The environmental actions of firms: Examining the role of spillovers, networks and absorptive capacity

Journal of Environmental Management 146 (2014) 150e163

Contents lists available at ScienceDirect

Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman

The environmental actions of firms: Examining the role of spillovers, networks and absorptive capacity* Facundo Albornoz, Matthew A. Cole, Robert J.R. Elliott*, Marco G. Ercolani Department of Economics, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK

a r t i c l e i n f o

a b s t r a c t

Article history: Received 9 December 2013 Received in revised form 6 July 2014 Accepted 8 July 2014 Available online

In the light of climate uncertainty and growing concern for the natural environment, an increasingly important aspect of global business is the environmental behaviour of firms. In this paper we consider the factors that influence firms' environmental actions (EAs). Our study of Argentinean firms concentrates on measures of environmental spillovers, informal and formal networks and absorptive capacity by testing four related hypotheses. We find that foreign-owned firms, large firms and those with a greater capacity to assimilate new environmental technologies are more likely to adopt EAs. We also show that formal and informal networks aid the adoption of EAs in the presence of traditional firm-level spillovers. Finally, we show that foreign-owned firms have different motives to domestic firms for undertaking EAs. © 2014 Elsevier Ltd. All rights reserved.

JEL classification: D21 Q20 Q56 Keywords: Multinational Environment Firm characteristics Management Motives

1. Introduction Understanding the factors that influence the environmental actions (EAs) of firms is of great importance to policy makers and business leaders attempting to control the environmental impact of the manufacturing sector. Industrial emissions pose a particular problem in developing and newly industrialising economies where concentrations of local air pollutants still regularly exceed World Health Organisation (WHO) guidelines. Estimates from the WHO (2006) are that urban air pollution kills more than a million people each year, predominantly in developing countries. Moreover, according to the WHO, over 80% of all diseases are wholly or partially attributable to environmental factors. For policymakers to lessen the environmental impact of industrial activity, a detailed understanding of the environmental actions of firms is required together with an understanding of what motivates firms to undertake actions that are likely to be beneficial to the environment.

* We gratefully acknowledge the support of Leverhulme Trust grant number F/ 00094/AG. * Corresponding author. Tel.: þ44 121 4147700; fax: þ44 121 414 7377. E-mail address: [email protected] (R.J.R. Elliott).

http://dx.doi.org/10.1016/j.jenvman.2014.07.011 0301-4797/© 2014 Elsevier Ltd. All rights reserved.

In a world that is increasingly concerned about the economic and social impact of anthropogenic climate change, foreign firms have a potentially pivotal role in the development of a sustainable future.1 In this paper we ask four distinct but related research questions centred on the ability of domestic firms in a newly industrialising country to learn and benefit from the EAs of foreign firms. To validate our research questions it is important that, firstly, foreign firms have superior environmental practices to their domestic counter-parts and, secondly, that domestic firms are able to learn from foreign firms to enable them to subsequently increase the number of environmental actions that they undertake. The key research questions are as follows. First, we ask whether foreign firms are more likely to adopt EAs than domestic firms. If we can confirm this relationship, we should then able to ask more detailed questions concerning the ability of domestic firms to learn from the EAs of foreign firms. We then investigate the role played by the absorptive capacity of domestic firms and how this influences their

1 In this paper we use the term environmental actions (EAs) rather than environmental management systems (EMS) which refer to a more systematic approach to dealing with a firm's environmental activities.

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

ability to learn. Next, we investigate the importance of formal and informal networks between firms and their peers, suppliers and customers, on the propensity of domestic firms to implement new green management and production technologies. Finally, we consider what motivates firms to take EAs and investigate why motives may differ across firms. A number of previous studies have argued that foreign owned firms in developing economies are likely to have superior environmental practices than domestically owned firms (Zarsky, 1999; Cole et al. 2008) although not all studies agree on this point (Eskeland and Harrison, 2003). However, little has been said on the issue of environmental spillovers from foreign to domestic firms, the possible channels through which such spillovers might occur and the potential roles of absorptive capacity and networks. Furthermore, while existing studies identify some of the characteristics of firms that undertake environmental actions (Cole et al. 2006; Nakamura et al. 2001), little has been said specifically about firms' motives for undertaking environmental actions. We therefore believe this study fills a significant gap in the literature and adds considerably to our understanding of firms' environmental behaviour.2 To address the main research questions we apply techniques from the traditional productivity spillover literature to the environmental behaviour of firms, using a highly detailed firmlevel dataset for 1087 firms in Argentina. We choose to focus on Argentina since it represents a middle-income economy facing the typical environmental challenges associated with rapid economic growth. Argentina is also a significant recipient of foreign direct investment and it undertook substantial liberalisation reforms in the 1990s making it an ideal country in which to examine the potential influence of foreign presence on firms' environmental actions. We hope our findings will be applicable to other outward-oriented middle-income economies. The richness of the data permits us to build on the existing literature to test a series of hypotheses not previously testable due to data limitations. Our analysis focuses on whether, on average, foreign firms undertake more environmental actions than domestic Argentinean firms.3 More specifically, we are interested in firms' ability to learn, as measured by their absorptive capacity. This, coupled with the existence of formal and informal networks, might determine how important these mechanisms are for good environmental practice to be transferred from foreign to domestic firms. Finally, we hope to show to what extent the motivation to adopt certain environmental practices differs by ownership and if local conditions play an important role in encouraging foreign firms to improve their environmental behaviour. The remainder of the paper is organised as follows. In Section 2 we explain the relationship between ownership and the environmental practice of firms and present our four testable hypotheses. In Sections 3 and 4 we describe the data and present our results. Section 5 concludes.

2 Although space does not permit a wider discussion, it should be noted that in the political ecology literature the integration of developing countries into global markets is seen as part of an unequal struggle that could result in local land owners degrading their environment (Blaikie and Brookfield, 1986). See Newall (2012) for a review of the political ecology debate and the power relationships between global ecology and the global economy. 3 A lack of data on the actual environmental performance of firms (e.g. pollution emissions) means that we have to focus on environmental actions such as whether or not firms have taken actions to improve the efficiency of natural resource use or whether they have adopted environmental certification (these actions are defined fully in Section 3). We acknowledge that the precise link between such actions and actual environmental performance is unclear.

151

2. Theory and hypotheses 2.1. Environmental spillovers In order to investigate the effect of foreign ownership on the adoption of EAs we have to establish the theoretical mechanisms through which foreign-owned firms may influence domestic environmental practices. Clearly, this requires an understanding of firms' learning processes and the manner in which knowledge is disseminated across firms. For domestic firms to learn and benefit from the good environmental practices of foreign firms two conditions must hold: Condition 1: Foreign firms must have superior environmental practices to domestic firms. Condition 2: Domestic firms must have the ability to learn from foreign firms. The suggestion that domestic firms can learn from foreignowned firms is implicit in the so-called ‘pollution halo’ hypothesis.4 The pollution halo hypothesis argues that if multinationals utilise more advanced technologies, cleaner production methods, and possess more developed environmental management systems (EMS) and organisational techniques, then these may yield substantial environmental benefits to developing countries. For example, it has been argued that OECD based multinationals typically utilise cleaner technologies and possess more sophisticated EMS than many domestic firms in developing countries motivated, it is argued, by a more stringent regulatory environment in the OECD (Zarsky, 1999).5 Multinationals may also feel pressured to continue to use such technologies in their overseas affiliates because a percentage of production may be exported back to OECD markets where the requirements of environmentally aware consumers must be met. Both Wallace (1996) and Zarsky (1999) note that such technologies may also be indirectly passed on to domestic firms via relationships with customers and suppliers. With the above in mind, we therefore assess whether Condition 1 holds by testing Hypothesis 1. Hypothesis 1. Foreign-owned firms are more likely to adopt EAs than domestic firms. If Condition 1 is empirically verified, then there is the potential for a flow of information from foreign to domestic firms in the form of positive environmental spillovers. These may arise for a variety of reasons: First, they may arise due to workers moving from foreignowned to domestic firms and bringing their experience and expertise with them. Second, domestic firms may adopt technologies utilised by foreign-owned firms through imitation or reverse engineering. Finally, spillovers can move up or down the supply chain if foreign-owned firms, concerned about their public image, require their suppliers or customers to adopt certain minimum environmental standards. This is more likely between foreign firms and their suppliers. A number of studies have examined the existence of economic spillovers, typically in the form of productivity improvements, from

4 The ‘pollution halo’ hypothesis contrasts with the ‘pollution haven’ hypothesis which argues that multinational firms may choose to locate in a developing country or region to take advantage of less stringent regulations and transfer environmentally inferior technologies and practices to their foreign affiliates or use these affiliates to market products that are banned or restricted in their home countries (Ives, 1985). The empirical evidence for the pollution haven hypothesis is mixed (see for example Eskeland and Harrison, 2003; Smarzynska-Javorcik and Wei, 2004; Cole and Elliott, 2005). 5 Christmann and Taylor (2001) argue that global ties themselves can increase self-regulation pressures on firms in low regulation economies.

152

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

foreign to domestic firms. They argue that domestic firms improve productivity by learning from foreign-owned firms in the sector and/or suppliers and customers (Aitken and Harrison, 1999; Liu et al. 2000; Smarzynska Javorcik, 2004; Wei and Liu, 2006; Driffield and Love, 2007).6 However, the previous literature has not focused on the related role of environmental spillovers or the transfer of green technologies and ideas from foreign-owned to domestic firms and how firm characteristics are related to the motives for the adoption of green technologies. Meyer (2003) concludes that the impact of foreign-owned firms on the host country’s environment can be positive or negative.7 This ambiguity suggests that environmental spillovers are not an automatic process and may actually require the active involvement of domestic firms. 8 More precisely, we argue that Condition 2 will depend on: (a) a firm’s relationship with these firms (peers, suppliers or customers) and (b) the ability of the firm to assimilate new ideas and technologies. To study (a) and (b) we bring together two strands of the literature, one on firms’ absorptive capacity and the other on information networks. Absorptive capacity. Cohen and Levinthal (1990) introduced the most influential conceptualization of absorptive capacities to define the skills of firms needed to absorb and exploit external knowledge. This is associated with the firms’ previous knowledge, which enables them to identify the value of new information and apply it in business. Previous attempts to empirically examine absorptive capacities have defined them in terms of a wide range of firms’ observable characteristics, reflecting the multidimensional nature of knowledge.9 These characteristics include R&D expenditure, measures of the skill intensity of the labour force, as well as other characteristics associated with firm performance such as productivity or integration in global markets by exporting. Our second hypothesis allows us to examine the role played by absorptive capacity in the context of environmental spillovers: Hypothesis 2. The impact of foreign presence on domestic firms’ EAs will be influenced by the degree of domestic firms’ absorptive capacity. The greater the degree of domestic firms’ absorptive capacity, the greater the impact of foreign firms’ presence on domestic firms’ EAs. Inter-firm linkages. Firms’ absorptive capacity alone is unlikely to suffice if domestic and foreign-owned firms do not have close ties. An early elaboration of the importance of inter-firm linkages in the learning process was made by Lall (1980, p.204), where firm linkages summarize the “relationships established by firms in complementary activities which are external to 'pure market

6 Within the broader international business literature there is a considerable body of research that examines the location choice of multinational firms (Dunning, 1998; Cantwell, 2009) and the impact of firm ownership, engagement in foreign markets and inter-firm linkages on firm performance (see e.g. Driffield and Munday, 2000; Doukas et al., 2003). 7 Albornoz et al. (2009) also discuss environmental spillovers but do not consider the motives for firms' environmental actions. The labour migration motive was tested by Cole et al. (2008) who find some evidence that spillovers pass through human capital acquirement (from foreign to domestic firms) and reduces the energy intensity of Ghanaian firms. 8 More broadly, the international business literature considers the role of corporate governance on FDI decisions in emerging markets (Lien et al. 2005; Young et al. 2008) and the role of technology and knowledge spillovers (Blomstrom and Kokko, 1998; Veugelers and Cassiman, 2004; Marin and Bell, 2006). 9 See George and Zahra (2002) for an influential early review of the literature.

transactions”. Since then, diverse strands of the literature have identified and attempted to quantify the importance of these linkages as a channel through which knowledge flows.10 While a discussion of the conceptual implications of the environmental learning effects of these relationships among firms is beyond the scope of this paper, we do exploit a unique attribute of our detailed Argentinean data, namely the inclusion of firms’ formal and informal networks. Their inclusion allows us to examine whether such networks can act as a catalyst to enable greater EA adoption and integration of new and cleaner practices over and above the usual environmental spillovers. Hence, we test a third hypothesis. Hypothesis 3. The impact of foreign firms ‘presence on domestic firms’ EAs is influenced by the extent of domestic firms’ formal and informal relationships with all firm types. Hence, we test whether domestic firms benefit more from foreign-owned firms if they have other direct and indirect networking relationships with firms in other sectors (including cooperation agreements or joint ventures) or informal relationships (frequent contact between employees or managers of different firms). Positive answers to Hypotheses 2 and 3 would suggest that Condition 2 holds in the Argentine manufacturing sector and thus would establish the existence of conditional environmental spillovers from foreign-owned firms. Before we consider the motives for adopting environmental actions it should be noted that a number of studies have examined more broadly the characteristics of firms that influence environmental behaviour, whether such behaviour is in the form of abatement expenditure, emissions of pollutants or environmental actions. It is therefore important in this paper to control for as many of these additional determinants as possible. For example, Pargal and Wheeler (1996) argue that water pollution depends on a firm’s output, productivity, ownership (state or private) and informal regulatory pressure from citizens local to the firm. Drawing on the organizational management literature, AragonCorrea and Sharma (2003) argue that the extent to which a firm’s resources and capabilities (which include technology, managerial skills and attitudes) affect environmental management is contingent upon a number of factors including the complexity and uncertainty of the business environment and the altruism of the firm in question. In a US study, DeCanio and Watkins (1998) found that size and shareholder structure affected the decision to participate in the voluntary Green Light pollution prevention programme. Again for the US, Arora and Cason (1995, 1996) show that firm size and the underlying characteristics of the firm’s industry were important determinants of a firm’s participation decision in the Environmental Protection Agency’s (EPA) voluntary 33/50 programme. Finally, for Japan Nakamura et al. (2001) and Cole et al. (2006) emphasize firm size, age of employees, export status, FDI activities, the physical capital intensity of firms and productivity as factors that positively influence the implementation of EMS.11

10 See for example, the literature on clusters (Schmitz, 1995), production networks (Albornoz and Yoguel, 2004) or value chains and the global commodity chain (Gereffi, 2002). In the economics literature, the importance of externalities through firm linkages has also largely been acknowledged and quantified in terms of productivity gains (Aitken and Harrison, 1999; Smarzynska Javornik, 2004). 11 Note that although these papers consider the factors that are internal to a firm (size, investment in innovation, management practices) and external to a firm (pressure from consumers and shareholders) a number of them (Pargal and Wheeler 1996, Nakamura et al. 2001; Cole et al. 2006) focus on fairly narrow definitions of environmental behaviour (e.g. the adoption of ISO 14001 certification).

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

2.2. Firms' motives for adopting environmental actions The second stage of the paper considers the motives for EA adoption. There is little previous research on the motives for a firm to adopt EAs, nevertheless several studies provide some insights. Khanna and Anton (2002), for instance, find that public recognition was an important motive for firms' participation in the EPA's 33/50 programme. Concern for a firm's public image is also identified as important by Cole et al. (2006) who find that a firm's marketing expenditure was a positive and significant determinant of EMS adoption. Levy (1995) and Henriques and Sadorsky (1996) emphasise the role played by consumer and shareholder pressure. Pargal and Wheeler (1996) and Dasgupta et al. (2000) found regulatory pressure was an important motive for improving plant-level environmental performance in Mexico. However, Halkos and Evangelinos (2002) found that regulatory pressure was not a driver of EMS take-up amongst Greek firms, perhaps because of a perception that regulatory enforcement is weak. Sheldon (1997) argues that firms who adopt EAs may experience cost savings through the more effective use of raw materials and energy and a reduction in expensive waste management. Porter and Van der Linde (1995) emphasise the potential stimulus to innovation that environmental regulations may provide. If firms become aware of such benefits this in itself may encourage the voluntary adoption of EMS. In a related paper, Diller (1997) suggests that the adoption of EMS may yield organisational benefits to firms while Laplante and Lanoie (1994) argue that one additional benefit, and hence motivation for EMS implementation, is an increase in market valuation. How multinational firms manage their overseas affiliates can also influence the adoption of environmental actions. For example, in a decentralized system affiliates are often free to choose their own environmental actions, which will tend to be the responsibility of the local CEO. More often, headquarters (HQ) attempt to ensure foreign affiliates comply with regulations in whichever country they operate. In a centralized system this usually means complying with the HQ's home country standards. According to UNCTAD (2002), in a survey of 153 firms, the main drivers of environmental performance for foreign-owned affiliates were: HQ policies, procedures and standards (42%), regulatory pressures, current and anticipated (34%), local management leadership (12%), consumer pressure (4%), rules and pressures from international organisations (3%), pressure from NGOs and media (3%) and finally fear of accidents (2%). This leads us to our last hypothesis. Hypothesis 4. More efficient firms (larger and foreign-owned firms) are more likely to implement EAs in order to prepare for environmental certification while foreign-owned firms are more likely to implement EAs because of HQ policy. Local regulation is likely to induce large firms to adopt EAs and, finally, access to foreign markets is likely to encourage good environmental practices. 3. Data and method 3.1. Sample and data Our analysis is based on firm-level data from the Argentinean National Institute of Statistics and Censuses (INDEC).12 The survey covers the years 1998e2001 and provides data for approximately 1200 firms. Although we have annual data for non-

12

See INDEC (2002) for details.

153

environmental variables for the years 1998e2001, the firms' environmental action questions span the entire period rather than being year-specific. This means firms were asked if they had undertaken any environmental actions between 1998 and 2001. The survey was undertaken in a single wave during a threemonth period in 2003. The questionnaires were distributed by hand to the headquarters of each firm by professional staff from INDEC and were collected 10 days later. The questions on environmental actions were in a separate questionnaire to the economic data and were answered at a different point in time, albeit still within the 3 month period. The firms were encouraged to contact INDEC staff with any queries. Once collected, the questionnaires were analysed by INDEC staff who contacted firms if there was any missing information or concern about the information provided, thereby helping to address the potential risk of informant fallibility. The individuals who completed the questionnaires varied by firm. In the case of smaller firms, the questionnaires were typically completed by the owner whilst in larger firms they tended to be completed by the relevant heads of section (e.g. Human Resources, R&D etc.). However, each firm was required to nominate one person with overall responsibility for the information provided. By combining these two surveys, we believe this reduces concerns about common method variance (CMV) which often causes problems in papers of this type. Another mitigating factor with respect to the CMV problem is that the responses are largely factual of the yes/no variety, requiring very little value judgement (Podsakoff et al. 2003). However, given the importance of CMV in the international business literature we do undertake Harman's one factor test to increase further our confidence that we do not suffer from CMV biases. The response rate was very high (about 95%) and there were no differences in the response rates across firms suggesting no strategic answering of questions. The data are a representative sample of all Argentinean manufacturing firms with 10 employees or more and account for more than 50 per cent of total sector sales and employment and account for 60 per cent of total exports.13 The literature on productivity spillovers, where the dependent variable is a measure of total factor productivity (TFP), has been criticised because of the difficulty with interpreting the direction of causality (Hanson, 2005). For example, does foreign ownership increase productivity or are foreign-owned firms attracted to industries which contain the most productive domestic firms? Not controlling for this reverse causality can bias the results. In our analysis, the dependent environmental variables are for the whole period 1998e2001 whilst our explanatory variables are typically for 1998.14 This helps us to overcome some

13 INDEC claim that the dataset is representative of the manufacturing sector in terms of employment, output and trade. We find it is also representative of foreign ownership. A recent study by the UN Economic Commission for Latin America and Caribbean (ECLAC) on foreign investment and multinational corporations calculates that foreign-owned Argentinean firms account for 28% of total manufacturing firms during 1991e2000 (Kulfas et al. 2002). If we calculate foreign ownership in a comparable way, we find 25.2% of firms in our sample are foreign owned. Further comparisons are limited because the vast majority emanate from the sample that we are using, however, where additional estimates of foreign activity have been made they are very similar to those in our sample. For example, the ECLAC (2001) report states that wholly owned foreign owned firms are responsible for 49.1% of manufacturing exports in 1998. The comparable value for 2003 from our sample is 52% and a slight increase would be expected between 1998 and 2003. 14 While this might introduce unknown bias, since firms are responding to 1998 characteristics with different lead times, in unreported estimations we replace 1998 explanatory variables with variables for 1999, 2000 and 2001. In each case the sign and significance of key variables was unaffected, suggesting that our main results are robust.

154

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

of the concerns about the direction of causality (mitigates endogeneity bias) as it is less likely that current EAs would affect the 1998 values of our explanatory variables. Furthermore, although foreign-owned firms might be attracted to more productive sectors, it seems less likely that foreign-owned firms would be attracted to industries merely because they contain firms who have undertaken EAs. That said, if foreign investment and EAs are both functions of the stringency of environmental regulations then the interpretation of our linkage variables may be biased. However, links with suppliers and customers should be unaffected since they refer to foreign presence in industries other than the one in which the firm resides.15 3.2. Environmental survey In order to test our four hypotheses we require information on firms' environmental actions and their motives for such actions. This information is provided by three main questions from the INDEC (2002) survey, which we outline below. (i) The types of environmental action implemented (Hypotheses 1, 2 and 3). To identify firm's characteristics that make them more likely to undertake EAs we use firms' responses to survey question 501 below. We create a dummy variable equal to zero if a firm does not undertake any form of EA (i.e. the firm answers yes to part (a) of Q.501) and equal to one if answers yes to any one of (b) to (i) in Q.501. We also create a further eight dummy dependent variables that equal one if that type of EA in Q.501 is adopted and zero otherwise. We define each of these as an environmental action and emphasise that such actions may, or may not, equate to actual improved environmental performance. Our data do not include any direct measures of environmental performance and so we are unable to compare these to EAs.16 Q. 501. Indicate if the firm has done any of the following between 1998 and 2001: (a). None of the following (b). Used systems and equipment for the treatment of residuals and effluents (c). Taken actions for the purposes of environmental remediation (d). Improve efficiency of the use of water, energy and other inputs (e). Replaced or modified pollution processes (f). Replaced inputs that are pollution intensive (g). Developed environmentally friendly products (h). Established internal or external recycling procedures (i). Obtained any environmental certification (ii) Motives for the implementation of environmental actions (Hypothesis 4). Question 502 asks those firms who have undertaken some form of EA to indicate their motives for doing so. Firms can select one or more of seven motives. Q. 502. What was the motivation for undertaking any of the actions outlined in Q. 501?

15 We acknowledge that lagging may not fully address the concern that unobserved firm characteristics correlated with both foreign-ownership and environmental action biases the results. The problem of unobserved heterogeneity/omitted variable bias is never completely avoidable so our use of causal language should be considered with caution. 16 We also acknowledge that these different EAs may have very different cost implications for firms and the magnitude of any environmental improvements may differ across EAs.

(a). (b). (c). (d). (e). (f). (h).

To improve the image of the firm To meet local environmental regulations To meet the requirements of local customers Merely acting in accordance with firm policy To meet the requirements of foreign markets To prepare for environmental certification To imitate competitors in the local market

3.3. Linkages and networking Hypotheses 2, 3, and 4 require us to measure the extent of any possible spillovers from foreign-owned to domestic firms and to consider the role played by networks. The methodology we use to measure foreign linkages and networks is outlined below. We follow Aitken and Harrison (1999), Smarzynska Javorcik (2004) and Albornoz et al. (2009) to capture the impact of foreign investment within a sector (Horizontal), on upstream sectors (Backward) and on downstream sectors (Forward). Relative to a representative firm, we assume customers are upstream and that intermediate producers are downstream. Hence, if customers put pressure on a firm to implement EAs it would be an example of a backward linkage. If, on the other hand, suppliers put pressure on consumers to introduce EAs we consider this a forward linkage. The simplest type of linkage a spillover from firms in the same industry. The variable Horizontaljt measures the presence of foreign-owned firms in a sector. More specifically.

P Horizontaljt ¼

ici2j P

FFijt Yijt

ici2j Yijt

where FFijt takes the value of 1 for foreign-owned firms, defined as those firms whose foreign ownership is greater than 10%, and Yijt is a firm's output. The value of Horizontaljt increases with the output and the number of foreign-owned firms in the sector and the fraction of the sector's output that is produced by foreign-owned firms. The variable Forward captures the presence of foreign-owned firms in industries that supply the sector to which the firm i belongs. This variable allows us to assess whether firms and their foreign suppliers or subsidiaries influence the implementation of EAs. Specifically:

Forwardjt ¼

X

dkj Horizontalkt

k if ksj

where the weight dkj is the proportion of sector k's output supplied to sector j as given by the inputeoutput matrix at the two-digit ISIC level in 1997.17 Forward is just a weighted average of the foreign percentage of sector j's suppliers, where the weights are the proportion of total inputs supplied. Similarly, Backward measures the presence of foreign-owned firms in those sectors that are being supplied by the sector to which the firm in question belongs. This variable allows us to capture the possible pressure exerted by foreign-owned firms or those from developed countries who might expect their suppliers to reach a certain level of good environmental management practice (whether driven by for example shareholders or government policy). Specifically:

17 The 1997 Input-Output matrix can be found in the Argentinean National Institute of Statistics and Censuses.

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

Backwardjt ¼

X

djk Horizontalkt

k if ksj

where dkj is the proportion of sector j's output supplied to sector k according to the inputeoutput matrix. Following from Hypothesis 2, given firms are likely to require a degree of absorptive capacity in order to benefit from spillovers, we combine our linkage variables with two proxies for absorptive capacity, namely the percentage of the workforce who are skilled and whether or not the firm is an exporter. As Cohen and Levinthal (1990) point out, absorptive capacities refer to the ability of firms to exploit and absorb external knowledge. This ability is associated with firms' previous knowledge (for example, basic skills, training, R&D development) which ensures that firms are able to identify the value of new information and apply it in business. Many studies use a measure of skill intensity to capture absorptive capacities. Malecki (1997) argues that technological capability embodied in skilled workers is necessary for firms to evaluate new external technologies. In addition, Varga (2000) finds that a concentration of high-tech employment is necessary for knowledge transfer from US universities. Furthermore, Audretsch and Feldman (1996) find evidence at the industry level that high skill-intensive industries are likely to agglomerate because they tend to benefit more from spillovers than low skillintensive industries. Other papers using measures of skill in-

155

tensity as proxies for absorptive capacity include Marin and Bell (2006), Haskel et al. (2007) and Saito and Gopinath (2011). Our justification for using export share as a proxy for absorptive capacity is the correlation between export activity and other relevant firm characteristics linked to absorptive capacity, for example, skill intensity, productivity and production quality (see Bernard and Jensen, 1999; Bernard et al. 2007). Furthermore, Escribano et al. (2009) find that absorptive capacity is an important source of competitive advantage while Barrios et al. (2004) explicitly capture absorptive capabilities using a dummy variable to indicate whether the firm is an exporter. They justify this by arguing that exposure to foreign markets is likely to be an indicator that a firm has a higher level of technology relative to a firm that only operates in the local market. In addition to our spillover variables, we include direct measures of contact with other sectors. Our networking measures are dummy variables that capture the existence of firm j's formal relationships (cooperation agreements or joint ventures) or informal relationships (frequent contacts established by members of the firms) with firms in the same sector (horizontal), suppliers (forward) and customers (backward). The three variables are NetworkSupp which captures forward networking at the local level, NetworkCust which captures backward networking at the local level and NetworkHoriz which provides a measure of horizontal networking at the local level.

Table 1 Description and summary statistics of variables. Variable

Description

EA

Dummy variable equal to 1 is the firm undertakes any form of EAs Dummy variable equal to 1 if the firm is more than ten-percent foreign-owned. Total number of workers. Whether a firm is independent or part of a larger group. Sales growth captures the idea that a growing firm is likely to be financially stronger. Labour productivity is included as a proxy for TFP. Exports as a percentage of sales R&D expenditure as a percentage of sales Percentage of workforce that are technical workers. Investment expenditure as a percentage of sales. Backward linkages measured at the 2-digit industry level. Forward linkages measured at the 2-digit industry level. Horizontal linkages at the 2-digit industry level. Firm-level measure of whether a firm has contact with local suppliers. Firm-level measure of whether a firm has contact with local customers. Firm-level measure of whether a firm has contact with other local firms in the same industry. Backward spillovers interacted with Perskilled Forward spillovers interacted with Perskilled Horizontal spillovers interacted with Perskilled Backward spillovers interacted with dexport Forward spillovers interacted with dexport Horizontal spillovers interacted with dexport Interaction term to capture whether backward linkages only work if the firm has close contact with customers. Interaction term to capture whether forward linkages only work if the firm has close contact with suppliers. Interaction term to capture whether horizontal linkages only work if the firm has close contact with the other firms in its sector.

FO10 Size Independent Salesgrowth Labprod perexport perRD Perskilled Invsales Backward Forward Horizontal NetworkSupp NetworkCust NetworkHoriz Backward*skilled Forward*skilled Horizontal*skilled Backward*export Forward*export Horizontal*export Backward *NetworkCust

Forward *NetworkSupp

Horizontal *NetworkHoriz

Mean

S.D.

0.61

0.49

0.21

0.41

227 0.73

470 0.45

2.49

11.00

0.15 10.38 2.64 34.0 6.00 0.11

0.31 20.76 7.38 28.51 17.1 0.11

0.10

0.60

0.40 0.25

0.22 0.43

0.18

0.38

0.08

0.27

Hypothesis/sign

H1: þ

H2: H2: H2: H2: H2: H2: H3:

þ þ þ þ þ þ þ

H3: þ

H3: þ

156

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

In order to test Hypothesis 3 we create a more sophisticated set of linkage variables by combining our Horizontal, Forward and Backward variable with our three networking variables. Hypothesis 3 argues that firms need established relationships with other firms in order to benefit from and be able to absorb any benefits from foreign-owned firms. Put another way, firms that do not have a local network and are hence more insular in their outlook are less likely to benefit from the presence of foreign-owned firms in Argentina. 3.4. Model specification and summary statistics The odds, or likelihood, that a firm undertakes EA can be expressed as the ratio of the probability that EA will be adopted (Pr) to the probability that it will not be adopted (1-Pr). We estimate a logistic transformation of this ratio:

  Pr logit ½PrðEAÞ ¼ 1 ¼ log 1  Pr

(1)

Our equation to be estimated is of the form;

logit ½PrðEAÞ ¼ 1 ¼ a þ lFO þ b0 X þ 40 Z þ εi

(2)

where FO is foreign ownership, X is a vector of network and linkage variables and Z is a vector of other firm characteristics (discussed below). We report our coefficients in the form of odds ratios. Equation (2) is estimated using the maximum likelihood method, which maximises the likelihood of the observed data given the estimated parameter. Our vector Z includes a number of variables likely to influence the decision of firms to undertake EAs, each defined in Table 1. Sales growth (Salesgrowth) is included given the possibility that a growing firm is more likely to be able to devote financial resources to environmental activities. Firm size may also be important since larger firms are more likely to dedicate human resources to environmental management and to benefit from economies of scale in pollution abatement (Cole et al. 2005; Nakamura et al. 2001). Some studies have suggested that the relationship between environmental behaviour and firm size may be subject to diminishing returns (Cole et al. 2013). We capture firm size using the total number of workers and include a squared term to allow for a non-linear relationship. A firm's EAs may also be influenced by whether the firm is independent or part of a large group (Independent), the general efficiency of the firm, measured using labour productivity (Labprod) and the firm's exports as a percentage of sales (PerExport) since firms entering certain export markets will often have to meet more demanding environmental standards. In addition, we might expect a firm's investment in innovation to affect EAs and hence include R&D expenditure as a share of sales (PerRD). We also include the percentage of skilled workers (Perskilled) as the adoption of EAs may to some extent hinge upon a firm's competencies in terms of skills and training. We use the level of investment expenditure (as a percentage of sales) to capture the possibility that newer technology may be less environmentally damaging (Cole et al. 2005). Finally, we control for the inherent pollution intensity of certain industries. Although we have no specific information on the pollution intensity of each sector in Argentina, in those countries for whom data exist the most pollution intensive sectors tend to be SIC 21 (Pulp and Paper), SIC 23 (Petroleum), SIC 24 (Chemicals), SIC 26 (Non-Metallic

Minerals), and SIC 27 (Steel and Aluminium). All estimations include industry sector dummies. Since our linkage variables are measured at the industry level, we are mixing plant-level and industry-level variables and therefore cluster the standard errors using the Moulton (1990) correction. To test Hypothesis 4, we undertake another Logit analysis in which we seek to identify the characteristics of those firms that selected each motive. We estimate each motive individually with the dependent variable taking the form of a dummy ¼ 1 if a firm selected that motive and 0 otherwise. Descriptions and summary statistics for our main control variables are given in Table 1 and include a full list of our interaction terms used to test Hypotheses 2 and 3. On average, firms in our sample export 10.38% of their sales and spend the equivalent of 2.64% of their sales on R&D. In addition, although not reported in Table 1, 53% of the firms are exporters and 60% are involved in R&D activities during the period of analysis. Investment expenditure represents 6.0% of sales. The average size of the firms in the sample is 227, measured by the total number of workers, and 34 per cent of the total workforce is classed as being ‘technical workers or above’, where technical worker corresponds to a low-level engineer or technician (or equivalent). Table 2 presents summary data by two-digit industries for our sample of 1187 firms. The sample contains a large number of firms from SIC 15 (Food and Beverages) and to a lesser extent SIC 24 (Chemicals) and SIC 29 (Machinery and Equipment). Table 2 indicates that these sectors have some of the highest EA adoption rates. Within a given industry there appears to be no obvious link between EA adoption and the percentage of foreign-owned firms. A correlation matrix is provided in the Appendix. Although our dependent variables emanate from a different survey to the independent variables, common method variance remains a theoretical possibility. In order to test for this we use Harman's one-factor test. A principal components factor analysis of

Table 2 Industry level summary statistics. % of firms with EAs

2 Digit SIC

Industry

No. of firms

15 16 17 18 19 20 21

Food and Beverages Tobacco Textiles Clothing Leather and Footwear Wood and Wood Products Pulp, Paper and Paper Products Publishing and Printing Petroleum Chemicals Rubber and Plastics Non-metallic Minerals Steel and Aluminium Metals Products, except mach & equip Machinery and Equipment Office Machines and Computers Electrical Machinery Radio, TV and Comm. Equip. Medical, Precision and Optical Equip. Automotive and Transport Equip. Other Transport Equip. Furniture and Other Manufacturing

245 4 92 31 34 25 36

66 50 46 10 59 64 67

18 25 11 6 9 8 31

64 9 126 74 61 32 60

45 100 81 70 64 72 55

14 44 40 18 25 16 23

105 2

60 50

21 0

49 16 14

61 38 36

20 50 14

49

67

47

24 35

32 53

8 6

22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

Source: INDEC (2002) and authors' own calculations.

% of foreign owned firms

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

157

Fig. 1. Environmental actions by ownership structure.

Fig. 2. Motives by ownership structure.

our dependent and explanatory variables provided 5 variables with eigenvalues greater than one while no general factor was apparent in the unrotated factor structure (factor 1 accounting for only 31% of the variance). We therefore do not believe that our results are unduly affected by common methods variance. Figs. 1 and 2 provide summary information for the firms in our sample. Each figure distinguishes between foreign-owned and domestic firms. Fig. 1 shows that the most common forms of EA are ‘the treatment of residuals and effluent’, ‘measures to achieve the efficiency of water, energy and other inputs’, and ‘recycling procedures’. In support of Hypothesis 1, domestic firms are considerably more likely not to have implemented any EAs during this period, although at this stage we are obviously not controlling for the effect of other firm characteristics. For each type of EA we generally observe only minor differences between foreign-owned and domestic firms, although foreign-owned firms appear considerably more likely to undertake recycling. Fig. 2 reveals that the most common motives for the adoption of EA (Hypothesis 4) are ‘to improve the image of the firm’, ‘to meet local environmental regulations’, ‘acting in accordance with firm

policy and to prepare for environmental certification’.18 However, in the latter two cases we see that foreign-owned firms are significantly more likely to hold such motives. There is generally little difference between foreign-owned and domestic firms for the other four motives. 4. Empirical results The results for our tests of Hypotheses 1, 2 and 3 are presented in Tables 3 and 4. Table 3 provides the estimated results where the dependent variable is a dummy variable equal to one if a firm undertakes any form of environmental action and zero otherwise. Columns (1) and (2) are estimated using all firms in the sample,

18 We were not able to find data on the environmental performance of Argentinean firms (e.g. emissions) or on the environmental regulations they face. However, local environmental regulations often appear to be strict and in part borrowed from the US or the EU. However, Argentina lacks a strong institutional framework which leads to weak enforcement and the authorities have insufficient resources to closely monitor the environmental performance of firms.

158

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

Table 3 Determinants of environmental actions (Hypotheses 1, 2 and 3). (1)

(2)

All firms FO10 Backward Forward Horizontal Backward*skilled Forward*skilled Horizontal*skilled

1.86*** (4.3) 1.0042 (0.3) 0.99 (0.3) 1.0023 (0.2) 1.00047* (1.7) 1.0015* (1.9) 1.00 (1.3)

Backward*export Forward*export Horizontal*export NetworkSupp NetworkCust NetworkHoriz Backward*networkCust Forward*networksupp Horizontal*networkHoriz Size Size2 Independent Salesgrowth Labprod Perexports PerRD Perskilled Invsales Wald Nagelkerke Pseudo R2 HosmereLemeshow C-statistic Observations

1.39 (0.8) 0.71 (1.6) 1.0048 (0.01) 1.024*** (2.8) 1.031 (0.9) 1.00 (0.2) 1.31*** (4.9) 0.99*** (4.2) 0.84 (1.1) 1.022** (2.3) 1.082 (0.5) 1.0043 (1.1) 1.041*** (3.2) 1.00 (0.5) 0.81 (0.5) 131.7*** 0.25 23.71** 0.76 1187

(3)

(4)

Domestic firms 1.90*** (4.1) 1.0058 (0.4) 1.034 (0.7) 1.00 (0.1)

1.034** (2.2) 1.016 (0.4) 0.99 (0.8) 1.29 (0.6) 0.72 (1.5) 1.0036 (0.01) 1.028*** (2.7) 1.036 (1.0) 1.00 (0.2) 1.30*** (4.5) 0.99*** (3.9) 0.83 (1.0) 1.016** (2.2) 1.081 (0.5) 1.0033 (0.7) 1.042*** (3.6) 1.0074** (2.5) 0.81 (0.6) 139.2*** 0.25 22.61* 0.76 1187

1.00 (0.2) 0.97 (0.7) 1.01 (0.5) 1.00050 (1.2) 1.0013 (1.3) 0.99 (0.7)

1.58 (1.1) 0.80 (1.0) 0.80 (0.3) 1.014 (1.3) 1.027 (0.8) 1.0043 (0.3) 1.31*** (3.9) 0.99*** (3.2) 0.79 (1.2) 1.013* (1.9) 1.077 (0.4) 1.0050 (1.0) 1.031** (2.2) 0.99 (0.4) 0.84 (0.4) 75.7*** 0.18 9.71 0.72 935

(5)

(6)

Foreign firms

0.99 (0.03) 0.99 (0.1) 1.0080 (0.5)

1.05*** (3.0) 1.033 (0.7) 0.98 (1.2) 1.47 (0.9) 0.80 (1.0) 0.82 (0.3) 1.019 (1.5) 1.033 (1.0) 1.0028 (0.2) 1.30*** (3.6) 0.99*** (3.1) 0.78 (1.1) 1.012* (1.7) 1.059 (0.3) 1.0046 (0.8) 1.031** (2.4) 1.0058* (1.9) 0.83 (0.5) 82.0*** 0.18 16.23* 0.72 935

1.015 (0.6) 1.10 (1.0) 0.97 (1.1) 1.00067* (1.7) 1.0018 (0.7) 0.99 (0.2)

0.21 (1.2) 0.39* (1.7) 1.59 (0.5) 1.079** (2.5) 1.18 (1.5) 0.98 (1.2) 1.35** (2.0) 0.99* (1.7) 0.93 (0.2) 1.031 (0.8) 1.13 (0.2) 0.99 (0.2) 1.17*** (2.9) 0.99 (0.3) 0.27 (1.0) 45.5*** 0.34 11.57 0.83 252

1.056* (1.7) 1.30*** (3.2) 0.94 (0.9)

0.98 (0.4) 0.87 (1.0) 1.043 (1.0) 0.22 (1.1) 0.37 (1.6) 1.72 (0.7) 1.084** (2.4) 1.19 (1.4) 0.98 (1.1) 1.37* (1.8) 0.99 (1.6) 1.0091 (0.02) 1.032 (0.8) 1.11 (0.2) 0.99 (0.2) 1.17*** (2.8) 1.017** (2.2) 0.33 (0.8) 46.5*** 0.34 10.14 0.82 252

Robust Z statistics in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%. The dependent variable is a dummy variable capturing whether or not firms undertake any form of EAs. The coefficients are reported in the form of odds ratios. Industry dummies are included.

columns (3) and (4) are estimated only for domestic firms and columns (5) and (6) only for foreign-owned firms.19 Table 4

provides results for all firms where the environmental actions are estimated individually.20 Hypothesis 1. Focusing first on all firms in Table 3, we see that foreign ownership is a positive and statistically significant

19 It may appear desirable to include firms' stated motives for adopting EAs as determinants of the decision to implement an EA. However, such an approach would not be meaningful. The inclusion of such a motive as an explanatory variable would mean we would be asking whether firms who undertook an EA for a particular reason were more likely to adopt an EA than firms who did not adopt an EA for that reason. Furthermore, motive data are only available for firms who engage in some form of EA.

20 For reasons of space, the results in Table 4 stem from a model in which absorptive capacity is captured using skilled workers. The results for our other measure of absorptive capacity, and for domestic and foreign firms separately, are available upon request.

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

159

Table 4 Determinants of Individual Environmental Actions (all firms) (Hypotheses 1, 2 and 3).

FO10 Backward Forward Horizontal Backward*skilled Forward*skilled Horizontal*skilled NetworkSupp NetworkCust NetworkHoriz Backward*NetworkCust Forward*NetworkSupp Horizontal*NetworkHoriz Size Size2 Independent Salesgrowth Labprod Perexports PerRD Perskilled Invsales Wald Nagelkerke Pseudo R2 HosmereLemeshow C-statistic Observations

Action b

Action c

Action d

Action e

Action f

Action g

Action h

Action i

1.35** (2.0) 0.98 (1.1) 1.00 (0.09) 1.016 (1.6) 1.00* (1.7) 1.00 (1.4) 1.00* (1.9) 1.054 (0.1) 1.00 (0.01 1.74 (0.7) 1.015 (0.9) 1.013 (0.5) 0.98 (0.9) 1.33*** (6.2) 1.00*** (5.0) 0.67** (2.1) 1.00 (0.7) 1.78 (1.6) 1.01*** (3.2) 1.041*** (3.4) 1.0029 (0.5) 1.26 (0.8) 168.0*** 0.26 8.26 0.77 1187

1.32** (2.2) 0.98 (1.0) 0.93 (1.7) 1.038** (2.3) 1.00*** (2.7) 1.00*** (3.0) 1.00*** (3.2) 1.50 (1.0) 1.17 (0.4) 0.71 (0.6) 1.016 (1.0) 0.95* (1.8) 1.01 (0.6) 1.096*** (2.7) 1.00 (1.6) 0.94 (0.2) 1.00 (1.2) 1.074 (0.6) 1.00** (2.2) 1.027** (2.3) 1.00 (0.04) 0.84 (0.5) 92.6*** 0.12 6.85 0.70 1187

1.42*** (2.9) 1.00 (0.07) 1.041 (1.2) 1.00 (0.2) 1.00* (1.7) 1.00 (1.5) 1.00 (1.6) 1.55** (2.0) 0.91 (0.4) 0.73 (0.8) 1.013 (1.1) 0.99 (0.6) 1.01 (1.2) 1.17*** (4.5) 1.00*** (3.2) 0.68*** (2.8) 1.00** (2.5) 1.16 (1.1) 1.01 (1.6) 1.023* (1.9) 1.00 (0.4) 1.21 (0.5) 137.3*** 0.19 9.75 0.72 1187

0.94 (0.3) 1.01 (0.6) 1.016 (0.4) 1.00 (0.1) 1.00** (2.4) 1.00 (3.2) 1.00* (1.8) 1.88* (1.8) 1.07 (0.3) 1.21 (0.3) 1.00 (0.16) 1.00 (0.1) 0.99 (0.6) 1.24*** (4.5) 0.99*** (4.9) 0.63*** (2.8) 1.00 (0.9) 1.19 (0.8) 1.01*** (4.0) 1.02 (1.6) 1.00 (0.02) 0.60** (2.0) 147.9*** 0.20 5.21 0.74 1187

1.05 (0.3) 0.98** (2.0) 1.03 (0.7) 0.99 (0.5) 1.00*** (2.9) 1.00** (2.0) 1.00 (1.4) 0.98 (0.05) 0.79 (0.9) 1.10 (0.2) 1.02 (1.2) 1.03 (1.4) 1.00 (0.4) 1.16*** (2.9) 1.00* (1.9) 0.81 (1.3) 0.99 (1.2) 1.08 (0.3) 1.00 (0.8) 1.02** (2.1) 1.00 (0.08) 0.68 (1.4) 117.9*** 0.15 1.61 0.71 1187

1.44* (1.8) 1.02 (1.2) 1.02 (0.5) 0.98 (1.5) 1.00 (1.1) 1.00** (2.1) 1.00 (0.6) 1.59 (1.3) 1.14 (0.4) 0.46 (0.8) 1.02 (1.2) 1.00 (0.3) 1.00 (0.1) 1.29*** (4.6) 0.99*** (3.5) 1.26 (1.0) 1.00* (1.8) 1.10 (0.3) 1.01* (1.9) 1.00 (0.1) 0.99 (1.5) 0.80 (0.5) 127.3*** 0.18 6.74 0.76 1187

1.79** (2.6) 1.00 (0.2) 1.02 (0.6) 0.99 (0.6) 1.00 (1.1) 1.00 (0.9) 1.00 (0.7) 1.50 (0.9) 1.50 (1.4) 0.33** (2.6) 1.00 (0.03) 0.97 (1.1) 1.018** (2.4) 1.15*** (6.3) 1.00*** (4.4) 0.66** (2.1) 1.00 (0.7) 1.38 (1.3) 1.00 (0.5) 1.03** (2.2) 0.99 (1.5) 0.91 (0.3) 181.1*** 0.23 10.03 0.76 1187

1.51** (2.3) 1.02 (1.0) 0.90 (0.7) 1.034** (2.4) 1.00 (1.6) 1.00*** (3.7) 1.00 (1.6) 0.45 (1.1) 1.12 (0.3) 0.27 (1.7) 0.99 (0.3) 1.034 (1.5) 1.028 (1.8) 1.20*** (3.3) 1.00** (2.6) 0.55*** (2.9) 1.00* (1.7) 1.42 (1.3) 1.01 (1.4) 1.016 (0.8) 1.01 (1.2) 1.37 (0.8) 134.6*** 0.28 3.96 0.81 1187

Robust Z statistics in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%. The dependent variable is a dummy variable capturing whether or not firms undertake each specific EA. The coefficients are reported in the form of odds ratios. Industry dummies are included. Where action (b) refers to the treatment of residuals and effluent, action (c) is to take action for environmental remediation, action (d) is to improve the efficiency of use of water, energy and other inputs, action I to replace or modify pollution processes, action (f) is to replace inputs that are pollution intensive, action (g) is to develop environmentally friendly products, action (h) is to establish internal or external recycling processes and action (i) is to obtain environmental certification.

determinant of EA adoption. These results support Hypothesis 1 by suggesting that foreign-owned firms are approximately 1.9 times more likely to implement EAs than domestic firms.21 If we consider the determinants of individual environmental

21 We also test Hypothesis 1 using two alternative measures of foreign ownership to FO10, by defining a firm as foreign owned if 25% (FO25) and 50% (FO50) of equity is foreign owned. For reasons of space we cannot report full results, but reestimating the models in columns (1) and (2) of Table 3 provides odds ratios (and z-statistics) for the foreign ownership variables of 1.85*** (3.7) and 1.88*** (3.6), respectively for FO25 and 1.79*** (3.2) and 1.86*** (3.1) for FO50. We therefore find our results to be consistent across all 3 measures of foreign ownership suggesting that it is not the degree of foreign ownership that is important but whether the firm has any foreign influence through the ownership structure that is important to the decision to implement EAs.

actions in Table 4 we see that foreign ownership is a statistically significant determinant of all but two of the environmental actions. Of most relevance for Hypotheses 2 and 3 is the role of environmental spillovers. In Table 3, we note that the individual linkage variables (Backward, Forward and Horizontal) are not statistically significant for the analyses using all firms or domestic firms on their own, but are significant for the analysis of foreign-owned firms in column (6). This suggests that foreign-owned firms benefit from each other but domestic firms do not benefit from the presence of foreign-owned firms. One explanation could be related to the absorptive capacity of firms as tested by Hypothesis 2. In Table 4 our results for individual actions for all firms again finds that all linkage variables are non-significant with the exception of horizontal

160

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

Table 5 Motives for EAs (Hypothesis 4).

FO10 Backward Forward Horizontal Backward*skilled Forward*skilled Horizontal*skilled NetworkSupp NetworkCust NetworkHoriz Backward*Networkcust Forward*Networksupp Horizontal*NetworkHoriz Size Size2 Independent Salesgrowth Labprod Perexports PerRD Perskilled Invsales Wald Nagelkerke Pseudo R2 HosmereLemeshow C-Statistic Observations

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Motive A

Motive B

Motive C

Motive D

Motive E

Motive F

Motive G

0.78 (0.8) 0.99 (0.2) 0.96 (0.9) 1.039** (2.1) 1.00037 (1.6) 1.0011 (1.5) 0.99** (2.3) 1.22 (0.8) 1.25 (1.3) 0.37* (1.8) 0.98* (1.6) 0.97** (2.0) 1.016 (1.4) 1.084** (2.4) 0.99** (2.4) 0.91 (0.5) 1.00020 (0.6) 0.91 (0.4) 1.010** (2.5) 1.025 (1.6) 1.010 (1.6) 0.50** (2.6) 39.8** 0.13 4.32 0.65 719

0.91 (0.5) 0.97** (2.0) 0.90* (1.9) 1.043** (2.1) 1.00051 (1.5) 1.00033 (0.4) 0.99 (1.3) 1.36 (0.9) 0.81 (0.6) 0.70 (0.8) 1.025 (1.1) 1.0089 (0.4) 0.99 (0.3) 1.088*** (2.8) 0.99*** (2.9) 1.052 (0.2) 1.00057* (1.8) 0.79 (1.0) 1.0061* (1.7) 1.0019 (0.1) 1.0082 (1.4) 0.66 (0.7) 37.7*** 0.11 3.71 0.64 719

1.085 (0.3) 1.031 (1.0) 1.18*** (3.3) 0.98 (1.1) 0.99** (2.1) 0.99 (1.3) 1.00049 (1.3) 0.77 (0.6) 1.76 (1.3) 1.91 (0.8) 1.025 (1.1) 1.0023 (0.1) 0.98 (0.9) 1.067 (0.9) 0.99 (0.7) 1.0028 (0.01) 1.00083** (2.1) 0.84 (0.4) 0.98* (1.9) 1.018 (1.1) 1.00010 (0.01) 0.26 (1.0) 48.7*** 0.11 3.57 0.71 719

3.51*** (3.8) 0.99 (0.5) 0.98 (0.3) 1.017 (1.0) 1.00059 (1.3) 1.0020** (2.0) 0.99 (0.9) 1.028 (0.06) 2.65 (2.4) 1.040 (0.08) 0.97* (1.9) 0.96 (1.4) 0.99 (0.2) 1.065 (1.4) 0.99 (0.7) 0.40*** (5.1) 0.99*** (4.2) 6.55*** (5.0) 0.99 (0.4) 1.028*** (2.7) 0.99 (0.3) 2.51 (1.1) 151.7*** 0.40 1.83 0.84 719

0.71 (1.1) 1.0 (0.00) 0.96 (0.7) 1.019 (1.6) 1.00042 (1.2) 1.0020* (1.7) 0.99* (1.8) 0.59 (0.9) 0.86 (0.3) 1.99 (0.8) 1.028 (1.1) 1.029 (0.8) 0.98 (1.6) 1.080* (1.8) 0.99 (1.1) 0.65** (2.4) 0.99** (2.4) 1.12 (0.4) 1.022*** (7.3) 1.029 (1.4) 1.0092 (0.6) 0.40 (1.0) 48.1*** 0.12 4.68 0.73 719

1.39** (2.3) 0.99 (1.3) 1.0038 (0.1) 1.013 (1.1) 1.0010** (2.5) 1.0017* (1.8) 0.99** (2.2) 0.81 (0.9) 1.68 (1.6) 0.99 (0.01) 0.98 (1.5) 1.019 (1.0) 0.99 (0.8) 1.084*** (3.8) 0.99** (2.0) 0.71** (2.1) 0.99 (1.6) 3.13*** (3.5) 1.0086*** (2.8) 1.015 (1.0) 1.0034 (0.4) 2.90** (2.0) 80.2*** 0.18 7.66 0.72 719

2.97 (1.4) 1.036 (1.5) 0.87* (1.7) 1.044 (1.6) 1.000044 (0.06) 1.0049** (2.1) 0.99* (1.8) 0.65 (0.5) 4.70** (2.5) 1.16 (0.2) 0.98 (1.2) 0.98 (0.5) 1.00030 (0.02) 1.56 (1.0) 0.89* (1.8) 2.94** (2.0) 0.98 0.9) 2.10 (1.2) 0.93* (1.8) 0.99 (0.00) 1.019 (0.9) 2.17 (0.6) 73.0*** 0.15 6.65 0.82 719

Robust z-statistics in parentheses, * significant at 10%, ** significant at 5%, *** significant at 1% Notes: The dependent variable is a dummy variable capturing whether or not firms gave that motive for undertaking EAs. The coefficients are reported in the form of odds ratios. Industry dummies are included. Motive A refers to ‘improving the image of the firm’; motive B refers to ‘meeting local environmental regulations’; motive C refers to ‘meeting the requirements of local customers’; motive D refers to ‘acting in accordance with firm policy’; motive E refers to ‘meeting the requirements of foreign markets’; motiveF refers to ‘preparing for environmental certification’ and motive G refers to ‘imitating competitors in the local market’.

linkages for action (i) (obtaining environmental certification). This result indicates that firms that come into contact with foreign firms within their industry are more likely to have environmental certification. Hypothesis 2. By interacting our spillover variables with our measures of absorptive capacity, we find that for ‘all firms’ in Table 3, Backward and Forward linkages are statistically significant when interacted with the percentage of skilled labour. The same finding is made for the majority of the individual environmental actions reported in Table 4. This suggests that firms who consume from, and supply to, sectors containing a large percentage of

foreign-owned firms are more likely to adopt EAs the greater the percentage of skilled workers within their workforce. Returning to Table 3, column (2) partially supports this result by finding Backward linkages to be significant for firms who export. No such finding is made for Forward linkages. When we consider foreignowned and domestic firms separately, we find that Backward interacted with skilled labour is significant for foreign-owned firms while Backward interacted with exports is significant for domestic firms. This suggests that absorptive capacity increases the likelihood of a firm experiencing environmental spillovers from foreignowned firms. So partial support for Hypothesis 2 is found.

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

Hypothesis 3. Turning to our network variables in Table 3, for ‘all firms’ we find that Backward interacted with local customer networks (NetworkCust) is positive and significant. This suggests firms are more likely to implement EAs if they have both formal and informal links with customers and if they supply customers in industries with a large proportion of foreign-owned firms. For domestic firms, the odds ratios for these interaction terms are greater than one although they are not statistically significant. However, such interactions are significant for foreign-owned firms suggesting that they may be influenced by the presence of other foreign-owned firms, particularly if the firm is part of a network with local customers. It is notable that the results for individual environmental actions in Table 4 do not suggest that our network variable interacted with the linkage variables are statistically significant. Finally, Table 3 suggests that being part of a network with other local firms does not, in itself, increase the likelihood of adopting EAs, although Table 4 indicates that being in a network with local suppliers does increase a firm’s likelihood of adopting two environmental actions. Overall, the evidence suggests that formal and informal links help with the adoption of environmental actions and provide partial support for Hypothesis 3. Turning to the other explanatory variables in Tables 3 and 4 we see that firm size (by number of workers) and size squared are significant across all specifications and have the expected odds ratio of greater than one and less than one for size and size squared respectively (although they are both close to one). That is to say, EA implementation increases with size but at a decreasing rate. Size appears to be more important for foreignowned firms. We find that sales growth is positively and significantly correlated with EA adoption, although not for foreign firms or for certain individual environmental actions. R&D expenditure is positive and significantly correlated for all firms, domestic and foreign-owned alike, and for the majority of environmental actions. Exports as a percentage of sales are not statistically significant.22 Labour productivity and the skill level of the workforce are not consistently significant across models, while independent firms are found to be less likely to undertake many individual environmental actions. Tables 3 and 4 also provide a number of diagnostic tests. The Wald test is statistically significant for all estimations suggesting that each model represents an improvement over the intercept only model. We also report Nagelkerke Pseudo R2 measures which, although not directly interpretable in the same way as a standard R2 measure, do not raise cause for concern. Furthermore, the HosmereLemeshow goodness of fit measure is not statistically significant in the majority of models suggesting that observed EAs match expected EAs in subgroups of the sample. Finally, we undertake validation of predicted probabilities via the C-statistic measure of association which, in Tables 3 and 4, varies from 0.70 to 0.83. A value of 0.83 indicates that for 83% of pairs of firms where one firm had EAs and the other did not, the model correctly assigned a higher probability to the firm with EAs. Hypothesis 4. Turning to hypothesis 4, Table 5 presents the results for our motivating question. The different motives are

22 In unreported estimations we find that an export dummy, indicating whether or not a firm exports, is a statistically significant determinant of environmental actions. This suggests that exporters are more likely to undertake environmental actions but this likelihood is not influenced by the degree of exports. We found that exporters are between 1.30 and 2.02 times more likely than non-exporters to adopt EAs.

161

labelled with letters corresponding to Q.502 (and in the notes at the bottom of the table). Table 5 shows firstly that foreignowned firms are more likely than domestic firms to adopt EAs due to firm policy (motiveD), presumably following orders from overseas headquarters, and to prepare for environmental certification (motiveF). We also find that firms who are in networks with local customers are more likely to adopt EAs to imitate local competitors (motiveG) and in accordance with firm policy (motiveD). Similarly, firms that simultaneously have a high percentage of skilled workers and backward and forward linkages are more likely to adopt EAs to prepare for environmental certification. Hence, there is broad support for Hypothesis 4. We find that large firms, as measured by the number of workers, are more likely than small firms to adopt EAs to improve their image (motiveA). They are also more likely to adopt EAs to meet local environmental regulations (motiveB) (perhaps because such firms are more visible to environmental regulators), to meet the requirements of foreign markets and to prepare for environmental certification (motiveF). This latter point is consistent with Table 3, which found large firms to be more likely than small firms to adopt environmental certification. Table 5 also indicates that independent firms are less likely to adopt EAs because of firm policy (motiveD), to meet the requirements of foreign markets (motiveE), and to prepare for environmental certification (motiveF) but are more likely to adopt EAs to imitate local competitors (motiveG). Reassuringly, we find exporters are more likely than non-exporters to adopt EAs to meet the requirements of foreign markets (motiveE). Exporters are also more likely to implement EAs in order to prepare for environmental certification (motiveF) and to improve the image of the firm (motive A). 23

4.1. Sensitivity analysis To test the sensitivity of our results to changes in our specification and choice of independent variables we implemented a range of robustness checks. For reasons of space we are unable to report full results, but all are available upon request. We show that our results are not unduly influenced by outliers (through the use of dfbetas24) and we also test alternative measures of our variables where available. For example, we tested a continuous measure of foreign ownership as well as dummies based on 25% and 50% ownership thresholds. We also measure size as output rather than total workforce and dummy variables of R&D and exports rather than continuous measures. We also tested national and regional measures of our network variables rather than the local measures reported here. TFP, measured following the procedure by Olley and Pakes (1996), was used in the place of labour productivity, albeit for a much smaller sample. A number of alternative estimation procedures were also tested, including the use of a Probit model instead of Logit. In all of the above cases, our key findings were unaffected leading us to conclude that our results are robust. Finally, although the correlation matrix in the Appendix indicates that most correlations between

23 The diagnostic tests in Table 5 are again supportive of our modelling approach. 24 Dfbetas focus on one coefficient and measure the difference between the regression coefficient when the ith observation is included and excluded, the difference being scaled by the estimated standard error of the coefficient. Bollen and Jackman (1990) argue that an observation is deserving of special attention if jdfbetaj > 1, implying that the observation shifted the estimated coefficient by at least one standard error. We find no dfbetas that exceed 1.

1.00 0.29 1.00 0.60 0.27 1.00 0.027 0.057 0.040 1.00 0.12 0.0075 0.017 0.092 1.00 0.047 0.0044 0.023 0.040 0.032 1.00 0.082 0.027 0.037 0.079 0.014 0.042 1.00 0.055 0.25 0.033 0.0004 0.084 0.029 0.025 1.00 0.0032 0.015 0.029 0.019 0.040 0.025 0.066 0.0048 1.00 0.12 0.014 0.21 0.0095 0.061 0.073 0.0004 0.028 0.014 1.00 0.21 0.049 0.0034 0.027 0.0045 0.010 0.096 0.029 0.025 0.011 1.00 0.027 0.18 0.088 0.043 0.24 0.054 0.14 0.068 0.052 0.015 0.033 1.00 0.21 0.017 0.021 0.074 0.014 0.10 0.026 0.10 0.12 0.017 0.011 0.071 1.00 0.20 0.45 0.036 0.16 0.12 0.071 0.26 0.043 0.11 0.19 0.081 0.039 0.056

NetHoriz NetCust NetSupp Backward Invsales Perskld PerRD PerExpts Labprod Salesgr Indpt Size FO10 EA

Appendix. Correlation matrix

This paper tests four complementary hypotheses related to the characteristics of firms who implement EAs and the motivation behind these decisions. We make a number of interesting findings. First, our results indicate that foreign-owned firms are more likely to implement EAs than domestic firms (Hypothesis 1). Second, we find evidence of positive environmental spillovers from foreign-owned firms to domestic ones. We also find evidence to suggest that absorptive capacity of domestic firms is important (Hypothesis 2). The effect of foreign presence in increasing the likelihood of EA adoption is greater the greater the domestic firm's absorptive capacity. Surprisingly, we also find strong evidence of spillovers from foreign-owned firms to other foreign-owned firms, again perhaps emphasising the role of absorptive capacity. Furthermore, domestic firms are more likely to implement EAs if they have formal and informal links with their customers and if they supply customers who are based in industries containing foreign-owned firms (Hypothesis 3). Finally, we also find that motives for undertaking EAs differ between foreign-owned and domestic firms. For example, foreign-owned firms are more likely than domestic firms to adopt EAs merely as a result of firm policy or in order to prepare for environmental certification (Hypothesis 4). While our results indicate that foreign owned firms in Argentina are more likely to adopt EAs than domestically owned firms, there are several reasons why we remain cautious about claiming that foreign investment is good for the environment more generally. First, the link between EAs and actual environmental performance remains unclear, largely due to data limitations. Second, even if foreign owned firms are cleaner per unit of output than domestic firms, if the output of the foreign-owned firms is in addition to that of domestic firms, rather than at the expense of domestic output, then foreign investment is still likely to result in a net increase in pollution and resource use. Finally, it is worth noting that our results do not preclude the possibility that foreign affiliates in less developed countries are still more pollution intensive than those in developed countries and hence may be taking advantage of lax or poorly enforced regulations. The implications of our findings for firms are that in order to mitigate against future government regulation costs it is important to learn cleaner production techniques from neighbouring firms and to ensure that skilled labour and capital are available to take advantage of this opportunity to learn. Locating close to foreign firms might also be beneficial for both firms and the local environment. An obvious deficiency of our dataset is the lack of a time-series dimension, something we hope to rectify in future research. We intend to contribute to the next survey design by including questions on actual emissions and the role of government via information provision, environmental finance or technology transfer facilitation. Such questions would allow us to better understand peer effects between firms and the ways in which policy makers can influence these linkages.

Forward

5. Conclusions

1.00 0.23 0.19 0.20 0.045 0.078 0.077 0.12 0.18 0.033 0.16 0.14 0.13 0.061 0.057

explanatory variables are relatively small (on average 0.055), some are larger with the largest being between the variables Independent and FO10 (0.45). As a result we therefore estimate Equation (2) omitting one explanatory variable at a time. In each case the sign and significance of the remaining variables was unaffected, suggesting that multicollinearity is not unduly influencing our key results.

1.00

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

EA FO10 Size Independent Salesgr Labprod Perexports PerRD Perskilled Invsales Backward Forward NetSupp NetCust NetHoriz

162

F. Albornoz et al. / Journal of Environmental Management 146 (2014) 150e163

References Albornoz, F., Yoguel, G., 2004. Competitiveness and production networks: the case of the argentine automotive sector. Ind. Corp. Change 13 (4), 619e642. Albornoz, F., Cole, M.A., Elliott, R.J.R., Ercolani, M.G., 2009. Search of environmental spillovers. World Econ. 32 (1), 136e163. Aitken, B.J., Harrison, A.E., 1999. Do domestic firms benefit from direct foreign investment? Evidence from Venezuela. Am. Econ. Rev. 89, 103e132. Aragon-Correa, J.A., Sharma, S., 2003. A contingent resource-based view of proactive environmental strategy. Acad. Manag. Rev. 28 (1), 71e88. Arora, S., Cason, T.N., 1995. An experiment in voluntary environmental regulation: participation in EPA's 33/50 program. J. Environ. Econ. Manag. 28, 271e286. Arora, S., Cason, T.N., 1996. Why do firms exceed environmental regulations? Understanding participation in EPA's 33/50 program. Land Econ. 72, 413e432. Audretsch, D.B., Feldman, M.P., 1996. R&D spillovers and the geography of innovation and production. Am. Econ. Rev. 86 (4), 253e273. Barrios, S., Dimelis, S., Louri, H., Strobl, E., 2004. Efficiency spillovers from foreign direct investment in the EU periphery: a comparative study of Greece, Ireland, and Spain. Rev. World Econ. 140 (4), 688e705. Bernard, A.B., Jensen, J.B., 1999. Exceptional exporter performance: cause, effect, or both? J. Int. Econ. 47, 3e31. Bernard, A.B., Jensen, J.B., Redding, S., Schott, P., 2007. Firms in international trade. J. Econ. Perspect. 21 (3), 105e130. Blaikie, P.M., Brookfield, H., 1986. Land Degradation and Society, London and New York: Methuen. € m, M., Kokko, A., 1998. Multinational corporations and spillovers. J. Econ. Blomstro Surv. 12, 247e277. Bollen, K.A., Jackman, R.W., 1990. Regression diagnostics: an expository treatment of outliers and influential cases. In: Fox, J., Long, J.S. (Eds.), Modern Methods of Data Analysis. Sage Publications, Newbury Park, California. Cantwell, J., 2009. Location and the multinational enterprise. J. Int. Bus. Stud. 40, 35e41. Christmann, P., Taylor, G., 2001. Globalization and the environment: determinants of firm self-regulation in China. J. Int. Bus. 32 (3), 439e458. Cohen, W., Levinthal, D., 1990. Absorptive capacity: a new perspective on learning and innovation. Adm. Sci. Q. 35, 128e152. Cole, M.A., Elliott, R.J.R., 2005. FDI and the capital intensity of ‘dirty’ sectors: a missing piece of the pollution haven puzzle. Rev. Dev. Econ. 9 (4), 530e548. Cole, M.A., Elliott, R.J.R., Okubo, T., Zhou, Y., 2013. The carbon dioxide emissions of firms: a spatial analysis. J. Environ. Econ. Manag. 65 (2), 290e309. Cole, M.A., Elliott, R.J.R., Shimamoto, K., 2005. Industrial characteristics, environmental regulations and air pollution: an analysis of the UK manufacturing sector. J. Environ. Econ. Manag. 50 (1), 121e143. Cole, M.A., Elliott, R.J.R., Shimamoto, K., 2006. Globalisation, firm-level characteristics and environmental management: a study of Japan. Ecol. Econ. 59 (3), 312e323. Cole, M.A., Elliott, R.J.R., Strobl, E., 2008. The environmental performance of firms: the role of foreign ownership, training and experience. Ecol. Econ. 65, 538e546. Dasgupta, S., Hettige, H., Wheeler, D., 2000. What improves environmental performance? Evidence from the Mexican industry. J. Environ. Econ. Manag. 39, 39e66. DeCanio, S.J., Watkins, W.E., 1998. Investment in energy efficiency: do the characteristics of firms matter? Rev. Econ. Stat. 80, 95e107. Diller, J.M., 1997. ISO 14001 certified: to be or not to be. Chem. Eng. Prog. 93 (11), 36e39. Doukas, J.A., Lang, L.H.P., 2003. Foreign direct investment, diversification and firm performance. J. Int. Bus. Stud. 34, (2), 153e172. Driffield, N., Love, J., 2007. Linking FDI motivation and host economy productivity effects: conceptual and empirical analysis. J. Int. Bus. Stud. 38 (2), 460e473. Driffield, N., Munday, M., 2000. Industrial performance, agglomeration and foreign manufacturing investment in the UK. J. Int. Bus. Stud. 31 (1), 21e37. Dunning, J.H., 1998. Location and the multinational enterprise: a neglected factor? J. Int. Bus. Stud. 29 (1), 45e66. ECLAC, 2001. Economic Survey of Latin America and the Caribbean, 2000e2001. LC/ G.2139-P/I.  , J.A., 2009. Managing external knowledge flows: the Escribano, A., Fosfuri, A., Tribo moderating role of absorptive capacity. Res. Policy 38 (1), 96e105. Eskeland, G.S., Harrison, A.E., 2003. Moving to greener pastures? Multinationals and the pollution haven hypothesis. J. Dev. Econ. 70 (1), 1e23. George, G., Zahra, S.A., 2002. Absorptive capacity: a review, reconceptualization, and extension. Acad. Manag. Rev. 27 (2), 185e203. Gereffi, G., 2002. Prospects for industrial upgrading by developing countries in the global apparel commodity chain. Int. J. Bus. Soc. 3 (1), 27e60. Halkos, G.E., Evangelinos, K.I., 2002. Determinants of environmental management systems standards implementation: evidence from Greek industry. Bus. Strategy Environ. 11, 360e375. Hanson, G., 2005. In: Moran, T.H., Graham, E.M., Blomstrom, M. (Eds.), Does Foreign Direct Investment Promote Development?. Institute for International Economics.

163

Haskel, J.E., Pereira, S.C., Slaughter, M.J., 2007. Does inward foreign direct investment boost the productivity of domestic firms? Rev. Econ. Stat. 89 (3), 482e496. Henriques, I., Sadorsky, P., 1996. The determinants of an environmentally responsive firm: an empirical approach. J. Environ. Econ. Manag. 30, 381e395. n y Conducta Tecnolo gica de INDEC, 2002. Segunda Encuesta Nacional de Innovacio las Empresas Argentinas 1998e2001. In: Serie de Estudios del INDEC Number 38. Ives, J.H., 1985. The Export of Hazard. Transnational Corporations and Environmental Control Issues. Routledge and Kegan Paul, , London. Khanna, M., Anton, W.R.Q., 2002. Corporate environmental management: regulatory and market-based pressures. Land Econ. 78, 539e558. Kulfas, M., Porta, F., Ramos, A., 2002. Foreign Investment and Multinational Corporations in Argentina. ECLAC Documents N 10. Lall, S., 1980. Vertical inter-firm linkages in LDCs: an empirical study. Oxf. Bull. Econ. Statistics 42 (3), 203e226. Laplante, B., Lanoie, P., 1994. The market response to environmental incidents in Canada: a theoretical and empirical analysis. South. Econ. J. 60, (3), 657e672. Levy, D.L., 1995. The environmental practices and performance of transnational corporations. Transnatl. Corp. 4 (1), 44e68. Lien, Y.-C., Poesse, J., Strange, R., Filatotchev, I., 2005. The role of corporate governance in FDI decisions: evidence from Taiwan. Int. Bus. Rev. 14 (6), 739e763. Liu, X., Siler, P., Wang, C., Wei, Y., 2000. Productivity spillovers from foreign direct investment: evidence from UK industry level panel data. J. Int. Bus. 31 (2), 407e425. Malecki, E., 1997. Technology and Economic Development: the Dynamics of Local, Regional and National Competitiveness, second ed. Addison Wesley Longman, London). Marin, A., Bell, M., 2006. Technology spillovers from foreign direct investment (FDI): the active role of MNC subsidiaries in Argentina in the 1990s. J. Dev. Stud. 42 (4), 678e697. Meyer, K.E., 2003. FDI Spillovers in Emerging Markets: a Literature Review and New Perspectives. DRC Working Paper, n.15. Centre for New and Emerging Markets, London Business School. Moulton, B.R., 1990. An illustration of a pitfall in estimating the effects of aggregate variables on micro units. Rev. Econ. Stat. 72 (2), 334e338. Nakamura, M., Takahashi, T., Vertinsky, I., 2001. Why Japanese firms choose to certify: a study of the managerial response to environmental issues. J. Environ. Econ. Manag. 42 (1), 23e52. Newall, P., 2012. Globalization and the Environment: Capitalism, Ecology and Power, Polity. Olley, S.G., Pakes, A., 1996. The dynamics of productivity in the telecommunications equipment industry. Econometrica 64 (6), 1263e1297. Pargal, S., Wheeler, D., 1996. Informal regulation of industrial pollution in developing countries: evidence from Indonesia. J. Polit. Econ. 104 (6), 1314e1327. Podsakoff, P.M., MacKenzie, S.B., Lee, J.Y., Podsakoff, N.P., 2003. Common method biases in behavioral research: a critical review of the literature and recommended remedies. J. Appl. Psychol. 88 (5), 879e903. Porter, M.E., Van der Linde, C., 1995. Toward a new Conception of the environmentcompetitiveness relationship. J. Econ. Perspect. 9 (4), 97e118. Saito, H., Gopinath, M., 2011. Knowledge spillovers, absorptive capacity, and skill intensity of chilean manufacturing plants. J. Reg. Sci. 51 (1), 83e101. Schmitz, H., 1995. Collective efficiency: growth path for small-scale industry. J. Dev. Stud. 34 (4), 529e566. Sheldon, C., 1997. ISO 14001 and Beyond: Environmental Management Systems in the Real World. Greenleaf Publishing, Sheffield. Smarzynska Javorcik, B., 2004. Does foreign direct investment increase the productivity of domestic firms? In search of spillovers through backward linkages. Am. Econ. Rev. 94 (3), 605e627. Smarzynska Javorcik, B.S., Wei, S.J., 2004. Pollution Havens and Foreign Direct Investment: Dirty Secret or Popular Myth?. Contributions to Economic Analysis & Policy: 3, (2): Article 8. UNCTAD, 2002. Managing the Environment Across Borders. United Nations, New York. Varga, A., 2000. Local academic knowledge spillovers and the concentration of economic activity. J. Reg. Sci. 40, 289e309. Veugelers, R., Cassiman, B., 2004. Foreign subsidiaries as a channel of international technology diffusion: some direct firm level evidence for Belgium. Eur. Econ. Rev. 48 (2), 455e476. Wallace, D., 1996. Sustainable Industrialization. Earthscan, London. Wei, Y., Liu, X., 2006. Productivity spillovers from R&D, exports and FDI in China's manufacturing sector. J. Int. Bus. Stud. 37 (4), 544e557. World Health Organisation, 2006. Preventing Disease Through Healthy Environments: Towards an Estimate of the Environmental Burden of Disease. WHO, Geneva. Young, M.N., Peng, M., Ahlstrom, D., Bruten, G.D., 2008. Corporate governance in emerging economies: a review of the principle perspective. J. Manag. Stud. 45 (1), 196e220. Zarsky, L., 1999. Havens, Halos and Spaghetti: Untangling the Evidence About Foreign Direct Investment and the Environment. OCED, Paris.