Dynamic Relationship Between Environmental Policy and Technological Innovation: Evidence from Firms in Korean Emission Trading Scheme

Abstract

To cope with radical global environmental regulation, it is necessary to find efficient ways to decrease carbon emissions with a low burden on firms and to design valid policies for a sustainable economy. This study investigates the link between the carbon emissions market and firms’ innovation activities and examines the impact of R&D investment caused by the carbon emissions market on the financial performance of firms participating in the Korean Emission Trading Scheme (KETS). This study assumes that a well-designed carbon emissions market can amplify the effect of R&D investment on improving the financial performance of firms and tests the efficiency of this market based on the KETS data. The results show that the indirect R&D elasticity that affects financial performance is simultaneously negative in the short and long run, except in Phase 1 (2015–2018). From this result, this study concludes that the carbon emissions market cannot induce technological innovation to increase financial performance in Korea, and the dynamic efficiency of the market is low. The industry comparison results confirm that the emissions market has a statistically significant effect on financial performance only in carbon-intensive industries. This study thus shows that the impact of technological innovation induced by the carbon emissions market varies depending on industrial characteristics.

1. Introduction

To cope with international environmental regulations, such as the United Nations Framework Convention on Climate Change (UNFCCC), high-energy-consuming countries have developed various policies and methods to reduce carbon emissions efficiently. Among various methods, emission trading schemes (ETSs) are widely adopted and applied as the most economical mechanisms for carbon reduction [1] because they provide incentives for firms to improve carbon reduction technologies and lower the marginal abatement costs that each firm has to bear in the long [2]. Korea also launched a national carbon emission trading system mechanism called the Korean Emission Trading Scheme (KETS) in 2015. During the first phase of the KETS, from 2015 to 2017, 525 firms participated in the mechanism with the designated CO2e cap of 1.69 billion tons from 23 industries. In the second phase, from 2018 to 2020, 609 firms participated in the KETS from the 30 industries under the CO2e cap of 1.64 billion tons. In the third phase, from 2021 to 2025, 687 firms in 32 sectors participated in the KETS under the CO2e cap of 2.68 billion tons. In the third phase, free allocation accounts for 90% of the total allowances, which is a significant decrease from 100% in the first phase and 97% in the second phase. The third phase aims to contribute to the achievement of the 2030 national greenhouse gas reduction goal by actively inducing market transactions to allow financial institutions and third parties to participate in the market.

In theory, a well-designed carbon emissions market can have a significant effect in inducing market participants to improve carbon mitigation technologies, eventually enhancing the financial performance of firms [3]. The carbon emissions market participants make efforts to reduce total costs by developing various carbon reduction technologies to cope with environmental regulations. If the emissions market is designed to attract participants in technological innovation, its implementation can eventually have a positive effect on improving the financial performance of firms based on the reduction of total costs in the long run. Therefore, a well-designed carbon emissions market can amplify the effect of R&D investment on improving the financial performance of firms.

The effect of the carbon emissions market on firm performance depends on the varying conditions of different industries and countries [4]. The impact of technological innovation caused by the emissions market on the financial performance of firms is not very clear and widely investigated and may vary according to the industry structure of the country and the design of the emissions market. Therefore, investigating the effects of the carbon emissions market using firm-level data is essential. Additionally, this paper estimates the impact of indirect R&D investment on financial performance by various industries and suggests policy implications.

While the carbon emissions markets are expected to amplify the impact of a firm’s innovation activities on its financial performance, the link between those is not actively investigated and controversial. Moreover, the indirect R&D effect caused by the carbon emissions market can be different depending on the country and the industry. This study analyzes the effect of technological innovation induced by the carbon emissions market on financial performance and compares the effects of these factors by industry. The contribution of this study to the existing literature can be summarized into the following three components.

First, this study provides empirical evidence to compare the impact of the KETS on financial performance based on specific firm-level data and provides empirical evidence to capture the relationship between the technological change caused by the carbon emissions market and financial performance. While existing research has analyzed the effect of innovation activity and the carbon emissions market separately, this study identifies the path through which the carbon market induces R&D investment. Finally, we analyze the impacts of the carbon market and technological innovation on financial performance simultaneously.

Second, this study provides industry-specific strategies for enhancing a firm’s performance by comparing the short- and long-term effects of the carbon emissions market on financial performance by industry. Different production processes and characteristics of varying industries can cause heterogeneous carbon market effects on a firm’s performance. Therefore, policies considering industry-specific characteristics are needed to develop an effective long-term strategy for enhancing carbon emissions market efficiency.

Third, the study offers insights for firm managers regarding what needs to be considered to align the carbon reduction performance and financial performance. The estimated results provides firms participating in the KETS with guidelines that are helpful in setting up long-term carbon reduction plans. Through the results of this study, it is possible for the government to establish plans for increasing the efficiency of that ETS market that take into account the characteristics reflecting the industry condition.

The structure of this paper is as follows: Section 2 investigates the literature and discusses our hypothesis structure regarding the relationship between the carbon emissions market and financial performance; Section 3 illustrates the data and model structure; Section 4 presents and discusses the empirical results; and Section 4 provides the conclusions and policy implications.

2. Literature Review

Various studies, including those by references [5,6,7,8] discuss the relationship between R&D and financial performance. Reference [5] confirmed that intensive R&D investment significantly lowers financial performance. Reference [9] tested the effect of innovation on one-year lagged financial performance based on Chinese listed IT firms for 2007 to 2009. The results indicated a significant correlation between R&D intensity and financial performance. Reference [6] discussed the impact of green R&D on environmental and financial performance using a sample of Japanese manufacturing firms from 2001 to 2010. The authors concluded that green R&D was positively related to financial performance, but that there was a negative relationship between green R&D and carbon emissions. Reference [7] found a relationship between government R&D and financial performance, reflecting green factory identification (GFI) shock. The authors concluded that there existed an inverted U-shaped relationship. Reference [8] investigated the relationship among R&D investment; environmental, social, and governance (ESG) performance; and financial performance based on Chinese manufacturing firms’ data from 2018 to 2022. From the estimation results, they concluded that innovation activities boosted firms’ performances.

References [10,11,12,13] analyzed the impact of the carbon emissions market on a firm’s financial performance. Reference [10] investigated the effect of cap-and-trade regulations on a firm’s profits using daily returns for 552 stocks from EUROSTOXX index data. The results revealed that stock values decreased because of the implementation of the EU ETS in carbon- and electricity-intensive industries. Reference [11] examined the relationship between environmental and financial performance under the increasing likelihood of environmental regulation based on data from 1095 U.S. corporations from 2004 to 2008. They concluded that improving corporate environmental performance caused a decline in short-term financial performance (return on assets) but increased potential long-term value. Reference [13] analyzed the impact of China’s pilot ETS on financial performance using the PSM-DID method based on Chinese firm-level data. The results showed that the implementation of the pilot ETS improved the firm’s financial performance. Reference [12] investigated the heterogenous relationship between environmental and financial performance based on firms participating in EU ETS. Based on a quantile regression estimation results, they concluded that the magnitude of the relationship differed by the financial performance distribution. While innovation activities and ETS are expected to affect the financial performance of firms, the theory supporting this relationship is not being actively investigated. In addition, the relationship between innovation effort of firms, environmental policy, and firm productivity can vary by different industries and countries. The contribution that this paper, setting it apart from the previous literature, can be addressed from two perspectives. First, this study simultaneously estimates the direct effect of environmental policies on financial performance and the indirect effect of the emissions market on enhancing the performance of firms by causing technological innovations. Second, this paper compares the relationship between the innovation activity of firms, ETS, and the financial performance of firms by industry. The following three hypotheses are investigated in this study to verify the factors influencing financial productivity in firms joining the KETS.

Hypothesis 1: 

The financial performance of a firm is influenced by innovation activity in Korea.

Hypothesis 2: 

The Korean Emissions Trading Scheme induces technological innovation in market participants.

Hypothesis 3: 

The factors influencing the financial productivity of firms vary by industry.

This study estimates the financial productivity of firms participating in the KETS to test our three hypotheses. A panel analysis based on firm-level data is conducted to investigate the link between innovation, ETS policy, and financial productivity in Korea.

3. Data and Methodology

3.1. Data

Two major sources are used for data in this study. The data on the firm’s carbon emissions between 2015 and 2022 were acquired from Korea’s Emission Trading Registry System (ETRS). The financial data of the firms were obtained from the KISVALUE database offered by NICE Information Services Co. The firm data consist of panel sample data on 687 firms disclosing carbon emissions annually and participating in the first-to-third-phase emission trading schemes in Korea (KETS). A total of 32 manufacturing industries are included in the sample data. Finally, we obtained 8244 firm year data points from Korean industries. For industry composition in the sample data, the general industry accounts for 15.9%, petrochemicals for 11.2%, the waste for 10.6%, and finally transportation for 7.2%.

Table 1. Summary statistics.

	Mean	Std. Dev.	Min	Max	Skewness	Kutosis
lnROE	1.981	1.212	−6.119	13.192	−0.384	7.687
lnROA	1.243	1.197	−7.243	5.824	−1.048	5.632
lnTobins_q	−0.467	0.747	−8.873	2.552	−1.451	12.585
lnCarbon intensity	−11.694	1.942	−24.549	−1.828	0.096	4.675
lnLabor	6.026	1.682	0.000	11.704	0.294	3.155
lnCapital intensity	−0.516	1.046	−8.066	10.640	0.646	14.756
lnDebt ratio	2.439	1.474	−3.090	11.455	0.539	4.201
lnR&D_profit	−5.785	2.060	−16.539	2.096	−0.976	4.658
lnFirm’s age	2.974	0.957	0.000	4.522	−0.963	3.597

presents summary statistics of the data. This study selects the return on equity (ROE), the return on assets (ROA), and Tobin’s Q to measure each firm’s financial performance. ROE is a measure of short-term financial performance, showing the ratio between net income and shareholders’ equity, while ROA represents the short-term performance of firms and is defined as net income divided by total assets. Tobin’s Q is measured using the firm’s market value scaled by its replacement cost and indicates the market’s expectation of its future growth. Therefore, in this study, Tobin’s Q represents the long-term financial performance of firms [14]. Carbon intensity is measured as the ratio between total amount of carbon emissions and total sales, and it is a key indicator with which to determine improvements in environmental efficiency. Capital intensity is measured by dividing capital by total sales. We also define the debt ratio as total firm debt scaled by total assets. The R&D_profit represents the firm’s technological innovation and is defined as the total firm’s R&D investments scaled by the total firm’s profit.

Table 2. Correlation matrix.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
lnROE (1)	1.000
lnROA (2)	0.911 *	1.000
lnTobins_q (3)	0.090 *	−0.085 *	1.000
lnCarbon intensity (4)	−0.451 *	−0.537 *	−0.113 *	1.000
lnLabor (5)	−0.073 *	−0.016	0.300 *	−0.435	1.000
lnCapital intensity (6)	−0.273 *	−0.013	−0.237 *	−0.163	−0.036 *	1.000
lnDebt ratio (7)	−0.714 *	−0.913 *	0.293 *	0.520	0.031 *	−0.209 *	1.000
lnR&D_profit (8)	0.060 *	0.056 *	0.147 *	−0.088	0.176 *	0.113 *	−0.052 *	1.000
lnFirm’s age (9)	−0.109 *	−0.009	0.091 *	−0.108	0.262 *	0.049 *	−0.053 *	−0.045 *	1.000

* denotes a significant correlation at the 5% significance level.

shows the pairwise correlations and illustrates that the R&D variable is positively correlated with a firm’s financial performance variables but negatively correlated with the debt ratio. Unlike other financial performance variables, Tobin’s Q is positively correlated with debt ratio and firm age in the data.

3.2. Model

We consider the following model to examine the effect of technology innovation caused by emission trading schemes on the financial performance of firms:

lnFP_it = α + β₁(lnFP)_it−1 + β₂(lnCarbon intensity)_it + β₃(lnLabor)_it + β₄(lnCapital intensity)_it
+ β₅(lnDeb tratio)_it + β₆(lnR&D_profit)_it + β₇(lnR&D_profit)_it−1 + β₈(lnR&D_profit_it*Phase1)
+ β₉(lnR&D_profit_it*Phase2) + β₁₀(lnR&D_profit_it*Phase3) + β₁₁(lnR&D_profit_it−1*Phase1) + β₁₂(lnR&D_profit_it−1*Phase2)
+ β₁₃(lnR&D_profit_it−1*Phase3) + β₁₄(lnAge)_it + μ_i + η_t + ε_it.

(1)

Financial performance (FP) is our dependent variable, which is the ROE, ROA, and Tobin’s Q at period t of each firm i. The number of employees represents the firm size in this model. In addition, capital intensity and debt ratio variables are included as explanatory variables to control for firm characteristics. Carbon intensity represents the firm’s environmental performance. lnR&D_profit captures the effect of technological innovation on the firm’s financial performance. Phases 1, 2, and 3 are modeled as dummy variables, and these variables capture the effects of the KETS. The value for Phase 1 is set as 1 from 2015 to 2017 and 0 otherwise. The value for Phase 2 is set as 1 from 2018 to 2020 and 0 otherwise. The value for Phase 3 is set as 1 from 2021 to 2022 and 0 otherwise. lnR&D_profit*Phase1, lnR&D_profit*Phase2, and lnR&D_profit*Phase3 capture the indirect and additional effects of innovative activities on improving financial performance due to the introduction of the KETS. If the implementation of the first, second, and third phases of the KETS induces additional innovation activities to improve firms’ financial performance in the short run, the coefficients β₈, β₉, and β₁₀ in Equation (1) will show statistically meaningful positive signs.

In reality, the effects of a firm’s investment in R&D due to the implementation of the carbon emissions market is difficult to capture in the short run. Firms facing environmental regulations invest in carbon reduction technologies to reduce their carbon costs. However, the effects of investments occur in the future and not in the present. Therefore, to reflect the cumulative and time lag effects of R&D investment, this model added explanatory variables multiplied by lagged R&D investment and Phases 1, 2, and 3. Based on these estimators, we estimate the long run effect of promoting R&D investment caused by the carbon emissions market. This study tests whether the emission trading scheme is inducing technological innovation in the short run based on the t-test for coefficients β₈, β₉, and β_10. and calculates the long-run elasticity of the R&D caused by the emissions trading scheme based on β₁, β₁₁, β₁₂, and β_13.

μ_i is the individual fixed effect, which represents the unique characteristics of firms that remain unchanged over time; η_t is the time-fixed effect; and ε_it is residual. In this model, the expectation value of the residual is not zero because of the existence of μ_i and η_t. Therefore, the estimation results based on OLS (ordinary least squares) are biased. Additionally, endogeneity problems may occur because the lagged dependent variable is included in the explanatory variables. If the lagged dependent variable is used as the independent variable in this model, the residual and the lagged dependent variable are highly correlated, and finally, the estimation results are biased. To solve this problem, this study applies the system GMM [15]. The system GMM combines the first difference and level equations with a wider set of instruments. The level of the dependent variable and the lagged value of the difference variable are used as additional instrumental variables to solve the endogeneity problem. Therefore, Equation (1) is converted into the final model in Equation (2) as follows:

∆lnFP_it = α + β₁(∆lnFP)_it−1 + β₂(∆lnCarbon intensity)_it + β₃(∆lnLabor)_it
+ β₄(∆lnCapital intensity)_it + β₅(∆lnDeb tratio)_it + β₆(∆lnR&D_profit)_it + β₇(∆lnR&D_profit)_it−1 + β₈(∆lnR&D_profit_it*Phase1)
+ β₉(∆lnR&D_profit_it*Phase2) + β₁₀(∆lnR&D_profit_it*Phase3) + β₁₁(∆lnR&D_profit_it−1*Phase1) + β₁₂(∆lnR&D_profit_it−1*Phase2)
+ β₁₃(∆lnR&D_profit_it−1*Phase3) + β₁₄(∆lnAge)_it + ∆μ_i +∆η_t +∆ε_it.

(2)

To apply the system GMM, the following three conditions must be satisfied [16]: first, the dependent variable does not have a unit root; second, the appropriateness of using instrumental variables should be verified using the Hansen test; third, the null hypothesis that there is no autocorrelation among error terms should be tested based on the Arellano–Bond test.

Table 3. Results of panel unit root tests.

Data	LLC	IPS	Fisher-ADF	Fisher-PP
lnROA	−59.419 ***	−20.228 ***	2105.52 ***	2416.49 ***
lnROE	−63.235 ***	−25.478 ***	2415.57 ***	2777.00 ***
lnTobins_q	−44.325 ***	−11.732 ***	1639.28 ***	1943.37 ***
∆ lnROA	−118.395 ***	−55.515 ***	4337.79 ***	5839.68 ***
∆ lnROE	−102.883 ***	−58.351 ***	4578.74 ***	6334.97 ***
∆ lnTobins_q	−80.742 ***	−43.329 ***	3498.72 ***	4236.87 ***

Note: ∆ is the first difference operator. *** show significance at the 1% levels, respectively.

summarizes the panel unit root test results of the dependent variables. In this study, four methods—Fisher-ADF, Fisher-PP, Im–Pesaran–Shin (IPS), and Levin–Lin–Chu (LLC)—were selected to analyze the stationarity. The test results are consistent, and they show that unit root tests for the panel are rejected at the 1% significance levels in the first difference. These results imply that ROE, ROA, and Tobin’s Q data are stationary.

The test results for the null hypothesis that over-identification of the instrumental variables is appropriate are summarized in

Table 4. Results of over-identification and error term autocorrelation tests.

Data	Depen.var	Hansen Test	Arellano–Bond Test
		Null hypothesis: The instruments are valid	Null hypothesis: No autocorrelation of order 1	Null hypothesis: No autocorrelation of order 2
All industries	lnROE	44.86 Prob > chi2 = 0.679	−4.17 Pr > z = 0.000	−1.50 Pr > z = 0.134
	lnROA	170.46 Prob > chi2 = 1.000	−3.98 Pr > z = 0.000	−1.84 Pr > z = 0.065
	lnTobins_q	178.37 Prob > chi2 = 1.000	−3.52 Pr > z = 0.000	−1.79 Pr > z = 0.074
Carbon intensive industries	lnROE	41.11 Prob > chi2 = 0.810	−2.73 Pr > z = 0.006	−0.83 Pr > z = 0.405
	lnROA	41.07 Prob > chi2 = 1.000	−1.48 Pr > z = 0.138	−1.60 Pr > z = 0.110
	lnTobins_q	51.22 Prob > chi2 = 0.995	−3.41 Pr > z = 0.001	0.61 Pr > z = 0.544

. According to Table 4, at a 5% significance level, the null hypothesis is not rejected. This result shows that all instrumental variables are used significantly in the model. Lastly, the test results of the null hypothesis that the error term has no autocorrelation are also reported in Table 4. From the error term autocorrelation test, the null hypothesis that there is no second-order autocorrelation of the error term is not rejected. Therefore, it is concluded that the error term of this model does not have the autocorrelation problem.

4. Empirical Results

4.1. System GMM Analysis—The Whole Industry

Table 5. System GMM estimation results (all industries).

	Depen.var: ∆lnROE	Depen.var: ∆lnROA	Depen.var: ∆lnTobins_q
∆lnFPit−1	0.064 *** (3.03)	−0.015 *** (−7.86)	0.832 *** (171.64)
∆lnCarbon intensityit	−0.105 *** (−5.58)	−0.050 *** (−19.36)	−0.028 *** (−16.30)
∆lnLaborit	−0.097 *** (−3.96)	−0.032 *** (−9.95)	0.003 (1.63)
∆lnCapital intensityit	−0.347 *** (−4.63)	−0.320 *** (−54.23)	−0.003 (−0.68)
∆lnDebt ratioit	−0.611 *** (−23.39)	−0.801 *** (−189.28)	0.053 *** (23.43)
∆lnR&D_profitit	0.472 ** (2.19)	0.052 *** (7.29)	0.001 (0.22)
∆lnR&D_profitit−1	−0.442 ** (−2.07)	−0.025 *** (−3.87)	0.034 *** (6.11)
∆lnR&D_profitit*Phase1	−0.538 ** (−2.17)	−0.020 ** (−2.34)	−0.019 *** (−2.97)
∆lnR&D_profitit*Phase2	−0.617 ** (−2.34)	−0.055 *** (−6.17)	−0.017 ** (−2.5)
∆lnR&D_profitit*Phase3	−0.671 * (−1.93)	−0.081 *** (−6.00)	−0.057 *** (−4.69)
∆lnR&D_profitit−1*Phase1	0.567 ** (2.28)	0.035 *** (3.81)	0.027 *** (3.62)
∆lnR&D_profitit−1*Phase2	0.584 ** (2.23)	0.036 *** (4.13)	−0.020 ** (−2.55)
∆lnR&D_profitit−1*Phase3	0.706 * (1.96)	0.065 *** (4.94)	0.004 (0.31)
∆lnAgeit	−0.241 * (−1.95)	0.008 (0.68)	−0.008 (−0.94)
constant	3.137 *** (6.8)	2.728 *** (46.26)	−0.592 *** (−13.57)
Year effect	Y	Y	Y
Industry effect	Y	Y	Y

***, **, * Significance at 1%, 5%, 10% respectively. Values in () show t-value.

illustrates the empirical results with three different dependent variables to analyze the effect of the R&D and emissions trading market on the financial performance of firms in Korea. Based on the estimation results, it is confirmed that carbon intensity negatively affects a firm’s financial performance. It shows that a 1% increase in carbon intensity decreases ROE, ROA, and Tobin’s Q by approximately 0.105%, 0.05%, and 0.028%, respectively. Carbon intensity, by dividing the carbon emission by total output, represents the environmental performance of a company. The estimation results suggests that the decreasing environmental performance of a firm has a negative effect on financial performance.

In addition, we conclude that R&D positively impacts ROE at the 5% significance level and ROA at 1% significance levels. This outcomes show that a 1% increase in R&D investment increases a firm’s ROE and ROA by 0.472% and 0.052%, respectively. However, the firm’s innovative activities are not statistically significant for Tobin’s Q. Lastly, our results imply that the indirect and additional R&D effect on improving financial performance due to the KETS is negative in the short run in all three models.

Table 6. Elasticity (all industries).

	Short Run			Long Run
	Depen.var: lnROE	Depen.var: lnROA	Depen.var: Tobins_q	Depen.var: lnROE	Depen.var: lnROA	Depen.var: Tobins_q
Direct R&D elasticity	0.472	0.052	0.001	0.032	0.026	0.206
Indirect R&D elasticity (Phase 1)	−0.538	−0.020	−0.019	0.031	0.015	0.049
Indirect R&D elasticity (Phase 2)	−0.617	−0.055	−0.017	−0.035	−0.019	−0.225
Indirect R&D elasticity (Phase 3)	−0.671	−0.081	−0.057	0.037	−0.015	−0.312

summarizes the results of the short- and long-run elasticities of the direct and indirect R&D effects caused by the KETS based on the coefficients in Table 5. In the long run, ∆lnFP_it−1 = ∆lnFP_it, ∆lnR&D_profit_it−1 = ∆lnR&D_profit_it, ∆lnR&D_profit_it−1*Phase_j = ∆lnR&D_profit_it*Phase_j. The long-run direct R&D elasticity and indirect R&D elasticity caused by the carbon market are derived as follows:

$$\mathrm{Long}\text{-}\mathrm{run} \mathrm{direct} \mathrm{R}\&\mathrm{D} \mathrm{elasticity}={\displaystyle \frac{(\mathsf{\beta }6+\mathsf{\beta }7)}{(1-\mathsf{\beta }1)}},$$

(3)

$$\mathrm{Long}\text{-}\mathrm{run} \mathrm{indirect} \mathrm{R}\&\mathrm{D} \mathrm{elasticity} \mathrm{caused} \mathrm{by} \mathrm{KETS} \mathrm{Phase} 1={\displaystyle \frac{\left(\mathsf{\beta }8+\mathsf{\beta }11\right)}{\left(1-\mathsf{\beta }1\right)}},$$

(4)

$$\mathrm{Long}\text{-}\mathrm{run} \mathrm{indirect} \mathrm{R}\&\mathrm{D} \mathrm{elasticity} \mathrm{caused} \mathrm{by} \mathrm{KETS} \mathrm{Phase} 2={\displaystyle \frac{(\mathsf{\beta }9+\mathsf{\beta }12)}{(1-\mathsf{\beta }1)}},$$

(5)

$$\mathrm{Long}\text{-}\mathrm{run} \mathrm{indirect} \mathrm{R}\&\mathrm{D} \mathrm{elasticity} \mathrm{caused} \mathrm{by} \mathrm{KETS} \mathrm{Phase} 3={\displaystyle \frac{(\mathsf{\beta }10+\mathsf{\beta }13)}{(1-\mathsf{\beta }1)}}.$$

(6)

Based on Table 6, direct R&D elasticity is positive in both the short and long run and is more elastic in the long run than in the short run. On the other hand, indirect R&D elasticity is negative in the short run but positive in the long run only in Phase 1. Theoretically, a well-designed carbon market can encourage market participants to actively develop carbon emission reduction technologies, allowing firms to effectively reduce carbon costs in the long run. Based on the results of the indirect R&D long-run elasticity in the KETS Phases 2 and 3, it is concluded that the KETS does not induce technological innovation of market participants, and the dynamic efficiency of this emissions market is low. The KETS has several limitations. First, recent carbon market prices are low. The average carbon emission price from 2021 to 2023 is 8.6 dollars, which is lower than that of other major countries’ emissions market prices. Therefore, market participants choose to purchase emission permits from the market rather than to reduce their carbon emissions by investing in carbon reduction facilities and technologies. Second, the proportion of free allocation during the third phase is approximately 90%, and industries subject to free allocation are designated based on their carbon intensity. In the case of the EU ETS, the proportion of free allocation varies depending on the industry, but it is 0–50%, showing a lower ratio of free allocation compared to the KETS. Third, the high price volatility of the emissions market hinders companies in their efforts to establish long-term roadmaps to develop carbon mitigation technologies. Owing to these problems, the price function of the carbon emissions market is not working properly, and the efficiency of the market is low. Therefore, the incentive for technological innovation in the market has decreased. These market problems reflect the long-run ETS elasticity results in Table 6.

4.2. System GMM Estimation for Carbon-Intensive Industries

The impact of technological innovation on a firm’s financial performance may vary, and elasticity can also differ by industry because firms have different characteristics by industry. The effect of the carbon market on carbon-intensive industries may differ from that on general industries. In the case of carbon-intensive industries, since carbon and energy use reduction have a significant effect on total cost reduction, the effect and magnitude of the KETS on a firm’s financial performance may differ from that of general industries.

Thus, the main objective of this study is to regroup the data for carbon-intensive industries, to examine the impact of technology innovation on financial performance by varying industries, and to suggest policy implications for enhancing the efficiency of the KETS. First, we classify carbon-intensive industries, including energy, steel, semiconductors, and petrochemicals, into two major perspectives: (1) produce large amounts of carbon emissions; (2) consume large amounts of energy.

Table 7. System GMM estimation results (carbon-intensive industries).

	Depen.var: lnROE	Depen.var: lnROA	Depen.var: Tobins_q
lnFPit−1	0.022 * (1.95)	−0.020 * (−1.9)	0.897 *** (47.52)
lnCarbon intensityit	−0.051 *** (−3.73)	−0.099 *** (−5.61)	−0.029 *** (−5.35)
lnLaborit	−0.054 ** (−2.54)	−0.356 *** (−14.79)	−0.017 ** (−2.61)
lnCapital intensityit	−0.461 *** (−8.36)	0.074 *** (3.46)	0.001 (0.13)
lnDebt ratioit	−0.708 * (−37.83)	−0.826 *** (−52.64)	0.051 *** (6.44)
lnR&D_profitit	−0.145 * (−1.71)	−0.028 (−1.26)	−0.040 *** (−3.88)
lnR&D_profitit−1	0.191 ** (2.36)	0.008 (0.47)	0.068 *** (6.55)
lnR&D_profitit*Phase1	0.250** (2.2)	0.069 *** (2.95)	0.023 (1.07)
lnR&D_profitit*Phase2	0.282 *** (2.95)	0.155 *** (4.15)	−0.147 *** (−4.64)
lnR&D_profitit*Phase3	0.137 (1.56)	0.026 (0.99)	0.018 (0.94)
lnR&D_profitit−1*Phase1	−0.224 * (−1.92)	−0.056 ** (−2.62)	0.027 (1.25)
lnR&D_profitit−1*Phase2	−0.296 *** (−2.89)	−0.129 *** (−3.96)	0.162 *** (5.18)
lnR&D_profitit−1*Phase3	−0.136 (−1.44)	0.010 (0.39)	−0.095 *** (−5.51)
lnAgeit	−0.170 *** (−3.06)	−0.075 * (−1.8)	−0.083 *** (−4.96)
constant	3.666 *** (21.04)	2.145 *** (5.58)	−0.009 (−0.07)
Year effect	Y	Y	Y
Industry effect	Y	Y	Y

***, **, * Significance at 1%, 5%, 10% respectively. Values in () show t-value.

summarizes the estimation results for carbon-intensive industries to analyze the effects of R&D and emission trading markets on the financial performance of firms in Korea. From Table 7, it is concluded that R&D has a negative impact on ROE and Tobin’s Q at the 10% and 1% significance levels, respectively. This is contrary to the results of the analysis based on all industry data. Among the carbon-intensive industries, energy, steel, and petrochemical industries are classified as process industries. Process industries require a change in the entire production process if they introduce new carbon reduction technologies. Therefore, the R&D investment costs are initially high. The results confirm that the R&D investment of these firms has a negative effect on the firm’s ROE and Tobin’s Q of firms in the short run.

In addition, it is concluded that the indirect and additional innovative activity effects on a firm’s financial performance are mostly positive in the short run among all three models in Phases 1 and 2. Owing to the introduction of Phase 1, the effect of R&D on improving a firm’s ROE and ROA increased by 0.25% and 0.069%, respectively. The effects of the implementation of Phase 2 are estimated to be 0.282% and 0.155%, respectively, which is 0.032% and 0.086% higher than that of those of Phase 1. This also means that the estimation results are contrary to those of the analysis using all industry data. This result shows that the KETS forces carbon-intensive industries to incentivize technological innovation in the short run.

Table 8. Elasticity (carbon-intensive industries).

	Short Run			Long Run
	Depen.var: lnROE	Depen.var: lnROA	Depen.var: Tobins_q	Depen.var: lnROE	Depen.var: lnROA	Depen.var: Tobins_q
Direct R&D elasticity	−0.145	−0.028	−0.040	0.047	−0.020	0.271
Indirect R&D elasticity (Phase 1)	0.250	0.069	0.013	0.026	0.023	0.481
Indirect R&D elasticity (Phase 2)	0.282	0.155	0.026	−0.015	−0.147	0.143
Indirect R&D elasticity (Phase 3)	0.137	0.026	0.036	0.001	0.018	−0.751

summarizes the short- and long-run elasticity results for carbon-intensive industries. Direct R&D elasticity is negative in the short run but mostly positive in the long run. Indirect R&D elasticity is mostly positive in the short and long run in all phases. This means that the carbon emissions market incentivizes carbon reduction technology innovation well; and, as a result, R&D has a statistically significant effect on improving the financial performance of carbon-intensive industries.

5. Conclusions and Policy Implication

This study analyzes the impact of environmental regulations on the financial performances of firms joining the KETS based on firm-level data. The system GMM analysis is applied to examine factors influencing the ROE, ROA, and Tobin’s Q and to find statistically meaningful variables. We also investigate the impact of innovation caused by the carbon emissions market on financial performance by industry and suggest policies for enhancing the efficiency of the carbon emissions market in Korea.

Our findings indicate that the KETS cannot induce technological innovation to increase financial performance based on all manufacturing industry data. Indirect R&D elasticity is negative in the short run but positive in the long run only in Phase 1. These results confirm that the KETS is not an efficient or well-designed market because it does not induce technical innovation in participants. To enhance the dynamic efficiency of the emissions market, it needs to be redesigned such that firms can participate actively in market transactions and invest in carbon reduction technologies by lowering the proportion of free allocation and price volatility in the carbon emissions market and increasing the value of the market price.

The empirical results for the carbon-intensive industries indicate that the impact of R&D caused by the carbon emissions market on financial performance varies across industries. Unlike the estimation results based on all industries, direct R&D elasticity is negative in the short run but mostly positive in the long run. These results reflect the characteristics of the process industry, which has high initial R&D investment costs. Indirect R&D elasticities caused by the KETS are mostly positive in the short and long run in all phases. This finding confirms that the carbon emissions market has a statistically significant effect on the financial performance of carbon-intensive industries. Therefore, we conclude that the KETS induces mid- to long-run technological innovation in carbon-intensive firms, which can help them increase their financial performance by lowering carbon reduction costs. We suggest that government policies for R&D funding and low-interest loans are needed for carbon-intensive firms that face difficulties in changing their production processes because of the huge initial costs of introducing carbon reduction technology development.

There are a few limitations of our study which could be addressed in further research. This study does not clearly identify the various paths through which environmental performance improves profitability. There are several ways to improve a firm’s financial performance through environmental performance, including acquiring an eco-friendly brand image, generating profits through the sale of carbon permits, and providing incentives to decrease carbon reduction costs [17]. As this study only considers the cost reduction incentives caused by the carbon emissions market, further empirical analysis is required to identify various other paths. Further, this study only focused on carbon-intensive industries. Since the factors and paths concerning the impact of the carbon emissions market on improving financial performance differ by industry [18], it is necessary to research several other industries by reflecting industry-specific characteristics. In this study, only carbon emissions were used as a dependent variable representing firms’ undesirable goods. In future research, based on the approach [19], it will be necessary to analyze other emissions to capture the relationship between innovation activity and the effect of environmental policy if the data become available.