Theoretical and Empirical Foundations of Energy Production Efficiency Activity

The objective of this paper is to explore the energy production efficiency activity of large R&D-intensive firms. Research methodology follows two steps: first, we describe the theoretical background through a firm level model and, second, we develop econometric techniques that explore spatial spillovers and deal with the endogeneity of the explanatory variables. The findings show a significant effect of energy innovation externalities on firms’ environmental performance.


Introduction
The transition to a low-carbon competitive economy is a central challenge of our time.The success of this aim would assure new economic opportunities, prosperity, welfare and growth, while the eventual failure may put our well being at stake.All countries need to step up their investments in energy efficiency and renewable technologies.According to Porter (1991), strict environmental regulations could enhance the competitive advantage against rivals.
The existence of energy spillovers as a market failure provides support to the Porter hypothesis (Ambec et al. 2013).Indeed, Mohr (2002) finds that firms may switch to equilibrium with higher R&D investments, when there are technology spillovers to competitors, while Gans (2012) explains that more rigid climate policies will not produce necessarily more innovation.Also Greaker (2003) identifies technological spillovers as a theoretical prediction of Porter hypothesis.Hence, knowledge spillovers play a potentially important role for innovation and productivity in environmental technologies.However, we lack empirical studies about knowledge spillovers in such technologies at the firm level, as discussed in Hoppmann (2016).
To close this gap, in this article we investigate the effects of spillovers from innovation in energy production efficiency activities on environmental performance for American, Japanese and European firms.In particular, we distinguish the intra-industry spillovers (Marshall, 1890;Arrow, 1962;Romer, 1986) and inter-industry externalities (Jacobs, 1969).
The paper is structured as follows.Section 2 explores the spatial energetic spillovers of firms.Section 3 illustrates the theoretical core of the paper.Data and empirical framework are presented in the section 4. Section 5 evidences the results and final section discusses the findings and concludes.

Energy Innovation and Spatial Analysis within the Triad
In order to investigate energetic innovation and its distribution, we use particular econometric tools (Pisati, 2008;Crow, 2015;Kondo, 2015 and2016).
In Table 1, we illustrate energy patents with IPC selected by the OECD or the World Intellectual Property Organization (WIPO), as in Marin and Lotti (2016).
As we may observe in Tables 2 and 4, the positive value of Moran-I indicates positive spatial autocorrelation across the American and European regions, that is, regions neighboring a region with high energy patents also show high energetic innovation rates, while in Japan ( Table 3) we do not find a significant autocorrelation across regions.
Figure 1.Energy patents in the USA In Figure 1, where the USA country is clustered into 51 states (see Appendix), we may observe that New Jersey, Connecticut, Michigan, California and Tennessee exhibit the hot spots, while New Hampshire, Ohio, Kentucky, Virginia, Oregon and Florida display the cold spots.In Figure 2, where Japan country is clustered into 47 prefectures (see Appendix), we observe that Tokyo exhibits hot spots, while Kanagawa and Saitama display cold spots.

Theoretical Framework
In order to better specify the relationships and sources of our empirical framework, this section develops a multi-region multi-sector theoretical model, and focuses on aspects related to different types of green energy (wind, solar and geothermal energy, integrated emissions control, lightning to quote some) knowledge diffusion and accumulation.Following Bretschger et al. (2017), in each region at time t final production   in sector i may be taken as the combination of two different outputs from two different production techniques: a first   where knowledge diffusion leads to a "greening" of economies improving the different types of green energy (wind, solar and geothermal energy, integrated emissions control, lightning to quote some), a second one   where companies' green investments are quite irrelevant.Knowledge diffusion, together with other targets such as water pollution abatement and solid waste collection, may have an important role for a drop of global climate policies cost in most of emerging countries.The final production   may be written as follows: The green output   may be taken as a sector-specific output   and a composite input from other sector where  ∈ [0,1] is a share parameter.Output   is produced by combining varieties of intermediate composite goods   according to: where   measures the total number of intermediate varieties in sector i region r at time t.If we assume symmetric production of intermediate composite goods:   =   we will have:  =     and (2) may be written as: from inspection of which we will observe that   may grow either by yielding more quantity (x) per firm, given ( J ), or by increasing the whole number of intermediate varieties.In this light, as well emphasized in Bretschger et al. (2017) each single variety is produced by a precise company and every new firm requires additional knowledge capital to be active.Different interpretations may be attributed to J:  a measure of knowledge green Capital,  the number of intermediate varieties,  total of intermediate Companies.
Let's now introduce a mechanism of knowledge accumulation and diffusion.We will assume that knowledge capital at time t in region r depends on investment at time t-1  −1 , knowledge stock at time t-1, and knowledge increment at time t-1 ∆ −1 : with  measuring a constant knowledge depreciation rate.In order to extend the knowledge diffusion to cover inter-sectorial domestic and foreign intra-sectorial knowledge spillovers, as in Bretschger et al. (2017) we can take assume: = ∑  ℎ ℎ≠ (7) where s and h are regional and sectorial indexes, while    and    stand for respectively domestic inter-sectorial and foreign intra-sectorial knowledge stocks.By combining the above conditions we may easily derive what follows: Since intermediate goods   are produced by combining labor   , energy   and physical capital   , and may be: condition (2) will be revised as: )(  ,   ,  )];(1 − )  } (11).
Shifting our attention to   , with irrelevant companies' green investments, we can assume that production derive by the combination of labor   , energy   , physical capital   , and finally knowledge capital   depending on knowledge green Capital   =   (  ) we may write that: ))]} (13) Inspection of condition (13), and the premise according to which each sector has a production structure as the one depicted in Figure 4, take us to suggest that Knowledge diffusion leading to a greening process of economies depends on accessibility and absorptive capacity, and conclude with the following testable result:

Result [H]:
The effect of spillovers due to diversified green technology fields concerning energy production efficiency activity (Jacobian externalities) on firms' environmental performance is positive.

Data and Methodology
OECD, REGPAT database ( 2017) is employed.We match the name of the same 240 firms to applicant's name from European Commission (2013), as in Aldieri (2013).Another source is the Environmental Accounts providing CO2 emissions variable by country and by year in the World Input Output Database (WIOD).
In order to measure the effect of energy technology on firms' environmental performance, we develop the following extended Cobb-Douglas production function: 2 =   +   +  1   +  2   +  3   +  1   +  2   +   (14) where  = natural logarithm; 2  = Environmental performance measured as ratio between net sales and CO2 for firm i and year t;   = physical capital stock for firm i and year t; = number of employees for firm i and year t;   = R&D capital stock of firm i and year t;   = firm's fixed effects;   = set of time dummies;   = vector of externalities relative to firms belonged to the same industry, computed on the basis of environmental proximity.This is the Jaffe's procedure (Aldieri and Cincera, 2009) with energy patents;   = vector of externalities relative to firms of different industries, as for the previous variable; ,  = vectors of parameters; The previous extended version Cobb-Douglas production function to consider also the spillover components is in line with the relative empirical literature (as discussed in Aldieri and Cincera, 2009;Aldieri andVinci, 2017a and2017b).
In Table 5, we display the summary statistics of our sample.Note: a) 1837 observations;

Results and Discussion
In order to address the endogeneity of the explanatory variables, we estimate equation ( 14) using the Generalized Method of Moments (GMM) 3 estimator.
In Table 6, we show the effects of intra-industry spillovers (INTRA) and inter-industry spillovers (INTER) on firms' environmental performance.Moreover, we include lagged energy spillover components by a year to mitigate contemporaneous effects.In order to identify the elements that change over time but not over the cross-sectional dimension of the sample, we include also time, country and industry dummies.The results of Hansen tests do not reject the null hypothesis of valid instruments, supporting the no-correlation of the instruments with the error term.
As far as our findings are concerned, intra-industry externalities (INTRA) have a negative effect, while the inter-industry externalities has a positive one, in line with the expected sign of the variables included in the model.Indeed, the diversification process in green activities leads to important environmental performance improvements.This result could be interpreted as an important instrument of policy makers: more additional incentives are required to improve the complementarity between the energetic sectors.

Concluding Remarks
The objective of this paper is to explore the energy production efficiency activity of large R&D-intensive firms.An original Environmental proximity matrix is constructed, on the basis of technological vectors for each firm.Since there are few empirical studies about knowledge spillovers in such technologies at the firm level, as discussed in Hoppmann (2016), we close this gap by investigating the effects of spillovers from innovation in energy production efficiency activities on environmental performance for large international firms.In particular, we distinguish the intra-industry externalities and inter-industry externalities.To address the endogeneity of the explanatory variables, we run the Generalized Method of Moments (GMM) estimator.From the empirical results, we may observe that energy intra-industry spillovers (INTRA) have a negative impact on environmental performance, while the energy inter-industry components (INTER) have a positive effect.This finding is useful as a relevant policy maker instrument: the full sustainability achievement requires more incentives to complete the integration between energy technology fields.
However, further research on this topic is needed.The analysis should focus on factors that affect heterogeneity in technology spillovers effects both in spatial context and on the basis of industrial sectors.

Figure
Figure 4. Firms' Production structure

Table 2 .
Moran scatterplot for the USA Note: Moran-I test: 0.467, p-value: 0.000 Figure2.Energy patents in Japan

Table 3 .
Moran Scatterplot for Japan Note: Moran-I test: 0.099, p-value: 0.460Figure 3. Energy patents in EuropeIn Figure3, where Europe is clustered into 42 countries (see Appendix), we observe that Germany exhibits hot spots, while Belgium displays cold spots.

Table 5 .
Summary statistics

Table 6 .
Environmental Performance of Spillovers effects: GMM estimates