE STATISTICS

The dataset includes a sample of 123 industrial listed firms from newly industrialized countries that operated continuously from the year 2005 to 2010. The firms considered, have recorded R&D expenditure in all the consecutive years of the selected periods. The data selection and descriptive statistics are presented in this section.

4.1 Data

The emerging economies considered in this paper are: Turkey, Malaysia, India and China. The remaining newly industrialized countries do not have the required sample size to be eligible to be considered. As mentioned previously, it is under researched how the financial markets of newly industrialized countries perceive investments in intangibles, wherein intangibles are specified as R&D expenditures. Furthermore, empirical evidence is lacking on the impact of internationalization on R&D efficiency in emerging economies.

This empirical research focuses solely on listed industrial firms in newly industrialized countries, as specified by the IMF, having a market value for at least six consecutive years (2005-2010). Accounting data and information concerning subsidiaries is derived from Bureau van Dijk (ORBIS)3. Market value data is derived from DataStream. Table 4.1 in Appendix II lists the sources of all the variables. The final sample contains 123 firms of which 21 are Turkish, 39 Malaysian, 15 Indian and 48 China. The sample integrated only these firms that recorded at least in one of the

3

selected periods (2005-2010) R&D expenditures. There are 681 observations distributed over six years and four countries.

A limitation of this paper is related to the size of the stock markets, which is intrinsic to the market themselves, and the fact that the sample is selected based on R&D availability. Therefore, the results should be interpreted with a word of caution as it does not represent the population of the whole sample. Therefore, I assess how the stock markets reward R&D investment. The sample requires a deeper investigation of individual countries, different periods, such as before and during the crisis and by industrial classification, as is further developed.

4.2 Descriptive statistics

The descriptive statistics for the key variables are divided in summary and subsample statistics. The summary statistics are discussed first. Previous research indicates that crisis; type of industry, and country of origin can have different effects on R&D efficiency. Therefore, the total sample is split into subsamples per country, period, and industry classification which are analyzed in the subsample statistics.

Summery statistics

Table 4.2 presents descriptive statistics and Table 4.3 in Appendix III a Pearson correlation matrix of all the variables. The descriptive statistics (Table 4.2) extant a much lower median than mean for the abnormal return indicating an asymmetry in the distribution of the data. An explanation for asymmetry is that the abnormal return shows a maximum of 10.10 although the mean is 0.24, the sample is skewed. Therefore in the robustness checks, further on, I leave these outliers out to see if the results are driven by these outliers. I find the same asymmetry for R&D investments, internationalization and the interaction term, as the mean is greater than the median.

internationalization and R&D expenditures, in the correlation matrix, indicate that the degree of internationalization has a negative effect on R&D efficiency. The result provides preliminary evidence to support H2, that internationalization has an effect on R&D efficiency.

Variable Mean Std. Dev. 1st quartile Median 3rd

quartile Maximum Minimum

Abnormal return 0.24 0.91 -0.22 0.05 0.43 10.10 -1.20 R&D Expenditures 1.23 3.36 0.10 0.36 1.09 35.90 0.00 Internationalization 0.81 1.61 0.00 0.25 0.92 9.14 0.00 Interaction (R&D*internat.) Size 0.82 12.82 2.47 2.05 0.00 11.32 0.03 12.66 1.01 14.42 24.06 19.37 0.00 7.34 Leverage 1.54 1.95 0.46 0.98 1.76 26.36 0.05 Ebit 7.79 9.35 3.28 7.18 12.03 54.74 -64.20 Crisis 0.69 Industry Low-tech 0.35 Medium-tech 0.11 High-tech 0.54 Country China 0.38 India 0.12 Malaysia 0.31 Turkey 0.18 Observations 681

Table 4.2 Descriptive Statistics of the full sample

Summery statistics of the subsamples

previous findings of Bae et al. (2008) and Lu and Beamish (2004) where firms with greater R&D efficiency tend to be more multinational. Hence, theory gives different implications. This makes the direction of the relation uncertain as stated in H2.

The distributions of industry reflect the different industrial structure of listed firms within the countries. Approximately two third of the observations in the China sample (65%) and more than a half in the Malaysia sample (56%) is concentrated in the high-tech industry. Although, China and Malaysia present a high concentration of high-tech firms it does not necessarily lead to higher return. This is different from the findings in Appendix IV, Table 4.6, as is explained further on. Substantial shares of the observations for India are half in the low (50%) and nearly a half (42%) in the high tech industry. However, Turkey presents a more even distribution among the classification of low, medium and high-tech industries. Size, as one of the main explanatory variables is more similar among the countries. Furthermore, the subsample per country illustrate that the median leverage in China and India is higher in comparison to Malaysia and Turkey (Appendix IV: Table 4.4 & 4.5).

5 RESULTS AND FINDINGS

In this section I test the hypotheses proposed in Section 2 and analyze the results. Because the focus in this study is on the under researched sample of emerging economies, I first report the results of the regression about the effect R&D expenditure has on market value for the whole sample. Second, I present the effect of internationalization on R&D efficiency. Third, I evaluate the R&D efficiency per country and the impact of internationalization thereon. Fourth I analyze if R&D efficiency exists during a period of crisis. Finally, I examine the robustness of the results.

5.1 Evidence of R&D efficiency

H1 is tested with Equation 1 and the results are displayed in model 1 to 13. The results for the estimation of model 1 to 12 are presented in Table 5.1 and the result for model 13 in Appendix V. Buy-and-hold AR is the dependent variable for the thirteen models. Model 1, is the baseline model, one which is not corrected for industry and country fixed effects and excludes the control variables. H1, R&D efficiency is strongly supported by model 1 (0.03**).

The models 2 to 13 are as well built on Equation 1. Model 2 includes the internationalization dimension, which indicates a significant negative effect on AR (-0.02***). The impact of internationalization is explained below. Model 3 includes almost all the control variable and model 4 includes all the firm specific control variables. I include these control variables to check if the coefficients and significance levels change for R&D and internationalization. In both models only size has a significant negative impact (-0.04** and -0.03*) on abnormal return. One reason may be that firm size reflects inflexibility as larger firms are characterized with more bureaucracy. A negative relation between size and abnormal return indicates that investors value smaller firms more optimistically due to the fact that they are potentially more flexible and risky. The conclusions concerning R&D efficiency do not alter when the control variables are added in model 3 and 4. Coefficients and significance levels only change marginally. Striking is the fact that internationalization loses its significance in model 4. So, changes in AR are driven by other factors than internationalization.

In model 6, I correct for industry fixed effects, I differentiate between low, medium, and high tech industry. In this way I correct for industry specific shocks that might have an effect on AR. The inclusion of these industry effects does not change R&D efficiency and the results stay significant. So, the support for H1 remains. Previous research has proven that the high tech industry affects firm performance the most (Kafouros, 2005; Gustavsson et al., 1999; Pegels & Thirumurthy, 1996). On the other hand, Ballardini et al. (2005) and Chan et al. (2001) argue that the first costs creating that enterprise value do not out weight expenditures in innovation generating that market value. Thereby, arguing that firms in the high or low-tech with different business models generate the same return. To test the robustness of the result in model 6, a second regression is done including 2digit-SIC code for the industries and exclude the variable low and high tech. The result of the regression show that agricultural production-crops (0.66**), engineering, accounting and research, management and other related services (0.43**) and measuring, analysing controlling instruments, photographic, medical and optical goods, watches and clocks (0.72*) affect firm performance positively (Appendix V). According to Yin (2011) these industries are reported as low and high-tech, confirming the arguments of Ballardini et al. (2005) and Chan et al. (2001). In other words, high-tech firms do not necessarily affect firm performance the most and have a bigger impact on return. However, the result should be interpreted with a word of caution as the sample size might be insufficient for some industries. If a comparison is made between model 6, 12 and 13 it can be argued that that the same significant results are find for R&D efficiency. Therefore, it does not matter if I define industry by a 2digit-SIC code or low, medium and high-tech. Concluding, the coefficient concerning R&D efficiency do not alter when a different industry dummy is added. Coefficients and significance levels only change marginally. Therefore, in the regressions for H2, H3 and H4, I neglect the 2digit-SIC code and control for low, medium and high-tech, adding low, medium and high-tech dummy increases the explanatory power (R-squared) and goodness of model fit (F-statistic).

Models 7 to 10 present four regressions which include in each regression a country dummy. China, India and Malaysia have a significant effect on market value. Whereof India (-0.42**) and Malaysia (-0.29***) show a negative and China (0.46***) a positive effect. The result for firms in Turkey is not significant. Therefore, in model 11 China, India and Malaysia are regressed together. When I control for country fixed effects in model 11 and country and industry fixed effects in model 12, I still witness a linear positive relationship between R&D and Return (0.02**) and (0.02**). The adjusted R-squared and F-statistic of Model 11 and 12 are substantially high, indicating that the regression models, including all the control variables, impose a better fit.

Abnormal Return Independent Model Model Model Model Model Model Model Model Model Model Model Model

variables 1 2 3 4 5 6 7 8 9 10 11 12 Intercept 0.21*** 0.23*** 0.66*** 0.63*** 0.81*** 0.39** 0.74*** 0.66*** 0.99*** 0.63*** 0.94*** 0.83*** (17.24) (27.31) (2.92) (2.61) (3.46) (1.93) (3.35) (2.83) (4.19) (2.62) (4.23) (5.79) R&D Expenditures 0.03** 0.03** 0.02** 0.02** 0.02** 0.02** 0.02** 0.03** 0.02* 0.02** 0.02** 0.02** (2.54) (2.52) (2.19) (2.19) (2.17) (1.99) (2.05) (2.32) (1.87) (2.22) (2.17) (2.12) Internationalization -0.02*** -0.01 -0.01 -0.01 0.01* 0.00 -0.00 -0.01 0.02*** 0.02*** (-2.98) (-1.02) (-0.52) (-1.25) (1.67) (0.67) (-0.44) (-1.06) (3.36) (4.28) Size -0.04** -0.03* -0.03* -0.02 -0.06*** -0.04* -0.05*** -0.03* -0.07*** -0.06*** (-1.98) (-1.68) (-1.74) (-1.04) (-2.95) (-1.84) (-2.82) (-1.65) (-3.40) (-3.82) Leverage 0.00 0.00 0.00 -0.00 0.00 0.01 -0.00 0.00 0.00 0.01 (0.34) (0.29) (0.19) (-0.08) (0.18) (0.76) (-0.06) (0.28) (0.39) (0.46) Ebit 0.00 0.00 0.00 0.00 0.01 0.01 0.00 0.00 0.01* 0.01* (0.63) (0.64) (0.44) (0.71) (1.62) (1.29) (0.83) (0.61) (1.84) (1.88) Crisis -0.26*** (-3.54) Industry Low-tech -0.03 0.10 (-0.43) (1.37) High-tech 0.16 0.14 (1.54) (1.26) Country China 0.46*** 0.31* 0.29* (2.65) (1.89) (1.76) India -0.42** -0.35*** -0.36*** (-2.40) (-2.77) (-2.75) Malaysia -0.29*** -0.18 -0.20* (-3.03) (-1.63) (-1.68) Turkey -0.05 (-0.52) Observations 685 685 681 681 681 681 681 681 681 681 681 681 Adjusted R-squared 0.054 0.055 0.058 0.057 0.025 0.062 0.113 0.075 0.075 0.056 0.120 0.120 F-statistic Durbin-Watson 7.60*** 2.03 6.67*** 2.03 5.63*** 2.05 5.09*** 2.05 3.94*** 2.10 4.74*** 2.06 8.85*** 2.18 6.02*** 2.10 5.98*** 2.09 4.65*** 2.05 8.13*** 2.20 7.14*** 2.21 Year fixed effects4 in all models except model 5. Robust t-statistics in parantheses, corrected for cross-sectional heteroskedasticity and correlation.

***/**/*indicates significance at 1%, 5%, and 10% level (2-tailed), respectively.

Table 5.1 Results of regression analysis of R&D efficiency

5.2 Internationalization negatively affects R&D efficiency

Theory implicates that internationalization has an effect on R&D efficiency. Bae et al. (2008) and Lu and Beamish (2004) find empirical evidence that internationalization has a positive effect on R&D efficiency, only they focused on one developed country. As results can differ, more research is needed on the impact of internationalization on R&D efficiency in different benchmarks. Therefore, the second hypothesis that is regressed and analyzed includes a product of degree of internationalization and R&D expenditures to abnormal return of a bundle of countries in emerging economies.

H2 is regressed using Equation 2. The product of degree of internationalization and R&D efficiency, from now on the interaction term, is the main focus and therefore of considerable importance. The regression also controls for possible market-wide effect, as crisis, through the use of year fixed effects (not shown in Table 5.2). However, my results deviate from previous research that focus on developed countries as the opposite is true for emerging countries. A higher degree of internationalization and more R&D lead to lower abnormal return. H2 is accepted as internationalization has a significant effect on R&D efficiency. The adjusted R-squared and F-statistic of Model 10 and 11 are substantially high, indicating that the regression models including all the control variables, impose a better fit in comparison to the other models in Table 5.1 and 5.2. In Table 5.2, models 1 to 12, present a strong significant negative impact between the interaction term and abnormal return for firms investing in R&D. Possible explanations for the fact that results in emerging economies differ with developed countries and suggestion for future research are denoted next.

Abnormal Return Independent Model Model Model Model Model Model Model Model Model Model Model Model

Year fixed effects5 in all models. Robust t-statistics in parantheses, corrected for cross-sectional heteroskedasticity and correlation.***/**/*indicates significance at 1%, 5%, and 10% level (2-tailed), respectively.

Table 5.2 Results of regression analysis of degree of R&D efficiency on internationalization

Literature states that FDI brings improved technology and access to international networks, all of which increase productivity and growth (Castellani & Zanfei, 2007; Barkema & Vermeulen, 1998; Barlett & Ghoshal, 1990). Geographical dispersion leads to exploitation and exploration benefits as economies of scale and scope. Although, Morck and Yeung (2003) highlighted the dimension that intangible assets as R&D investment are informative intensive. They denote that internationalization can have a negative influence on return, as processes for innovation are not internalized. Concluding, as emerging economies are characterized with lack of international experience and a low resource base it can be more difficult to internalize innovative processes by developing a global network for sharing knowledge or extending their domestic innovation process to foreign operations. Future research is needed in the internalization of innovation processes. Moreover, as firms internationalize they should take advantage of their unique resource which is a cost advantage rather than an innovation advantage in emerging economies. Therefore, the negative relation of the interaction term is supported by the resource-based theory where managers are better off focusing on their core-resources. Hence, R&D investment and multinationality has a negative impact on market value. Rather, a firm‟s degree of multinationality has a positive impact on market value.

Since, H2 is accepted as Table 5.2 proposes a negative relationship the sample is split according to the median of the degree of internationalization into two subsamples. The regression also controls for crisis and possible market, industry, and country-wide effect through the use of year fixed effects, industry and country dummies (not shown in Table 5.3). To examine if the coefficient for R&D is different for more and less internationalized firms. The subsample partly accepts H2 (Table 5.3). The regression shows that for less internationalized firms investors value R&D efficiency positively (0.09***), but is insignificant for highly internationalized firms.

This phenomenon of degree internationalization can be explained in the light of literature about size. Meyer (2011) and Duqi and Torlucci (2010) and Lewin and Massini (2003) state that: “in the valuation by investors for large firms, return is more important than R&D”. This is confirmed for high internationalized firms, as earning instead of R&D investment has a positive effect on AR. Small firms are given higher stock valuation as they invest a large proportion in innovations, which is risky and reflects growth potential to the investor (Tutticci et al., 2007). Table 5.3 confirms that for less internationalized firms R&D efficiency is strongly positive. In other words, firms whose operations are geographically diversified are not able to fully benefit from their R&D investment.

5

Another explanation is that less internationalized firms have a sort of flexibility (Knight & Cavusgil, 2004). For less internationalized firms, in comparison to more internationalized firms, less bureaucratic layers are needed to internalize information. Intangible assets as R&D investment are information intensive and therefore hard to internalize (Morck & Yeung, 2003; Kotabe et al., 2002) Therefore, it can be that high internationalized firms in emerging economies, who are in the early stage of internationalization, did not have the time to internalize processes of innovation. Because processes for innovation need to be internalized and are time consuming in large and complex organizations. Concluding, a higher degree of internationalization can lead to more time before R&D investment affect financial performance through operational performance.

AR AR

Independent

variables High internationalization Low internationalization

Intercept 0.46* 0.72*** (1.88) (2.91) R&D Expenditures 0.01 0.09*** (1.07) (3.91) Size -0.04** -0.05** (-2.07) (-2.40) Leverage 0.02 -0.01 (0.86) (-0.28) Ebit 0.02* 0.00 (1.83) (0.90) Observations 296 385 Adjusted R-squared 0.164 0.112 F-statistic Durbin-Watson 5.13*** 2.00 4.47*** 2.35

Year fixed effects6, industry and country dummies in all models. Robust t-statistics in parantheses, corrected for cross-sectional heteroskedasticity and correlation.***/**/*indicates significance at 1%, 5%, and 10% level (2-tailed), respectively.

Table 5.3Results of regression analysis of R&D efficiency for more and less internationalized firm

H2 is accepted. Moreover it is proven that the product of degree of internationalization and R&D expenditure has a negative influence on abnormal return for R&D intensive firms in newly industrialized countries. In the subsample for less internationalized firms R&D efficiency is positively valued whereas for more internationalized firms this is not significant.

5.3 The existence of R&D efficiency per country

Differences in R&D efficiency per country can be expected. Moreover, previous researcher has already proven that country of origin has an impact on R&D efficiency (Duqi & Torlucci, 2010;

6

Hall & Oriani, 2006; Hofstede, 2010, 2002, 2001 1988). To further analyze the country and crisis effects, firms are analyzed before and during the crisis according to their country of origin in Table 5.4. Methodology applied is panel analyses, whereby corrected for year fixed and industry effects (not shown in Table 5.4).

Only China (0.18***) and India (0.10*) show a significant positive linear relation for R&D efficiency before the crisis. A possible clarification for this is that multinationals investments in knowledge creation and high technology takes place in emerging countries as China and India which are according to Santos-Paulino et al. (2008) seen as two of the top ten destinations for foreign R&D expansion. These countries are characterized by well-trained engineers and scientists and endowment of low-cost which is complemented with fast growing domestic market and increasing foreign direct investments (Sun et al., 2007).

During the crisis R&D efficiency was only significant for China and Malaysia. It may be explained by the fact that in the past, Malaysia lacked high-skilled workers and was incapable of commercializing R&D outputs. The government started a second National Science and Technology Policy in 2002 to stimulate R&D investments and improve capacity and capability needed for R&D. However, these investments did not show an increase in R&D investments, only some laudable patents (Govindaraju & Wong, 2011). Due to the poor causal link among institutions promoting research and failure to expand their own capabilities in process and product technologies Malaysia has limited success. However, my result shows that R&D investment in the period during the crisis has a positive impact on R&D investment. The fact that R&D efficiency exist during the crisis can indicate that policy reforms affect not the real side but the financial system in Malaysia as policy reform yield a strong effect on the behaviour of investors. Unlike the other countries, China shows R&D efficiency in both periods. A justification can be that China is less affected by the crisis as government stimulated domestic demand during the period of crisis and China has a big home market which makes it attractive to let R&D investment go on (Bruche, 2009).

India did not show any significant relationship during the crisis. The paper of Das et al. (2011) describes the decomposition of India‟s GDP growth rate, quarterly from 2007 to 2009. According to Das et al. (2011) the manufacturing industry presents a sharp decline in the growth rates of real GDP, from 8.5% to -0.5%, which becomes even negative in 2009. The growth in the service industry on the other hand decreases only little, from 10.8% to 8.4%. As most R&D expenditures are in the industrial sector and not in the service sector, the enormous decreases in real GDP growth for industrial sector might explain why I did not find any significant results for India. The negative sentiments concerning real GDP consequentially suggests that investors do not know how to value R&D investments.

their own (Huq, 2004). Second, Turkey has a low level employment (Cetindamar et al., 2006). Therefore, it can be argued that Turkish firms are weak in adopting and implementing R&D. This weak absorptive capacity seems to be valid in the sample of Turkey as firms that invest in R&D are valued negatively. Concluding, investors are sceptic about R&D investments that may be due to the fact that most firms in Turkey do not have the institutional structure and resources for R&D, there is likely to be R&D inefficiency.

Concluding, H3 is partly confirmed as in some countries R&D efficiency exists and others not. Interesting is that Turkey deviates from others as it presents R&D inefficiency. Different R&D efficiency per country can be mainly addressed to different stages of development and the institutional context (Beine & Candelon, 2011; Tsang et al., 2008; Huq, 2004). Therefore, government policy can exerts strong effects on behavior of investors towards R&D efficiency.

Abnormal Return Abnormal Return

Before crisis '05-'06 During crisis '07-'10 Independent Model Model Model Model Model Model Model Model variables China India Malaysia Turkey China India Malaysia Turkey Intercept 2.55*** -6.12*** -0.53* 1.62*** 0.54*** -0.69 -0.31 0.19 (6.60) (-6.86) (-1.97) (22.42) (4.18) (-1.62) (-0.85) (0.45) R&D Expenditures 0.18*** 0.10* 0.09 -0.54*** 0.09* 0.01 0.10* -0.00 (2.71) (1.88) (0.32) (-7.51) (1.90) (0.14) (1.82) (-0.04) Internationalization 0.15 -0.08 0.09*** -0.19 0.05 -0.04 0.04*** 0.06 (1.61) (-1.75) (12.71) (-1.10) (0.81) (-1.07) (7.51) (0.79) Interaction -0.20*** -0.08*** -0.41** -0.09* -0.01 -0.01 -0.28*** -0.11 (R&D*internat.) (-5.05) (-3.53) (-2.40) (-1.81) (-0.13) (-0.13) (-5.40) (-1.00) Size -0.22*** 0.48*** 0.01 -0.05*** -0.04* 0.03 0.00 -0.01 (-23.13) (6.17) (1.54) (-3.22) (-1.78) (1.24) (0.01) (-0.27) Leverage 0.19*** 0.23*** 0.09 -0.13 -0.01 0.03 -0.00 -0.00 (3.35) (78.60) (0.89) (-0.50) (-0.94) (0.57) (-0.12) (0.24) Ebit 0.07*** 0.00 0.01** 0.01*** 0.00 0.01 0.01* 0.00 (4.52) (0.11) (2.19) (2.54) (1.14) (1.04) (1.66) (0.58) Observations 83 26 65 40 178 56 149 84 Adjusted R-squared 0.224 0.392 -0134 0.286 0.247 -0.011 0.094 -0.038 F-statistic Durbin-Watson 3.62*** 2.48 2.79** 3.56 0.16 2.73 2.74** 2.61 6.26*** 2.36 0.95 2.27 2.40*** 2.67 0.72 2.83 Year fixed effects7 and industry dummies in all models. Robust t-statistics in parantheses, corrected for cross-sectional

heteroskedasticity and correlation.***/**/*indicates significance at 1%, 5%, and 10% level (2-tailed), respectively.

Table 5.4 Results of regression analysis of R&D efficiency per country, including periods

Besides the fact that R&D efficiency differs, other explanatory variables differ per country as is explained further on. First, I explain the interaction and internationalization variable. Only the Malaysian sample confirms a positive and significant relation between internationalization and

7

abnormal return before and during the crisis. Possible explanations can be that in Malaysia the government played an important role in stimulating internationalization by National Economic Policy and the Malaysian Development Plans. In these plans, Malaysian firms are stimulated to produce high quality and value added products for the domestic and export market with the use of modern technology and management (Chelliah, 2010). Furthermore, Malaysian firms tend to internationalize to markets which are almost similar to them in cultural aspects and this enables easy market penetration (Hofstede et al., 2010). These factors can partly explain the positive relationship between internationalization and AR in Malaysia.

The product of internationalization and R&D, known as the interaction variable, indicates that internationalization has a negative effect on R&D efficiency for firms in China, India, Malaysia and Turkey. As stated earlier, building a resource base, such as technological resources, is a stage-dependent process, requiring extensive effort and time (Oviatt & McDougall, 2005). Second, newly industrialized countries lack experience in foreign markets. Another factor which can strengthen the negative effect of the interaction variable is that newly industrialized countries have a cost advantages vis-a-vis developed economies (Aulakh et al., 2000). According to the resource-based theory companies have to prioritize their investments as to reap benefits from their core-competence. The core-competence of these firms lies in the cost advantages not in the differentiation strategy by R&D investment. R&D investment is not a crucial resource in this case. Firms in emerging economies, which internationalize can better compete on cost-leadership strategy rather than investing in R&D. The results of the interaction term strongly support the resource-based and internationalization theory.

Additionally, Table 5.4 presents a negative relation between size and abnormal return before and during the crisis for China and before the crisis for Turkey. This confirms previous empirical evidence that small firms have higher returns to compensate their associated risk. India indicates a positive relationship between size and abnormal return. According to Majumdar (1997), this can be due to the industrial policy instruments and framework of the Indian economy. The government did not create incentive for survival in a competitive environment, leading to less motivation to earn superior profitability. Therefore, large firms had a rent-seeking perspective, which is unproductive profit seeking. In other words large firms are indicated as more profitable but less productive which can explain the positive relationship between size and AR.

abnormal return, an explanation can be that highly leveraged firms suffer more in periods of economic downturn. Probably, because they are unable to go on with their R&D investments due to a decrease in the cash flow (Opler & Titman, 1994). Earnings are positively valued, which is in line with previous theoretical predictions.

Overall, the results should be interpreted with a word of caution as the sample size is small. Whereas, the sample size might be insufficient, results can be driven by outliers. Moreover, some models do not indicate a goodness of model fit as they have no significant F-statistic. Despite this fact, it is interesting to adhere to the differences per country. Differences per country can be mainly addressed to different stages of development and the institutional context. Moreover, the fact that R&D efficiency is only affirmed by the Chinese sample in both periods may be explained by the insufficient size of the other country samples. For that reason, future research is needed, using a bigger sample size.

5.4 R&D efficiency exists during crisis

In Section 4, the descriptive statistics, I find implications, which advocate that R&D investments go on during a period of crisis and can be positively valued, adhering to H4. Table 5.4 provided partly evidence as, only in China and Malaysia, R&D is valued during the crisis. The regression controls for possible market, industry and country-wide effect through the use of year fixed effects, industry and country dummies (not shown in Table 5.5).