ARTICLE IN PRESS Journal of Environmental Management 88 (2008) 1552–1561 www.elsevier.com/locate/jenvman Environmental performance and spillover effects on productivity: Evidence from horticultural firms Emilio Galdeano-Gómeza, José Céspedes-Lorenteb,, Javier Martı́nez-del-Rı́ob a Departamento de Economı´a Aplicada, Universidad de Almerı´a, Cañada de San Urbano s/n, 04120 Almerı´a, Spain Departamento de Dirección y Gestión de Empresas, Universidad de Almerı´a, La Cañada de San Urbano, s/n, 04120 Almerı´a, Spain b Received 17 January 2007; received in revised form 8 July 2007; accepted 31 July 2007 Available online 6 September 2007 Abstract This study investigates the effect of environmental investment and related spillover effects on productivity in the agricultural sector by using a panel data of horticultural firms in Andalusia (Southern Spain). The results indicate a positive relationship between firm investment in environmental practices and productivity improvement, also showing the presence of positive environmental spillovers. In a second-stage of analysis, the incidence of environmental factors in firm specific individual technical efficiency is estimated. This analysis also shows the link between environmental knowledge diffusion and horticultural firms’ performance. r 2007 Elsevier Ltd. All rights reserved. Keywords: Environmental performance; Spillovers; Productivity; Cobb–Douglas function; Efficiency; Horticultural firms 1. Introduction The increasing awareness of environmental problems as a result of economic activity has led to greater political and social demands on companies to reduce their environmental impact. One argument supporting the diffusion of environmental innovations on a company level is the socalled ‘‘win–win hypothesis’’. According to this hypothesis, firms that increase their investment in environmental technology can obtain a competitive advantage, while reducing their negative environmental impact (Porter and van der Linde, 1995). Previous studies on environmental management strategy has been based on theoretical approaches such as the resource-based view to argue that investment in green technology may foster the development of firm’s resources and capabilities which form the basis for a competitive advantage (Hart, 1997; Aragón-Correa and Sharma, 2003). For instance, Sharma and Vredenburg (1998) point out that Corresponding author. Tel.: +34 950 015523; fax: +34 950 015178. E-mail addresses: galdeano@ual.es (E. Galdeano-Gómez), jcespede@ual.es (J. Céspedes-Lorente), jamartin@ual.es (J. Martı́nez-del-Rı́o). 0301-4797/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvman.2007.07.028 investment in proactive environmental practices (e.g. pollution prevention) contributes to the development of valuable capabilities such as innovation, organizational learning and stakeholder integration. Those organizations that develop these capabilities related to environmental management are able to obtain greater financial performance. Most of the above-mentioned studies assume that the origin of capabilities associated with environmental investment is internal, including complementary with other assets or resources (e.g. innovation; see Christmann, 2000). Therefore, they do not analyse the moderating role of the characteristics of the companies’ environment (e.g. spillover effects in a network of companies) to explain the relationship between environmental investment and financial performance (Aragón-Correa and Sharma, 2003). One important external factor that moderates the relationship between environmental investment and firm performance is the diffusion of knowledge associated to networks and industrial agglomeration (spillover effects). As noted by Mazzanti and Zoboli (2006), environmental innovations are particularly important in local industrial frameworks since they may give rise to a ‘‘double externality’’: (1) the reduction of environmental externalities (Jaffe et al., 2005) and (2) the typical R&D spillover effect. ARTICLE IN PRESS E. Galdeano-Gómez et al. / Journal of Environmental Management 88 (2008) 1552–1561 Spillovers are usually connected with R&D stock as main knowledge capital. In the framework of productivity analysis, spillovers may be considered as another input of the knowledge production process of a firm, industry or country (e.g. Griliches, 1992; Mairesse and Mohnen, 2002). Likewise, spillover effects associated to the transfer of environmental knowledge, which is facilitated by the proximity of firms operating in the same geographical region, may be considered as an input of a firm production function. Most studies on environmental management and spillovers tend to focus on the industrial sector. Nevertheless, when the agricultural sector is considered, and especially fresh consumption products (whose transformation for marketing is minimal), environmental factors may have important implications on the production function (McNally, 2002). In this line, some works have related environmental practices and management as one of the main factor of innovation in recent years (e.g. Chambers and Eisgruber, 1998; Brouwer et al., 2000). The objective of our empirical analysis is to analyse environmental practices spillovers in horticultural firms and their impact on firm’s productivity. To this end, we use insights from previous works which stress the role of interaction between companies located in the same geographical area (e.g. geographical clusters) in the development of resources and capabilities in two levels: the cluster or group of companies and the individual company (Tallman et al., 2004; McEvily and Zaheer, 1999; McEvily and Marcus, 2005). Whilst the knowledge exchange through technology spillovers or informal exchange is critical to define performance at cluster level, the firm’s specific knowledge is critical to explain the generation of competitive advantage for individual firms (Tallman et al., 2004). This study makes two main contributions. First, it studies the relationship between firms’ environmental management and their individual financial performance (as measured by productivity), taking into account the direct and moderating roles that the environmental innovation spillover effect has on the local agricultural production framework. We assumed that heterogeneity characterizes the development of strategic capabilities related to environmental investment in companies belonging to the same geographical area and network resources are assumed to form part of the origin of these capabilities (McEvily and Zaheer, 1999). Second, the empirical analysis centres on the agricultural sector, which has traditionally received scant attention in studies on environmental management and financial performance (Carpentier and Ervin, 2002). Analysis of this sector can complement previous studies by relating the strategic characteristics of companies dedicated to the production and marketing of agricultural produce to the development of strategic capabilities associated with environmental management. 1553 2. Environmental management in the agricultural and horticultural sector Environmental management and performance in the agricultural sector have received scant attention compared to the industrial sector (Carpentier and Ervin, 2002). The scale of firms, the nature of operations (diffuse pollution in comparison with larger industries), the relative lack of environmental regulation and the voluntary character of the policy programs, among other factors, may explain the lack of studies on this topic. Nevertheless, the OECD (Organization of Economic Cooperation and Development), identifies five drivers that promote the implementation of environmental practices in agri-business, including government policies and regulations, commercial and economic considerations, corporate image, codes of conduct, and growing pressure from the financial/investment community (OECD, 1998). Carpentier and Ervin (2002) show six types of motivation: (1) improvement in productivity associated to the creation of integrated production and marketing systems and other tasks necessary to implement an environmental management program (Esty and Porter, 1998, Reinhardt, 2000); (2) differentiating products to meet demand requirements (for example, green consumers); (3) mitigating future environmental regulations; (4) strategically managing competitors (incurring additional costs to improve environmental performance may increase some firms’ profits if the actions cause competitors’ costs to rise even further than their own); (5) capturing more value in the market, combining cost reduction, product differentiation, and competitor management to shift market conditions and capture more value along the supply and marketing chains (Reinhardt, 2000), and (6) managing risk and uncertainty more effectively. Evidence on most of these motivations is sparse in agriculture (Céspedes and Galdeano, 2004). Only a few recent empirical studies have referred to environmental management in agri-business. A review of the empirical evidence by the OECD (2001) concludes that overall there appears to be a positive correlation between environmental and commercial/financial performances. For example, Asmundson and Foerster (2001) show that companies that invest in business environmental management can do at least as well economically as other companies. In capturing more market value, Chambers and Eisgruber (1998) describe how farm managers lessened their ‘‘environmental footprint’’ by using diverse crop rotations, growing nitrogen sources, reducing or eliminating pesticides, protecting ground water, engaging in soil and product residue testing, water reuse, recycling and composting. Horticultural firms from Andalusia, in Southern Spain form the basis of our empirical analysis. Traditionally, this horticultural system, based on greenhouses, has implied intensive use of soil, water, fertilizers, insecticides, and raw materials (plastics, packaging, etc.) generating high levels of waste. Environmental management control in Spanish ARTICLE IN PRESS 1554 E. Galdeano-Gómez et al. / Journal of Environmental Management 88 (2008) 1552–1561 horticultural firms was intensified during the 1990s, when different European Common Agricultural Policy programmes were applied to environmental management. Representative horticultural firms in our empirical setting have a cooperative or associative nature, and according to Community Regulation, they are classified as Organizations of Producers (OP).1 The main objective of these firms is to manipulate and commercialize their associates’ (farmers) products in their warehouses and installations linking the farming and marketing activities. They are proving to be key elements in the development of environment-friendly practices due to their direct contact with farmer members, which makes it easier to explain environmental regulation and to apply controls and audits. These entities can develop environmental investment practices, which result in increases in productivity, but which the growers as individuals would be unable to make due to the small scale of their firms. In addition, most of the OPs are integrated in local associations. These and other organizations within the agricultural production system play different representation roles of the collective (for instance, negotiating with public administrations the elaboration of incentive policies for the adoption of environmental practices), research centres, exports promotion, etc. These entities act as the central organization of a geographical cluster (McEvily and Zaheer, 1999) and take on the tasks of training and the diffusion of information on ‘‘best practices’’(e.g. of environmental and food quality). Thus, a number of horticultural firms are increasing their investment in the implementation of environmental management practices, which are classified into two main groups by the Andalusian Council of Agriculture. One set of investment aims at preventing environmental impacts through recycling vegetable and plastic waste, reducing the use of fertilizers, monitoring pesticide concentration throughout the production process, diminishing soil pollution and water consumption (Downward and Taylor, 2007), etc. The other set of practices aims to instill ecological principles in agricultural production through the implementation of integrated pest management, the renewal of irrigation systems, the implementation of technologies that prevent soil pollution (e.g. mesh nets to prevent the action of insects), as well as the introduction of control mechanisms for continual monitoring of soils, water, and waste. Another consideration in our empirical framework refers to the category of these environmental investments (i.e. ‘‘preventive’’ or ‘‘end-of-pipe’’), since each type may have different implications on the results (e.g. Sharma and Vredenburg, 1998; Aragón-Correa and Sharma, 2003). In 1 In the European agricultural model, expectations for attaining sustainable and competitive agriculture lie to a great extent on the cooperative sector’s ability to adapt to the new market conditions. In the European Union (EU), cooperative entities are responsible for over 60% of the harvest, handling and marketing of agricultural products, with a turnover of approximately 210 000 million euros (GCAC, 2000). our case, considering the characteristics of the activity in horticultural firms (for example, diffuse pollution in comparison with larger industries, intensive use of specific natural resources such as water and soil, etc.) investment in ‘‘end-of-pipe’’ management practices are relatively low. In addition, the recent application of environmental management programs (and the voluntary nature of their adoption) and the scale of firms have led to a proactive application (e.g. Carpentier and Ervin, 2002). Thus, as we have described, much of the environmental investment can be considered as preventive practices. 3. Hypotheses 3.1. Environmental management and performance There is a wide body of research addressing the relationship between environmental practices or capabilities and firms’ performance in different industries (e.g. Klassen and Whybark, 1999; Russo and Fouts, 1997; Schaltegger and Synnestvedt, 2002) and also in agricultural companies (Nijkamp and Vindrini, 2002). The distinction between cost and differentiation advantages can be useful to discuss these effects in the horticultural sector (Shrivastava, 1995). Cost advantages typically arise from the adoption of practices that improve the production process (Hart, 1995), increasing its efficiency and reducing input and waste disposal costs (Hart, 1995; Shrivastava, 1995). As Christmann (2000; p. 664) pointed out, ‘‘these process-focused practices include redesigning of production processes to be less polluting, substituting less polluting inputs, recycling by-products of processes, and innovating less polluting processes’’. In horticulture typical examples of cost advantages from environmental practices are selling to recycling plants the plastic greenhouses are made of, the use of smaller amounts of more accurate fertilizers and pesticides, improvements in the irrigation system in order to save water, and avoiding product take-back costs derived from not fulfilling environmental standards (Ecker and Coote, 2005). Differentiation advantages typically arise from the customers’ perception that the product is more valuable. Thus it usually depends on the fit of product characteristics and market needs, and on companies’ ability to market environmental characteristics of their products and services. Organic horticultural products are a typical example of value added by environmental practices. Environmentally sound agricultural products have better market access (Ecker and Coote, 2005), and customers perceive those products as healthier and fulfilling higher quality standards (Walley et al., 2000). Therefore, given the potential of environmental practices in horticultural production to generate competitive advantages and superior performance derived from product differentiation and cost advantages, we propose the following hypothesis: ARTICLE IN PRESS E. Galdeano-Gómez et al. / Journal of Environmental Management 88 (2008) 1552–1561 Hypothesis 1. The stock of a horticultural firm’s investments in environmental practices is positively related to its performance. 3.2. Knowledge spillovers and environmental management practices It has been suggested that geographical proximity increases the probability of knowledge diffusion between organizations (Maskell and Malmberg, 1999; Tallman et al., 2004). Similar concepts have been proposed by researchers to describe the development of specific knowledge in geographical clusters. For instance, constructs such as ‘‘knowledge in the air’’ (Malmberg and Maskell, 2002), ‘‘social capital’’ (Bowles and Gintis, 2002), ‘‘untraded interdependencies’’ (Storper, 1997), or ‘‘learning networks’’ (Maskell, 2001) have been suggested in the fields of geographical economy and business strategy to explain the competitive advantages of industrial districts (Tallman et al., 2004). Storper (1997; p. 12) defines spillover effects as ‘‘relations of interdependency between one type of knowledge and another, whose frontiers change, (y) requiring interpretation and communication’’. Spillovers are dynamic processes of knowledge diffusion between different organizations. Interactive feed-back between actors plays an important role in those processes, making this knowledge evolve and improve through interactions over time. Firms can gather tacit and explicit information through formal and informal interactions with other organizations in the area—and the people working in them—about issues such as technological paths followed by competitors or the success or failure of tactical and strategic decisions of competitors. Those interactions between managers and specialized workers of different organizations may be due to idiosyncratic circumstances derived from sharing geographical location (Malmberg and Maskell, 2002; McEvily and Zaheer, 1999), e.g. casual personal relations, accidental personnel exchange, providing to the same customers and having the same providers, contacts with the same local institutions—universities, regional associations, public administrations, etc.—or attending industry meetings and fairs. Thus companies share tacit knowledge that is used in the decision process (Maskell, 2001, Pouder and St John, 1996; Tallman et al., 2004). For example, if a company chooses a technological option for an environmental problem which turns out to be a failure, its competitors will probably prefer another option, whose success or failure will also be known by its rivals and providers in the same geographical area. Therefore, repeated interactions between all the companies in the industry, and the resulting spillover effects, generate a collective ‘‘knowledge-in-the air’’ that is widespread for the companies in the area, but is very difficult to observe for firms located anywhere else (Tallman et al., 2004; Pouder and St. John, 1996). 1555 Tallman et al. (2004) argue that companies in geographical clusters share firm specific and cluster specific knowledge that helps them to develop capabilities which lead to superior performance compared to the firms not placed in the geographical area. Bansal and Roth (2000) found positive externalities of environmental practices in highly polluting industries. They suggest that if an industry implements environmental practices which go beyond pollution control and fulfilling legal regulations, this industry can obtain competitive advantages resulting from a better reputation. Thus, if most agricultural companies in a geographical area implement environmental practices, all the firms in the area—implementing environmental practices or not— could benefit from a better area-related reputation. The contrary also holds true. If most horticultural companies in the area develop a negative reputation for massive use of pesticides, no control of product quality or contamination, extensive use of highly polluting plastics and fertilizers, etc., all the companies in the region will suffer negative bias from customers. This argument is reinforced in the case of the ecological production of horticultural products. Firstly, geographical identification of agricultural products is frequent. Thus a positive area-of-origin reputation might have an important influence on performance. Secondly, besides their reputation for not polluting the environment, organic foods are very often associated by consumers with superior product quality, which is not always the case in other industries (e.g. chemical industries, auto-industry, paper, etc.). The geographical concentration of environmental horticultural production supposes a bigger market for providers of technical solutions for environmental problems. Hence it implies greater attention from them and better access to the best solution available to the companies that are trying to implement such practices. Industry concentration in a geographical area also makes more probable the flow of knowledgeable employees and engineers from one company to one another (e.g. Almeida and Kogut, 1999; Maskell, 2001), spreading valuable information and capabilities about environmental management between organizations. Maier and Finger (2004) suggest that organic agricultural foods also tend to be challenged by marketing problems such as supply regularity and product labelling. These problems can be better faced by a company if it is surrounded by many others in the same geographical area producing the same products. One single company cannot provide a market every day, but many companies grouped together can guarantee constant supply, at least during the season. Local companies can also reach agreements in order to use a homogeneous style of labelling that can be recognized by wholesalers and consumers. All of these arguments suggest the following hypothesis: Hypothesis 2. Environmental spillover effect is positively related to single horticultural firm’s performance. ARTICLE IN PRESS 1556 E. Galdeano-Gómez et al. / Journal of Environmental Management 88 (2008) 1552–1561 An industry’s particular characteristics may moderate the potential of environmental capabilities to generate competitive advantages for the company and therefore improve its performance. For example, environmental uncertainty, complexity, and munificence (e.g. AragonCorrea and Sharma, 2003; Russo and Fouts, 1997) have been studied as moderators of the effect of environmental practices on firms’ profitability. Spillovers reduce uncertainty about the implementation of environmental practices in horticultural production. Consider an individual who would like to implement a concrete environmental practice—e.g. pollinization by natural means (bees) instead of ‘‘artificial’’ ones—and is willing to risk at least one crop season to do so. Typical issues he or she will confront include finding reliable providers, finding information on which moment of the season is the most appropriate, learning about optimal conditions for bees, uncertainty about the whole process productivity, etc. Similar issues could arise with other environmental practices, such as using natural compost instead of industrial fertilizers or using little or no pesticide amounts. Without information on these issues, this individual would have to face considerable uncertainty, which diminishes the probability of implementation of environmental practices (Aragón-Correa and Sharma, 2003). If there is a greater number of surrounding firms in the area, formal and informal networks will emerge (Inkpen and Tsang, 2005), producing a dissemination of information. This information can overcome hurdles and uncertainty for the introduction of environmental practices, resulting in wider and more efficient implementation, and leading better performance of local firms compared with those located elsewhere. McEvily and Zaheer (1999) and McEvily and Marcus (2005) found empirical support for the influence of formal and informal local networks of managerial contacts on the development of environmental management capabilities associated with superior performance. Russo (2003) found increasing geographical concentration over time of wind energy companies in California, USA due to the development of social capital and industry-specific knowledge. The dissemination of environmental knowledge on specific practices within the network of horticultural companies facilitates the development of ‘‘green competencies’’ on this collective level. According to the particular characteristics of each company (e.g. presence of complementary assets associated with environmental innovation), this general knowledge will be integrated with more or less efficiency in such a way that green competencies at firm level which give rise to a competitive advantage can evolve with varying degrees of intensity (Tallman et al., 2004). Thus, it is to be expected that horticultural companies which invest extensively in the implementation of environmental practices and therefore develop environmental capabilities can obtain greater benefit from the spread of specific environmental knowledge within the network. This discussion suggests that the environmental spillover effect moderates the relationship between firm’s investment in environmental practices and performance. Hypothesis 3. The industry investment in environmental practices in the region, moderates the relationship between a single horticultural firm’s investment in environmental practices and its performance. 4. Methods 4.1. Sample The empirical analysis has been based on balanced panel data using the annual financial reports of 56 farmingmarketing cooperatives located in Andalusia (South of Spain), over the period 1995–2002.2 Horticultural production of this region is representative of Spanish horticulture (particularly for vegetables) and it accounts for 33% of the total national volume. In terms of output the sample accounts for 62% of regional production and 20% of national horticultural production, taking the average figures of the period under study. 4.2. Empirical model Due to the specific characteristics of horticultural firms, we argue that a horticultural firm’s production function includes environmental performance and the related spillovers as the main components in the development of new technologies and management methods. We assume that the production function can be approximated by a Cobb–Douglas function expanded here with a measure of knowledge capital (see, for instance, Mairesse and Hall, 1996): Y it ¼ Aelt C ait Lbit K git eit , (1) where Y is a measure of output (value added in this case), C is the physical capital (usually plant and equipment), L is a measure of labour (as hours worked) and K is the knowledge capital (which can include the correspondent of the firm and spillovers); the subscripts refer to the firm i and the current year t; l is the rate of disembodied exogenous technical change (the time trend lt is usually replaced by time dummies in the estimation); a, b and g, the corresponding elasticities of the three defined inputs, and e is the random error term for the equation, reflecting the effect of unknown factors, differences in technologies across firms and other disturbances. Taking logs of Eq. (1), the equation to estimate is yit ¼ a þ lt þ acit þ bl it þ gkit þ it . (2) The exogenous variations of environmental spillover can influence the environmental investment decision of a firm and then have an impact on productivity. When the effect of knowledge spillover to firm i from an external source is taken into account, it is partly determined by firms and 2 These sample firms are quite homogeneous in size: the number of workers ranges between 65 and 315. ARTICLE IN PRESS E. Galdeano-Gómez et al. / Journal of Environmental Management 88 (2008) 1552–1561 serves as an endogenous variable (Jaffe et al., 1995; Chen and Yang, 2005; Adams, 2006). Thus, the expression for Kit (now KEit) should be a function including investment in environmental practices (as knowledge) and environmental spillover shown as follows: KE it ¼ f ðEPit ; ES it Þ, (3) where EPit is the stock of knowledge capital generated by firm i indexed by environmental investment, and ESit is the environmental spillover from other firms in the same industry. Since the interactive effect between EPit and ESit is unknown, following the approach of Basant and Fikkert (1996), we adopt a general functional form for knowledge KEit and modify the production function in Eq. (1) as follows: Y it ¼ Aelt eKEit C ait Lbit K git eit , (4) where the stock of environmental knowledge, KEit, takes the generalized Leontief linear functional form as follows (Basant and Fikkert, 1996; Chen and Yang, 2005): KE it ¼ gep ðEPit Þ1=2 þ ges ðES it Þ1=2 þ geps ðEPit Þ1=2 ðESit Þ1=2 . (5) Here we have three new parameters, gep, ges and geps, related to environmental factors. As regards environmental spillovers, we can consider ges as ‘‘direct effect’’ while geps reflects an ‘‘indirect effect’’ (moderation effect) on productivity (Chen and Yang, 2005). To implement the estimation, taking logs of Eq. (4) and substituting for KEit from Eq. (5) we obtain the following form3: yit ¼ a þ lt þ acit þ bl it þ gcp ðEPit Þ1=2 þ geps ðEPit Þ1=2 ðES it Þ1=2 þ it . ð6Þ Here cit and lit represent the natural logarithms of Cit and Lit, respectively. On exploring the relationship between environmental knowledge and productivity growth, we allow for the existence of individual effects, which are potentially correlated with the right-hand side regressors. The error term may be decomposed as it ¼ mi þ xit , (7) where mi stands for the firm (time-invariant) specific effect that accounts for the possible heterogeneity across firms (for example, in their technological efficiency), whereas xit is a white noise error term and reflects temporary effects with finite moments, and in particular: E(xit) ¼ E(xitxjs) ¼ 0 for t 6¼ s and i 6¼ j. In this way, the possibility of existing individual effects (mi), whether fixed or random, can be considered in the econometric analysis of panel data. The suitability of dealing with these effects as fixed or as random depends on their correlation or not with the explanatory variables. If 3 Nevertheless, the variables of environmental knowledge are not taken in logarithms and this specification does not lend itself to the constraint of constant returns to scale (Chen and Yang, 2005). 1557 the mi presents the above-mentioned correlation, the OLS (ordinary least squares) estimator is inconsistent and the WG (within group) estimator should be considered; in such a case, the mi constitutes a whole of additional coefficients (within group) in the model. If the mi represents a random variable (independent of explanatory variables) the OLS estimator would be consistent but inefficient because the error term is not white noise, and the GLS (generalized least squares) estimator should be suitable. Another important issue is that the estimations of production function are often affected by biases due both to simultaneity and to measurement errors in the inputs. In order to get the robust estimation of standard errors we can use the ‘‘long-difference’’ approach developed by Griliches and Hausman (1986).4 This consists of regressing the log difference of firm’s output between the starting and ending period of the ‘‘long’’ log difference in levels of capital input, labour input, and environmental knowledge variables. This will be applied in each estimation method. In a second-stage, it is worth here estimating the individual fixed effect to determine the incidence on the efficiency of each firm (especially the impact of environmental variables). To this end, we can consider a linear estimation method of mi (Novales, 1996): ^ m^ i ¼ ½s2m =ðTs2m þ s2 Þ  1T ðyi  X i bÞ, (8) where T is the sample period number (t ¼ 1, y, T), 1’T is a column vector (inverse) constituted by T ones, Xi is the explanatory variable vector and b represents the parameter vector estimated. This estimation can be interpreted as a residual proportion of GLS assigned to mi and determined by the relative variances sm2 and se2 (whose values are given through GLS and WG estimations). When the mi estimations are obtained, they may be used as technical efficiency measures (TE) making the following normalisation in which TE of firm i is bounded in the [0,1] interval: TE i ¼ expðm^ i  maxm^ j Þ (9) where max m^ j is the maximum sample of the estimated fixed effects, and m^ 2 is the estimated fixed effect of firm i. In order to determine the impact of different inputs considered, we suggest a linear model as follows5: TE i ¼ y0 þ y1ci þ y2 l i þ y3 ðEPi Þ1=2 þ y4 ðES i Þ1=2 þ y5 ðEPÞ1=2 ðES i Þ1=2 þ vi , ð10Þ 4 Another approach widely used is the Generalized Method of Moments. Nevertheless, for the sake of simplicity in estimations we consider that ‘long-difference’ may be a feasible remedial method. 5 We only consider the influence of the explanatory variables of the production function due to the main objective of the present work, i.e. the inference and the influences of environmental investment and spillover effects. An appropriate extension related to TE would be to consider other exogenous variables that determine the (in)efficiency of the firm, which can be considered in a first-stage approach (e.g. Battese and Coelli, 1995) or in a second-stage approach (e.g. Sickles, 2005). ARTICLE IN PRESS 1558 E. Galdeano-Gómez et al. / Journal of Environmental Management 88 (2008) 1552–1561 where ci, li, EPi and ESi are the firm means (i ¼ 1,...56) in the analysed period.6 Table 1 Descriptive statistics of variables Variables Mean Standard deviation Minimum Maximum ln VA ln C ln L (ES)1/2 (EP)1/2 (EP)1/2(ES)1/2 3.26 5.81 7.39 25.02 179.21 4572.70 4.08 8.23 5.64 41.29 92.16 1527.03 0.97 2.13 3.05 11.52 113.04 2858.13 5.22 11.39 9.82 63.96 204.07 5382.11 4.3. Variables specification In Eq. (6) the dependent variable, y, denotes a value added (VA) as firm’s output.7 This has been obtained from the gross value added (value of sales minus purchases, mainly the farmers’ production and packaging materials). The VA has been corrected for inflation using the output deflator constructed from the sector data contained in the National Accounts of Spain (base year 1995). The capital stock, C, is a constructed estimate of equipment and plant, also adjusted for inflation. The labour input, L, is calculated from the total hours worked in each year.8 In this case we have considered the existence of workers specialized in environmental management (technicians, engineers, etc.) of the firms analysed. In order to avoid excess environmental investment elasticity (as pointed out by Mairesse and Hall, 1996, regarding R&D elasticity) we have corrected the labour measure. To construct the variable EP, we can consider the accumulative effect on the value added, as is usual for R&D expenditures. Following this methodology, the EP as stock for period t, is obtained from the annual expenditure on environmental practices in the period t plus the accumulative expenditure up to period t1, applying a deflator (Z): EPit ¼ AEPit+(1  Z) EPi(t1). This can be developed as follows (Hall and Mairesse, 1995): EPi1 ¼ AEPi0 þ ð1  ZÞAEPið1Þ þ ð1  ZÞ2 AEPið2Þ þ . . . ¼ AEPil =ðgi þ ZÞ, ð11Þ where g represents the growth rate of environmental expenditure. As regards the deflator, Z, assuming that the depreciation may be different in comparison with traditional R&D capital, we opt to follow recent works on environmental input in Spanish firms (Garcés and Galve, 2001) who apply a deflator of 10%, considering two lagged years. In constructing the environmental spillover, ES, we adopt a similar method to EP, measuring the investments in environmental practices by Andalusian horticultural 6 In this Eq. (10) we assume that ni  IN (0,s2n) and the regression function of the estimated firm’s individual effects (time-invariant),m^ i , as well as the TEi, are linear in the means of all the characteristics in the inputs of production function (cross-section of firms), although they are independent of their time-variant inputs (i.e. for all i and t), relaxing the possible endogeneity problem (Novales, 1996). These assumptions can be considered reasonable especially when WG panel data estimators are used, as the correlation of the inputs with error term is corrected by demeaning all explanatory variables (e.g. Garcés and Galve, 2001). 7 We use the value added as dependant variable, as a better indicator of the environmental quality incorporated to the final product (Tyteca et al., 2002). 8 We opt for this measure due to the important seasonality in the number of employees in this sector, especially in handling workers. Number of observations ¼ 448. The variables in levels are measured in thousands of euros, except hours worked (L). firms other than firm i in this sector, as follows: ESit ¼ SJj ¼ 1 EPjt 8 j 6¼ i. 5. Estimation and results Descriptive statistics are shown in Table 1. The OLS, WG and GLS regressions have been previously estimated in order to select the most suitable estimation. The econometric programme used is Limdep (vs. 7.0); several groups of firms have been made by this programme for WG estimator, selecting a model with five groups (based on F and likelihood ratio tests) which show a more homogeneous behaviour in the sample of firms. WG (R2-adjusted ¼ 0.857) performs the estimation better than OLS (R2-adjusted ¼ 0.693) and GLS (R2-adjusted ¼ 0.649) estimations.9 Taking as reference the Breusch-Pagan test we have to assume the existence of firm’s individual effects,10 and considering the value of Lagrange multiplier ratio (Lagrange multiplier ¼ 118.45; p ¼ 0.000) these can represent a relevant differentiation across sample. Haussman test (chi-square ¼ 19.24; p ¼ 0.0028) supports the hypothesis that these effects are firm’s fixed effects. We therefore focus on the results of WG estimators. Table 2 shows WG estimation of three models. Model 1 excludes the spillovers variables; Model 2 excludes the interaction term (moderation effect), and Model 3 includes all regressors. Results suggest the physical capital (C) has a positive impact on the value added, although the result (0.227) is rather low compared to other analyses in the Spanish or international context.11 This coefficient is certainly higher for the OLS and GLS estimations, so it indicates that part of the capital variable impact may be reflected in the firm’s individual effects, mi, in the WG estimation. 9 These results are available upon request. Despite the mentioned homogeneity in firm size and performance in the sample, the results reveal that there are possible differences in the corporate culture, the know-how, technological opportunity or other factors, particularly efficiency, which affect input use specificity. 11 In analyses of industrial sector from different countries (USA, France, Japan...), this parameter is about 0.3 (Mairesse and Hall, 1996; Wakelin, 2001, among others). In the case of Spain, for instance, Raymond (1989) obtains an elasticity of 0.389, on consolidate data in the country economy. 10 ARTICLE IN PRESS 1559 E. Galdeano-Gómez et al. / Journal of Environmental Management 88 (2008) 1552–1561 Table 2 Within group (WG) estimations (dependant variable: ln VA) Table 3 Technical efficiency regressions (dependant variable: TEi) Estimation Model 1 Model 2 Model 3 Explanatory variables Model 4 ln C ln L (EP)1/2 (ES)1/2 (EP)1/2(ES)1/2 R2 (adjusted) D R2 F-test for D R2 Observations 0.364***(4.012) 0.071** 1.863) 0.127*** (2.619) – – 0.736 – – 448 0.285*** (3.216) 0.064** (1.951) 0.103*** (2.711) 0.087** (1.965) – 0.794 0.058 4.021** 448 0.227***(2.451) 0.058* (1.749) 0.099***(2.538) 0.083** (1.892) 0.071** (1.795) 0.857 0.063 5.862*** 448 Constant ln C ln L (EP)1/2 (ES)1/2 (EP)1/2(ES)1/2 R2 (adjusted) DR2 F-test for D R2 Observations Notes: t-tests are reported in parentheses. *** Significant at 1%;** significant at 5%; * significant at 10%. All the regressions are carried out including the dummy temporal variables, from 1996 to 2002 (the omitted variable corresponds to 1995). The labour coefficient is positive and significant, but the parameters (in the three estimations) also reflect less impact than the referenced studies. This is probably due to the adjustment made to avoid simultaneity (in hours worked) with the environmental performance variable. A significant (po0.01) positive relationship is observed between value added and firm’s environmental investment, supporting Hypothesis 1. Consistently with Hypothesis 2, spillover effects increase (po0.05) firm value added. The estimated relationship between the interaction term and productivity is significant (po0.05) and positive (Model 3). Additionally, the increase in R2 from models 2 to 3 is highly significant (po0.01). These results support Hypothesis 3. To determine the environmental variables incidence on the efficiency of each firm, the individual fixed effect is estimated. Thus, the normalized technical efficiencies have been estimated (Eq. (9)). Results of regression are shown in Table 3. The results show that physical capital (C) is positively related to individual efficiency (po0.01). In a relative sense, the estimated parameter value may complement the one obtained previously (model 3). If we consider the sum of both, 0.356 (as the addition of the temporal variability and the individual fixed effect obtained with models 3 and 6), this value is similar to that obtained in previous studies (e.g. Wakelin, 2001). The coefficient of labour variable (L) is positive but not significant in model 6. Regarding environmental variables, the EP variable has a positive and significant impact on managerial efficiency considering the three models (4,5, and 6), although this impact is less than the effect on productivity obtained previously. The whole impact (temporal variability and firm’s fixed effect) reaches a value of 0.153, considering the complete models 3 and 6. This value can be considered relatively high in comparison with other related studies, but this may be associated with the horticultural firms’ adaptation to new environmental requirements. The environmental spillover variables indicate that there is a positive and significant spillover direct effect (ES), improving the estimation’s efficiency (from models 4 to 5). 2.306***(7.219) 0.205*** (3.026) 0.050*(1.753) 0.081***(2.708) – – 0.527 – – 56 Model 5 Model 6 1.825***(6.381) 0.134*** (2.822) 0.049*(1.740) 0.051**(2.208) 0.035**(1.912) – 0.609 0.082 4.319** 56 1.178***(5.206) 0.119*** (2.614) 0.025(0.851) 0.062***(2.570) 0.033**(1.847) 0.029*(1.651) 0.648 0.039 2.925* 56 Notes: t-tests are reported in parentheses. *** Significant at 1%; ** significant at 5%; * significant at 10%. These results support Hypothesis 2. The estimations of model 6 indicate that the inclusion of the interaction effect also improve the coefficient of determination, although the impact (in terms of parameter significance) is relatively low in comparison with productivity estimations. Thus, we can deduce that the moderation effect of environmental spillover has more impact on value added (productivity) than on individual efficiency, which can be partially affected when a greater intensification of environmental expenditure takes place in the sector. These results support Hypothesis 3. 6. Discussion and conclusions In this paper the concept of spillover effects is applied to assess the ability of investment in environmental practices to influence horticultural firms performance. Our results show that the stock of a firm’s investments in environmental practices is positively related to its performance (value added), supporting studies on the same issue in other industries (Klassen and Whybark, 1999; Russo and Fouts, 1997) and in the agricultural industry in particular (Nijkamp and Vindigni, 2002; Shrivastava, 1994). They also confirm that total industry investment in environmental practices in the region is positively related to an individual firm’s performance, obtaining empirical evidence of the influence of spillover effects on competitiveness. Finally, empirical evidence has also been obtained showing the total investment in the sector on environmental practices has a slight but significant moderating effect on the relationship between firm-level investment in environmental practices and performance. The findings about the role of spillover effects within the industry, and therefore outside the company, in generating competitive advantages based on environmental capabilities has major implications for future research both in the field of environmental management strategy (e.g. AragónCorrea and Sharma, 2003), and for the study of industry agglomerations and geographical clusters (e.g. Maskell, 2001). ARTICLE IN PRESS 1560 E. Galdeano-Gómez et al. / Journal of Environmental Management 88 (2008) 1552–1561 By providing evidence on a business sector which has traditionally been afforded a secondary role in studies on environmental management, the present study shows that the development of strategic capabilities associate with environmental management is not exclusive to large-scale industrial companies. Horticultural companies characterized by, among other factors, their relatively small size and little experience of applying environmental practices can also generate green competitive advantages. These advantages are favoured by the inclusion of companies within a regional network, via which knowledge of the most suitable environmental practices can be diffused. There are some limitations of this study, which may encourage further work. One of these is its focus on a single industry in a given geographical area, which affects the generalizability of the findings. Although other research works have studied the dissemination of innovation capabilities by the spillover effects in different industries (Almeida and Kogut, 1999; Audretsch and Feldman, 1996), researches should interpret these results with care when extrapolating them to other organizational contexts, with different levels of uncertainty and different regulatory, competitive and technological conditions. The intense consumption of natural resources by most horticultural firms affects their ability to generate value and to develop environmental resources through the implementation of environmental practices. For instance, most firms in this sample are high consumers of water, which is at a premium in the region (Downward and Taylor, 2007). Therefore, there are good opportunities to improve organizational efficiency through reductions in consumption of natural resources. We argue that deploying environmental protection measures in these highly intensive agricultural firms may produce greater benefits than those obtained by traditional agricultural firms (Céspedes and Galdeano, 2004). Horticultural firms are as yet recent adopters of environmental protection practices. Thus, firms in our sample that deploy environmental management practices may be regarded as proactive and first movers in environmental issues. As well as the possibility to obtain competitive advantages from the increase in environmental investment, other conclusions on management practices can be drawn from the present work. In particular, the effect of industrial agglomeration should be borne in mind by managers when taking decisions on location. This is especially true in these industries in which environmental proactivity is related to the generation of sustainable competitive advantages. Our results show that spillover effects moderate the relationship between investment in environmental practices and performance in horticultural firms. The profitability of investing in this type of capabilities will therefore depend on the location of said investment among other factors. For public administrations, this article suggests the importance of encouraging environmental awareness (Tallman et al., 2004) and resources and competitive capabilities on a local or regional level. It also highlights the importance of encouraging spillovers effects by means of tools such as contact networks (e.g. McEvily and Zaheer, 1999; McEvily and Marcus, 2005), business associations (McEvily and Zaheer, 1999), the transfer of qualified staff among organizations (Almeida and Kogut, 1999), or the companies’ capacity to imitate (Maskell and Malmberg, 1999; Maskell, 2001). References Adams, J.D., 2006. Learning, internal research, and spillovers. The Economics of Innovation and New Technology 15, 5–36. Asmundson, P., Foerster, S.R., 2001. Socially responsible investing. Better for your soul or your bottom line? Canadian Investment Review 14 (4), 9–18. Almeida, P., Kogut, B., 1999. Localization of knowledge and the mobility of engineers in regional networks. Management Science 45 (7), 905–917. Aragón-Correa, J.A., Sharma, S., 2003. A contingent resource-based view of proactive corporate environmental strategy. Academy of Management Review 28 (1), 71–88. Audretsch, D.B., Feldman, M.P., 1996. R&D spillovers and the geography of innovation and production. American Economic Review 86 (3), 630–640. Bansal, P., Roth, K., 2000. Why companies go green: a model of ecological responsiveness. Academy of Management Journal 43 (4), 717–736. Basant, R., Fikkert, B., 1996. The effects of R&D, foreign technology purchase, and domestic and international spillovers on productivity in Indian firms. Review of Economics and Statistics 78, 187–199. Battese, G.E., Coelli, T.I., 1995. A model for technical inefficiency effects in a stochastic frontier production function for panel data. Empirical Economics 20, 325–332. Bowles, S., Gintis, H., 2002. Social capital and community governance. Economic Journal 112, 419–436. Brouwer, F., Baldock, D., Carpentier, C.L., Dwyer, J., Ervin, D., Fox, G., Meister, A., Stringer, R., 2000. Comparison of Environmental and Health-related Standards Influencing the Relative Competitiveness of European Union Agriculture vis-à-vis Main Competitors in the World Market. European Commission, Bruxelles, Belgium. Carpentier, C.L., Ervin, D.E., 2002. Business Approaches to Agrienvironmental Management: Incentives, Constraints and Policy Issues. OECD, Paris. Céspedes, J., Galdeano, E., 2004. Environmental practices and the value added of horticultural firms. Business Strategy and the Environment 13, 403–414. Chambers, K., Eisgruber, L., 1998. Green marketing and green competitive. In: Batie, S., Ervin, D., Schulz, M. (Eds.), Initiatives in Environmental Management: The Next Generation of Policy? Special Report 92, East Lansing, MI, Michigan State University, Department of Agricultural Economics, pp. 25-36. Chen, J.R., Yang, C.H., 2005. Technological knowledge, spillover and productivity: evidence from Taiwanese firm level data. Applied Economics 37, 2361–2371. Christmann, P., 2000. Effects of ‘‘best practices’’ of environmental management on cost advantage: the role of complementary assets. Academy of Management Journal 43 (4), 663. Downward, S.R., Taylor, R., 2007. An assessment of Spain’s Programa AGUA and its implications for sustainable water management in the province of Almerı́a, southeast Spain. Journal of Environmental Management 82, 277–289. Ecker, S., Coote, L., 2005. Best farms-an integrated approach to environmentally sustainable farming in the south-west region of Western Australia. Australian Journal of Environmental Management 12 (4), 215. ARTICLE IN PRESS E. Galdeano-Gómez et al. / Journal of Environmental Management 88 (2008) 1552–1561 Esty, D., Porter, M., 1998. Industrial ecology and competitiveness: strategic implications for the firm. Journal of Industrial Ecology 2 (1), 35–43. Garcés, C., Galve, C., 2001. The effects of environmental regulations on the productivity of large companies: an empirical analysis of the Spanish case. Journal of Management and Governance 5 (2), 129–152. GCAC (General Committee for Agricultural Cooperation), 2000. The agricultural cooperation in the European Union: tendencies and current issues. GCAC, Brussels. Griliches, Z., 1992. The search for R&D spillovers. Scandinavian Journal of Economics 94, 29–47. Griliches, Z., Hausman, J.A., 1986. Errors in variables in panel data. Journal of Econometrics 31, 93–118. Hall, B.H., Mairesse, J., 1995. Exploring the relationship between R–D and productivity in France manufacturing firms. Journal of Econometrics 65, 263–293. Hart, S.L., 1995. A natural-resource-based view of the firm. Academy of Management Review 20 (4), 986–1014. Hart, S.L., 1997. Beyond greening: strategies for a sustainable world. Harvard Business Review January–February, 67–76. Inkpen, A.C., Tsang, E.W.K., 2005. Social capital, networks, and knowledge transfer. Academy of Management Review 30 (1), 146–165. Jaffe, A.B., Peterson, S.R., Portney, P.R., Stavins, R.N., 1995. Environmental regulation and the competitiviness of US manufacturing: What does the evidence tell us? Journal of Economic Literature XXXIII (1), 132–163. Jaffe, A.B., Newell, R.G., Stavins, R.N., 2005. Environmental policy and technological change. Environmental and Resource Economics 22, 41–69. Klassen, R.D., Whybark, D.C., 1999. The impact of environmental technologies on manufacturing performance. Academy of Management Journal 42 (6), 599–615. Maier, S., Finger, M., 2004. Constraints to organizational change processes regarding the introduction of organic products: case findings from the Swiss food industry. Business Strategy and the Environment 10 (2), 89. Mairesse, J., Hall, B.H., 1996. Estimating the productivity of research and development in France and United States manufacturing firms: an exploration of simultaneity issues with GMM. In: Wagner, K., van Ark, B. (Eds.), International Productivity Comparisons. North-Holland, Amsterdam, pp. 285–315. Mairesse, J., Mohnen, P., 2002. Accounting for innovation and measuring innovativeness: an illustrative framework and an application. American Economic Review 92 (2), 226–230. Malmberg, A., Maskell, P., 2002. The elusive concept of localization economies: towards a knowledge-based theory of spatial clustering. Environmental Planning A 34, 429–449. Maskell, P., 2001. Knowledge creation and diffusion in geographic clusters. International Journal of Innovation Management 5 (2), 213–237. Maskell, P., Malmberg, A., 1999. Localised learning and industrial competitiveness. Cambridge Journal of Economics 23 (2), 167–185. Mazzanti, A., Zoboli, R., 2006. Economic instruments and induced innovation. The European policies on end of life vehicles. Ecological Economics 58 (4), 867–896. McEvily, B., Marcus, A., 2005. Embedded ties and the acquisition of competitive capabilities. Strategic Management Journal 26 (11), 1033–1148. 1561 McEvily, B., Zaheer, A., 1999. Bridging ties: A source of firm heterogeneity in competitive capabilities. Strategic Management Journal 20 (12), 1133–1158. McNally, S., 2002. Are ‘Other Gainful Activities’ on farms good for the environment? Journal of Environmental Management 66, 57–65. Nijkamp, P., Vindigni, G., 2002. Food security and agricultural sustainability: An overview of critical success factors. Environmental Management Health 13 (5), 495–510. Novales, A., 1996. Econometrı́a, third ed. McGraw-Hill, New York. OECD, 1998. Cooperative Approaches to Sustainable Agriculture. OECD, Paris. OECD, 2001. Encouraging Environmental Management in Industry. OECD, Paris. Porter, M.E., van der Linde, C., 1995. Green and competitive: ending the stalemate. Harvard Business Review 73 (5), 120–134. Pouder, R., St John, C.H., 1996. Hot spots and blind spots: geographical clusters of firms and innovation. Academy of Management Review 21 (4), 1192–1225. Raymond, L., 1989. Productividad de los factores y expansión del sector público en España. Papeles de Economı́a Española 41, 159–171. Reinhardt, F., 2000. Down to Earth: Applying Business Principles to Environmental Management. Harvard Business School Press, Boston. Russo, M.V., 2003. The emergence of sustainable industries: building on natural capital. Strategic Management Journal 24 (4), 317–331. Russo, M.V., Fouts, P., 1997. A resource-based perspective on corporate environmental performance and profitability. Academy of Management Journal 40 (3), 534–559. Schaltegger, S., Synnestvedt, T., 2002. The link between ‘green’ and economic success: environmental management as the crucial trigger between environmental and economic performance. Journal of Environmental Management 65, 339–346. Sharma, S., Vredenburg, H., 1998. Proactive corporate environmental strategy and the development of competitively valuable organizational capabilities. Strategic Management Journal 19 (8), 729–753. Shrivastava, P., 1994. Sustainable development with special reference to environment and agriculture. Journal of Rural Development 13 (1), 47–55. Shrivastava, P., 1995. Environmental technologies and competitive advantage. Strategic Management Journal 16, 183–200. Sickles, R.C., 2005. Panel estimators and the identification of firm-specific efficiency levels in parametric, semiparametric and nonparametric settings. Journal of Econometrics 126, 305–334. Storper, M., 1997. The city: centre of economic reflexivity. Service Industries Journal 17 (1), 69–90. Tallman, S., Jenkins, M., Henry, N., Pinch, S., 2004. Knowledge, clusters, and competitive advantage. Academy of Management Review 29 (2), 258–271. Tyteca, D., Carlens, J., Berkhout, F., Hertin, J., Wehrmeyer, W., Wagner, M., 2002. Corporate environmental performance evaluation: evidence from the MEPI report. Business Strategy and the Environment 11, 1–13. Wakelin, K., 2001. Productivity growth and R&D expenditure in UK manufacturing firms. Research Policy 30 (7), 1079–1090. Walley, K., Custance, P., Parsons, S., 2000. UK consumer attitudes concerning environmental issues impacting the agrifood industry. Business Strategy and the Environment 9 (6), 355–366.