Available online at www.sciencedirect.com
Energy Economics 30 (2008) 2517 – 2536 www.elsevier.com/locate/eneco
The energy efficiency paradox revisited through a partial observability approach Kostas Kounetas, Kostas Tsekouras ⁎ Department of Economics, University of Patras, University Campus — Rio, P.O. Box 1391, Patras 26500, Greece Received 29 November 2005; received in revised form 8 March 2007; accepted 8 March 2007 Available online 25 April 2007
Abstract The present paper examines the energy efficiency paradox demonstrated in Greek manufacturing firms through a partial observability approach. The data set used has resulted from a survey carried out among 161 energy-saving technology firm adopters. Maximum likelihood estimates that arise from an incidental truncation model reveal that the adoption of the energy-saving technologies is indeed strongly correlated to the returns of assets that are required in order to undertake the corresponding investments. The source of the energy efficiency paradox lies within a wide range of factors. Policy schemes that aim to increase the adoption rate of energy-saving technologies within the field of manufacturing are significantly affected by differences in the size of firms. Finally, mixed policies seem to be more effective than policies that are only capital subsidy or regulation oriented. © 2007 Elsevier B.V. All rights reserved. JEL classification: Q42; Q49; L6 Keywords: Energy efficiency paradox; Adoption; Partial observability
1. Introduction Policy makers of most OECD countries in Europe and in the USA have been developing schemes in order to achieve better environmental conditions. One major component of these schemes is the adoption of energy efficiency technologies (EET) by manufacturing firms. This adoption, in turn, requires a wide range of investment projects, the rate of which however is quite low and constituted the well-known energy efficiency gap (Jaffe and Stavins, 1994). ⁎ Corresponding author. Tel.: +30 261 0996376; fax: +30 261 0996130. E-mail address:
[email protected] (K. Tsekouras). 0140-9883/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.eneco.2007.03.002
2518
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
A sizeable technical and scholarly literature in the last fifteen years presents an imposing body of evidence that the typical rate of return, usually measured by the Net Present Value (NPV) available from energy-saving investments, is quite higher than the discount rate for projects of comparable risk (Lovins and Lovins, 1991; Ayers, 1993; DeCanio, 1993; Jaffe and Stavins, 1994). In spite of this evidence, however, research has empirically recorded the fact that firms display a relative reluctance to adopt such EET, and obviously they do not become involved in the appropriate capital investment projects. These seemingly conflicting findings have been described in the relevant literature as the energy efficiency paradox. More specifically, the energy efficiency paradox can be defined as the case in which firms, presumed to behave rationally and to be economically efficient, do not undertake capital investment projects on EET, although they are preferable in terms of profitability and risk to other non-related to EET projects. According to DeCanio and Watkins (1998) in an energy efficiency paradox situation, firms may involve in decision making processes which do not imply profit maximization, since, DeCanio (1998) defines the energy efficiency paradox as the situation where “there is abundant evidence that highly profitable energy-saving opportunities exist, yet the technologies embodying these opportunities have not spread universally throughout the economy”. In other words we are talking about a situation where cost-effective technologies at current prices appear to be un-adopted by many business firms, or, to use a more inclusive approach, we can describe the energy efficiency paradox as a phenomenon of an inadequate diffusion of apparently cost-effective EETs (Shama, 1983). From the economic theory point of view, therefore, it is evident that the question which may be posed here regards the criteria under which the adoption of an EET is examined by the possible adopters. When firms undertake a decision making process to adopt such a technology, do they take into account, at least substantially, the profitability dimensions of their decision making output? Although the content of this assumption brings forward a deviation from the strong neoclassical theory of investment, it does not imply that firms necessarily exhibit non-optimizing behaviour (Howarth and Stanstad, 1995). On the contrary, this may be interpreted by the crucial role that firm-specific characteristics may play in this decision making process (DeCanio, 1998).1 Our first and main research question, therefore, concerns the formulation and testing of the following hypothesis: the decision of the firms to adopt an EET or not is correlated to their profitability, under the condition that several other factors may also play a crucial role. In order to approach the question at hand we need to include in our analysis those factors that have been recorded in the relevant literature as possible sources of the energy efficiency paradox. This leads us to the second objective of the present paper (and we believe an important contribution to the existing literature) which is to examine the influence these factors may exert on a firm's decision to adopt an EET in conjunction to its profitability. In order to accomplish this, we implement unique data available to us that should overcome any misspecification problems of the omitted variable types that seems to characterise the examination of the joint influence of all those factors up to date. To the best of our knowledge it is the first time that a one-shot examination is realized concerning the influence of all the factors that are identified, according to the relevant literature, as possible sources of the energy efficiency paradox. The first step in our approach is to develop a partial observability model that studies the adoption of an EET and the firm's profitability. Then, the unique data set that resulted from a survey conducted in 2004 among Greek manufacturing firms that adopted energy-saving
1
We owe this to an anonymous referee.
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
2519
technologies and have been granted capital subsidies for undertaking such investments will allow us to examine the influence of the factors that may lead to the energy efficiency paradox. The remainder of the paper is organized as follows. In the next section we present a review of literature focusing on the factors which may cause the energy efficiency paradox. In Section 3, we develop a model that combines the adoption of an EET and the firms' profitability. In Section 4 we present the institutional environment for EET adoption in Greek manufacturing as well as the survey process from which we derive the data and the used variables in the empirical estimation of our model. The following section discusses the empirical results and the last section concludes the paper, summarizes the research findings, points to future research questions and puts forth some policy implications. 1.1. Factors affecting the adoption of energy-saving technologies — a review of literature A plethora of factors that aim to interpret the adoption of energy-saving technologies and thus, directly or indirectly, to shed light on the energy efficiency paradox may be identified in the relevant literature. Factors affecting the adoption of energy-saving technology may be classified into four distinct categories. The first group focuses on issues that arise from potential market failures such as market structure and concentration (Hannan and McDowell, 1984), information problems, principal-agent slippage, unobserved costs, or the stochastic rate of technological progress, which implies technological uncertainty (DeCanio, 1993; Van Soest and Bulte, 2001; DeGroot et al., 2003). The second category includes factors that although they are wider economic environmentspecific-variables, they do not represent market failures. Such economic factors (Stoneman and David, 1986; Khanna and Zilberman, 2001) are private information costs, demand uncertainty (Faria et al., 2002), level of discount rates, heterogeneity among potential adopters, capital subsidies. Non-economic factors (Jung et al., 1996) include labels and standards, investment incentives, taxes and permits (Verhoef and Nijkamp, 1997; Jung et al., 1996), environmental regulations and community environmental impact (Florida and Davison, 2001). In the same direction, a growing body of recent research shows that non-formal regulatory pressure, known as “informal regulation”, exercised by private-sector groups such as neighborhood organizations, non-governmental organizations, and trade unions can substitute for formal pressure (Pargal and Wheeler, 1996). Blackman and Bannister (1998) find that informal regulation is correlated with the adoption of a clean technology. Firm-specific-variables constitute the third group of factors that may influence the decision for the adoption of energy-saving technologies (DeCanio and Watkins, 1998). As such firm-specific characteristics may be regarded: the firm's size in relation to market concentration and ownership structure (Antonelli and Tahar, 1990; Karshenas and Stoneman, 1993; Rose and Joskow, 1990; Pizer et al., 2002); the financial structure of the firm (Pizer et al., 2002); the scarcity of managerial time or skilled personnel (De Almeida, 1998; Gabel and Sinclair-Dsgagne, 1992); institutional Table 1 Time distribution of investment projects in energy-saving technologies
[1987–1996] [1996–2000] [2000–2003] Total
Frequency
Percent
Cumulative percent
5 63 93 161
3.11 39.13 57.76 100.00
3.11 42.24 100.00
2520
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
and organizational barriers (De Almeida, 1998; DeCanio, 1998); the firm's age, which captures learning-by-doing effects, and consequently the level of accumulated knowledge, R&D and innovative activities (Conrad, 1997); the firm's capital vintage (Geroski, 2000). The last group of factors that may influence the adoption of energy-saving technologies pertains to technology and investment and includes relative advantage, compatibility, complexity, trial ability and observability (Rogers, 1995; Weiss, 1994; Van Soest and Bulte, 2001). Pindyck, (2000) provided an insight into decision making processes concerning innovations when he found that the market for innovation at a current point is limited by expectations of greater future improvements in the field. Since the investment in new technology, included energy-saving technology, may turn out to be inferior in the face of a subsequent innovation a degree of irreversibility, even small, may be present here (Van Soest and Bulte, 2001). Such being the case it will bear weight on the firm's adoption decision and subsequent performance. 2. Theoretical and modelling issues Let us assume that there is a set of firms N that becomes involved in a decision making process to examine the possibility to adopt or not an EET through the realization of capital investment. The i-th (i = 1,2,…N) firm poses objectives that it intends to pursue through this adoption in the postadoption period. Firms may pursue strategies based mainly on cost reduction and the production of price competitive products, the so-called cost leadership pathway of development (Porter, 1980, 1985). It is reasonable to assume that a cost leadership strategy should be reflected on profitability (π) indicators of performance. One could argue that post-adoption objectives may be different from profitability. For example, adopters may pursue strategies that create resources or combine existing resources in new ways to develop and commercialize new products and service new customers and markets, the so-called strategic entrepreneurship pathway of development (Lu and Beamish, 2001; Hitt et al., 2001). However, the energy efficiency paradox becomes manifest in a world of firms that interpret their performance in terms of profitability, so it is only reasonable to assume that their post-adoption objectives should also be interpreted in such terms. The profitability of the i-th firm that examines the possibility to adopt the EET may be described as: pi ¼ β Vxi þ epi
ð1Þ
where xi is a vector of the firm's specific characteristics as well as the characteristics of the industry in which the prospective adopter is examined, β a vector of parameters to be estimated and eiπ an error term that is associated with the firm's profitability. As it has already been said, according to the energy efficiency paradox logic the adoption of an EET demands the realization of a capital investment, which it is reasonable to assume would be one among a number of capital investment projects that would maximize the firm's profitability. The approach we follow in the context of the present paper allows for multiple EET investment projects under consideration out of which the firm decides to implement one, more than one or none. Capital investments, in general, permit a firm to improve its competitiveness and to gain profitability. On the other hand capital investments incur costs. Thus the firm's optimal level of investment (INV⁎) is the outcome of an underlying utility maximization process in which: INV⁎ ¼ INVðzi Þ
ð2Þ
where zi is a vector of the firm's specific characteristics including capital cost, capital subsidies, expectations, risk preferences, input and knowledge constraints as well as characteristics of the
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
2521
industry it belongs to. Regarding the economic notion which is conveyed by Eq. (2), an anonymous referee pointed out that one of the clearest consequences of the neoclassical theory of investment is that the discount rate used to calculate NPV's should be the available return of all the capital investment projects in the same risk class. Thus, the variable INV⁎ is not necessarily consistent with neoclassical profit maximization. However, in our point of view, an investment decision is not the culmination of a financial criterion. It includes firms' characteristics (DeCanio and Watkins, 1998), constraints and barriers (DeCanio, 1993; Jaffe and Stavins, 1994; etc.), risk preferences and inherent uncertainty (DeGroot et al., 2003). Although the inclusion of variables of this kind in our model does not arise directly from the neoclassical theory of investment, their incorporation in our approach is justified by the weakness of the neoclassical approach to interpret the energy efficiency paradox. In other words in our view, the investment decision is not just the output of a financial criterion, but a crucial factor affecting the firm's objectives, capabilities and weaknesses in a wider sense. Among its other investments the firm decides to invest in energy-saving technologies, and becomes an adopter A or an element of a subset of adopters (A ⊆ N) if the expected benefits minus the expected cost of this adoption are greater than a certain threshold INVic.2 This threshold is determined, mainly, by (a) the firm's specific characteristics (financial constraints, risk preferences, technological uncertainties, level of information, organizational characteristics, accumulated knowledge etc.), (b) the characteristics of the economic environment (environmental and production regulations, incentive schemes for adopting EET, level of competition and market structure) and (c) the specific characteristics of the examined investment projects. Thus the firm in its decision making process moves towards the adoption if: c INV⁎ i NINVi
ð3Þ
which implies that the difference of the expected gains, from the expected costs, when both have been adjusted for risk preferences, DINV⁎ is positive:
⁎ c DINV⁎ i uINVi INVi N0:
ð4Þ
Eq. (4) may be expressed as a linear model: inv DINV⁎ i ugzi þ ei
ð5Þ
where γ is a vector of parameters to be estimated and eiinv is a normally distributed error term. In practice, DINVi⁎ is not observed. However, we can easily construct an indicator variable denoted by ADPiinv which equals 1 if the firm decides to adopt the EET and 0 if the firm decides otherwise. Thus: inv 1; if DINV⁎ i N0; gzi z ei : ADPinv ¼ ð6Þ ⁎ i inv 0; if DINVi V0; gzi V ei Apparently, ADPiinv which reflects the outcome of the i-th firm decision making process to adopt or not an EET is a dichotomous variable and thus a probit model applies to it. In usual
2 At this point it is not worthless to mention that the approach, which is based on the value of the discounted net benefits, is under the condition that the risk factor has been taken into account usually through the selection of an appropriate discount rate.
2522
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
circumstances, the estimation of profitability (Eq. (1)) would be carried out under a selectivity bias framework taking into account the adoption decision observation mechanism described in Eq. (6). Eq. (6), however, describes a partially observable selection rule and thus the usual estimation procedures under selectivity bias do not apply. The selection rule is partial because there are no observations for those firms which decide not to adopt the EET, i.e., the zeros in Eq. (6). This model is valid as long as the correlation coefficient between eiπ and eiinv is non-zero, which in turn brings to the foreground the basic rationale of the energy efficiency paradox that the adoption of EET is closely related to the firm's performance in terms of profitability. The model, under the condition that eiπ and eiinv are correlated, can be estimated by the maximum likelihood estimator devised by Bloom and Killingsworth (1985). The full specification is represented by Eqs. (1), (2), (6) and, in addition, we assume that eiπ and eiinv have a bivariate normal distribution with zero means and correlation ρeπeinv. That is,3 fN ½0; 0; r2ep ; r2einv ; qep einv : epi ; einv i
ð7Þ
The data consist of a set of observations on (π), x, and z, known for every firm that has adopted EET (ADPiinv = 1) and unknown if it has not. Thus, the profitability of firms is estimated subject to the condition that ADPiinv = 1, i.e., for those firms that have adopted EET through capital investments. The likelihood for the truncated sample of EET adopters only, conditional on their ADPiinv, is (Bloom and Killingsworth, 1985): h i zi gþlep einv p 1 F p f e =r r p inv e i h e e i Lðpi jADPinv ð8Þ i ¼ 1; xi Þ ¼ j zi g iaA rep 1 F r inv e
where F is the standard normal cumulative density function, μeπeinv = σeπeinveiinv / σeinv is the mean of eiπ conditional on eiinv and σeπeinv = σeinv[1 − (σ2eπeinv / σeπσeinv)] is the variance of eiinv conditional on eiπ. Estimation of (8) by maximum likelihood yields consistent estimations of the parameters β, γ, σe2π and ρeπeinv. 3. The incentives schemes, data and variables definition 3.1. The Greek incentives schemes for the adoption of energy-saving technologies The empirical implementation of the above-described theoretical model in the present paper is based on a unique data set, which consists of two parts. The first part is the outcome of a survey that we have conducted among firms that have adopted EET in Greece. The second part of our data set regards firm-specific data that come from the business database maintained by the private financial and business information service company called ICAP.4 From the annual directories of ICAP we designed a database of Greek manufacturing firms that we previously identified as adopters of EET for the period 1987–2001 based on the official records of the Greek Ministry of Development.
At this point we should refer to Bloom's and Killingsworth's caveat that “in models like the one developed here, parameter estimates may be sensitive to distributional assumptions”. 4 The annual ICAP directories provide key production, employment and financial information from the published balance sheets of almost all Plc and Ltd firms operating in all sectors of economic activity in Greece. 3
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
2523
Table 2 Industry and size distribution of the energy-saving technologies adopters Industry
Chemicals Primary metals Non-metallic mineral products Machineries Drugs Oil refining Plastics & rubber Textile & underwear Food & beverages products Paper & wood Construction materials Miscellaneous Total
Number of employees Micro firms [1–9]
Small firms [10–49]
Medium size firms [50–249]
Large firms N250
Total
0 0 1
3 4 5
1 4 5
3 2 2
7 (4.35%) 10 (6.21%) 13 (8.07%)
0 0 0 0 0 1
2 0 0 3 7 7
2 4 0 5 22 22
0 3 5 1 12 15
4 (2.48%) 7 (4.35%) 5 (3.10%) 9 (5.59%) 41 (25.46%) 45 (27.95%)
0 0 0 2 (1.24%)
3 3 0 37 (22.98%)
6 2 1 74 (45.96%)
1 1 1 48 (29.81%)
10 (6.21%) 6 (3.73%) 2 (1.24%) 161 (100.00%)
The Greek government has already recognized the need to conserve energy in manufacturing and to reduce dangerous emissions in order to meet the criteria of the Kyoto Protocol. In the past two decades, energy policies that aim to promote conservation and induce firms to adopt environment-friendly technologies have been subsidized by the central or regional authorities and formulated based on (i) the Support Frameworks for Regional and Industrial Development, (ii) the Energy Operational Program (OPE), which was part of the second European Union Support Framework (1994–2000) and (iii) the Operational Program “Competitiveness”, which is part of the third European Union Support Framework (2000–2006) (Table 1). The OPE as well as the Competitiveness programs have been assessed by the Directorate of Renewable Energy and energy-saving in the Ministry of Development, while the Support Frameworks for Regional and Industrial Development were assessed by the Ministry of National Economy. Although government efforts aiming at more efficient energy consumption by manufacturing firms date back to 1982,5 the first worthwhile attempt took place ten years later under the Regulation Framework 1892/90. As a result, energy consumption was reduced by about 26.3 Ktoe6 a year, while conventional energy sources were replaced by biomass and natural gas resulting in energy conservation of 37.4 Ktoe per year (IEA, 2002). However, the most substantial incentive for the adoption of EET in the Greek manufacturing industries was provided by an OPE sub-program, launched in January 1994 and lasting to the end of 2001. As a follow-up to OPE, the third European Union Support Framework has provided financial resources so that firms will be able to invest in EET through the Competitiveness Operational Program. 3.2. The sampling framework and survey process The records provided by the Ministry of Development and the Department of Energy and Natural Resources show a total number of 396 firms to have invested in energy efficiency 5 6
Incentives Schemes Framework 1262/82. One Ktoe equals 1000 tonnes of oil equivalent.
2524
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
technologies. If we exclude those firms that belong to the Services sector and therefore implement a different energy-saving technology, the manufacturing firms pertinent to our investigation come to 325. A further communication with the firms' managers and chief-engineers indicated that only 298 firms had actually accomplished the transition to energy-saving technologies and 293 had been granted capital subsidies as an investment incentive for that purpose. In order to test our theory, we designed a rather extensive questionnaire staggering all the theoretical aspects discussed earlier into eight sections dealing with: the firms' general and organizational and managerial characteristics, characteristics of the energy conservation technology project they have undertaken, market competition, their strategic orientation, their knowledge and degree of implementation of energy-efficient technologies, their investment behaviour, their attitude towards energy-saving technologies and towards barriers to invest in energy-efficient technologies, and finally their evaluation of the investment and their potential expectations concerning further investments in energy efficiency projects. The questionnaire was tested, for about two months, in a pilot study in the region of Western Greece for further evaluation and improvement. The final questionnaire was addressed to the 298 firms across the country, 161 of which (54% of the whole EET adopters population) agreed to be interviewed. Face to face interviews took place in the first six months of 2004. This response rate is quite high and acceptable, compared to other similar survey processes (Velthuijsen, 1993; Vicini, 1998; DeGroot et al., 2001). In addition, the distributional characteristics (firm size, industry, location, age and the amount of the invested capital) of the firms which eventually responded positively to the survey process were not statistically different compared to the corresponding distributional characteristics of the 298 firms which constitute the entire population of EET adopters in Greek manufacturing. This allowed us to use our data set without further stratification (Table 2). 3.3. Data and variables definition As it has been already mentioned, the profitability equation and the EET adoption equation are estimated jointly to formulate our empirical model. In determining the most appropriate set of possible explanatory variables, we took into account the relevant literature, for both the profitability equation and the EET adoption equation. The data regarding these variables come from (i) the published firms' (balanced sheets) data and (ii) the information we collected through the survey process. Then, we looked for the model with the best econometric properties among alternative models. This implies that variables with no statistically significant results have been included in our final model, because this is also an important finding and they improve the overall fitness of the model as it is expressed by the value of the maximum likelihood corrected, each time, with the corresponding degrees of freedom. Separate tests examining the null hypothesis that individual coefficients are zero, and a joint test of the null hypothesis that all the parameters associated with the explanatory variables are equal to zero has been made. Maximum likelihood estimated coefficients, their corresponding asymptotic standard errors, the chi-square test and the estimated ρeπeinv parameter are presented in the lower part of Table 9. The value of the chi-square test, which corresponds to the finally selected model, is quite high (χ2 = 156.916 with corresponding critical value at 5% level equal to 43.733). In addition, we should point out that the omission of a significant variable, in the context of a binary, dichotomous choice model, implies that even if the omitted variable is uncorrelated with the included one, the coefficient of the included variable will be inconsistent (Yatchew and Criliches, 1984). The theoretical context according to which we constructed the possible explanatory variable sets is presented next.
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
2525
3.4. The profitability equation The dependent variable (π) in the profitability equation is defined as the ratio of net profits to the firm's assets, i.e., the simple ROA index which is the baseline in order to determine the logic under which a firm decides to adopt an energy-efficient technology and to capture the effect of energy policies on the firms' ROA. The variable (π) has been computed as the simple three year average of the post-adoption period in order to avoid the well-known time fluctuations of the financial indices. Although traditional approaches evaluating the performance implications of several exogenous factors rely heavily upon the use of the accounting rate of return (Fedenia and Hirschey, 1992), they constitute a rather controversial issue.7 One could argue that despite their obvious imperfections, a number of recent empirical studies strongly suggest that both accounting and market value-based rate of return data provide useful evidence of economic profitability (McFarland, 1988). This paper considers the accounting ROA as the best available indicator of a firm's profitability or in other words of the management's use of the firm's assets. Because it measures the accounting rate of return on total assets, it is relatively unaffected by reporting errors tied to leverage, or to recapitalizations tied to share repurchases (Jacobson, 1987). Finally, we should point out that in Greece, as in most European Countries, the Ministry of Development requires firms to adjust the book value of their assets every three years so that they approximate their true (market) value. This adjustment is ruled by certain directives, which are common for all firms. Regarding the set of explanatory variables in the profitability equation, one may identify two main strands of research that have attempted to explore the determinants of a firm's profitability. The first is Industrial Organization and the second strategic management. In industrial economics the traditional approach, that is the Structure-Conduct-Performance (SCP) paradigm, focuses mainly on industry-level determinants of competition, but also on variables which have a mixed character, that is, they are industry- and firm-level determined (Bain, 1956; Porter, 1980; Slater and Olson, 2002). In the first group of drivers the main variable used is the industry's concentration, while in the second group we have variables which capture economies of scale, product differentiation and entry and exit barriers. The theoretical foundation for the determination of profitability in the context of the SCP paradigm has been discussed by Feeny et al. (2005). The strategic management literature emphasizes the role of internal resources specific to the firm as determinants of variations in profitability (Teece, 1981; Barney, 2001; Levinthal, 1995). The resource-based view is that organizational structures and management practices represent the main source of heterogeneity in performance between firms. Internal resources which can be tangible (financial and physical factors of production) or intangible (technology, age as a proxy for accumulated knowledge which arises from learning-by-doing effects) reflect the firm's core capabilities (Winter, 2003). Taking the above into account and having performed a large number of econometric tests, of the type we have described above, we have finally included the following variables in the profitability equation. From the SCP paradigm we have included the industry's Herfindahl concentration ratio (HI), and the firm's size as it is captured by the variable of the firm's market share (MS). From the strategic management approach we have included variables that depict a 7
In the relevant literature we can identify, on the one hand, those who find fault with the ROA use as a profitability proxy. In this category we can encompass, among others, the papers by Fisher and McGowan (1983), Jensen and Zimmerman (1985), and Hirschey (1985, 1986). On the other hand, there are numerous studies which depict a series of arguments in favour of ROA's use (Fedenia and Hirschey, 1992; Watts and Zimmerman, 1980; Schwert, 1981).
2526
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
firm's internal resources, such as the introduction in the business process of a product or process innovation (INNOVD), the existence within the firm of research and development activities (RDD), the cost or other type of advantages that the firm may possess that stem from learning-bydoing effects and are captured by its age (AGE) and by its superior technological characteristics (automation for instance) which are proxied by the ratio of the firms' capita to labor CAPLAB.8 3.5. The adoption equation Turning now to the EET adoption equation we should recall to mind that the relevant literature records six group of drivers which may affect the decision making process to adopt or not an EET. More specifically these groups are (i) the technological and economic uncertainty that is associated with the particular technology under adoption (Van Soest and Bulte, 2001; DeGroot et al., 2003), (ii) the uncertainty that springs from the economic environment and is mainly macroeconomic and industry-specific (Pindyck, 2000, 2002; Faria et al., 2002; Hasset and Metcalf, 1993), (iii) risk (DeGroot et al., 2003) and transaction costs that are associated to the specific investment (Sanstand and Howarth, 1994; Howarth and Stanstad, 1995; DeCanio, 1998), (iv) the regulations that are imposed on the firms' operation regarding the use of energy (Florida and Davison, 2001; Blackman and Bannister, 1998), (v) the institutional, informational, organizational and financial barriers that are firm- and investment- or adopted technology-specific (De Almeida, 1998; DeCanio, 1993, 1998; DeGroot et al., 2001) and (vi) the firms' specific technological and financial characteristics (DeCanio and Watkins, 1998; Velthuijsen, 1993). We should recall that one of the objectives of our study is to bring under examination all the possible determinants that may be the cause of the energy efficiency paradox. In other words we aim to avoid misspecification or the well-known problem of omitted variables. Taking into account that we had to capture twenty-two possible answers, twenty-two possible sources for the six groups of factors, it would not be econometrically feasible to record every single answer as an explanatory variable in our model. Consequently, we constructed a dummy variable, which takes the value of 0 if the deviation of a firm's answers from the median of all answers is negative and the value of 1 if it is positive. By doing so, it is possible that some of the available information will not be used, but the possibilities of overspecification and multicollinearity are minimised in our empirical model. Accordingly, therefore, we constructed firstly the variable (UNCTECN), which depicts technological and economic uncertainty associated with the specific technology under adoption as recorded in the first category of possible reasons for the energy efficiency paradox. Firms were asked to evaluate the sources of this uncertainty as regards their decision to adopt energy-saving technologies. Frequencies are presented in Table 3. The second category of the energy efficiency paradox causes concerns in the uncertainty associated with the macroeconomic and industrial environment. Firms were asked to express their expectations about the variation of input prices in the three years following their decision to adopt the specific energy-saving technology. Frequencies are presented in Table 4. Following the same simple technique that we described above in the case of the (UNCTECN) variable, we constructed 8 The variables (INNOVD), (RDD), (AGE) and (CAPLAB) may also be viewed as determinants of profitability in the context of industrial organization in the sense that they depict barriers to entry. The first two are considered proxies of product differentiation (Slade, 2004), the age variable a relative cost advantage (CSBRC, 1992; Dunne and Hughes, 1994) and the ratio of capital to labour an incumbents' capital commitment, which may deter entry (Ghemawat and Caves, 1986).
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
2527
Table 3 Evaluation of technology- and investment-specific uncertainties Source of uncertainty
Importance Low
Medium
High
Total
Expected costs
23 (14.3%) 28 (17.4%) 17 (10.6%) 69 (42.9%) 40 (24.8%) 78 (48.4%)
31 (19.3%) 26 (16.1%) 13 (8.1%) 31 (19.3%) 39 (24.2%) 58 (36%)
107 (66.5%) 107 (66.5%) 131 (81.4%) 61 (37.9%) 82 (50.9%) 25 (15.5%)
161 100% 161 100% 161 100% 161 100% 161 100% 161 100%
Existence of other technology Environmental effects Size of subsidies Hidden costs Investment's financial benefits
the (UNCENVRM) variable that captures the role of macroeconomic and industry-specific uncertainties on the decision to adopt energy efficiency technologies. The variables (RISK), (BARR) and (REGUL) were also constructed along the same principles and based on the firms' answers when they were asked to evaluate (i) the sources of the risk they undertook in their decision to invest in the specific energy-efficiency technology, (ii) the various barriers they had to overcome that might have influenced their decision and (iii), in the case of the (REGUL) variable, to comment on their readiness to adopt energy-saving technologies when regulations regarding the extent of energy consumption were imposed on their production process. In other words, the variables (RISK), (BARR) and (REGUL) convey information that capture the third, fourth and fifth category of factors from which the energy efficiency paradox may steam. Frequencies of raw information are presented in Tables 5–7, respectively. We should make a note here concerning the (BARR) variable. The list of possible barriers the firms were called upon to evaluate, in the context of the survey process, consisted of sixteen different barriers that have appeared in the relevant literature. Table 6 presents the categories of barriers that firms face when they adopt energy-saving technologies. However, there is no discrete categorization of these barriers in the relevant literature. A review indicates that each of them may have an “institutional” or “organizational,” “behavioural” or “financial” character. For our analysis purposes we reviewed the existing literature regarding the barriers or constraints that Table 4 Evaluation of macroeconomic- and industry-specific uncertainties
Energy costs Labor costs Capital costs Material costs Other costs
[0–10%]
[10–20%]
N20%
Missing
134 (82.7%) 97 (59.9%) 125 (77.2%) 126 (77.8%) 139 (85.8%)
24 (14.8%) 64 (39.5%) 33 (20.4%) 32 (19.8%) 17 (10.5%)
3 (1.9%) 0 (0%) 3 (1.9%) 3 (1.9%) 5 (3.1%)
1 (0.6%) 1 (0.6%) 1 (0.6%) 1 (0.6%) 1 (0.6%)
2528
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
Table 5 Evaluation of technology- and investment-specific risk Risk source
Importance Low
Medium
High
Total
Costs of adjustment before the investment
68 (42.2%) 80 (49.7%) 84 (52.2%) 78 (48.4%)
66 (41%) 58 (36%) 58 (36%) 58 (36%)
27 (16.8%) 23 (14.3%) 19 (11.8%) 25 (15.5%)
161 100% 161 100% 161 100% 161 100%
Costs of adjustment after the technology's adoption Investment's irreversibility Limited information
discourage firms from investing in energy-saving technologies. Taking into account the work of several authors (indicatively, DeCanio, 1993, 1998; Sutherland, 1991; Hirst and Brown, 1990; Jaffe and Stavins, 1994), we were actually motivated by the work of DeGroot et al. (2001) to formulate our classification. In the survey these barriers were addressed to the firms in separate questions to explore their response to them. The barriers which were included in the survey process are presented in Table 6 and were finally classified into four categories. Finally, we tested the effect of a firm's financial health, government subsidies and cost of investment on its decision to adopt an EET. Indeed several case studies indicate the importance of those variables. Accordingly, we point out the empirical study of Velthuijsen (1993), who reports that financing possibilities, government subsidies and the investment costs of an EET can be used as criteria for adopting those technologies. Moreover, there is DeCanio and Watkins' survey (1998) that demonstrates through their analysis of the factors that influence companies to join EPA's Green Light program that a wide range of company specific characteristics are associated
Table 6 Evaluation of barriers' importance Barriers
Barriers confronting firms
Character Behaviouralinformational
Institutional
Organizational
Financial
Energy efficiency has low priority Asymmetric information Information gaps concerning EET High initial costs Low subsidy rates Regulatory policies concerning energy efficiency Fluctuations of energy prices Incentives are not sufficiently attractive Lack of executives experienced in EET Lack of equipment-trained personnel Executive personnel unrelated to EET Incompatibility for EET implementation Existence of other EET investments Energy costs are not important Difficult to implement due to firm size Realization of other investments
Importance Low
Medium
High
Total
59 (36.6%)
33 (20.5%)
69 (42.9%)
161 (100%)
56 (34.8%)
38 (23.6%)
67 (41.6%)
161 (100%)
76 (47.2%)
29 (18.0%)
56 (34.8%)
161 (100%)
74 (46.0%)
12 (7.5%)
75 (46.5%)
161 (100%)
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
2529
Table 7 Regulations' importance Source of regulation
Importance Low
Medium
High
Total
National energy tax
101 (62.7%) 110 (68.3%) 61 (37.9%) 91 (56.5%)
34 (21.1%) 27 (16.8%) 49 (30.4%) 37 (23%)
26 (16.1%) 24 (14.9%) 51 (31.7%) 33 (20.5%)
161 100% 161 100% 161 100% 161 100%
European energy tax Technological standards Use of maximum energy
with EET adoption. Finally, we quote Pizer et al. (2002) who also suggested that plant and firm characteristics, such as energy prices, plant size and financial health have a statistically significant effect on the adoption of an EET. Table 8 Descriptive statistics of the used variables in the estimation of the incidental truncation model Explanatory variables Profitability equation MS HI AGE CAPLAB INNOVD RDD
Adoption equation SUBSD SUBSDSZ PRFM FIXT INVFIX BARR RISK REGUL UNCTECN AGEFIX EINTD UNCENVRM a b
Mean a
Standard deviation
Minimum
Maximum
0.066 0.170 44.595 0.267 0:54.67% 1: 45.33% 0:53.55% 1: 46.45%
0.105 0.044 25.972 2.162
0.000 b 0.112 1.000 0.002
0.321 0.262 143.000 14.570
0.422 0.4589 0.257 0.449 0.060 0: 39.66% 1:60.34% 0: 74.46% 1: 25.24% 0:63.58% 1: 36.42% 0:54.53% 1:45.47% 802.294 0:30.47% 1: 69.53% 0:40.40% 1: 59.60%
0.062 0.3614 0.388 0.151 0.188
0.000 0.000 − 0.348 0.084 0.000 b
0.640 1.553 19.183 0.877 31.617
136.624
0.007
5776
Frequencies are reported for dummy variables. Actually smaller than 0.0001.
2530
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
Table 9 Results of the incidental truncation model ML estimation Profitability equation
Adoption equation a
Explanatory variables
Value
CONSTANT MS HI INNOVD RDD AGE CAPLAB LL ρeπeinv σ2eπ N
− 0.065(− 5.175) − 0.025 (− 0.029) 0.302 (4.335) 0.0359 (6.848) − 0.023 (− 3.534) 0.001 (8.789) 0.003 (1.953) 306.224 0.873 (70.141) 0.062 (28.014) 161
a
Probability
Explanatory variables
Value
Probability
0.000 0.395 0.002 0.004 0.001 0.000 0.051
CONSTANT SUBSD SUBSDSZ PRFM FIXT INVFIX BARR RISK REGUL UNCTECN AGEFIX EINTD UNCENVRM
0.381 (0.608) 2.046 (2.038) − 2.189 (−4.890) − 3.340 (−10.549) − 0.513 (−1.399) 0.621 (1.345) − 0.791 (−2.334) − 0.142 (−0.543) − 1.075 (−3.062) 0.254 (1.286) 0.049 (4.976) 1.485 (3.331) 0.774 (4.759)
0.543 0.041 0.000 0.000 0.162 .178 0.020 0.587 0.002 0.545 0.000 0.001 0.000
0.000 0.000
Numbers in parentheses are the ratios of estimated coefficients to their standard errors.
Basic descriptive statistics of all the used variables, both in the profitability and adoption equations, are presented in Table 8. 4. Results and discussion The equations of returns to assets and the decision to adopt were jointly estimated as an incidental truncation model using the LIMDEP v.8 software (Greene, 2002). The maximum likelihood parameter estimates are shown in Table 9. What we notice first is that the correlation coefficient, ρeπeinv, of the disturbance terms eπ and einv is quite high and statistically different from zero allowing us to argue that the two equations are correctly jointly and not individually fitted, which was not known a priori. From an economic point of view, the specific finding suggests that the adoption decision of an EET is related to the firms' return on invested assets. Thus, the interpretation of the low rate of EET adoption which constitutes the energy efficiency gap (Jaffe and Stavins, 1994) in a larger scale or the energy efficiency paradox in particular, should be sought along different lines. In other words, the specific finding does not support the argument that the EET adoption decision making process ignores the profitability dimension of the decision outcome. We should also point out here9 that the specific finding does not answer the question of whether the firms' optimizing behaviour is false (Howarth and Stanstad, 1995). There is strong evidence, in our point of view, that energy efficiency is correlated to a firm's profitability and thus the energy efficiency gap has to be interpreted in terms of firm-specific characteristics that may influence its decision on the subject of adoption. In addition, the specific finding reveals that the approach of the present paper (i.e. the joint estimation of the firms' profitability and their adoption decision) is valid. 4.1. The decision to adopt an EET equation Accordingly, in the adoption equation we can distinguish two groups of factors (right-hand part of Table 9); one with variables that tend to influence this decision to adopt positively and one 9
We owe this to an anonymous referee comment.
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
2531
that discourages firms from adopting. In the first group we note the capital subsidy variable (SUBSD), the enforcement of regulations as it is captured by the (REGUL) variable, the combined effect of firm age and the size of its fixed capital (AGEFIX), the uncertainties that arise from the economic environment regarding principally the future evolution of inputs and produced output prices (UNCENVRM), and finally the specific energy consumption technological characteristics of the firms, captured by the dummy variable (EINTD). The positive signs of the (SUBSD) and (EINTD) variables (right-hand part of Table 9) are those expected. Firms that are granted greater capital subsidies and whose production technology is energy demanding are more likely to become energy-saving technology adopters. The same holds for the (REGUL) variable, as it is concluded from the positive sign and statistical significance of the corresponding coefficient. Thus we could argue that the establishment of regulations may be an effective policy instrument in the direction of energy consumption reduction and environmental protection. It is worthwhile here to analyse the results regarding the (AGEFIX) and (UNCENVRM) variables, which both exert positive and statistically significant influence on the probability that a firm will become an EET adopter. Considering that the product of the firms' age and their fixed capital (AGEFIX) depict their capital vintage (Geroski, 2000), the probability that firms will eventually decide to adopt energy-saving technologies increases as their own equipment becomes old and worn out, and thus replacement costs are unavoidable. This is particularly true for capital equipment that is so specialized that the cost incurred on installing it is sunk, since in this case there are no second hand markets where the old equipment might be disposed of. This finding confirms the arguments of DeGroot et al. (2003) and is also in accordance with Geroski's (2000) and Ernst's (1997) arguments. It is quite interesting to note the positive and statistically significant influence of the (UNCENVRM) variable on the decision to adopt energy-saving technologies (right-hand part of Table 9). It seems that the increase of uncertainties that are associated with the broader economic environment lures firms to adopt energy-saving technologies as a refuge in case of future unfavourable evolutions of input prices. The second group of variables encompasses all those drivers contributing negatively to the decision to adopt an EET and, consequently, positively to the energy efficiency paradox or to the energy efficiency gap. This group of variables includes the multiplicative effect of the size of the capital subsidy and the size of the firm proxied by the product of these two variables (SUBSDSZ), a proxy of the firms' profit margin (PRFM) and a variable capturing various types of institutional, informational, organizational and financial barriers (BARR) that in all probability confront a firm when the energy-saving investment project is realized. The results indicate some interesting relationships and implications. Firstly, the multiplicative effect of a firm's size by the subsidized capital which was granted to the firm in order to undertake the investment in EET (SUBSDSZ) is statistically significant and negative. This indicates that the size of the substitution effect between own and subsidized capital varies between small and large firms. Indeed, more general results on the effect of capital subsidies on firm performance and/or employment creation have shown that smaller or financially constrained firms are more susceptible to capital incentives than their larger or less constrained counterparts (Skuras et al., 2006; Wren, 1994). The probability to adopt energy-saving technologies becomes greater as the economic, environmental and technological uncertainties associated with the specific investment are reduced. Regarding the negative influence of the (PRFM) variable, we may argue that firms that exert market power, which in turn is depicted in their profit margins, are less willing to adopt energy-saving technologies. This finding is analogous to De Almeida's (1998) argument that public intervention is a necessary condition for organizing the market and promoting energy
2532
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
efficiency. At this point we have to mention that although our methodological approach is closer to the probit models (Geroski, 2000), this particular finding is consistent with the leading alternative, the epidemic models of technology diffusion, which correlate the extent of the adoption of new technologies to the processes of competition, resulting in the well-known Scurve of technology diffusion (indicatively, Stoneman, 1987; Karshenas and Stoneman, 1993; Baptista, 1999; Lucchetti and Sterlacchini, 2004). The ratio of the fixed to total assets (FIXT) and the ratio of the cost of the investment project to the fixed assets (INVFIX) have been found to have no statistically significant influence on the adoption decision. The same is concluded if we look at the risk and uncertainties that are associated with the examined energy-saving technologies that are captured by the variables (RISK) and (UNCTECN) respectively. None of the estimated coefficients of these two variables has been found to have statistically significant influence on the possibility that a firm will adopt an EET (right-hand part of Table 9). This surprising result may be due to the superior level of information that the adopters possess about the characteristics of the examined technologies as it is concluded from their answers to the relevant questions in the survey, as well as to the high capital subsidies that they are granted for adopting such technologies and which minimise the risk that would threaten their own funds. 4.2. The profitability equation In the estimation results of the (ROA) equation (left part of Table 9) we can see that five out of the six explanatory variables are statistically significant. Only the firms' size as it is captured by the market share (MS) variable was found to be a non-significant factor. We should note that we have tested the inclusion of the size variable both in the (ROA) and in the adoption equation under a large number of different specifications. In no case has the firms' size appeared to have significant effects either on the adoption decision or on the returns on assets. We have finally decided to include the (MS) variable as an explanatory variable in the adoption equation based on the value of the maximum likelihood function. Market concentration as it is captured by the Herfindahl-index (HI) has a predictably positive influence on returns on assets as it is expected by the indications of the SCP paradigm (Slade, 2004). The influence of the (INNOVD) dummy variable, which depicts the undertaking or not of innovative activities, products or processes in the last five years, is also found to be positive. Apparently, innovation activities result in increased product differentiation (product innovation) or in cost reductions (process innovation), which in turn reward firms with vast profits and consequently increase return on assets. The significant and positive influence of the (AGE) variable on the (ROA) may reveal two facts: first, that significant learning-by-doing gains are apparent and thus older firms perform better than firms that are in their infancy or in their adolescence period; second, that the analysed firms operate in a routinized technological regime instead of an entrepreneurial technological regime, which is in accordance with the aforementioned slow diffusion of energy-saving technologies among Greek manufacturing firms (Malebra and Orsenigo, 1993). The routinized technological regime is unfavourable to firms that are early adopters of new technologies. Achs and Audretsch (1988, 1990) and Audretsch (1991, 1995) correlate the nature of these two regimes to firm size, which is not, however, confirmed in our case. Firms that are capital instead of labor intensive appear to be more profitable as the estimated coefficient of the (CAPLAB) variable reveals. This may be due to their superior technological characteristics and/or to their high capital requirements, which also serve as barriers to entry, reducing competition and increasing profit margins (Ghemawat and Caves, 1986).
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
2533
Finally, the variable (RDD) that captures the research and development activities of the firms has a negative effect on returns on assets, which is contrary to the Yasuda's (2005) empirical finding. This may be attributed to the fact that although R&D activities are cost demanding, their outcome is highly uncertain and is realized, if ever, in the long run. In addition, this is in accordance with the aforementioned argument of the routinized technological regime in which the firms operate and is unfavourable to the introduction of innovations and R&D activities. 5. Conclusions and policy implications In the present paper we have developed a framework that addresses the energy efficiency paradox in terms of determinants of the energy-saving technology adoption in correlation to the returns of assets through a partial observability approach. For our analysis we use a unique data set, the outcome of a survey conducted among Greek manufacturing firms that were granted capital subsidies in order to invest in energy-saving technologies. Our empirical findings, which result from the joint estimation of an adoption and a returns on assets equation as an incidental truncation model, confirm that investment projects in energysaving technologies, at least in the case of Greek manufacturing firms in the past decade, are correlated to the returns of invested assets. This empirical finding combined with the low diffusion rate of EETs reveals that firm-specific characteristics may play a crucial part in a firm's decision to adopt such a technology, alongside uncertainties of an economic, social and political nature. Along these lines, we examine the influence of several factors that are not directly related to financial returns and costs, but may lead to the reduction of EETs adoption rate. Even though the role of most of these factors has already been explored in the relative literature, it is the first time, to the best of our knowledge, that their joint influence is investigated, thus avoiding misspecifications. According to our empirical findings, therefore, firms that receive relatively large subsidies, firms whose production technology is energy intensive and firms which present increased fixed capital vintage are more likely to adopt an EET. A flexible and effective internal organization which allows firms to cope with a wide range of barriers related to innovative technologies, such as human capital, information gathering and accumulated knowledge, process flexibility, and financial constraints also plays a positive part in this adoption process. The enforcement of regulations seems to be an effective policy measure, as is the reduction of bureaucratic obstacles to ensure firm participation in energy efficiency and pollution abatement or environmental conservation programs. A particularly interesting finding stems from the relatively greater effectiveness that capital subsidies exert on smaller scale firms. This may be due to the difficulty such small firms face to cope with the cost of replacing their mechanical equipment on the one hand and the internal cost of adjustment that results from such a substitution on the other. A note should be made here that the size of the substitution effect between own and subsidized capital may vary between small and large firms. Finally, a surprising but very interesting policy implication may arise from the fact that market imperfections leading to the reduction of competition on the one hand and the reduction of the expected volatility of input and produced output prices on the other, function as a restraint for the adoption of energy-saving technologies. Overall, our analysis concludes that the firms' decision to adopt an EET is positively related to their profitability, while the diffusion rate of the EET is influenced by specific firm characteristics. In addition, policy measures may be effective by accommodating the firms' access to such EETs and reducing the costs, financial and others, which arise from a positive adoption decision.
2534
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
Aknowledgment We thank the European Social Fund (ESF), Operational Program for Educational and Vocational Training II (EPEAEK II), and particularly the Program HERAKLITOS, for funding the above work. We also would like to thank Bill Greene and two anonymous referees for their helpful comments. References Achs, Z.J., Audretsch, D.B., 1988. Innovation in large and small firms: an empirical analysis. American Economic Review 78, 678–690. Achs, Z.J., Audretsch, D.B., 1990. Innovation and Small Firms. MIT Press, Cambridge, MA. Antonelli, C., Tahar, G., 1990. The adoption of process innovations and the competitive selection mechanism. The case of the Italian cotton weaving industry 1973–1982. Economic Notes 15, 1–19. Audretsch, D.B., 1991. New firms survival and the technological regime. Review of Economics and Statistics 73, 441–450. Audretsch, D.B., 1995. Innovation and Industry Evolution. The MIT Press, Cambridge, MA. Ayers, R.U., 1993. On Economic Disequilibrium and Free Lunch. Working Paper, vol. 93/45/EPS. Centre for the Management of Environmental Resources INSEAD, Fontainebleau, France. Bain, J., 1956. Barriers to New Competition. Harvard University Press, Cambridge, MA. Baptista, R., 1999. The diffusion of process innovations: a selective survey. International Journal of the Economics of Business 6, 107–130. Barney, J., 2001. Resource-based theories of competitive advantage: a ten year retrospective of the resource based view. Journal of Management 17, 99–120. Blackman, A., Bannister, G.J., 1998. Community pressure and clean technology in the informal sector: an econometric analysis of the adoption of propane by traditional Mexican brick makers. Journal of Environmental Economics and Management 35 (1), 1–21. Bloom, D.E., Killingsworth, M.R., 1985. Correcting for truncation bias caused by a latent truncation variable. Journal of Econometrics 27, 131–135. Cambridge Small Business Research Centre (CSBRC), 1992. The State of British Enterprise: Growth, Innovation and Competitive Advantage in Small and Medium-Sized Firms. Small Business Research Centre, University of Cambridge. Conrad, K., 1997. An Econometric Model of Production with Endogenous Improvement in Energy Efficiency 1970–1995. Manheim University Department of Economics. De Almeida, E.L.F., 1998. energy efficiency and the limits of market forces: the example of the electric motor market in France. Energy Policy 26 (8), 643–653. DeCanio, S.J., 1993. Barriers within firms to energy-efficient investments. Energy Policy 21, 906–914. DeCanio, S.J., 1998. The efficiency paradox: bureaucratic and organizational barriers to profitable energy-saving investments. Energy Policy 26 (5), 441–454. DeCanio, S.J., Watkins, W.E., 1998. Investment in energy efficiency: do the characteristics of firms matter? Review of Economics and Statistics 80, 95–107. DeGroot, H.L.F., Verhoef, E.T., Nijkamp, P., 2001. Energy saving by firms: decision-making, barriers and policies. Energy Economics 23, 717–740. DeGroot, H.L.F., Mulder, P.,Van Soest, D.P., 2003, ‘Subsidizing the Adoption of Energy-Saving Technologies’. Tilbergen Institute, Disc. Paper TI 2003-019/3. Dunne, P., Hughes, A., 1994. Age, size, growth and survival: UK companies in the 1980s. Journal of Industrial Economics XLII, 115–140. Ernst, H., 1997. The use of patent data for technological forecasting: the diffusion of CNC-technology in the machine tool industry. Small Business Economics 9 (4), 361–381. Faria, A., Fenn, P., Bruce, A., 2002. Determinants of adoption of flexible production technologies: evidence from Portuguese manufacturing industry. Economics of Innovation and New Technology 11, 569–580. Fedenia, M., Hirschey, M., 1992. Ownership structure and profitability. In: Audretsch, Siegfried, J.J. (Eds.), Empirical Studies in Industrial Organization, Essays in Honor of Leonard W. Weiss. Kluwer, London, pp. 237–251. Feeny, S., Harris, M.N., Rogers, M., 2005. A dynamic panel analysis of the profitability of Australian tax entities. Empirical Economics 30, 209–233.
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
2535
Fisher, M.F., McGowan, J.J., 1983. On the misuse of accounting rates of return to infer monopoly profits. American Economic Review 73, 82–97. Florida, R., Davison, D., 2001. Why do firms adopt advanced environmental practices, and do they make a difference? In: Cogliancese, C., Nash, J. (Eds.), Going Private: Environmental Management Systems and the New Policy Agenda. Resources for the Future, Washington, DC, pp. 82–104. Gabel, H.L., Sinclair-Dsgagne, B., 1992. Corporate responses to environmental concerns. In: Folmer, H., Gabel, H.L., Opschoor, H. (Eds.), Principles of Environmental and Resources Economics. Edward Elgar, Aldershot, pp. 347–361. Geroski, P.A., 2000. Models of technology diffusion. Research Policy 29, 603–625. Ghemawat, P., Caves, R.E., 1986. Capital commitment and profitability: an empirical investigation. Oxford Economic Papers 38, 94–110. Greene, W., 2002. LIMDEP Version 8.0. Econometric Software. New York. Hannan, T.H., McDowell, J.M., 1984. The determinants of technology adoption: the case of banking firm. The Rand Journal of Economics 15 (3), 328–335. Hasset, K.A., Metcalf, G.E., 1993. Energy conservation investments. Energy Policy 21 (6), 710–716. Hirschey, M., 1985. Market structure and market value. Journal of Business 58, 89–98. Hirschey, M., 1986. Mergers, buyouts and fakeouts. American Economic Review 76, 317–322. Hitt, M.A., Ireland, R.D., Camp, S.M., Sexton, D.L., 2001. Strategic entrepreneurship: entrepreneurial strategies for wealth creation. Strategic Management Journal 22 (6–7), 479–491. Hirst, E., Brown, M., 1990. Closing the efficiency gap: barriers to the efficient use of energy. Resources, Conservation and Recycling 3, 267–281. Howarth, R.D., Stanstad, A.H., 1995. Discount rates and energy efficiency. Contemporary Economic Policy 13, 101–109. IEA, 2002. Energy Policies of IEA Countries: Greece 2002 Review. International Energy Agency/Organisation for Economic Co-operation and Development, Paris. Jacobson, R., 1987. The validity of ROI as a measure of business performance. American Economic Review 77, 470–478. Jaffe, A.B., Stavins, R.N., 1994. The energy paradox and the diffusion of conservation technology. Resource and Energy Economics 16, 91–122. Jensen, M.C., Zimmerman, J.L. (Eds.), 1985. Symposium on Management Compensation and the Managerial Labour Market in Honour of William H. Meckling. Journal of Accounting and Economics, vol. 7, pp. 1–257. Jung, C., Krutilla, K., Boyd, R., 1996. Incentives for advanced pollution abatement technology at the industry level: an evaluation of policy alternatives. Journal of Environmental Economics and Management 30 (1), 95–111. Karshenas, M., Stoneman, P., 1993. Rank, stock, order and epidemic effects in the diffusion of new process technology. RAND Journal of Economics 24 (4), 503–527. Khanna, M., Zilberman, D., 2001. Adoption of energy efficient technologies and carbon abatement: the electricitygenerating sector in India. Energy Economics 23, 637–658. Levinthal, D., 1995. Strategic management and the exploration of diversity. In: Montgomery, C.A. (Ed.), Resource-based and Evolution Theories of the Firm. Kluwer, Norwell, MA. Lovins, A.B., Lovins, L.H., 1991. Least-cost climatic stabilization. Annual Review of Energy and Environment 16, 433–531. Lu, J.W., Beamish, P.W., 2001. The internationalisation and performance of SMEs. Strategic Management Journal 22 (6–7), 565–586. Lucchetti, R., Sterlacchini, A., 2004. The adoption of ICT among SMEs: evidence from an Italian survey. Small Business Economics 23 (2), 151–168. Malebra, F., Orsenigo, L., 1993. The dynamics and evolution of industries. Industrial and Corporate Change 5, 51–87. McFarland, H., 1988. Evaluating q as an alternative to the rate of return in measuring profitability. Review of Economics and Statistics 70, 614–622. Pargal, S., Wheeler, D., 1996. Informal regulation of industrial pollution in developing countries: evidence from Indonesia. Journal of Political Economy 104 (6), 1314–1327. Pindyck, R.S., 2000. Irreversibilities and the timing of environmental policy. Resource and Energy Economics 22, 233–259. Pindyck, R.S., 2002. Optimal timing problems in environmental economics. Journal of Economics Dynamics & Control 26, 1677–1697. Pizer, W.A., Harrington, W., Kopp, R.J., Morgenstern, R.D., Jhih-Shyang, S., 2002. Technology Adoption and Aggregate Energy Efficiency. Resources for the Future. Discussion Paper 02–52. Porter, M.E., 1980. Competitive Strategy: Techniques for Analyzing Industries and Competitors. The Free Press, New York. Porter, M.E., 1985. Competitive Advantage: Creating and Sustaining Superior Performance. Free Press, New York. Rogers, 1995. Diffusion of Innovations, 4th Ed. The Free Press, New York.
2536
K. Kounetas, K. Tsekouras / Energy Economics 30 (2008) 2517–2536
Rose, N.L., Joskow, P.L., 1990. The diffusion of new technologies: evidence from the electric utility industry. RAND Journal of Economics 21 (3), 354–373. Sanstand, A., Howarth, R.B., 1994. Normal markets, market imperfections and energy efficiency. Energy Policy 22 (10), 811–818. Schwert, G.W., 1981. Using financial data to measure effects of regulation. Journal of Law and Economics 24, 121–158. Shama, A., 1983. Energy conservation in US buildings: solving the high potential/low adoption paradox from a behavioural perspective. Energy Policy 11 (2), 148–167. Skuras, D., Tsekouras, K.D., Dimara, E., Tzelepis, D., 2006. The effects of regional capital subsidies on firms; productivity growth. Journal of Regional Science 46 (2), 353–379. Slade, M., 2004. Competing models of firm profitability. International Journal of Industrial Organization 22, 289–308. Slater, S., Olson, E., 2002. A fresh look at industry and market analysis. Business Horizons 15–22 (sp suppl.). Stoneman, P.L., 1987. The Economic Analysis of Technology Policy. Oxford University Press, Oxford. Stoneman, P.L., David, P.A., 1986. Adoption subsidies vs information provision as instruments of technology policy. The Economic Journal 96, 142–150 (Supplement: Conference Papers). Sutherland, R., 1991. Market barriers to energy–efficiency investments. The Energy Journal 12 (3), 15–34. Teece, D., 1981. Internal organization and economic performance: an empirical analysis of the profitability of principal firms. Journal of Industrial Economics 30, 173–199. Van Soest, D.P., Bulte, E.H., 2001. Does the energy-efficiency paradox exist? Technological progress and uncertainty. Environmental and Resource Economics 18, 101–112. Velthuijsen, J.W., 1993. Incentives for investments in energy efficiency, an econometric evaluation and policy implications. Environmental and Resource Economics 3, 153–169. Verhoef, E., Nijkamp, P., 1997. The adoption of energy efficiency enhancing technologies. Market performance and policy strategies in case of heterogeneous firms. NRP-II project on “Policy Instruments for Energy Efficiency Improvement” nr-953215. Vicini, G., 1998. How Italian companies deal with the environment. The results of a survey. FEEM Newsletter, vol. 2-1998, pp. 17–18. Watts, R.L., Zimmerman, J.L., 1980. On the irrelevance of replacement cost disclosures for security prices. Journal of Accounting and Economics 2, 95–106. Weiss, A., 1994. The effects of expectations on technology adoption: some empirical evidence. Journal of Industrial Economics 42 (4), 341–360. Winter, S., 2003. Understanding dynamic capabilities. Strategic Management Journal 24, 991–996. Wren, C., 1994. The build-up and duration of subsidy-induced employment: evidence from U.K. regional policy. Journal of Regional Science 34, 387–410. Yatchew, A., Criliches, Z., 1984. Specification error in probit models. Review of Economics and Statistics 67 (1), 610–619. Yasuda, T., 2005. Firm growth, size, age and behavior in Japanese manufacturing. Small Business Economics 24 (1), 1–15.
