3.1 Productivity

The first important task of our paper is to appropriately estimate firms' productivity. We adopt a semi-parametric method introduced by Olley and Pakes (Reference Olley and Pakes1996) to address this question. This method mitigates the endogeneity problem caused by unobserved productivity shock. In business practices, a firm will analyze its current productivity before making profit-maximizing selections of labor and materials that are combined with the quasi-fixed input of capital. A more productive firm tends to hire more workers, acquire more materials and assets. Due to this endogeneity of the choices of production inputs, the estimation of the input coefficients is very likely to be biased. To mitigate this endogeneity problem, we use the investment decision as an intermediate input which increases monotonically with productivity, conditional on capital input. This method helps to obtain unbiased coefficients of production inputs, and hence deliver accurately estimated productivity.

3.2 Environmental management

The next important task is to properly define environmental management and innovation activities. Let us first focus on environmental management. In early studies such as Pethig (Reference Pethig1976), Siebert (Reference Siebert1977), and McGuire (Reference McGuire1982), environmental management is perceived as stringent environmental legislation of the government. Recently it is considered as ‘a complex process that requires cross-departmental coordination and major changes in present operational processes’ (Ma et al., Reference Ma, Zhang and Yin2020) – for example, a mandatory monitoring and reporting system on greenhouse gas (Nishitani et al., Reference Nishitani, Kaneko, Komatsu and Fujii2014), international ISO14001 or organic certification (Delmas and Pekovic, Reference Delmas and Pekovic2013), or non-productive investments that aim to enforce environmental practices via renovating the organizational or human management of firms (Lannelongue et al., Reference Lannelongue, Gonzalez-Benito and Quiroz2017).

Following the literature, we define environmental management as the firms' managerial, non-productive measures to mitigate their environmental impacts. We further define two categories of management. One is internal management, which is undertaken by firms making changes to their internal managerial structure. Five measures are considered to pertain to this category. First, a firm monitors its own energy consumption including water usage, as well as its emissions of CO₂ and other pollutants. Second, a firm has strategic objectives that involve environmental or climate change issues. Third, a firm appoints a manager who is specifically responsible for environmental and climate change issues. Fourth, a firm sets targets for its energy consumption and emissions of CO₂ and other pollutants. Last, a firm aims to develop its own renewable energy such as solar, wind, hydro, biomass or geothermal energy. To sum up, internal management means that a firm incorporates environmental concerns into its managerial content.

The other is external management, which is enforced by external factors such as the government, the market, or the global community in terms of stringent environmental regulations. Another five measures are categorized under this type of management. First, a firm receives external audit of its energy consumption, water usage, and emissions of CO₂ as well as other pollutants. Second, a firm participates in an emission trading scheme such as the EU (for firms in EU countries only) or Kazakhstan (for Kazakhstan firms only) Emission Trading Schemes. Third, a firm is subject to energy taxation levied by the government. Fourth, a firm is subject to environmental standards set by the government. Last, a firm's customers require environmental certifications or adherence to certain environmental standards as a condition to do businesses with the firm. To sum up, external management captures the environmental awareness externally exerted by the government, the market, and the community.

A firm i is considered to implement internal management ($\textrm{internal} \textrm{managemen}{\textrm{t}_{it}}$ = 1) or external management ($\textrm{external managemen}{\textrm{t}_{it}}$ = 1) if it reports any internal or external management measures in year t; $\textrm{internal/external} \textrm{managemen}{\textrm{t}_{it}}$ = 0 if no environmental management is reported. We employ probit model equation (1) to examine the determinants of environmental management:

(1)\begin{equation}\textrm{Environmental management}_{it}^j=\left\{ {\begin{array}{@{}ll} 1& {\textrm{if } \gamma {X_{it}} + {\delta_1}{\theta_i} + {\delta_2}{\theta_t} + {\varepsilon_{it}} \ge 0}\\ 0& {\textrm{if otherwise}} \end{array}} \right.\end{equation}

where $\textrm{environmental management}_{it}^j \,{=}\, [{\textrm{internal management}_{it}^j, \textrm{external management}_{it}^j} ]$. j indicates each environmental management category, internal or external. ${X_{it}}$ includes all explanatory variables for firm i at year t, including $i$'s relationship with the international market and the government, labor input, and market condition. ${\theta _i}$ indicates industry and country fixed effects, which control for unobserved determinants of the outcomes and thus can account for the effects of a firm's difficult-to-quantify attributes, such as industry-specific policies, geographic features, and institutional services. ${\theta _t}$ is a year fixed effect that helps to control the time-specific attributes of the outcome. ${\varepsilon _{it}}$ is the error term.

We use equation (1) to estimate the probability of each environmental management j.Footnote ¹ We then create two integrated indices that respectively measure a firm's overall internal and external management. Specifically, we rank each estimated management probability from the lowest to the highest (five probabilities for internal management and another five for external management), sum up the ranks, and standardize the result to construct the management indices for each firm. Given this construction, a smaller index, or being top-ranked in our integration, means that a firm has a lower likelihood of environmental management.

3.3 Environmental innovation

The common feature shared by the environmental management practices is that they all directly affect a firm's managerial structure. Next, let us turn to environmental innovation. Rennings (Reference Rennings2000: 322) defines environmental innovation as ‘…measures of relevant actors (firms, …, private households) which: (i) develop new ideas, behavior, products and processes, apply or introduce them, and (ii) which contribute to a reduction of environmental burdens or to ecologically specified sustainability targets’. Following this definition, we consider the following activities that directly relate to the production process as environmental innovation: heating and cooling improvements; lighting system improvements; machinery and equipment upgrades; vehicle upgrades; climate-friendly energy generation on site; energy and water saving; pollution control, waste minimizing, and recycling during production. We then define two dummy variables that respectively represent two innovation statuses. Environmental innovation inputs equals 1 if a firm exhibits any of the above activities or 0 if no such activities have been reported.

At the same time, we are also interested in the outcome of a firm's environmental innovation. While environmental investments are supposed to lead to beneficial outcomes, economists and policy makers are also concerned about the possibility that the reduced costs and prices due to environmental innovation inputs, which are in terms of improved material and energy efficiency, will increase the consumption of the new efficient goods and cause a negative impact on the environment via a rebound effect (Binswanger, Reference Binswanger2001; Jaffe et al., Reference Jaffe, Newell and Stavins2005; Marin, Reference Marin2014). Therefore, we define that Environmental innovation outputs equals 1 if a firm's environmental impactsFootnote ² have been successfully reduced due to their innovation inputs, or 0 if otherwise.

(2)\begin{equation}\textrm{Environmental innovation}_{it}^j=\left\{ {\begin{array}{@{}ll} 1& {\textrm{if } \gamma {X_{it}} + {\delta_1}{\theta_i} + {\delta_2}{\theta_t} + {\varepsilon_{it}} \ge 0}\\ 0& {\textrm{if otherwise}} \end{array}} \right.\end{equation}

where $\textrm{environmental innovation}_{it}^j \,{=}\, [\textrm{environmental innovation inputs}_{it}^j, \textrm{environmental} \textrm{innovation outputs}_{it}^j ]$. We then use probit model equation (2) to estimate the probabilities of both innovation inputs and outputs.Footnote ³

3.4 Control variables

4.1 Environmental management and environmental innovation

We further provide figure 1 to exhibit the percentages of firms with environmental management and innovation in all 41 countries. ‘% Innovation Inputs (Outputs)’ indicates the percentage of firms with environmental innovation inputs (outputs) in each country; ‘% Internal (External) Management’ indicates the percentage of firms with internal (external) environmental management in each country. It is straightforward that a positive correlation between environmental management and innovation, which are both direct reflections of strong environmental caution, exists in all countries. If a country has a large percentage of firms implementing environmental management, it is very likely to have many firms conducting environmental innovation at the same time. This positive correlation between environmental management and innovation has been firmly supported by the literature, as discussed in section 2. Relatively speaking, Mongolia, Latvia, and Uzbekistan are the three countries with most firms engaging in both environmental management and innovation. Belarus, Ukraine, Czechia, and Slovenia have the highest percentages of firms with environmental management; Malta and Cyprus are the two countries featuring the highest percentages of environmental innovators.

Figure 1. Percentages of firms with environmental management and innovation by countries.

5.1 Baseline tests

The main purpose of this study is to investigate the influences of environmental management and innovation on productivity. We therefore run a baseline regression using equation (3):

(3)\begin{align} {a_{it}} & ={\beta _0} + {\beta _1}{\widehat {\textrm{environmental management}}_{it}} + {\beta _2}{\widehat {\textrm{environmental innovation}}_{it}}\nonumber\\ & \quad + {\beta _3}{\widehat {\textrm{environmental management}}_{it}} \times {\widehat {\textrm{environmental innovation}}_{it}}\nonumber\\ & \quad + \gamma {X_{it}} + {\delta _1}{\theta _i} + {\delta _2}{\theta _t} + {\varepsilon _{it}} \end{align}

where ${a_{it}}$ represents the estimated productivity from section 3.1 and ${\widehat {\textrm{environmental management}}_{it}}$ includes the internal and external management indices estimated in section 3.2. ${\widehat {\textrm{environmental innovation}}_{it}}$ includes the probabilities of the two innovation statuses estimated in section 3.3. Interaction terms are also included to examine their interactive influence on productivity. ${X_{it}}$ still includes the same explanatory variables for firm i at year t; industry, country, and year fixed effects are all included to address firms' heterogeneity.

Also notice that when estimating equation (3), an endogeneity problem arises. On the one hand, productive firms can overcome difficulties such as limited production conditions and exhibit active green managerial or innovating behavior. In other words, as shown in our statistics, productive firms self-select themselves to better manage and innovate; this sample selection bias leads to endogeneity. On the other hand, endogeneity may also occur because firms' management, innovation, and productivity are likely affected by the same elements of unobserved heterogeneity. For example, a firm's culture, history, or institutional structure that simultaneously affects its environmental awareness and productivity. For another example, the unpredictable outcomes and strategic confidentiality of innovation projects can create extra productivity shocks.

To address this endogeneity problem, notice that we intentionally make all the major explanatory variables in equation (3) be either in forms of or derived from estimated probabilities. This method is adopted from the Heckman correction, a two-step estimation that corrects bias caused by non-randomly selected samples and the consequential endogeneity explained above. Instead of using dummies (for example, whether to environmentally manage or innovate) as explanatory variables, we predict the probability of these variables in the first step and apply the results into the second step, a linear regression shown by equation (3). Hussinger (Reference Hussinger2008) uses this two-step method to analyze the effect of government subsidies on German firms' private innovation inputs. Wang and Huynh (Reference Wang and Huynh2014) also use this method and found that quality management played an important moderating role in how a firm's knowledge management affects its performance in Vietnam. We therefore follow the literature and adopt this two-step Heckman estimation on determinants of productivity.

5.2 Baseline results

Table 2 reports the results of equation (3), determinants of firms' productivity. The fact that a firm implements environmental management has a positive yet inconsistently significant effect on its productivity. This questionable influence of environmental management, therefore, requires further investigation, and we intend to look into how it interacts with innovation later in this study. In addition, environmental innovation, in terms of outputs or inputs or both, always promotes productivity.

Table 2. Determinants of productivity

Note: Standard errors are reported in parentheses.

A 1 per cent higher foreign ownership percentage brings up productivity by 0.2–0.3; foreign capital assets incorporate new knowledge and techniques from the global market, and therefore bring in technology spillover. However, the export percentage does not display significant coefficients. Firms' accession to the global market proves to help their productivity growth but only to a limited extent. Additionally, a closer relationship with the central government also features insignificant coefficients.

The more employees a firm has, the more productive it becomes. In addition, having more employees with an advanced education level exerts significantly positive influences on firm productivity; 1 per cent more employees with university degrees will significantly increase productivity by 0.3–0.4. Hence, a large employment size as well as a high-skilled workforce substantially promotes firm productivity. The existence of informal competition suggested a less regulated market; it significantly hinders firms' productivity growth.

Furthermore, interaction terms between environmental innovation and management reveal very important findings. These interaction terms all feature significantly negative coefficients, suggesting that environmental innovation and management are reducing each other's positive influence on productivity. Those firms that implemented environmental management will see a declining positive effect of its innovation on productivity; those that performed environmental innovation will also expect a diminishing positive effect of management. In other words, when it comes to promoting productivity, environmental management and innovation dilute each other's effect.

A potential problem with the results in table 2, however, is that the high correlation between environmental management and innovation may cause multicollinearity and hence bias the results. A firm is very likely to implement both environmental management and innovation at the same time. The positive correlation between both types of activities has been depicted in figure 1 and received abundant support from the literature. To mitigate the multicollinearity problem and further study the relationship between environmental management and innovation, we conduct subsample studies on firms with different management or innovation statuses.Footnote ⁸

5.3 With versus without environmental innovation

Table 3 reports the results of firms that have performed environmental innovation activities, and those that do not innovate. Columns (i) and (ii) respectively show firms with and without innovation inputs; columns (iii) and (iv) respectively show firms with and without innovation outputs. Interestingly, the coefficients of internal and external environmental management lose significance for innovators but remain significantly positive for non-innovators. This echoes the finding regarding the interaction terms in table 2. Among firms conducting innovation activities that enhance environmental friendliness, their environmental management does not affect productivity. We can only observe a positive influence of environmental management on productivity of firms not innovating. The substitutability instead of complementarity between environmental management and innovation is confirmed.

Table 3. Determinants of productivity – with versus without environmental innovation

Note: Standard errors are reported in parentheses.

There is another interesting finding on firms' connection with the international market. A 1 per cent higher foreign ownership significantly raises innovators' productivity by over 0.3, but insignificantly reduces non-innovators' productivity. One per cent more exports significantly increases productivity of innovators by 0.2–0.3, but significantly decreases productivity of non-innovators. According to the learning-by-exporting theory, a firm's participation on the international market will allow access to worldwide state-of-the-art technologies and facilitate its growth due to technology spillover. Nonetheless, we find that foreign-owned capital assets and exports do not always promote firms' productivity growth, especially for those which do not innovate. Therefore, the learning-by-exporting effect is assured via innovation. Without innovation, globalization cannot effectively improve productivity.

A similar conclusion also applies to firms' relationship with the government. Operating with 1 per cent more capital assets owned by the government significantly increases innovators' productivity by 0.6–0.7 while it decreases non-innovators' productivity by 0.4–1. Securing a contract to do businesses with the government significantly increases innovators' productivity by 0.12 but decreases non-innovators' productivity by 0.4–0.5. Therefore, government influence promotes firm productivity if and only if innovation activities have been performed.

Echoing the baseline findings in table 2, the size and quality of the workforce strongly affect productivity no matter whether firms innovate or not. Hiring more employees significantly raises productivity of both innovators and non-innovators. One per cent more employees with university degrees significantly increases all firms' productivity by 0.3–0.4. Informal competition significantly decreases innovators' productivity by over 0.1 and non-innovators' productivity by 0.3–0.4.

5.4 With versus without environmental management

Next, we group firms by their environmental management statuses and report the results in table 4. Columns (i) and (ii) are results of firms with internal management; columns (iii) and (iv) are those of firms without internal management. Similarly, the next two columns are firms with external management while the last two columns are those without external management. Among firms with internal or external management, environmental innovation insignificantly affects productivity; among firms without internal management, innovation inputs increase productivity by 0.1 while outputs increase productivity by 0.15; among firms without external management, both innovation inputs and outputs increase productivity by 0.1.Footnote ⁹

Table 4. Determinants of productivity – with versus without environmental management

Note: Standard errors are reported in parentheses.

Combining the findings from tables 3 and 4, the substitutability between environmental innovation and management is reaffirmed. Although both environmental activities often occur together and positively affect firm productivity, they tend to offset each other's effect. Instead of the common understanding that environmental innovation and management may augment each other's effect on productivity, we prove otherwise. Among firms with environmental innovation, their environmental management does not encourage productivity growth; the positive influence of environmental management can only be observed among the non-innovating firms. Meanwhile, the positive influence of innovation can only be observed among firms without environmental management.