*Article* **Medium- and High-Tech Export and Renewable Energy Consumption: Non-Linear Evidence from the ASEAN Countries**

**Cong Khai Dinh <sup>1</sup> , Quang Thanh Ngo <sup>1</sup> and Trung Thanh Nguyen 1,2,\***


**Abstract:** Sustaining economic growth while reducing dependence on fossil fuels remains a challenge for our world to fight against climate change and therefore finding a way to promote economic growth and increase renewable energy use is needed. This paper uses a 22-year panel dataset (1994– 2015) of 9 countries in the Association of Southeast Asian Nations provided by the World Bank World Development Indicators to examine the impact of medium- and high-tech export on renewable energy use. We employ a fixed-effects regression model with the Driscoll–Kraay nonparametric covariance matrix estimator to account for sectoral and temporal dependence. We also control for inflation, employment, population growth, and gross domestic product per capita in our estimations. Our results demonstrate a U-shaped association between medium- and high-tech export and renewable energy consumption of these economies. The results propose that enhancing medium- and high-tech export could be a feasible solution for promoting renewable energy consumption.

**Keywords:** renewable energy; medium- and high-tech export; economic growth; employment; inflation; ASEAN

#### **1. Introduction**

Sustaining economic growth while reducing dependence on fossil fuels remains a challenge in our era of climate change. In addition to the need for reducing emissions, continuingly increasing fossil fuel prices, fears of unaffordable and rapidly depleting sources of fossil fuels, and the desire to transitioning into a low carbon economy have combined to heighten the importance of renewable energy use [1]. Several countries have set a target of specific renewable energy share in their total energy consumption. For example, Germany aims to supply electricity solely from renewable energy sources by 2045. China also pledges to be carbon neutral by 2060 and sets the share of non-fossil fuels in primary energy consumption to around 25% by 2030 from a previous commitment of 20%. However, increasing renewable energy consumption is not easy for many developing countries, especially for rapidly growing economies as their demand for energy is increasing and their technical and financial capacities for a large-scale supply of renewable energy are limited. In this regard, looking for possible ways to increase renewable energy consumption while maintaining economic growth is required.

Industrialization generally results in a structural transformation from fossil fuelbased and low technology to clean energy-based and medium- and high technology. Medium- and high-tech industries are the value-added manufacturing sectors with higher technological intensity and productivity. They are referred to the level of technology that companies and industries producing goods with innovative qualifications and advanced technologies [2,3]. High technology industries include, for example, aviation and spacecraft industry, pharmaceutical industry, accounting and information processing technologies,

**Citation:** Dinh, C.K.; Ngo, Q.T.; Nguyen, T.T. Medium- and High-Tech Export and Renewable Energy Consumption: Non-Linear Evidence from the ASEAN Countries. *Energies* **2021**, *14*, 4419. https:// doi.org/10.3390/en14154419

Academic Editor: Surender Reddy Salkuti

Received: 11 June 2021 Accepted: 19 July 2021 Published: 22 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

radio, television and communication equipment industry, and medical and optical devices industry [4]. Thus, the production and export of medium- and high-tech products are an important source of export-oriented growth and development, and of the transition to a low-carbon economy [5–7].

There have been several studies on socio-economic factors affecting renewable energy consumption [8–15]. However, new drivers such as medium- and high-tech export have been given much less attention, especially in the context of developing countries [16]. In addition, existing research mainly explores the competitiveness benefit of the policies on renewable energy use in conventional industrial sectors, such as in iron, steel, paper, and glass industries [17]. Globalization has facilitated trade among countries, and the export of medium- and high-tech products have been promoted in rapidly growing economies. However, so far, the causal effects of trade in general, on renewable energy use in both short- and long- terms are weak and scattered [16,18,19], and the effect of medium- and high-tech export on renewable energy use has not been investigated, especially in the context of rapidly growing economies. This is the main motivation for our study.

Theoretically, high- and medium-tech export affect renewable energy use through three channels. First, higher export of high- and medium-tech goods stimulate domestic production of these exported goods and hence economic growth. Increases in domestic production of these goods and economic growth change the energy demand as energy is a key input for production. This is referred to as the scale effect. Second, trade openness allows countries to exchange energy-saving and cleaner energy technologies, which are exported by developed economies and imported by developing economies [20–23]. Such exchange facilitates technological advancement. This is referred to as the technique effect. Third, economic growth leads to economic structural transformation which means that at the beginning of the transformation when the economy is largely agricultural-based, energy intensity is low. However, at a later stage when industrialization starts, energy intensity increases. At the same time, economic growth makes people better-off and increases their awareness of the environment, thus demand more medium and high-tech goods from environmentally friendly producers. This is referred to as the composite effect. The net effect depends on the stage of economic growth and the changes in consumption patterns of consumers. While developed economies have advantages in improving technologies for promoting renewable energy, developing countries are less able to do so and most of them may rely on technology transfers from developed countries, but for various reasons, technology transfers might be constrained. Therefore, at an earlier stage medium- and high-export from developing countries may still demand more fossil-based resources. Later on, renewable energy use will increase.

We focus our analysis on the nine countries of the Association of Southeast Asian Nations (ASEAN), including Cambodia, Indonesia, Laos, Malaysia, Myanmar, the Philippines, Singapore, Thailand, and Vietnam (hereafter referred to as ASEAN-9 countries). The ASEAN has 10 member countries but we purposely exclude Brunei as its energy consumption solely depends on fossil fuels. Even though these countries differ in several aspects, especially in terms of income per capita (Figure 1), they are commonly known as rapidly growing economies and thus have an increased energy demand. It is projected that the overall primary energy supply will increase from 621 Mtoe in 2015 to 1.544 Mtoe in 2050, an annual increase of 2.6%; and the gross final energy consumption will increase with an annual rate of 2.4%, from 436 Mtoe in 2015 to 1006 Mtoe in 2050. However, ASEAN remains highly reliant on fossil fuels. Nearly 80% of the global primary energy supply by 2050 are projected to adhere to fossil fuels. The heavy reliance on fossil fuels along with the decreasing domestic fossil fuel stocks would force the ASEAN's member states to import more fossil fuels. ASEAN is currently the 3rd largest economy in the Indo-Pacific and the 5th largest economy in the world. It has a combined gross domestic product (GDP) of \$US 2.8 trillion, and is also the 3rd fastest-growing major Indo-Pacific economy in the past decade, after China and India (Figure 2). As a critical hub for global trade, over \$3.4 trillion in global trade transits through the ASEAN region each year [24]. Their export of

medium- and high-tech has also been growing. However, the share of renewable energy in the total energy consumption of these countries is still modest. In this regard, examining the effect of medium- and high-tech export on renewable energy in ASEAN is of particular interest and relevant for policymakers and the public. We use a 22-year panel dataset (1994–2015) of these ASEAN-9 countries provided by the World Bank World Development Indicators to empirically examine the impact of medium- and high-tech export on the share of renewable energy use in total energy consumption of these countries. These countries are diverse in many aspects. For example, Singapore is an advanced economy, Indonesia and Thailand are upper-middle-income countries, Vietnam is a lower-middle-income country, and Cambodia and Laos belong to the group of the least developed countries. Following the arguments in the previous paragraph, we hypothesize that the relationship between medium- and high-tech export and renewable energy consumption in these ASEAN-9 countries is U-shaped. over \$3.4 trillion in global trade transits through the ASEAN region each year [24]. Their export of medium- and high-tech has also been growing. However, the share of renewable energy in the total energy consumption of these countries is still modest. In this regard, examining the effect of medium- and high-tech export on renewable energy in ASEAN is of particular interest and relevant for policymakers and the public. We use a 22-year panel dataset (1994–2015) of these ASEAN-9 countries provided by the World Bank World Development Indicators to empirically examine the impact of medium- and high-tech export on the share of renewable energy use in total energy consumption of these countries. These countries are diverse in many aspects. For example, Singapore is an advanced economy, Indonesia and Thailand are upper-middle-income countries, Vietnam is a lowermiddle-income country, and Cambodia and Laos belong to the group of the least developed countries. Following the arguments in the previous paragraph, we hypothesize that the relationship between medium- and high-tech export and renewable energy consumption in these ASEAN-9 countries is U-shaped. over \$3.4 trillion in global trade transits through the ASEAN region each year [24]. Their export of medium- and high-tech has also been growing. However, the share of renewable energy in the total energy consumption of these countries is still modest. In this regard, examining the effect of medium- and high-tech export on renewable energy in ASEAN is of particular interest and relevant for policymakers and the public. We use a 22-year panel dataset (1994–2015) of these ASEAN-9 countries provided by the World Bank World Development Indicators to empirically examine the impact of medium- and high-tech export on the share of renewable energy use in total energy consumption of these countries. These countries are diverse in many aspects. For example, Singapore is an advanced economy, Indonesia and Thailand are upper-middle-income countries, Vietnam is a lowermiddle-income country, and Cambodia and Laos belong to the group of the least developed countries. Following the arguments in the previous paragraph, we hypothesize that the relationship between medium- and high-tech export and renewable energy consump-

tion in these ASEAN-9 countries is U-shaped.

**Figure 1.** Per capita GDP of the ASEAN economies as compared to the US in 2017. Source: East-West Center, 2019 [24]. **Figure 1.** Per capita GDP of the ASEAN economies as compared to the US in 2017. Source: East-West Center, 2019 [24]. **Figure 1.** Per capita GDP of the ASEAN economies as compared to the US in 2017. Source: East-West Center, 2019 [24].

**Figure 2.** Real GDP growth of ASEAN as compared to some other economies. Source: East-West Center, 2019 [24]. **Figure 2.** Real GDP growth of ASEAN as compared to some other economies. Source: East-West Center, 2019 [24]. **Figure 2.** Real GDP growth of ASEAN as compared to some other economies. Source: East-West Center, 2019 [24].

#### **2. Literature Review**

As natural resources in general and fossil fuels in particular, have been depleting, renewable energy use and economic growth are highlighted as major concerns for sustainable development of the global economy [25–29], given the fact that the sustainability of the economic system is dependent on the environment and natural resources [30]. Due to an increasing level of globalization, global energy demand has changed over time and has been driven by trade-related factors. Trade openness, export-oriented policies, and internationalization are considered crucial in the longterm development strategy of many developing countries [31,32]. In this context, recent literature has considered new determinants of energy demand and energy intensity such as export or import product portfolio [33,34], trade openness, and technological advancement [35–38]. In general, there is a controversy on the effect of trade on energy intensity. On the one hand, many studies find that trade openness positively affects energy intensity [39–45]. On the other hand, the effect can be adverse [46], or ambiguous [47] depending on competitiveness, factor price, and technology and infrastructure factors ([35,48]). Regarding the spectrum of trade and innovation, Samargandi [35] reveals that trade openness and innovation are significant factors for reducing energy intensity. Beser and Soyyigit [2] indicate that high-tech export has a strong impact on CO<sup>2</sup> emission in developed economies.

The importance of technology in determining energy use and energy efficiency in developing countries is increasing due to a growing level of internationalization and integration [22,49,50]. Technology innovation induced by investment in research and development and by foreign direct investment is supposed to increase energy use efficiency [40]. Domestic innovation is also an important contributor to technical development [49,51]. To mitigate the negative effects of climate change, technological progress is crucial [52], and increased R&D is associated with more technical innovation and renewable energy adoption [53] in both developed and developing countries, where renewable energy sources such as biomass, solar, wind and hydropower are adopted [54,55]. Recent studies on the relationship between technological advancement and carbon emission demonstrate positive effects of technology innovations on carbon emission reduction [56–58]). Liu, Xia, Tao, and Chen [56], for instance, analyze carbon emission in China and find that increasing technological expenditure could in turn increases carbon emission efficiency. Wang, Zhao, Wang, Guo, Kan, and Yuan [57] discover that investments in technology decrease carbon emissions. Zeng, Lu, Liu, Zhou, and Hu [58] find that foreign trade, foreign capital, and technological progress have positive effects on carbon emission reduction.

Several studies examine the drivers of renewable energy use. Most of these indicate the causal interactions between economic growth and renewable energy use [59–62], between renewable energy use and sustainable development [63–65], between energy use and trade openness, and between technological progress and renewable energy use. For example, Apergis and Payne [61] investigate the relationship between economic growth and renewable energy use of 20 OECD countries during 1985–2005 and find a bidirectional link between economic outcome and energy use. Fang and Chang [62] analyze the causality between energy use and economic performance in 16 countries of the Asia Pacific and finds a long-run cointegrating relationship. Kahia et al. [66] examine the relationship between energy use and economic growth, using a sample of 7 MENA Net Oil Importing Countries (NOICs) during 1980–2012; and their empirical results confirm the bidirectional causality between renewable energy use (and non-renewable energy use) and economic growth. Le and Sarkodie [67] investigate the nexus between renewable and conventional energy and economic growth, using panel data of 45 Emerging Market and Developing Economies (EMDEs) from 1990 to 2014. They find that renewable energy and GDP growth impact each other. Marques and Fuinhas [68], using a sample of 24 European countries during 1990–2006, find that the current level of renewable energy use is positively dependent on the previous level of renewable energy use. However, income and prices of fossilbased fuels are not significant for the development of renewable energy during the studied period. Ahmed et al. [69] investigate the interactions between renewable and non-renewable energies, CO<sup>2</sup> intensity, and economic growth in Myanmar during 1990– 2016; and their results reveal that renewable energy use significantly promotes economic growth.

Few studies have also examined the relationship between trade openness, technological innovation, and (renewable) energy use (Alam and Murad [16], Sohag, Begum, Abdullah, and Jaafar [22], Sbia et al. [70,71], Khan et al. [72], Cole [47], Shahbaz, Nasreen, Ling, and Sbia [21]). Alam and Murad [16] reveal that economic growth, trade openness, and technological progress significantly influence renewable energy use in 25 OECD countries. Sohag, Begum, Abdullah, and Jaafar [22] employ a dataset during 1985–2012 in Malaysia and find that while economic growth and trade openness are the main determinants of energy use, technological innovation reduces energy use in manufacturing sectors. Sbia, Shahbaz, and Hamdi [70] examine the impacts of foreign direct investment, trade openness, clean energy price, carbon emissions, and economic growth on the demand for energy in the United Arab Emirates. Their findings indicate that trade openness and foreign direct investment reduce energy use because energy-efficient technologies have been employed. By comparing upper-middle-income countries in Asia, Europe, Africa, and America, Khan, Yaseen, and Ali [72] indicate that trade can induce technology transfer for renewable energy. Shahbaz, Nasreen, Ling, and Sbia [21] use the data from 91 high, middle, and low-income economies and conclude that domestic energy use is affected by trade openness through several channels such as technological transfers, economies of scale, and input factors. In high-income economies, an inverted U-shaped relationship between trade openness and energy consumption is found. According to Shahbaz, Nasreen, Ling, and Sbia [21], the U-shaped relationship between trade openness and energy consumption exists when low and middle-income countries import or use energy-efficient technologies from developed countries to lower energy consumption, on the one hand; and on the other hand, when developed countries allow to release those technologies and share profits for low and middle-income countries that have limited access to technology and capital.

A most recent study that is close to our work in terms of geographical coverage (for ASEAN with a 22-year span of time) is Azam, Khan, Zaman, and Ahmad [25], who find that trade openness has a positive relationship with energy consumption in Thailand, Malaysia, and Indonesia. Apart from trade openness, they discover that population growth increases the energy consumption in Malaysia, while it decreases energy consumption in Indonesia. Real GDP is found to have a positive relationship with energy consumption in Thailand, Malaysia, and Indonesia.

The causal effects between international trade (exports or imports) and renewable energy use in both short- and long- terms that have been found so far are weak [16]. The results from Sadorsky [20] for a sample of Middle Eastern countries show that international trade increases domestic use of energy. In addition, Shahbaz, Nasreen, Ling, and Sbia [21] conclude that a U-shaped relationship exists for high-income countries, whereas an inverted U-shaped relationship is found for middle- and low-income countries for the relationship between international trade and energy use.

In sum, none of the previous studies examine the impact of medium-and high-tech export on renewable energy use. Our study is thus aimed to contribute to filling this gap. The contribution of our study to the literature are two-fold. First, our study is the first effort to examine the effects of medium- and high-tech export on renewable energy share in total energy consumption for this group of rapidly expanding economies. To our best knowledge, only Shahbaz, Nasreen, Ling, and Sbia [21] discover non-linear relationships between trade openness and energy consumption for two country groups, high-income countries and middle- and low-income countries. Second, from a methodological perspective, we use panel data and employ a fixed-effects regression model with the Driscoll–Kraay nonparametric covariance matrix estimator to account for unobservable time-invariant

factors and sectoral and temporal dependence. These concerns have not been successfully addressed in many previous studies (Azam et al. [25]).

#### **3. Data and Methods**

#### *3.1. Data Source*

We extract the data needed for our study from the World Bank World Development Indicators (WDI). This database has been established for years in many countries. For the ASEAN countries, the data are available from 1994 to 2015. This allows us to establish a balanced panel dataset for these nine ASEAN countries. As explained above, we exclude Brunei from the sample since it is an outlier in terms of renewable energy use. Our variables of interest are medium- and high-tech export, and renewable energy use of these countries over time as we would like to investigate the association between these two important variables. From the literature review presented in the previous section, it is clear that a direct relationship between them can be established.

We use the share of renewable energy in the total energy consumption of each country in each year (% of total final energy consumption) to represent renewable energy use. It is the explained variable. For explanatory variables, the share of medium- and high-tech export in total manufactured export (% of manufactured export) is our key variable of interest. In addition, we control for inflation, employment, population growth, and GDP per capita. Inflation is measured as the change in the consumer price index (%, year 2010 is the base year, 2010 = 100), employment is measured as the share of employment in the industrial sectors in total employment of an economy in each year (% of total employment) while population growth is also reported annually for each country. The GDP per capita is measured in purchasing power parity (PPP, constant 2017). These variables are coded as follows: *REU* for renewable energy share (%), *MHTE* for medium- and high-tech export share in total manufactured export (%), *INF* for inflation (%), *POPG* for population growth (%), *EMP* for the employment share in industry (%), and *GDPPC* for GDP per capita (PPP, constant 2017). All these variables are annual for these ASEAN-9 countries from 1994 to 2015. A more detailed definition of these variables is in Appendix A. The description of the data for each country is presented in Table 1 whereas Table 2 summarizes the data for the whole block. These tables show the *REU* mean value is 42.673 (%), the mean value of *MHTE* is 37.168 (%), while the mean value of *INF* is 77.034 (%). On average, *POPG* is 1.575%, *EMP* is 17.915 (%), and *GDPPC* is 13,417 (USD PPP 2007). The correlation coefficients between these variables are in Appendix B which show that *NHTE*, *INF*, *POPG*, *EMP*, and *GDPPC* all have a negative association with *REU* at a 1% level of significance.

**Table 1.** Descriptive statistics of variables of each ASEAN country (1994–2015).


Source: Authors' estimation based on WDI database; SD: Standard deviations.


**Table 2.** Descriptive statistics of variables of nine ASEAN countries (1994–2015).

Source: Authors' estimation based on WDI database; SD: Standard deviations; No. observations: 198; No. of countries: 9; No. of years: 22.

#### *3.2. Methods*

We employ an econometric analysis to examine the effect of the explanatory variable (medium- and high-tech export) on the explained variable (renewable energy share). Following previous studies, we control for GDP per capita, inflation, employment, and population growth as these factors have been found to influence energy use. They are the factors affecting energy demand. GPD per capita, inflation, and employment are key drivers of changes in purchasing power parity; and for population, Samargandi [35] argues that population growth positively influences energy usage and energy intensity, which might be harmful to the environment. The employment share of the industry might have either a positive or a negative on renewable energy use [73–75]. For the effects of these factors on energy use, see [7,16,20–22,35] for GDP, [73,74] for inflation, [75] for employment, and [35,76] for population growth. A factor affecting renewable energy use is its price. The declining price of renewable energy driven by technological progress could be important in increasing the renewable energy share in total energy consumption. Unfortunately, we cannot collect sufficient data. Oil price could be another significant factor. However, as some studies find the least influence of oil price on renewable energy production [7,8,11,20], or insignificant [77], especially in the case of oil-exporting countries like these ASEAN-9 countries, and/or in the case that these countries have subsidized oil prices to avoid any adverse effect of oil price fluctuations on the economy [11]. In addition, incorporating oil prices in our analysis could lead to an endogenous issue that must be addressed. Therefore, we excluded prices factors in our analysis.

Econometrically, the causal relationship could be examined using either a pooled ordinary least square (OLS) technique or the panel data method, including either a fixedeffects model (FEM) or a random-effects model (REM) [78]. To choose either the OLS technique or the panel data method, we conducted an F-test for the joint significance of differing group means. Results of the F-test presented in Appendix C (F (8, 184) = 76.97 with *p*-value 0.0000) indicate the null hypothesis that the pooled OLS model is appropriate is rejected. Thus, panel data analysis was chosen. We advanced further to choose either FEM or REM by performing a Hausman test. Results of this test in Appendix C indicate the FEM is a more suitable specification. From a theoretical point of view, FEM has the advantage of controlling for time-invariant unobservable factors. An alternative test to choose either FEM or REM was the overidentifying restriction test [79,80] which also indicates that the FEM model is a more suitable specification (results of this test in the last two rows in Appendix C). In addition, FEM is also recommended to estimate the parameters for a small cross-sectional sample [81] which is our case as we have only 9 countries. Our specification is also in line with previous studies on factors affecting renewable energy use such as Bamati and Raoofi [7] for 25 developed and developing countries, Alam and Murad [16] for 25 OECD countries, Azam, Khan, Zaman, and Ahmad [25] for three ASEAN countries (namely Indonesia, Malaysia, and Thailand); Kahia, Aïssa, and Lanouar [66] for 7 MENA Net Oil Importing Countries; Marques and Fuinhas [68] for 24 European countries; Sadorsky [20] for eight Middle Eastern countries; Bashir, Sheng, Do˘gan, Sarwar, and Shahzad [31] for 29 OECD countries; Beser and Soyyigit [2] for G20 countries (except Russia), Chen, Du, Huang, and Huang [5] for 34 industrial sectors in China, and Waheed et al. [82] for 6 Gulf Cooperation and Council countries.

Therefore, our econometric regression model is specified as follows (Equation (1)):

$$REIL\_{\rm il} = \mathfrak{a}\_0 + \mathfrak{f}\_1 MHTE\_{\rm il} + \mathfrak{f}\_2 INF\_{\rm il} + \mathfrak{f}\_3 POPG\_{\rm il} + \mathfrak{f}\_4 EMP\_{\rm il} + \mathfrak{f}\_5 PGDPC\_{\rm il} + \mathfrak{v}\_{\rm i} + \mathfrak{e}\_{\rm il} \tag{1}$$

where *REU* is the share of renewable energy in total energy consumption; *MHTE* is the share of medium- and high-tech export in total export; *INF* is the inflation rate; *POPG* is the population growth rate; *EMP* is the employment share of industry (%), and *GDPPC* is GDP per capita (PPP, constant 2017) as defined in Section 3.1 (see Tables 1 and 2). Subscripts *i* and *t* denote country and time, respectively; *α* is the fixed country effect while *v<sup>i</sup>* is the country-specific effect, and *eit* is the error term.

Several tests were undertaken to ensure the validity of our regression model. First, to control for possible multicollinearity between explanatory variables, the variance inflation factor (VIF) values were checked [83,84] and the results documented in Appendix D do not signal that problem. Second, as our sample is small in terms of both the number of countries (9 countries) and the number of time periods (22 years), we undertook the following tests: (i) Modified Wald test for group-wise heteroskedasticity, (ii) the Breusch– Pagan LM test for cross-sectional dependence [85,86]; (iii) the slope homogeneity test introduced by Swamy [87], and (iv) the stationarity test for each variable.

Cross-sectional heterogeneity should be controlled for when conducting a panel data empirical analysis [88]. Swamy [87] proposes the homoskedasticity assumption test for the slope homogeneity assumption. Results of this test presented in Appendix E show that we can reject the null hypothesis of the slope homogeneity for our sample. In addition, results of the Breusch–Pagan LM test of independence (also in Appendix E) indicate that the null hypothesis of no cross-sectional independence is rejected at the 1% level of significance, indicating strong cross-sectional dependence.

Once there is cross-sectional dependence across countries in the panel, it is needed to perform the cross-sectionally augmented Dickey-Fuller (CADF) procedure from Pesaran [89]. Results presented in Appendix F show that we were not able to reject the null unit root hypothesis for the *GDPPC* series; but when taking the first difference, the null hypothesis of the unit root is rejected for variables POPG and INF. However, this is not sufficient for us to conclude that there is a long-run equilibrium relationship among the concerned variables, namely *REU*, *MHTE*, *INF*, *EMP*, *POPG*, and *GDPPC*.

Given the presence of cross-sectional dependence and/or heteroscedasticity, we adopted our FEM with Driscoll–Kraay standard errors. According to Hoechle [90], the Driscoll–Kraay standard errors are heteroskedasticity-, and autocorrelation-, consistent and robust to general forms of cross-sectional and temporal dependence.

#### **4. Findings and Discussion**

#### *4.1. Findings*

Table 3 presents the results of our estimation, including both a normal fixed effects specification and the fixed effects specification with the Driscoll–Kraay standard errors. The R squared value of 0.718 from these two specifications indicates that shows our model can predict about 72% of the variation in the share of renewable energy in total energy consumption.

Regarding the effect of medium- and high-tech export, Table 3 shows that the mediumand high-tech export has a significant U-shaped relationship with the share of renewable energy use of these ASEAN-9 countries. This U-shaped relationship implies that at the beginning stage of economic development, a higher level of medium- and high-tech export would lead to a lower share of non-renewable energy in total energy consumption. However, once the economy has reached a certain level of medium- and high-tech export, then the higher the level of medium- and high-tech export, the higher the share of renewable energy in the total energy consumption of that economy. In our case, the threshold value for

the turning point of the U-shaped relationship is 64.47%. Within these ASEAN-9 countries, Malaysia and Singapore have passed this turning point since 1994, and the Philippines since 1995.

With regards to the controlled variables, inflation, employment in the industry sector, and GDP per capita are significantly and negatively associated with the share of renewable energy in total energy consumption, meanwhile, population growth is significantly and positively associated with the share of renewable energy.


**Table 3.** Impact of medium- and high-tech export on renewable energy share.

Source: Authors' estimation, standard errors in parentheses, \*\*\*, \*\*, and \* denote statistical significance at the 1%, 5%, and 10% levels, respectively.

#### *4.2. Discussion*

With respect to the influences of medium- and high-tech export on renewable energy use, our results indicate that medium- and high-tech export has a U-shaped association with renewable energy use in the ASEAN-9 countries. A couple of previous studies have not found any short- and long-term effects of trade on renewable energy use [8,91]. Our finding is a contribution to a new strand of literature on the non-linear effects of trade on resource use. This strand includes, for example, Shahbaz, Nasreen, Ling, and Sbia [21] who find the pattern of a U-shaped relationship exists in high-income countries, and an inverted U-shaped relationship in the middle- and low-income countries for the relationship between international trade and energy use.

Most previous studies so far have found only a positive effect of high-tech export. It is generally argued that under high-tech export orientation, there is peer-to-peer lending which enables them to adopt such technology that uses lower energy; in addition, different technologies and resources can be shared. Thus, energy consumption is also shared and reduced and it helps save energy costs, including renewable energy [2,7]. Bamati and Raoofi [7] provide an analysis of the influence of high-tech export on renewable energy production by levels of development and find that high-tech export increases renewable energy production for 10 developed countries, but for 15 developing countries the effect is insignificant. A stronger effect of high-tech export on developed countries rather than developing countries is also confirmed by Beser and Soyyigit [2] with a sample of the G20 countries (except Russia).

Our result can be explained by the fact that, in terms of medium- and high-tech export, most of the ASEAN-9 countries do not seem to have high shares of high-tech export or at least technologies that could affect promoting renewable energy use. This means, at a lower level of development, the effect is negative but later on, it turns out to be positive.

Renewable energy use is upon the spending and expectations toward the behaviors of consumers that are affected by different aspects of inflation [92–94]. While depicting the

concern of medium- and high-tech export toward renewable energy use, the controlled factors have a specific influence on renewable energy use in the ASEAN-9 countries. This is widely stated by the variables like GDP per capita and employment. Inflation is also counted as a major factor that influences renewable energy use. Inflation has a dominant impact on economic performance due to the implications of targeting the environment by high inflation [73]. Our results indicate that inflation has a negative relationship with renewable energy use. These results are supported by Kosai, Yuasa, and Yamasue [74]. Although prices of various products are associated with inflation and wide interpretation enhances the role of renewable energy use in the ASEAN-9 countries, a hike in prices of various goods and services induces a significant impact on renewable energy use. Therefore, inflation with the relevance of medium- and high-tech export inserts a negative role toward the renewable energy use of many ASEAN-9 countries [95]. It could be due to the devaluation of the currency that renders renewable energy use with a hike in prices. Inflation could help in uplifting the economies of various countries but could lead to some adverse effects too.

Our regression model shows that population growth has a positive effect on renewable energy use. In fact, the introduction of technology and innovation in renewable energy use has increased the demand of the population for renewable energy. While enumerating the ecological problems with energy-related factors, population growth is the main driver of environmental degradation [96]. Therefore, population growth might result in a negative contribution toward renewable energy use. Our results contribute, to some extent, to this disputable discussion. Recall that Azam, Khan, Zaman, and Ahmad [25] when examining the impact of various factors on energy consumption in three ASEAN countries (Indonesia, Malaysia, and Thailand) in the period 1980–2012, find that population growth increases the energy consumption in Malaysia, while it decreases energy consumption in Indonesia.

The advantage of renewable energy use is indisputable. It is upon the governments of these ASEAN-9 countries to induce needed measures to promote energy use. Potential impacts of employment on renewable energy use in various industries are evident with local renewable resources [97]. Many ASEAN-9 countries have contributed to a significant rise in the employment rate due to its influence on renewable energy use. For the regional development policy of renewable energy use, the employment regimes and challenges insert an important role [98]. The improper sharing of the economy has been eliminated by the positive enclosure of employment elements that widely induce technological innovation in ASEAN-9 countries.

The empirical results also show that GDP per capita has a negative but insignificant effect on renewable energy use. The result is in line with Marques and Fuinhas [68] who investigate drivers promoting renewable energy in 24 European countries and find that the per capita income (in natural logarithm) is not statistically significant in explaining the use of renewables. A negative effect is found by Waheed, Sarwar, and Mighri [82], and Samargandi [35]. Samargandi [35], for example, examines the impacts of trade openness, technological innovation, and energy price on energy intensity in OPEC countries, in which GDPPC is used as a controlling variable.

#### **5. Conclusions**

Our study investigates the influences of medium- and high-tech export on the renewable energy use of nine ASEAN countries in the period 1994–2015. We control for inflation, population growth, employment, and GDP per capita. Our findings suggest that mediumand high-tech export reduces the share of renewable energy in total energy consumption during an earlier stage of economic development but then increases the share of renewable energy consumption during the later stage of economic development. This seems to be a characterized feature observed in these ASEAN-9 countries, contributing to the complexity of trade-renewable energy nexus in the literature. Our study also elaborates that under high inflation, individuals and firms cannot afford costly high technology, effective techniques, and skilled human resources, which consume a smaller amount of renewable energy in

production. Growth in the population provides human resources and at the same time promotes renewable energy use. High employment opportunities indicate high economic growth, which can help to save energy.

Our study has some limitations despite its theoretical and empirical importance. First, due to the availability of the data provided by the World Bank in the World Development indicator, we are not able to control for other factors such as oil prices and/or price of renewable energy that might have significant effects on renewable energy use. Therefore, our results should be interpreted with care. Second, our study is at the macro level and thus not on the behavior of individual energy users (i.e., firms and businesses). The changes in the behavior of energy users should be examined as well to provide a better understanding of energy transition. Third, our sample is small with only nine ASEAN countries in a short-term period. Expanding the study to cover more countries in a longer time period would provide a more comprehensive picture of the effect of medium- and high-tech export on renewable energy use. Last, we are unable to undertake a measurement uncertainty analysis [99]. These issues should be themes for future studies.

**Author Contributions:** Conceptualization, T.T.N., and Q.T.N.; methodology, T.T.N.; data curation, Q.T.N.; writing—original draft preparation, T.T.N., Q.T.N., C.K.D.; writing—review and editing, T.T.N., Q.T.N., C.K.D.; project administration, T.T.N.; funding acquisition, Q.T.N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research is funded by the University of Economics Ho Chi Minh City, Vietnam.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.


**Appendix A. Definition of Variables Used in the Analysis**


**Appendix B. Correlation Matrix of the Variables Used in the Analysis**

\*\*\*, \*\*, and \*: Significant at the 1%, 5% and 10% level, respectively. Source: Authors' estimation.

#### **Appendix C. F test, Hausman Test, and Test of Overidentifying Restrictions**


#### **Appendix D. Variance Inflation Factor (VIF) Value**


Source: Authors' estimation.

#### **Appendix E. Cross-Sectional Dependence, Heteroskedasticity, and Slope Homogeneity Tests**


Note: The null hypothesis of the cross-sectional dependence test is no cross-sectional dependence. The null hypothesis of the slope homogeneity test is slope homogeneity. The Cross-sectional dependence and slope homogeneity tests are conducted by using respectively 'xttest2' [100] and 'xtrchh2' [101] commands in Stata. Source: Authors' estimation.


**Appendix F. Panel Unit Root Test**

Note: \*\*\* and \* denote statistical significance at the 1% and 10% levels, respectively. The null hypothesis is nonstationarity. The panel unit root test is conducted by using the 'xtcips' command in Stata [102]. Source: Authors' estimation.

#### **References**


**Jamal Mamkhezri 1,\*, Leonard A. Malczynski <sup>2</sup> and Janie M. Chermak <sup>3</sup>**


**Abstract:** State-mandated renewable portfolio standards affect substantial portions of the total U.S. electricity supply. Renewable portfolio standards are environmentally motivated policies, yet they have the potential to greatly impact economy. There is not an agreement in the literature on the impact of renewable portfolio standards policies on regional economies, especially on job creation. By integrating various methodologies including econometrics, geographic information system, and input–output analysis into a unique system dynamics model, this paper estimates the economic and environmental impacts of various renewable portfolio standards scenarios in the state of New Mexico, located in Southwestern U.S. The state is endowed with traditional fossil fuel resources and substantial renewable energy potential. In this work we estimated and compared the economic and environmental tradeoffs at the county level under three renewable portfolio standards: New Mexico's original standard of 20% renewables, the recently adopted 100% renewables standard, and a reduced renewable standard of 10%. The final one would be a return to a more traditional generation profile. We found that while the 20% standard has the highest market-based economic impact on the state as a whole, it is not significantly different from other scenarios. However, when environmental impacts are included, the 100% standard yields the highest value. In addition, while the state level economic impacts across the three scenarios are not significantly different, the county-level impacts are substantial. This is especially important for a state like New Mexico, which has a high reliance on energy for economic development. A higher renewable portfolio standard appears to be an economic tool to stimulate targeted areas' economic growth. These results have policy implications.

**Keywords:** renewable portfolio standards; employment; economic output; water use; greenhouse gases; emissions; social benefits

#### **1. Introduction**

Electric utilities in the United States (U.S.) are integrating more renewable energy (RE) sources in their energy mix. In May 2020, 24.3% of electricity generation in the U.S. came from renewable sources (Energy Information Administration, Form EIA-860M). This is partly a result of policies and regulations aimed at mitigating greenhouse-gas (GHG) emissions through programs such as the Regional Greenhouse Gas Initiative in the northeastern part of the U.S., and through the renewable portfolio standard (RPS) at the state level. While the primary objective of these regulations is to address global warming, there can be potential impacts on the economy at a microlevel (i.e., state and county levels). For rural western states, this has become increasingly important, as they strive to diversify their economies.

Debates are ongoing in the literature as to whether RPS policies have a positive (i.e., job creation, GHG and air pollution reduction), negative, or no impact on an economy and the environment (e.g., [1–5]). The main reason for the divergent findings is the inclusion or

**Citation:** Mamkhezri, J.; Malczynski, L.A.; Chermak, J.M. Assessing the Economic and Environmental Impacts of Alternative Renewable Portfolio Standards: Winners and Losers. *Energies* **2021**, *14*, 3319. https://doi.org/10.3390/en14113319

Academic Editor: George Halkos

Received: 13 May 2021 Accepted: 1 June 2021 Published: 5 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

exclusion of market failure due to environmental amenities in the analyses. For example, NYSERDA [1] assessed New York's RPS impact and found a gain of 24,000 job-years from 2002 to 2037. Divounguy et al. [2] investigated Ohio's 12.5% by 2025 RPS and found that it would result in a loss of more than 134,000 jobs. Upton and Snyder [3] evaluated states with an RPS versus those without, and they found that an RPS standard has no significant impact on increasing RE or reducing GHGs. Zhou and Solomon [4] found that more stringent RPSs result in lowering RE capacity additions, while Carley et al. [5] found the opposite. Further, most of the existing literature focused on either an aggregate scope (e.g., nation-wide) or state-specific assessments and has not considered impacts at lower-level jurisdictions (e.g., county level). Lastly, much of the literature overlooked the fundamental dynamics within the energy sector. The objective of this paper is to contribute to this line of research and assess the economic and environmental impact of renewable energy and the tradeoffs on a regional economy.

In particular, we are interested in answering the question of what are the economic and environmental impacts of varying RPSs on regional economies. This is a rather complex question, and answering it is aided by the use of system simulation [6–11]. Thus, in this work we develop, validate, and utilize a system dynamics (SD) based simulation model that integrates results from various methodologies such as input–output analysis, econometrics, and Geographic Information System (GIS). Combining these methodologies in an innovative approach to analyze the SD model is one key contribution of this paper. We execute our analysis in our case study of New Mexico, a southwestern state in the U.S. with an RPS and high potential for both fossil fuel (traditional) and RE sources. We hypothesize that RPS levels have substantial environmental and economic impacts on regional economies. This paper is an attempt to quantify those impacts.

Our findings suggest a net increase of jobs in rural counties that are most suitable for future RE installations. Depending on the scenario, our model estimated increasing 137–156 thousand cumulative, full-time equivalent jobs, 19 to 24 billion USD (2017\$) cumulative gross economic output, and 12,987 to 13,219 and 974 to 1122 billion liters of cumulative water withdrawal and consumption respectively from 2017 to 2050. These scenarios also resulted in increasing millions of avian mortalities, as well as millions of tonnes of GHG emissions and thousands of tonnes of air pollutants, each of which leads to billions of dollars in climatic and air-quality costs (social costs). Lastly, we found that higher RPS standards lead to greater benefits to the state when externalities and social benefits/costs are taken into account.

#### **2. Background**

The burning of fossil fuel (i.e., coal, natural gas, and oil) is the main source of GHG emissions in the U.S., and this contributes to climate change. Combusting fossil fuels for electricity generation not only emits air pollution but also requires an immense amount of water. There is extensive literature that demonstrates the correlation between air pollution and premature mortality/morbidity [12–18]. Maupin et al. [19] showed that roughly 40% of all of the U.S. freshwater withdrawal was used for thermoelectric power plants in 2010. Policymakers, as a result, are seeking to promote policies that lead to integrating more environmentally friendly generation sources with less externalities.

Electricity generation is moving towards integrating a higher level of RE and a lower level of fossil fuels in the U.S. due to regulatory mandated laws such as the RPS as well as cost competitiveness. Thirty states and the District of Columbia currently have an RPS in place. RPSs mandate that electric utilities source a portion of their generation from RE within a certain timeframe. Although the main goal of an RPS is environmentally oriented—that is, to mitigate GHG emissions and/or save water—these policies have the potential to impact economies. Previous research on the impact of RPSs shows that the policies are capable of yielding positive economic impacts if positive externalities (zero or close to zero water usage, zero emission, etc.) are taken into account [20,21]. Barbose et al. [20] demonstrated that meeting requirements mandated by RPSs led to supporting 200 thousand jobs and a reduction of 59 million tonnes of CO<sup>2</sup> in the U.S. in 2013. Wiser et al. [21] quantified positive externalities of RE and estimated that existing RPS policies lead to improving air quality and reducing climatic damages (258 billion USD), which not only compensates the increase in electric system costs (23 to 194 billion USD) but also exceeds those costs over the period of 2015–2050.

There are a handful of peer-reviewed papers and national laboratory reports that look at the feasibility of providing global energy through RE (e.g., [22–25]). For example, Jacobson and colleagues [22] estimated a portfolio mix that enables the United States to sustain its entire energy needs—including electricity, transportation, heating/cooling, and industry—using renewable energy by 2050. Similarly, Cole et al. [25] assessed different scenarios of achieving various levels of RE in only the power sector by 2050. Further, previous economic impact studies of constructing and operating RE projects suggested that the economic impacts to states are considerable [17,26,27]. Similarly, studies on environmental impact of RE showed significant climate and air quality benefits [28–31]. For instance, Millstein et al. [29] found that solar and wind development resulted in benefits of 30–113 billion USD (2015) and 5–107 billion USD from air quality and climate, respectively, while avoiding 3000–12,700 premature mortalities in 2007–2015. Most of these studies produced state-level or nation-wide job/environmental impact estimates, which in turn meant less understanding of lower-level dynamics such as job/environmental impacts across counties. These studies also did not consider the underlying dynamics within the energy sector.

To address the aforementioned gaps in the literature, we combine various methodologies to develop an SD model. SDs are a derivative of the work developed by Forrester [32], in which he introduced a novel approach to integrate multiloop feedback systems. So long as relationships among variables are known, this approach makes modeling complex systems possible [33]. The SD model of this paper integrates results from input–output analysis, econometrics, and GIS to form a unique framework that provides both the public and policymakers improved information with which to make decisions. The model is developed at a monthly time-step from January 2004 through January 2050. Multiple programs are used to analyze the complex electricity problems common to most utilities. Specifically, Jobs and Economic Development Impact (JEDI) coupled with Impact Analysis for Planning (IMPLAN) are used to calculate job multipliers by energy type and by county; Stata is used to estimate electricity demand; ArcGIS is utilized to estimate the potential of renewable and natural gas electricity generation by county, as well as the optimal location for siting additional power plants; lastly, results from previous models are all embodied in Powersim Studio, which is used to analyze various energy mix scenarios.

The objective of the SD model is to estimate electricity generation and consumption by different fuel sources and various sectors respectively. We provide a roadmap to assess the explicit and implicit impacts of various energy mix scenarios at the state and county level and at different points in time. Explicit impacts may include potential jobs and economic gross output associated with current and potential future electricity generation, and implicit impacts may include positive health effects and social benefits of reducing ambient emissions. We apply this roadmap to our case study of New Mexico.

#### *2.1. Study Area: New Mexico*

New Mexico has considerable potential for both fossil fuel and RE resources. It holds about 3%, 4%, and 5% of the United States' total estimated recoverable coal reserves, proved crude oil, and natural gas respectively and it possesses the second-largest uranium reserves in the nation. Most of the state's natural gas and crude oil are located in the San Juan and Permian Basins in the northwestern and southeastern part of the state, respectively, while coal reserves are mainly located in the San Juan and Raton Basins in the northern part of the state. The vast areas of New Mexico with available geophysiological landmass that receives high wind and sunlight levels are optimal for increasing RE usage. New Mexico ranks third in both solar and wind potential in the U.S. [34].

New Mexico's economy is ranked 46th in the nation. The energy industry, especially oil and natural gas extraction, is a main contributor to New Mexico's economy. The state receives approximately 2 billion and 300 million USD per year in direct (e.g., severance, property taxes, royalty, and rental income) and indirect (sales and income taxes) revenues, respectively, from oil and gas production. Depending on the state of the economy, based on recent state finance facts, revenues from oil and gas can contribute about 40% to New Mexico's general fund tax revenue. Thus, fluctuating oil and gas prices affect New Mexico's economy immensely.

On one hand, the energy industry is responsible for emitting GHG and ambient pollution as well as increased water usage in New Mexico. GHG contributes to climate change, while air pollution causes premature mortality and morbidity, and freshwater has historically been insufficient in New Mexico. On the other hand, RE is becoming more and more cost-competitive compared to fossil fuel technologies. Thus, it makes logical and economic sense for policymakers to promote policies such as an RPS in order to integrate more RE into New Mexico's energy mix.

At the time of analysis, New Mexico's RPS required all large electric utilities to generate 20% of their in-state electricity sales from RE resources by 2020. Although it did not pass, a bill (Senate Bill 312) was introduced to increase New Mexico's RPS previous level to 25% by 2020, 35% by 2025, 50% by 2030, 65% by 2035, and 80% by 2040 in the 53rd legislative session in 2017. A modified version of this bill was reintroduced in January 2019 (House Bill 15) and was passed in the 54th legislative session in March 2019 (Senate Bill 489). In addition to the requirements set by Senate Bill 312, Senate Bill 489 sets a 100% RPS by 2045 that is sourced from zero carbon resources. This makes New Mexico the third state in the U.S. after California and Hawaii and before Washington, New York, Maine, and Virginia to mandate a 100% RPS. Thus, New Mexico's current RPS policy requires 20% in-state electricity sales from RE resources by 2020, 40% by 2025, 50% by 2030, 80% by 2040, and 100% by 2045.

Currently, there are three large electric utilities in New Mexico: the Public Service Company of New Mexico, El Paso Electric, and Xcel Energy, with the first serving the largest customer pool in the state. Further, as New Mexico has considerable potential in both wind and solar energy, the Public Regulation Commission set diversity targets (carve-outs) for different types of RE to create a diversified portfolio. Based on this portfolio, the utilities are required to source at least 30%, 20%, and 3% of their in-state electricity sales from wind, utility-scale photovoltaic solar (PV), and residential photovoltaic solar (RPV), respectively, by 2020 (see Table 1). RPS requires the New Mexico's rural electric distribution cooperatives to generate 10% of their in-state electricity sale from renewable sources. We did not consider a rural cooperatives constraint in our analysis.



#### *2.2. Scenario Construction*

Our analysis investigated the number of jobs and their locations by energy source, as well as environmental impact based on thirty-four prices, three technological-costs, and three RPS scenarios. Each of these scenarios are described briefly below.

We adopted 34 price scenarios—i.e., electricity price by sector, Henry Hub natural gas price, and electric sector fuel cost (coal and natural gas)—developed by the Energy Information Administration's (EIA) Annual Energy Outlook 2018 (AEO2018), along with three technology cost scenarios developed by the National Renewable Energy Laboratory (NREL) [25]. A list of AEO2018 scenarios are summarized in the supplementary document (Table S1). The cost scenarios includes low, mid, and high (constant) cost and performance

estimates for wind, PV, RPV, natural gas (both baseload (combined-cycle; NGb) and peaker (single-cycle; NGp)), and coal from 2016 to 2050. Low-cost wind and solar scenarios utilize low-cost estimates for land-based wind, along with PV and RPV technologies, while high-cost scenarios use constant costs at or near the 2018 cost estimates. The mid-case scenario assumes prospective advances in the RE arena technology. The low- and highcost scenarios for fossil fuel beyond 2016 relies on two case estimates from AEO2018, i.e., the high oil and gas resource and technology case and the low oil and gas resource and technology case, respectively. The mid-case scenario serves as a reference case for fossil fuel technology costs adopted from AEO (2018). Overall, the SD model is capable of assessing 918 (34 × 3 × 3 × 3) different scenarios. For the purpose of brevity, we focus on the three most plausible future scenarios: the new RPS, the previous RPS, and a future where integrating RE in the electric grid is discouraged. Under the first scenario, i.e., 100% RPS, we assume a future with scarce natural resources with costly fossil fuel and cheap RE technologies that make 100% RPS by 2050 possible. The second scenario, i.e., 10% RPS, is the opposite of the first scenario, in that we assume abundant natural resources with cheap fossil fuel and expensive RE technologies, hence a decreased RPS (10% by 2050). Lastly, we implement a status quo scenario, i.e., 20% RPS, that assumes reference case AEO prices with mid-case technology cost of fossil fuel and constant RE technology cost, along with business-as-usual RPS (RPS 20% by 2020 and on). Below we summarize each scenario.


#### **3. Materials and Methods**

Our model consists of five submodels: (1) demand; (2) supply; (3) gap between supply and demand (hereafter "gap"); (4) jobs; and (5) environmental impact, with more than 1200 variables. The first submodel consists of two modules that together estimate electricity demand beyond 2016. The second submodel includes six modules that altogether project megawatt-hour (MWh) generation. The gap and the jobs submodels each contain seven modules. Finally, the environmental impact submodel contains only one module. A detailed description of the model is provided in the Supplementary Materials (Section B) and a related work [35]. Here, we briefly describe the overarching dynamics of the model.

Our model is based on a series of stocks and flows. Stocks can change from period to period; the changes are governed by "flows". The flows, based on either natural sciencebased rules, human choice, or policies, or a combination thereof, are the set of rules that dictate the change in the stocks. Figure 1 provides a schematic of the model. Arrows provide the connections between stocks and flows. In all cases, the arrows represent the direction of interaction. Associated with each stock, flow, and connecting arrow is a set of quantifiable relationships and rules that allow us to model the system and assess the impact and tradeoffs between sectors within a time period as well as over time.

The basic structure of the modeling components is the physical market for electricity, which (in the figure) is at the intersection of supply and demand and is governed by an exogenous price path. As mentioned in the scenario definitions section, we implemented 34 price scenarios developed by EIA's 2018 AEO report. Thus, "exogenous" here does not mean a fixed value over time but rather is a given independent variable that fluctuates by month and over time. Supply depends on capacity and capacity utilization, which is aggregated from individual generation sources of capacity, utilization, and net exports into/out of state, while demand depends on in-state (domestic) consumption. In-state demand is the aggregation of individual sectoral demands, which can be impacted by market conditions (price) and population impacts.

**Figure 1.** Modeling schematic. **Figure 1.** Modeling schematic.

The basic structure of the modeling components is the physical market for electricity, which (in the figure) is at the intersection of supply and demand and is governed by an exogenous price path. As mentioned in the scenario definitions section, we implemented 34 price scenarios developed by EIA's 2018 AEO report. Thus, "exogenous" here does not The electricity market outcome at each time step maps into economic activity, estimated in dollars, which is one part of our macroeconomic module. The level of economic activity impacts the job outcome through changes in the demand for workers. It should be noted that population can also be impacted by the job impact as it could result in net outor in-migration.

quantifiable relationships and rules that allow us to model the system and assess the im-

pact and tradeoffs between sectors within a time period as well as over time.

mean a fixed value over time but rather is a given independent variable that fluctuates by month and over time. Supply depends on capacity and capacity utilization, which is aggregated from individual generation sources of capacity, utilization, and net exports into/out of state, while demand depends on in-state (domestic) consumption. In-state demand is the aggregation of individual sectoral demands, which can be impacted by market conditions (price) and population impacts. The electricity market outcome at each time step maps into economic activity, estimated in dollars, which is one part of our macroeconomic module. The level of economic The linkage between the electricity market and the environment (potential external impacts) is represented through a pollution component, where emissions during a time step add to the concentration level of the pollutant. We depict direct impacts of pollution through impacts on economic activity and through population (e.g., health impacts). It should be noted that there are a number of potential indirect links through, for example, consumer groups. In addition to pollution, our basic model includes water resources and human and avian mortalities. Further, RPS policies are included. Depending on the policy, the generation capacity, supply, demand, market prices, emissions, economic activity, or jobs could be impacted. Finally, all of these modules and methodologies are gathered in a unique SD model. Figure 2 summarizes the causal loop diagram utilized in developing the SD model.

activity impacts the job outcome through changes in the demand for workers. It should be noted that population can also be impacted by the job impact as it could result in net out- or in-migration. The linkage between the electricity market and the environment (potential external impacts) is represented through a pollution component, where emissions during a time step add to the concentration level of the pollutant. We depict direct impacts of pollution In order to read the causal loop diagram depicted in Figure 2, we begin by imagining the variable at the base of the arrow increasing in value; the sign at the arrowhead indicates the increase (+) or decrease (−) in the variable at the arrowhead, all other variables unchanged. Lastly, parallel lines crossing an arrow indicate delay in impact from the variable at the base of the arrow to the variable in the head of the arrow. The causal loop diagram presents the logic behind our SD model. The following is an explanation of one path in the diagram.

through impacts on economic activity and through population (e.g., health impacts). It should be noted that there are a number of potential indirect links through, for example, consumer groups. In addition to pollution, our basic model includes water resources and human and avian mortalities. Further, RPS policies are included. Depending on the policy, the generation capacity, supply, demand, market prices, emissions, economic activity, or jobs could be impacted. Finally, all of these modules and methodologies are gathered in a unique SD model. Figure 2 summarizes the causal loop diagram utilized in developing

the SD model.

*Energies* **2021**, *14*, x FOR PEER REVIEW 7 of 24

The required generation to achieve a certain level of RPS increases as in-state electricity demand increases, which increases the need for additional RE capacity to meet the corresponding RPS level. The higher the need for additional RE capacity, the higher the new capacity of RE. As the new capacity of RE rises, the total RE capacity rises, and the capacity that is decommissioned in the future increases with a delay. A higher level of RE capacity that is to be decommissioned decreases the total RE capacity, creating an enforcing loop (see Figure 2). On one hand, the higher the RE capacity, the higher the RE generation, hence the higher the need for peaker natural gas, storage, and transmission lines. On the other hand, if we assume that a higher level of RE generation replaces fossil fuel generation, then a higher level of RE generation results in lower GHG and air pollution, thereby lowering population mortality and morbidity (social cost). A higher level of RE generation can also decrease the gap caused by a discrepancy between supply and demand for electricity and/or RPS requirement. The same logic holds true for the remaining components of the causal loop diagram.

#### *Data*

Data were obtained from numerous sources including, the U.S. Energy Information Administration (various survey forms, AEO2018, and Layer Information for Interactive State Maps shapefiles), Emissions and Generation Resource Integrated Database (eGRID) of the U.S. Environmental Protection Agency (EPA), the National Renewable Energy Laboratories (JEDI, Annual Technology Baseline, wind data, and solar data), the New Mexico Public Regulation Commission, the United States Geological Survey, the U.S. Bureau of Economic Analysis, the United States Census Bureau, and the Western Electricity Coordinating Council, as well as the energy literature. Except for RPV data, we obtained generation data from EIA-923 and EIA-861 (annual and monthly data). The data includes historical nameplate capacity of the existing power plants, generation, power plants' locations (county and latitude/longitude), operating and planned retirement year times, and capacity factors. The data for the existing RPV capacity were obtained from the New Mexico Public Regulation Commission. Further, IMPLAN 2016 data were used to calculate jobs and output multipliers for each energy source. Lastly, economic benefit/cost of air pollution and GHG reduction multipliers came from the energy literature. Table S2 of the Supplementary Materials summarizes the key data sources.

#### **4. Results**

In this section, we present our results. We first review electricity generation under the three modeled scenarios. Next, we discuss state-level and county-level economic and environmental impacts. Economic impact results are presented for full-time equivalent employment and gross economic output. Environmental impacts, on the other hand, are reported in terms of GHG emissions, air pollution, water usage, and human and avian mortality associated with each of our three modeled scenarios. These impacts are experienced once the plants are in the O&M phase. Thus, environmental impact results are reported for the operations period solely and on a state- and county-level basis. Finally, we compare results across scenarios to expose whether results are statistically significantly different. If they are, we then identify state and county levels that experience job gains (winners) and job losses (losers).

#### *4.1. Generation*

Figure 3 shows the total electricity generation under the three modeled scenarios, and Figure 4 presents the generation mix through 2050. Based on the 20% RPS scenario, as with the other two scenarios, RE and fossil fuel generations encompassed respectively 17% and 83% of total generation in 2017. In 2030, generation shares are 15% and 85% for RE and fossil fuel, respectively. Compared to the 20% RPS scenario, RE generations are 9% higher in the generation mix under the 100% RPS scenario (24%) and 5% lower under the 10% RPS scenario (12%). All scenarios estimated a dip in electricity generation from 2036 until

Thousands GWh

2004

the end of 2037. This is due to the decommissioning of the existing coal-fired power plants in that period. The dip in the overall electricity generation is expected to be compensated by importing nuclear energy from Arizona. The figures depict generation from both within the state and not imported into the state. the 10% RPS scenario (12%). All scenarios estimated a dip in electricity generation from 2036 until the end of 2037. This is due to the decommissioning of the existing coal-fired power plants in that period. The dip in the overall electricity generation is expected to be compensated by importing nuclear energy from Arizona. The figures depict generation from both within the state and not imported into the state. 2036 until the end of 2037. This is due to the decommissioning of the existing coal-fired power plants in that period. The dip in the overall electricity generation is expected to be compensated by importing nuclear energy from Arizona. The figures depict generation from both within the state and not imported into the state.

Figure 3 shows the total electricity generation under the three modeled scenarios, and Figure 4 presents the generation mix through 2050. Based on the 20% RPS scenario, as with the other two scenarios, RE and fossil fuel generations encompassed respectively 17% and 83% of total generation in 2017. In 2030, generation shares are 15% and 85% for RE and fossil fuel, respectively. Compared to the 20% RPS scenario, RE generations are 9% higher in the generation mix under the 100% RPS scenario (24%) and 5% lower under the 10% RPS scenario (12%). All scenarios estimated a dip in electricity generation from

Figure 3 shows the total electricity generation under the three modeled scenarios, and Figure 4 presents the generation mix through 2050. Based on the 20% RPS scenario, as with the other two scenarios, RE and fossil fuel generations encompassed respectively 17% and 83% of total generation in 2017. In 2030, generation shares are 15% and 85% for RE and fossil fuel, respectively. Compared to the 20% RPS scenario, RE generations are 9% higher in the generation mix under the 100% RPS scenario (24%) and 5% lower under

**.** 

2040

2044

2048

*Energies* **2021**, *14*, x FOR PEER REVIEW 9 of 24

*4.1. Generation* 

*4.1. Generation* 

*Energies* **2021**, *14*, x FOR PEER REVIEW 9 of 24

**Figure 3.** Total annual electricity generation under the three modeled scenarios. **Figure 3.** Total annual electricity generation under the three modeled scenarios. **Figure 3.** Total annual electricity generation under the three modeled scenarios.

**Figure 4.** Annual electricity generation (in thousand GWh or TWh) by all six energy sources: (**a**) 10% RPS; (**b**) 100% RPS; (**c**) 10% RPS scenarios. **Figure 4.** Annual electricity generation (in thousand GWh or TWh) by all six energy sources: (**a**) 10% RPS; (**b**) 100% RPS; (**c**) 10% RPS scenarios. **Figure 4.** Annual electricity generation (in thousand GWh or TWh) by all six energy sources: (**a**) 10% RPS; (**b**) 100% RPS; (**c**) 10% RPS scenarios.

As presented in Figure 4, for the scenarios we estimated the amount and type of energy source to replace coal generation. By 2050, RE generation increases to 52%, while fossil fuel generation drops to 48% under the 20% RPS scenario. The 100% RPS scenario and the 10% RPS scenario result in a 11% higher and a 48% lower RE generation, respectively, when compared with the 20% RPS scenario. As mentioned, RPS requires utility companies to generate a portion of their in-state sales from RE. Thus, it is possible to have fossil fuel

(**c**)

generation even under the 100% RPS scenario. The takeaway here is that different energy scenarios lead to different energy mixes, which therefore means different environmental and economic impacts. fossil fuel generation even under the 100% RPS scenario. The takeaway here is that different energy scenarios lead to different energy mixes, which therefore means different environmental and economic impacts.

As presented in Figure 4, for the scenarios we estimated the amount and type of energy source to replace coal generation. By 2050, RE generation increases to 52%, while fossil fuel generation drops to 48% under the 20% RPS scenario. The 100% RPS scenario and the 10% RPS scenario result in a 11% higher and a 48% lower RE generation, respectively, when compared with the 20% RPS scenario. As mentioned, RPS requires utility companies to generate a portion of their in-state sales from RE. Thus, it is possible to have

#### *4.2. Economic Impacts 4.2. Economic Impacts*

*Energies* **2021**, *14*, x FOR PEER REVIEW 10 of 24

Our model is capable of estimating employment and gross economic output by three categories: direct (onsite), indirect, and induced. Total impact is the sum of direct, indirect, and induced impacts. Since direct, indirect, and induced impacts are a fixed fraction of total impact, we only discuss total impacts here. In what follows, we first discuss employment impact at the state and county level. We then compare results across scenarios and identify whether there are winners or losers. Next, we summarize total economic output results in a similar approach. Further discussion of the results, especially more granular level results (e.g., different types of energy sources during different phases), can be found in the Supplementary Materials (Section C). Our model is capable of estimating employment and gross economic output by three categories: direct (onsite), indirect, and induced. Total impact is the sum of direct, indirect, and induced impacts. Since direct, indirect, and induced impacts are a fixed fraction of total impact, we only discuss total impacts here. In what follows, we first discuss employment impact at the state and county level. We then compare results across scenarios and identify whether there are winners or losers. Next, we summarize total economic output results in a similar approach. Further discussion of the results, especially more granular level results (e.g., different types of energy sources during different phases), can be found

Figure 5 summarizes the cumulative total employment impact by the 20% RPS scenario and the other two modeled scenarios. We estimated a total employment impact on New Mexico in construction and O&M to be as follows: 151,857 (42,517 RE and 109,340 fossil fuel), 151,284 (112,593 RE and 38,691 fossil fuel), and 155,520 (26,271 RE and 129,248 fossil fuel) full-time equivalent jobs according to the 20% RPS, 100% RPS, and 10% RPS scenarios, respectively, from January 2017 to January 2050. Thus, compared to the 20% RPS scenario, the 100% RPS one (RE intensive scenarios) results in 573 fewer cumulative (construction and O&M) full-time equivalent jobs, while the 10% RPS one (most fossil fuel intensive scenario) yields 3663 more cumulative jobs. Note that these results are based on the assumption that all labor is provided locally. This assumption, which is on a 0–100% scale, can also be changed in the original SD model. What is important here is that this assumption does not impact the dynamics within modules and only results in lower direct economic impact (labor and economic output) across scenarios. in the Supplementary Materials (Section C). Figure 5 summarizes the cumulative total employment impact by the 20% RPS scenario and the other two modeled scenarios. We estimated a total employment impact on New Mexico in construction and O&M to be as follows: 151,857 (42,517 RE and 109,340 fossil fuel), 151,284 (112,593 RE and 38,691 fossil fuel), and 155,520 (26,271 RE and 129,248 fossil fuel) full-time equivalent jobs according to the 20% RPS, 100% RPS, and 10% RPS scenarios, respectively, from January 2017 to January 2050. Thus, compared to the 20% RPS scenario, the 100% RPS one (RE intensive scenarios) results in 573 fewer cumulative (construction and O&M) full-time equivalent jobs, while the 10% RPS one (most fossil fuel intensive scenario) yields 3663 more cumulative jobs. Note that these results are based on the assumption that all labor is provided locally. This assumption, which is on a 0–100% scale, can also be changed in the original SD model. What is important here is that this assumption does not impact the dynamics within modules and only results in lower direct

economic impact (labor and economic output) across scenarios.

**Figure 5.** Temporal cumulative jobs (construction and O&M) by modeled scenarios from January 2017 to January 2050. **Figure 5.** Temporal cumulative jobs (construction and O&M) by modeled scenarios from January 2017 to January 2050.

As demonstrated in Figure 5, all scenarios estimate a boost in energy employment after 2037. This is because existing coal-fired power plants are expected to retire in 2037, meaning there should be no new installation. Depending on the scenario, coal generation As demonstrated in Figure 5, all scenarios estimate a boost in energy employment after 2037. This is because existing coal-fired power plants are expected to retire in 2037, meaning there should be no new installation. Depending on the scenario, coal generation is expected to be replaced by either renewables or natural gas after 2037, and thus jobs related to coal are also likely to be replaced by renewable or natural gas jobs. Although the 100% RPS scenario yields fewer cumulative total jobs than the 20% RPS case, its impact fluctuates and is more diverse throughout the timespan of the study. Figure 3 depicts the employment distribution by the three modeled scenarios from 2017 through January 2050. Any spikes in the estimated employment numbers can be due to whether RPS and RE carve-

out requirements are met. We performed nonparametric tests such as the Kolmogorov– Smirnov tests to compare the equality of distributions of total employment across scenarios. The test results suggest that the null hypothesis of equality of distributions across the three scenarios cannot be rejected. In other words, at the state level, the employment impact of these three scenarios are not statistically significantly different. Thus, we found that the state of New Mexico is not a winner or a loser in terms of job gains or losses at the state level under all of the assessed scenarios. Temporal and cumulative employment impacts during construction and O&M phases are provided in the Supplementary Materials (Section C).

Now, we turn our attention to county-level employment results. Table 2 summarizes the annual average employment for all three scenarios by county. This table demonstrates an important result of the current study: some counties will be winners, and others will be losers. Figure 6 puts these results in perspective; it shows employment gain and loss per 10,000 labor force for the 100% RPS case versus the reference case of 20% RPS. Lastly, the Kolmogorov–Smirnov test results at the county level support the statistically significantly different employment distributions as well.


**Table 2.** Annual average employment by county and modeled scenarios \*.

\* Average values are from 2017 to 2050.

Economic output follows the employment results closely: when there is employment impact, there is economic output impact as well. Construction and O&M employees, depending on type of energy source, earn an average annual salary (with benefit) of 35,000 to 58,000 USD (2017\$) and 56,000 to 76,000 USD (2017\$) per year, respectively [36]. Under the 20% RPS scenario, these employments result in a cumulative (sum of construction and

O&M) total economic output of 24 billion USD (2017\$) (18% RE and 49% O&M) per year from 2017 to 2050. The 100% RPS and 10% RPS scenarios respectively lead to roughly 4 (20 USD: 94% RE and 54% O&M) and 2 (22 USD: 4% RE and 45% O&M) billion USD (2017\$) per year less than the 20% RPS case. In other words, the 20% RPS scenario yields a cumulative economic output that is 20% and 9% higher than the 100% RPS and 10% RPS scenarios, respectively. Figure 7 summarizes these results. Torrance 155 235 136 Union 81 163 61 Valencia 306 84 322 \* Average values are from 2017 to 2050.

*Energies* **2021**, *14*, x FOR PEER REVIEW 12 of 24

2050.

populous counties).

replacement.

4.3.1. Water Usage

*4.3. Environmental Impacts* 

**Figure 6.** Employment gain and loss per 10,000 labor force under the 100% RPS case compared to the 20% RPS case. Note: Positive values indicate job gains; negative values are job losses. **Figure 6.** Employment gain and loss per 10,000 labor force under the 100% RPS case compared to the 20% RPS case. Note: Positive values indicate job gains; negative values are job losses. *Energies* **2021**, *14*, x FOR PEER REVIEW 13 of 24

**Figure 7.** Total annual economic output by energy source and modeled scenarios from 2017 to **Figure 7.** Total annual economic output by energy source and modeled scenarios from 2017 to 2050.

Similar to the state-level employment, economic impact distributions under the three assessed scenarios are not statistically significantly different. However, at the county level,

Based on all of the three modeled scenarios, coal-fired power plants are assumed to fully retire after 2037. This is mainly due to the fact that the existing coal-fired power plants are aging (>40 years), and fuel contracts with coal mines are ending; more importantly, it is highly likely that coal will no longer be cost-competitive. Given these situations, we expect that there would be no new coal-fired power plants constructed in the future (see Figure 4). Note that these power plants are the most water-intense and polluting technologies in our set of energy sources (see Table S6). Eliminating coal from New Mexico's energy mix would result in fewer negative externalities (GHG, ambient pollutions, and water usage) from fossil fuel overall. Different technology costs along with RPS requirements drive the energy source that would eventually replace coal. The more RE replaces coal, the fewer negative externalities and the higher the social benefit from the

In what follows, we first discuss cumulative water withdrawal and consumption results at the state and county level. We then compare results across scenarios and identify whether there is water saved at the state and county levels. We take a similar approach in explaining emissions. Finally, we discuss the social benefit/cost of different scenarios.

Figure 8 depicts the temporal water withdrawal and consumption from 2017 to 2050. The 20% RPS scenario suggests a cumulative 13,178 and 1096 billion liters of water withdrawal and consumption throughout the study timeline. Compared to the 20% RPS scenario, the 100% RPS scenario uses less water for withdrawal and for consumption by 190 and 122 billion liters, respectively. The 10% RPS scenario, with the highest level of fossil fuels in the energy mix, uses 41 and 26 billion liters of water more than the 20% RPS sce-

nario for water withdrawal and consumption, respectively.

infrastructure in place can benefit from the fossil fuel intensive scenarios (namely the more

Similar to the state-level employment, economic impact distributions under the three assessed scenarios are not statistically significantly different. However, at the county level, rural counties can benefit under the RE intensive scenario, and counties with fossil fuel infrastructure in place can benefit from the fossil fuel intensive scenarios (namely the more populous counties).

#### *4.3. Environmental Impacts*

Based on all of the three modeled scenarios, coal-fired power plants are assumed to fully retire after 2037. This is mainly due to the fact that the existing coal-fired power plants are aging (>40 years), and fuel contracts with coal mines are ending; more importantly, it is highly likely that coal will no longer be cost-competitive. Given these situations, we expect that there would be no new coal-fired power plants constructed in the future (see Figure 4). Note that these power plants are the most water-intense and polluting technologies in our set of energy sources (see Table S6). Eliminating coal from New Mexico's energy mix would result in fewer negative externalities (GHG, ambient pollutions, and water usage) from fossil fuel overall. Different technology costs along with RPS requirements drive the energy source that would eventually replace coal. The more RE replaces coal, the fewer negative externalities and the higher the social benefit from the replacement.

In what follows, we first discuss cumulative water withdrawal and consumption results at the state and county level. We then compare results across scenarios and identify whether there is water saved at the state and county levels. We take a similar approach in explaining emissions. Finally, we discuss the social benefit/cost of different scenarios.

#### 4.3.1. Water Usage

county level.

Figure 8 depicts the temporal water withdrawal and consumption from 2017 to 2050. The 20% RPS scenario suggests a cumulative 13,178 and 1096 billion liters of water withdrawal and consumption throughout the study timeline. Compared to the 20% RPS scenario, the 100% RPS scenario uses less water for withdrawal and for consumption by 190 and 122 billion liters, respectively. The 10% RPS scenario, with the highest level of fossil fuels in the energy mix, uses 41 and 26 billion liters of water more than the 20% RPS scenario for water withdrawal and consumption, respectively. *Energies* **2021**, *14*, x FOR PEER REVIEW 14 of 24

**Figure 8.** Water withdrawal and consumption over time by the electric sector under the three modeled scenarios. (a) Water withdrawal; (b) Water consumption. **Figure 8.** Water withdrawal and consumption over time by the electric sector under the three modeled scenarios. (**a**) Water withdrawal; (**b**) Water consumption.

Considering an average price of 0.00689 USD/liter for water consumption by each energy source [37], the 20% RPS scenario results in a total cost of 527 million USD (\$2017) in water consumption for electricity generation. Compared to the 20% RPS scenario, the 100% RPS scenario results in saving 58 million USD for water savings, while the 10% RPS scenario increases costs by 13 million USD, as it is more water intense. Considering an average price of 0.00689 USD/liter for water consumption by each energy source [37], the 20% RPS scenario results in a total cost of 527 million USD (\$2017) in water consumption for electricity generation. Compared to the 20% RPS scenario, the 100% RPS scenario results in saving 58 million USD for water savings, while the 10% RPS scenario increases costs by 13 million USD, as it is more water intense.

To compare water consumption distributions across scenarios, we performed Kolmogorov–Smirnov tests. Test results provided us with evidence to reject the null hypothesis of equality of distributions between the 100% RPS and 20% RPS scenarios even at the state level. On the whole, we did not find similar results when comparing the 10% RPS

Table 3 summarizes the annual average million liters of water consumption by county and the three scenarios; Figure 9 translates this information to per capita (county) water consumption saved or lost. While the majority of counties see no changes, the majority of impacts are the savings. Our nonparametric test results further support the alternative hypothesis of unique water consumption distributions across scenarios at the

**Table 3.** Annual average water consumption by county and modeled scenarios.\*

**County 20% RPS 100% RPS 10% RPS**  Bernalillo 46 16 53 Catron 0 0 0 Chaves 0.08 0.68 0.04 Cibola 0 0 0 Colfax 0.23 0.23 0.23 Curry 0 0 0 De Baca 0 0 0 Dona Ana 119 89 126 Eddy 31 1 38 Grant 34 4 41 Guadalupe 0 0 0 Harding 0 0 0 Hidalgo 39 8 45 Lea 217 187 223 Lincoln 0 0 0 Los Alamos 13 7 14 Luna 140 109 146 Mc Kinley 203 173 209 Mora 0 0 0 Otero 0.08 0.68 0.04

To compare water consumption distributions across scenarios, we performed Kolmogorov–Smirnov tests. Test results provided us with evidence to reject the null hypothesis of equality of distributions between the 100% RPS and 20% RPS scenarios even at the state level. On the whole, we did not find similar results when comparing the 10% RPS scenario against the 20% RPS one.

Table 3 summarizes the annual average million liters of water consumption by county and the three scenarios; Figure 9 translates this information to per capita (county) water consumption saved or lost. While the majority of counties see no changes, the majority of impacts are the savings. Our nonparametric test results further support the alternative hypothesis of unique water consumption distributions across scenarios at the county level.


**Table 3.** Annual average water consumption by county and modeled scenarios \*.

\* Average values are in million liters and from 2017 to 2050; "0" means no change.

#### 4.3.2. Air Pollution and Greenhouse-Gas Emissions

Figures 10 and 11 depict the cumulative impact of air pollution and GHG emissions, along with consecutive social benefit to the state from 2017 to 2050. Cumulatively, the RE intensive scenario emits roughly 91 million tonnes GHG less than the 20% RPS scenario throughout the study timeline, leading to more than 6.8 billion USD (2010\$) in cumulative climate benefit. The fossil fuel intensive scenario, on the other hand, emits 3% (19 million tonnes) higher GHG than the 20% RPS one, which causes more than 1400 million USD (2010\$) social cost compared to the 20% RPS one. Each one million tonnes of GHG emissions is equivalent to GHG emissions by approximately 2250 million miles driven by an average passenger vehicle. Table 4 summarizes the county level results only for GHG. Based on

the Kolmogorov–Smirnov test results, we can reject the null hypotheses of equality of GHG emission distributions when comparing both the 100% RPS and 10% RPS scenarios against the 20% RPS scenario. In other words, the 100% RPS scenario results in statistically significantly lower GHG than the 20% RPS scenario, while the opposite holds true for the 10% RPS scenario. We found similar results at both state and county levels. Socorro 0 0 0 Taos 0 0 0 Torrance 0 0 0 Union 0 0 0 Valencia 50 20 56 \* Average values are in million liters and from 2017 to 2050; "0" means no change.

Quay 0 0 0 Rio Arriba 0 0 0 Roosevelt 0.08 0.68 0.04 Sandoval 0 0 0 San Juan 1873 1843 1879 San Miguel 0 0 0 Santa Fe 0.08 0.68 0 Sierra 0 0 0

*Energies* **2021**, *14*, x FOR PEER REVIEW 15 of 24

**Figure 9.** Per capita water consumption saved/lost under the 100% RPS scenario compared to the 20% RPS scenario. Note: Negative value indicates water saved; 1 gallon is ≈3.79 liters. **Figure 9.** Per capita water consumption saved/lost under the 100% RPS scenario compared to the 20% RPS scenario. Note: Negative value indicates water saved; 1 gallon is ≈3.79 liters. ity of GHG emission distributions when comparing both the 100% RPS and 10% RPS scenarios against the 20% RPS scenario. In other words, the 100% RPS scenario results in statistically significantly lower GHG than the 20% RPS scenario, while the opposite holds

3


100%-RPS 10%-RPS

true for the 10% RPS scenario. We found similar results at both state and county levels.

**Figure 10.** State-level cumulative tonnes of GHG and air emission under the three modeled scenarios from 2017 to 2050. **Figure 10.** State-level cumulative tonnes of GHG and air emission under the three modeled scenarios from 2017 to 2050.

123

**Table 4.** Annual average thousand tonnes of GHG emissions by county and modeled scenarios.


**County 20% RPS 100% RPS 10% RPS**  Bernalillo 38 18 43 Catron 0 0 0 Chaves 0.09 0.78 0.02 Cibola 0 0 0 Colfax 0.24 0.24 0.24 Curry 0 0 0 De Baca 0 0 0 Dona Ana 90 69 95

**Figure 11.** Social impact of air pollution and GHG emission reduction for the 100% RPS and 10% RPS scenarios compared

CO2(x10^3) NOx PM SO2

CO2(×103) NOx PM SO2

to the 20% RPS scenario from 2017 to 2050.


0

50

100

Million 2010\$

150

7 7


470


400

900

1400

Tonne

574

555

**Figure 10.** State-level cumulative tonnes of GHG and air emission under the three modeled scenarios from 2017 to 2050.

1087

1096

CO2(×106 tonnes) Mercury NOx(×103 tonnes) PM(×102 tonnes) SO2(×103 tonnes)

1094

3

3

3

4.3.2. Air Pollution and Greenhouse-Gas Emissions

Figures 10 and 11 depict the cumulative impact of air pollution and GHG emissions, along with consecutive social benefit to the state from 2017 to 2050. Cumulatively, the RE intensive scenario emits roughly 91 million tonnes GHG less than the 20% RPS scenario throughout the study timeline, leading to more than 6.8 billion USD (2010\$) in cumulative climate benefit. The fossil fuel intensive scenario, on the other hand, emits 3% (19 million tonnes) higher GHG than the 20% RPS one, which causes more than 1400 million USD (2010\$) social cost compared to the 20% RPS one. Each one million tonnes of GHG emissions is equivalent to GHG emissions by approximately 2250 million miles driven by an average passenger vehicle. Table 4 summarizes the county level results only for GHG. Based on the Kolmogorov–Smirnov test results, we can reject the null hypotheses of equality of GHG emission distributions when comparing both the 100% RPS and 10% RPS scenarios against the 20% RPS scenario. In other words, the 100% RPS scenario results in statistically significantly lower GHG than the 20% RPS scenario, while the opposite holds true for the 10% RPS scenario. We found similar results at both state and county levels.

236 253

305 254

292 254

100%-RPS 10%-RPS 20%-RPS

**Figure 11.** Social impact of air pollution and GHG emission reduction for the 100% RPS and 10% RPS scenarios compared to the 20% RPS scenario from 2017 to 2050. **Figure 11.** Social impact of air pollution and GHG emission reduction for the 100% RPS and 10% RPS scenarios compared to the 20% RPS scenario from 2017 to 2050.


**Table 4.** Annual average thousand tonnes of GHG emissions by county and modeled scenarios. **Table 4.** Annual average thousand tonnes of GHG emissions by county and modeled scenarios.

Note: Average values are in thousand tonnes and from 2017 to 2050; "0" means no change.

Since coal is the only energy source that emits mercury and since it stays unchanged throughout our study period, mercury is therefore assumed to be the same amount in all three scenarios, i.e., 3 tonnes. The 100% RPS scenario results in a roughly 500 tonne reduction in SO<sup>2</sup> emissions (approximately 3 million USD (2010\$) in social benefit) compared to the 20% RPS scenario, while the 10% RPS scenario results in an increase of more than

100 tonnes of SO<sup>2</sup> (1 million USD (2010\$) in social cost) cumulatively from 2017 to 2050. NO<sup>x</sup> emissions in the RE intensive scenario are reduced by 6649 tonnes and a 7 million USD (2010\$) increase in social benefits compared to the 20% RPS scenario, while the fossil fuel intensive scenario yields 1990 tonnes more NO<sup>x</sup> and 2 million USD (2010\$) more in social costs. Lastly, PM emission in the 100% RPS scenario is reduced by 5612 tonnes, resulting in a 123 million USD (2010\$) increase in social benefits compared to the reference case scenario, while the 10% RPS scenario yields 1215 tonnes more and 27 million USD (2010\$) in social costs.

Table 5 summarizes the cumulative avoided air pollution, social benefit, and the premature mortality and morbidity impact of air pollution under the 100% RPS and the 10% RPS scenarios from 2017 to 2050 compared to the 20% RPS scenario. The 20% RPS scenario is estimated to have 408 to 924 adult fatalities caused by a combination of SO2, NOx, and PM pollutants. The 100% RPS scenario has the potential to avoid 23 to 52 premature mortality incidences, while the 10% RPS scenario increases 5 to 11 fatalities due to exposure to ambient pollution, when compared to the reference scenario. While the majority (>90%) of social benefits for each scenario comes from avoiding premature mortality [12], we also estimated a number of additional morbidity benefits, from avoiding nonfatal heart attacks, hospital visits for asthma, or other cardiopulmonary conditions, to fewer lost work or school days. For example, the 100% RPS scenario is estimated to result in avoiding 19 visits to the emergency department or hospital for cardiopulmonary conditions as well as approximately 3000 fewer lost work or school days from 2017 to 2050.


**Table 5.** Accumulated emissions, social benefits, and mortality and morbidity incidence reductions compared to the reference case scenario (20% RPS) using SO<sup>2</sup> , NOx, and PM reductions as a result of RE installation from 2017–2050.

Note: Positive value means reduction, whereas negative value indicates addition. a,b Multipliers from these studies are used to calculate the mortality and morbidity incidences.

> Fossil fuel and RE power plants are contributors to avian mortality; fossil fuel plants induce fatality through plant operation, acid rain, mercury, and climate change, while bird fatality associated with wind and PV power plants is mainly due to bird colliding with turbine blades and panels respectively [15,38,39]. Figure 12 summarizes avian mortality caused by different energy sources (i.e., coal, NG, wind, and PV) under different scenarios. The 20% RPS scenario leads to 5.131 million avian fatalities, of which fossil fuel

to 2050.

is responsible for approximately 99% of the overall number of deaths. Compared to the 20% RPS scenario, the 100% RPS scenario has the potential to save 441 thousand deaths, while the 10% RPS scenario leads to 106 thousand more avian fatalities. Lastly, the RE intensive scenario leads to more than 4.69 million bird deaths, with fossil fuel sources being responsible for 97% of the overall number. *Energies* **2021**, *14*, x FOR PEER REVIEW 19 of 24

**Figure 12.** Avian mortality caused by coal, NG, wind, and PV power plants under the three modeled scenarios from 2017 **Figure 12.** Avian mortality caused by coal, NG, wind, and PV power plants under the three modeled scenarios from 2017 to 2050.

Dissanayake and Ando [40] conducted a choice experiment survey in Illinois and found that their respondents are willing to pay between 1.11 and 1.13 USD for each extra bird per year, and between 7.72 and 10.22 USD for each endangered species annually. Since we were unable to discern different types of birds (generic versus endangered species) in our analysis, we utilized the mean value of the upper-level estimates as to how much each bird death is worth. We estimated that the 100% RPS scenario is capable of saving 3 million USD in bird mortality, while the 10% RPS scenario costs the state 1 million USD more in avian mortality, when compared with the 20% RPS scenario. We performed Dissanayake and Ando [40] conducted a choice experiment survey in Illinois and found that their respondents are willing to pay between 1.11 and 1.13 USD for each extra bird per year, and between 7.72 and 10.22 USD for each endangered species annually. Since we were unable to discern different types of birds (generic versus endangered species) in our analysis, we utilized the mean value of the upper-level estimates as to how much each bird death is worth. We estimated that the 100% RPS scenario is capable of saving 3 million USD in bird mortality, while the 10% RPS scenario costs the state 1 million USD more in avian mortality, when compared with the 20% RPS scenario. We performed Kolmogorov– Smirnov tests and *t*-tests on human and avian mortality, and also on air pollutants (except mercury), and we found similar results to those of GHG and water consumption.

#### Kolmogorov–Smirnov tests and t-tests on human and avian mortality, and also on air pol-*4.4. Summary of Cumulative Results*

lutants (except mercury), and we found similar results to those of GHG and water consumption. *4.4. Summary of Cumulative Results*  Our analysis seeks to investigate the economic and environmental impacts of the status quo scenario, along with two future scenarios. Without considering environmental impacts such as water usage, air pollution, GHG, and avian mortality, our results suggest that the reference case and the fossil fuel intensive scenarios lead to higher economic output and total employment impacts than the RE intensive scenario, though not statistically significant. Once the environmental impacts are included, these results no longer hold. Compared to the 20% RPS scenario, cumulatively, the 100% RPS scenario results in 3095 million USD (2017\$) higher benefit and the 10% RPS scenario in 3325 million USD (2017\$) more cost to the state. This makes the 100% RPS the best scenario, 20% RPS the second Our analysis seeks to investigate the economic and environmental impacts of the status quo scenario, along with two future scenarios. Without considering environmental impacts such as water usage, air pollution, GHG, and avian mortality, our results suggest that the reference case and the fossil fuel intensive scenarios lead to higher economic output and total employment impacts than the RE intensive scenario, though not statistically significant. Once the environmental impacts are included, these results no longer hold. Compared to the 20% RPS scenario, cumulatively, the 100% RPS scenario results in 3095 million USD (2017\$) higher benefit and the 10% RPS scenario in 3325 million USD (2017\$) more cost to the state. This makes the 100% RPS the best scenario, 20% RPS the second best, and 10% RPS the worst-case scenario, when both environmental and economic impacts are taken into account. Thus, the higher the RPS level, the higher the overall benefit to the state. Table 6 summarizes the state cumulative results in relation to the 20% RPS scenario. At the county level, compared to the 20% RPS case, we found that RE suitable counties are net gainers (in terms of both economic and environmental impacts), while fossil fuel counties suffer economically and benefit environmentally under the 100% RPS scenario. The opposite holds true when comparing the 10% RPS scenario against the 20% RPS scenario.

best, and 10% RPS the worst-case scenario, when both environmental and economic impacts are taken into account. Thus, the higher the RPS level, the higher the overall benefit to the state. Table 6 summarizes the state cumulative results in relation to the 20% RPS scenario. At the county level, compared to the 20% RPS case, we found that RE suitable counties are net gainers (in terms of both economic and environmental impacts), while fossil fuel counties suffer economically and benefit environmentally under the 100% RPS

**Table 6.** Summary of cumulative results in relation to the 20% RPS scenario from 2017–2050.

Economic Output −3962 (−120) −1881 (−57) Water benefit 59 (2) −13 (−0.4) CO2 6865 (208) −1402 (−42) SO2 3 (0.1) −1 (−0.03) NOx 7 (0.2) −2 (−0.1) PM2.5 123 (4) −27 (−1)

**Outcome 100% RPS, in Million USD 10% RPS, in Million USD** 

RPS scenario.


**Table 6.** Summary of cumulative results in relation to the 20% RPS scenario from 2017–2050.

Numbers in parentheses are annual values. <sup>a</sup> Employment monetary values are based on salary of 46,500 USD for construction and 66,000 USD for O&M jobs [36], calculated based on the employment count (see Section 4.2) of −573 (−17 annual) jobs and 3663 (111 annual) jobs for the 100% RPS and 10% RPS scenarios, respectively.

#### **5. Conclusions and Policy Implications**

Legislators across the globe are supporting policies that move toward electricity generation from renewable resources. To this end, some jurisdictions in the U.S. have enacted regulations, such as the RPS. These provide a mechanism that can result in not only GHG emission reduction but also water preservation. This is especially prudent in geographic locations with limited water resources. Moreover, RPS can support jobs, although the primary policy target of an RPS is not focused squarely on job creation.

This study provided a roadmap of how to quantify the economic and environmental impacts of three scenarios, in which not only the RPS level varies but also the energy sector dynamics, technological cost, and price of energy. Specifically, we modelled New Mexico's newly enacted RPS policy, where it increases from the status quo of 20% by 2020 to 100% by 2050. We also studied a scenario where RPS decreases to 10% by 2050. In so doing, we combined results from input–output (JEDI and IMPLAN) analyses, econometrics (Stata), and GIS (ArcGIS), and we created a unique SD model that enabled us to assess regional economic and environmental impacts of different scenarios. Our contribution to the current body of literature is twofold: not only did we assess different RPS scenarios by considering the underlying dynamics within the energy sector, but we also assessed these impacts at a lower granular level (i.e., county level).

Under the status quo scenario, the estimates in our model accounted for 152 thousand cumulative full-time equivalent jobs, 24 billion USD in economic output, 3648 million USD in air quality cost, 36 billion USD in climatic cost, 527 million USD worth of water use, 5 million avian mortality, and 409–924 premature mortality. Compared with this status quo scenario, our analysis suggests that the RE intensive scenario (100% RPS) leads to less cumulative employment and economic output, but much higher social benefits compared to the 20% RPS scenario, i.e., 500–15,000 fewer cumulative jobs, 3–4 billion USD less in cumulative economic output, 132 million USD less in air quality cost, 7 billion USD less in climatic cost, 58 million USD less in value of water use, 441–485 thousands less in avian mortality, and 23–53 less in premature mortality. The 10% RPS scenario leads to approximately 4000 more jobs, 2 billion USD less in cumulative economic output, 29 million USD more in air quality cost, 1 billion USD more in climatic cost, 13 million USD more in value of water use, 100 thousand more in avian mortalities, and 5–11 more in premature mortality than the 20% RPS scenario. Considering the environmental impacts, our analysis finds that the Senate Bill RPS scenario (100% RPS) is the best scenario, followed by the status quo scenario, and the 10% RPS scenario is the worst case.

Higher levels of RPS policy aligns with support from New Mexicans. In separate work by the co-authors [41,42], we estimated that a sample of New Mexicans are willing to pay 5.4 USD per year on top of their annual electricity bill for each 1% increase in the current level of RPS (20%). To achieve a 100% RPS by 2050, we extrapolate that, all else equal, New Mexicans are willing to pay 58, 180, 373, 581, 803, and 1144 million USD (2017\$) in 2020, 2025, 2030, 2035, 2040, and 2050, respectively. Note that the wide range of willingness to

pay is due to the way the bill requirements of achieving 80% RPS by 2040 were designed. Under this bill, electric utility companies were required to increase current RPS level to 25% by 2020, 35% by 2025, 50% by 2030, 65% by 2035, and 80% by 2040. The higher the percentage, the higher residents are willing to pay.

Although scenarios with a lower level of RPS might result in supporting a higher number in employment (in the fossil fuel sector), these scenarios lead to much higher social cost of GHG and ambient pollution (i.e., premature mortality and morbidity) and water usage. This suggests that coming up with an overarching policy that benefits both the environment and economy is not an easy task. Policymakers seeking to promote energy policies may need to consider not only the economic benefit associated with energy development but also social welfare. In other words, RPS policies are more desirable when internalizing external costs and hence correcting for market failure [21,43,44].

Further, the most decisive conclusion that can be drawn from job comparison across different scenarios is that the higher the RE development level, the more disperse and rural the employment impact. On the contrary, the higher the level of fossil fuel deployment, the less diverse and rural the job impact. San Juan County among all is expected to experience a net negative (loss) in O&M jobs, i.e., a loss of 780 jobs from coal-fired power plant retirements after 2037 and depending on the scenario, a gain of an annual average of 84 (100% RPS) to 601 (10% RPS) jobs. Concurrently, the state is estimated to experience nearly 686 billion USD (100% RPS) in social benefits, particularly from the coal power plants retirement. The disparity in job and economic output distribution across counties and energy sources suggests that counties with varying energy potential and population density may experience variation in impacts. In other words, some counties are likely to be net gainers while others may suffer.

The results of this study are broadly consistent with that in the literature [20,26,29,45–49]. We do recognize that the majority of these studies had explicit research questions only on wind energy. For example, some studies sought to measure the actual economic impact of a particular wind installation at county level (e.g., [48]), while others estimated a wind vision for the U.S. (e.g., [26,49]) or the environmental and economic impact of RPS policies nationwide for solely one year (i.e., [20]). Similar to Barbose et al. [20], Millstein et al. [29], and Wiser et al. [21], our model suggests that RPS policies have the potential to yield billions of dollars in climatic and air-quality benefits as well as economic benefits. Similar to preceding studies, we found that increasing RPS does not result in stimulating the economy of a state [3,4], but it does impact the environment positively [29,45,50]. Our contribution to the literature is that we demonstrated that increasing RPS does stimulate the economy of the state at the more granular levels (especially rural counties).

The tools and theories integrated for the analysis in this research are broadly transferable across a wide range of topics and/or regions. For example, a similar approach can be taken to evaluate RPS policies in each one of the other 28 states with such regulations. Our model can be modified and used for states with existing 100% RPS policies (Hawaii, California, Washington, Maine, New York, and Virginia), and those with promises for 100% clean electricity (Colorado, Connecticut, Massachusetts, Illinois, Oregon, New Jersey, Nevada, Wisconsin, and Puerto Rico). Additionally, our state-of-the-art modeling and set of methods are applicable to other topics, such as the impact of decarbonization through a battery of smart grid (e.g., smart meter), transportation (e.g., electric vehicle), and energy-efficient buildings; 100% RE for all sectors (i.e., electricity, heating/cooling, transportation, and industry); oil and natural gas extraction; and the agriculture sector on regional economies. Another expansion of this analysis could include developing nations, as well as other developed countries with similar regulatory mandates. One potential limitation of this work is that our model does not calculate electricity rates for each scenario and takes rates as independent. More expensive scenarios could potentially result in higher electricity rates, which can impact economic activity. This is also important as it has the potential to impact customers' perspective and willingness to pay towards higher level of RE diffusion. Another caveat is that we assume that employment impacts are fully

provided (100%) by local residents, which is not typically the case in real-world settings; although the model is capable of varying this assumption, we chose not to include this here for the purpose of brevity. Future research should account for data uncertainty and present results as confidence intervals rather than precise values. This can be done by using Monte-Carlo simulations. This study's results provided improved information for state policymakers seeking to alter RPS policies and can also be extrapolated to states with similar energy policies.

**Supplementary Materials:** The following will be available online at www.jamalmamkhezri.weebly. com/research.html.

**Author Contributions:** J.M.: Conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, resources, software, validation, writing—original draft preparation; L.A.M.: methodology, software, validation, formal analysis, writing—review and editing; J.M.C.: conceptualization, supervision, resources, funding acquisition, writing—review and editing. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partially funded by the U.S. National Science Foundation awards #IIA-1301346 and OIA-1757207. Jamal Mamkhezri acknowledges support received from the University of New Mexico Center for Regional Studies.

**Institutional Review Board Statement:** Not Applicable.

**Informed Consent Statement:** Not Applicable.

**Data Availability Statement:** Publicly available datasets were used in this study. Readers are directed to Table S2 for an overview of the key variables' data sources.

**Acknowledgments:** We thank the late Ronald Cummings, Jennifer Thacher, and Robert Berrens, as well as participants at the 2019 Southern Economic Association (SEA) conference, the 2020 Association of Environmental and Resource Economists at Western Economic Association International conference, the 2020 Energy Policy Institute workshop, the 2020 System Dynamics Society Economics special interest group, and at the 37th and 38th International Conference of the System Dynamics Society for on an earlier version of this paper. We also thank the four anonymous reviewers for their insightful comments. Any errors are, of course, our own.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **Abbreviations**


#### **References**


## *Article* **Comprehensive Benefit Analysis of Port Shore Power Based on Carbon Trading**

**Yang He and Yun Zhu \***

School of Electrical Engineering, Guangxi University, Nanning 530004, China **\*** Correspondence: zhuyun@gxu.edu.cn

**Abstract:** The concept of "oil to electricity" is crucial for expanding the share of electricity in final energy consumption as well as for encouraging energy efficiency and emission reduction. Initially, a multidimensional strategy analysis is conducted for the government, ports, and ships concerned. From an economics perspective, a mathematical model of electricity substitution benefit analysis based on multiagent cooperative game theory under cap and trade and carbon tax policies is constructed, and the effect of carbon emissions caused by ships on the environment and society is converted into economic value. How several variables, such as transformation costs, ship electricity consumption, subsidy rates, carbon tax prices, and the ratio of shore power usage time to berthing time, affect the functioning of shore power is analyzed. The best electricity price under various circumstances is determined while considering the benefits of the three parties to maximize social welfare. The reduction in carbon dioxide and pollutant emissions is calculated. Meanwhile, the environmental advantages of the "replacement of oil with electricity" procedure are estimated. An example supports the claim that the suggested modeling approach can successfully resolve the economic benefits of each participant for the period that fosters the growth of electricity replacement projects and offers a sound scientific foundation for the formation of pertinent legislation.

**Keywords:** carbon trading; cooperative game; pollutant emission; shore power

#### **1. Introduction**

Seaborne trade stalled in 2020 amid the COVID-19 epidemic and the anticipated downturn in global economic growth. Surveys conducted by UNCDAT have revealed that the world container throughput declined by 1.2% to 815.6 million 20-foot TEUs [1]. Governments advocated for citizens to stay inside to prevent contact, which tremendously accelerated the rise of international e-commerce. The distribution of vaccines has slowed the epidemic's growth and deaths, enabling the recovery of international trade. The beginning of the economic recovery was heralded in 2021, with seaborne trade predicted to increase by 4.3% [2]. The maritime sector's quick ascent has resulted in significant emissions of pollutants such as CO2, SO2, and NOx. As of 2012, shipping was responsible for 972 million tons of greenhouse gas release, or 2.5% of all releases worldwide [3]. It not merely raises the global temperature but also leads to respiratory illnesses in those who live close to ports and coastlines [4]. There is a consensus among experts that 60% to 90% of the diffusion in ports stems from ships, which also account for 70% of marine diffusion [5].

Several nations and international organizations are pursuing numerous explorations and research to lessen the issue of pollution discharge from ships. Based on the "IMO 2020" guideline, ships operating outside specified emission-control areas can diminish sulfur oxide outflow by 8.5 million tons while exploiting low-sulfur oil with a sulfur content of 0.50% m/m [6]. More than 570,000 residents will die prematurely if the SO<sup>x</sup> limit reduction is postponed from 2020 to 2025 [7]. Even though low-sulfur oil modestly reduces NO<sup>x</sup> discharge by 10% [8], the maritime industry still contributes to 250,000 fatalities [9]. The usage of shore power minimizes pollution emissions by 94–97% when berthing at the

**Citation:** He, Y.; Zhu, Y. Comprehensive Benefit Analysis of Port Shore Power Based on Carbon Trading. *Energies* **2023**, *16*, 2755. https://doi.org/ 10.3390/en16062755

Academic Editor: George Halkos

Received: 7 February 2023 Revised: 12 March 2023 Accepted: 13 March 2023 Published: 15 March 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

port [10]. It satisfies fundamental needs such as lighting, cooling, and communication, notably improving air quality [11], decreasing carbon emissions by 800,000 tons, and elevating environmental benefits [12,13].

The present literature on shore power is confined and centers especially on technology [14,15], economy [16,17], management [18,19], policy promotion [20], etc. One study by Qi [8] observed the trend in obstacles to the upgrading of shore power in China, focusing on the economic evaluation for different stakeholders. Zhao [21] considered the effects of port size, fines, and subsidies on the evolutionary game to analyze the financial relationship between the government and the port. A mathematical model was constructed by Wu [22] to investigate how government subsidy schemes might help shorten the outflow from ship berths. Song [23] set up four parties, the government, the port, the ship, and the power grid, then pondered the cost-effectiveness of each in the shore power system to calculate the optimal shore power price. Through quantitative evaluation, Tseng [24] demonstrated that environmental policies levying pollution taxes can immensely suppress pollutant discharge.

The global community has agreed on limiting carbon emissions since the "Kyoto Protocol" took effect. On the one hand, developed nations such as the European Union, Japan, and Australia were the first to adopt cap-and-trade and carbon tax policies, which victoriously decreased CO<sup>2</sup> emissions [25]. On the other hand, China pledged to achieve carbon neutrality by 2060 at the 75th UN General Assembly. However, there are few surveys on the carbon trade mechanism of shore power, and the majority of studies concentrate on economic factors. Murray [26] discovered the carbon price legislation lowered British Columbia's emissions by 5–15% through modeling. The EU Emissions Trading Scheme's implementation has resulted in a 0.5–2 million ton CO<sup>2</sup> reduction [27]. Song [28] developed a stochastic model to explore the effects of various carbon tax rates on the growth of logistical capacity. A dual-objective optimization model was developed by Liu [29] to discuss the liner's best performance under the carbon tax policy. Chen [30] described a social optimal welfare model to assess how carbon taxes affect production, consumption, and redistribution.

This study uses multiagent game behavior as its research object in the "oil-to-electricity" conversion process. Economically speaking, the government, the port, and the ship are strategically examined, and the cost of the port includes the value of the carbon. The whole social welfare is maximized by the favorable shore electricity price. The environmental advantages of "oil-to-electricity" are counted simultaneously.

#### **2. Electricity Substitution Multiagent Game Model**

#### *2.1. Multiagent Game Analses*

Figure 1 depicts the evolution of port shore power based on a cooperative game. Due to the contradictory objectives being sought by the government, port, and ships while replacing electricity, a game of interests has developed.

Ships consuming fuel oil pollute the environment, harm locals' health, and make it more difficult to achieve carbon neutrality, all of which run counter to the government's stated environmental objectives. From an economic and market standpoint, shipping corporations feel there are few ports with the ability to supply shore power, whereas ports think there are not many ships with the capability to use shore power. Nobody wants to take the initiative and make themselves passive. Now is the time for the government to take a two-pronged strategy and develop policies that would encourage the implementation of shore power projects through incentives and sanctions.

The government subsidizes port renovation and applies carbon emission controls to increase port operating costs. The government encourages the development of the energy structure by funding ship retrofits while imposing environmental protection tariffs on pollutants released through the use of fuel oil. By giving priority to berthing for ships using shore power, ports can entice ship retrofits. Ships can decide whether to employ shore power depending on their financial circumstances.

**Figure 1.** Multiagent game of oil-to-electricity conversion.

Consequently, the three parties' game is impacted by the subsidy rate, carbon mechanism, environmental protection tax, and shore electricity price. The ideal shore power pricing can be attained in the endless game, allowing the electric energy replacement effort to continue.

This article proposes a simple method to calculate the carbon emissions of ports, which only considers the carbon emissions of ships. Other equipment will not be considered, such as harbor railway, quayside container crane, locomotive, and other special machinery. We calculate the carbon emissions of ships using fuel oil and shore power and add the carbon emissions to the economic value, including the port cost. Then, the tripartite economic game models under the two carbon mechanisms are established, respectively, and the impact of the power supply service price, the carbon price, and the proportion of the time using shore power to the docking time on the social welfare, government benefits, port benefits, and ship benefits is discussed.

#### *2.2. Assumptions*

To simplify the problem and facilitate modeling and subsequent analysis and discussion, the following assumptions are proposed for the research content:


#### *2.3. Methodology*

In order to address climate change, the Ministry of Ecology and Environment of China has formulated "Implementation Plan for Setting and Allocating the Total Amount of Carbon Emission Trading Quotas for 2019–2020 (Power Generation Sector)", which includes enterprises or other economic organizations that emit 26,000 ton of carbon dioxide equivalent or more in any one year from 2013 to 2019. As carbon emissions trading management has only just started, the central government is currently only regulating the power sector. However, in places such as Guangdong Province and Shanghai, it has already started to cover industries such as steel, chemicals, cement, paper, and aviation. The allocation methods are mainly the historical intensity method and the historical emissions method. The former is applicable to industrial enterprises with a high correlation between product output and carbon emissions, and well measured. The latter is suitable for industrial enterprises where the boundary has changed significantly in recent years and it is difficult to apply the industry baseline method or the historical intensity method.

In the historical intensity method, the annual base quota for enterprises is equal to the historical intensity base multiplied by the annual business volume. The historical intensity base is the weighted average of the enterprise's annual business volume of carbon emissions in the previous three years. The annual business volume is the business volume data of the enterprise for the current year verified by a third party verification agency and validated and confirmed by the relevant departments. In the historical emissions approach, an enterprise's annual base allowance is equal to the historical emissions base. In this article, port companies use the historical intensity method and carbon emissions from ports only consider carbon emissions from ships, not from other equipment such as shoreside cranes and locomotives. Thus, carbon emissions are only relevant to the activities of the ship, and the oil or electricity consumed per unit of power has a relevant carbon emission factor to calculate carbon emissions.

#### *2.4. Parameter Descriptions*

The following is a description of the parameters that appear in the mathematical model of the three-way game.

$$\begin{cases} \mathbf{P}^{k,i} = \mathbf{W}\_{\mathbf{sp}}^{k,i} / T\_{\mathbf{sp}}^{k,i} \\\\ \mathbf{W}\_{\mathbf{sp}}^{k} = \sum\_{i=1}^{N\_i} \mathbf{W}\_{\mathbf{sp}}^{k,i} \\\\ \mathbf{W}\_{oil}^{k} = \sum\_{i=1}^{N\_i} \mathbf{P}^{k,i} \left( T^{k,i} - T\_{\mathbf{sp}}^{k,i} \right) \\\\ \mathbf{W}^{k} = \mathbf{W}\_{\mathbf{sp}}^{k} + \mathbf{W}\_{oil}^{k} \end{cases} \tag{1}$$

where *k* denotes the year; *i* denotes the type of ships; *P k*,*i* is the power of the ship's auxiliary engine, kWh; *W<sup>k</sup> sp* is the power consumed by the ship using shore power, kWh; *W<sup>k</sup> oil* is the power consumed by the ship using fuel oil, kWh; *W<sup>k</sup>* is the total power consumption of the ship, kW h; *T k*,*i* is the berthing time of the ship, h; *T k*,*i sp* is the time when the ship uses shore power, h.

$$\mathsf{C}\_{dj}^{k} = \mathsf{C}\_{\text{grid}}^{k} + \mathsf{C}\_{\text{serve}}^{k} \tag{2}$$

$$\begin{cases} \mathsf{C}\_{\text{e}}^{k} = \mathsf{W}\_{\text{sp}}^{k} \mathsf{C}\_{\text{grid}}^{k} \\\\ \mathsf{C}\_{\text{sp}}^{k} = \mathsf{W}\_{\text{sp}}^{k} \mathsf{C}\_{dj}^{k} \\\\ T\_{\text{e}}^{k} = \mathsf{C}\_{\text{e}}^{k} / (1 + 16\%) \times 16\% \\\\ T\_{\text{s}}^{k} = \mathsf{W}\_{\text{sp}}^{k} \mathsf{C}\_{\text{serve}}^{k} / (1 + 6\%) \times 6\% \end{cases} \tag{3}$$

where *C k grid* is the electricity basic price, CMY/kWh; *C k serve* is the electricity service price, CMY/kWh; *C k dj* is the electricity actual price, CMY/kWh; *C k e* is the cost of purchasing electricity for the port, CMY; *C k sp* is the ship paying the port for electricity, CMY; *T k e* is to pay value-added tax to the government as the power grid provides electricity to the port, CMY; *T k s* is the port that makes a profit from providing shore power service to the ship and pays value-added tax to the government, CMY.

$$\begin{cases} \mathsf{C}\_{oil,sp}^{k} = 10^{-6} \mathsf{W}\_{sp}^{k} E\_{1} \mathsf{C}\_{per-oil} \\\\ \mathsf{C}\_{oil,oil}^{k} = 10^{-6} \mathsf{W}\_{oil}^{k} E\_{1} \mathsf{C}\_{per-oil} \end{cases} \tag{4}$$

where *C k oil*,*sp* represents the fuel cost savings by the ship using shore power, CMY; *C k oil*,*oil* represents the cost of fuel oil used by the ship, CMY; *E*<sup>1</sup> is the fuel consumption per unit of electricity emitted by the auxiliary engine, g/kWh; *<sup>C</sup>per*−*oil* is the price of marine fuel oil, CMY/Mt.

$$\begin{cases} \begin{aligned} T\_{ep,sp}^k &= 10^{-3} \text{C}\_{dl} \mathcal{W}\_{sp}^k \sum\_{n=1}^{N\_n} \text{F}\_n / E\_{dl}^n \\\ T\_{ep,oil}^k &= 10^{-3} \text{C}\_{dl} \mathcal{W}\_{oil}^k \sum\_{n=1}^{N\_n} \text{F}\_n / E\_{dl}^n \end{aligned} \end{cases} \tag{5}$$

where *n* denotes the type of pollutant; *T k ep*,*sp* is the environmental protection tax saved by ships using shore power, CMY; *T k ep*,*oil* is the environmental protection tax paid by ships using fuel oil, CMY; *Cdl* is the pollution factor pollutant discharge fee standard per unit of pollution equivalent, CMY/equivalent; *F<sup>n</sup>* is the emission factor of pollutants discharged from the fuel oil of the ship's auxiliary engine, g/kWh; *E n dl* is the pollution equivalent value of pollutants, kg.

$$\begin{cases} \begin{aligned} V\_{sp}^k &= 10^{-6} \mathcal{W}\_{sp}^k F\_\varepsilon \\ V\_{oil}^k &= 10^{-6} \mathcal{W}\_{oil}^k F\_\varepsilon \\ V\_{actual}^k &= V\_{sp}^k + V\_{oil}^k \end{aligned} \tag{6}$$

$$\begin{cases} F\_{\rm cf}^k = 10^6 V\_{\rm actual}^k / W^k \\\\ F\_{\rm wef}^k = 10^6 \frac{\Sigma \left( V\_{\rm actual}^{k-3} + V\_{\rm actual}^{k-2} + V\_{\rm actual}^{k-1} \right)}{\Sigma \left( W^{k-3} + W^{k-2} + W^{k-1} \right)} \end{cases} \tag{7}$$

where *V k sp* and *V k oil* are the CO<sup>2</sup> emissions of the ship using shore power and fuel oil, respectively, Mt; *V k actual* is the total CO<sup>2</sup> emissions of the ship, Mt; *F<sup>e</sup>* is the annual average power supply emission factor of the regional power grid, g/kWh; *F<sup>c</sup>* is the emission factor of carbon dioxide pollutants emitted by marine auxiliary engine fuel oil, g/kWh; *F k cef* is the comprehensive CO<sup>2</sup> emission factor of the ship, g/kWh; *F k wef* is the ship's weighted CO<sup>2</sup> emission factor, g/kWh. Equation (6) gives the calculation of the carbon emissions from the use of shore power and the use of oil, respectively, as well as the total emissions for the year. In Equation (7), the ship CO<sup>2</sup> weighted emission factor is taken as the weighted average of the carbon emissions per unit of business of the port in the previous three years, which is used as the base of the historical carbon emission intensity.

$$\begin{cases} \begin{aligned} \boldsymbol{V}\_{\text{cap}}^{k} &= 10^{-6} \boldsymbol{\eta} \boldsymbol{F}\_{\text{unf}}^{k} \boldsymbol{W}^{k} \\ \boldsymbol{T}\_{\text{c1}}^{k} &= \left( \boldsymbol{V}\_{\text{actual}}^{k} - \boldsymbol{V}\_{\text{cap}}^{k} \right) \times \boldsymbol{P}\_{\text{c1}}^{k} \end{aligned} \tag{8}$$

where *V k quato* is the port's carbon emission cap, Mt; *η* is the annual decline coefficient, taking 1; *T k* c1 is the total amount of carbon trading, CMY; *P k c*1 is the carbon price in the carbon trading market, CMY/Mt CO2. Since only the power sector is currently subject to government regulation and other sectors are not yet subject to excessive restrictions, the annual decline coefficient in Equation (8) is set to 1. This gives the port room to strengthen its efforts to reduce carbon emissions. Carbon emission allowances for the year were calculated, as well as the fees paid in excess of the allowances.

$$T\_{c2}^{k} = V\_{actual}^{k} \times P\_{c2}^{k} \tag{9}$$

where *T k* c2 is the carbon emission tax paid by the port, CMY; *P k c*2 is the carbon tax price, CMY/Mt CO2.

$$T^k\_{\mu} = \mathcal{W}^k\_{sp} / \mathcal{S}\omega \cos \varphi \tag{10}$$

where *T k u* is the annual utilization hours, h; *S* is the total installed capacity of the shore power system, kVA; *ω* means that there is a certain margin between the actual use of the shore power capacity of the terminal and the planned construction capacity of shore power; cos *ϕ* is the comprehensive power factor of the shore power frequency conversion equipment and the ship's load. We experimentally set *ω* = 0.8 and cos *ϕ* = 0.7. *ω* is needed to make sure that the shore power equipment is working properly under the load. Part of the shore power equipment must be suspended during equipment repairs and maintenance in order to prolong the equipment's useful life. cos *ϕ* is related to the nature of the load: different devices have different power factors, so, here, the integrated power is used instead. When *W<sup>k</sup> sp* and *S* are fixed, obviously, the larger *ω* and cos *ϕ*, the smaller *T k u* .

#### **3. Game Research under Two Carbon Mechanism**

#### *3.1. Game Research on the Use of Shore Power under Cap and Trade*

Figure 2 explains the three-way game model in the case of cap and trade. The government defines carbon emission rights as a commodity and establishes carbon emission caps for ports. By incorporating the value of carbon emissions into port costs, ports can directly buy or sell allowances in the carbon trading market. Ships can voluntarily choose to use shore electricity while they are docked in port and the amount of time they utilize it has been rising every year.

**Figure 2.** Tripartite benefit analysis under cap and trade.

1. Government Benefit Analysis Model

$$\begin{cases} \begin{aligned} B\_{\mathcal{S}} &= T\_{\mathcal{e}} + T\_{\mathcal{s}} + T\_{\mathcal{ep},sp} + T\_{\mathcal{ep},\rho\mathcal{il}} \\ &\mathcal{C}\_{\mathcal{S}} = \mathcal{C}\_1 \mathfrak{a}\_1 + \mathcal{C}\_2 \mathfrak{a}\_2 + T\_{\mathcal{ep},\rho\mathcal{il}} \\ &F\_{\mathcal{S}} = B\_{\mathcal{S}} - \mathcal{C}\_{\mathcal{S}} \end{aligned} \end{cases} \tag{11}$$

where *B<sup>g</sup>* is the government income; *C<sup>g</sup>* is the government cost; *F<sup>g</sup>* is the government profit; *α*<sup>1</sup> and *α*<sup>2</sup> are the subsidy rates for shore power equipment transformation given by the government to the port and the ship, respectively; *C*<sup>1</sup> and *C*<sup>2</sup> are the transformation costs of the port and the ship, respectively, CMY.

2. Port Benefit Analysis Model

$$\begin{cases} \mathcal{B}\_p = \mathsf{C}\_{sp} + \mathsf{C}\_1 \mathsf{a}\_1 \\\\ \mathsf{C}\_p = \mathsf{C}\_1 + \mathsf{C}\_\varepsilon + T\_s + T\_{c1} \\\\ F\_p = B\_p - \mathsf{C}\_p \end{cases} \tag{12}$$

where *B<sup>p</sup>* is the port income; *C<sup>p</sup>* is the port cost; *F<sup>p</sup>* is the port profit.

3. Ship Benefit Analysis Model

$$\begin{cases} \text{ } B\_{\text{s}} = \text{C}\_{2}\text{a}\_{2} + \text{C}\_{\text{oil},sp} + T\_{ep,sp} \\\\ \text{ } \text{C}\_{\text{s}} = \text{C}\_{sp} + \text{C}\_{2} + \text{C}\_{\text{oil},oil} + T\_{ep,oil} \\\\ F\_{\text{s}} = B\_{\text{s}} - \text{C}\_{\text{s}} \end{cases} \tag{13}$$

where *B<sup>s</sup>* is the ship income; *C<sup>s</sup>* is the ship cost; *F<sup>s</sup>* is the ship profit.

$$SW\_1 = F\_p + F\_s + T\_e + T\_s - T\_{c1} \tag{14}$$

where *SW*<sup>1</sup> refers to social welfare, including consumer surplus, producer surplus, government tax, and environmental benefits [31]. *Tep*,*sp* represents the environmental benefit, which not only reduces the environmental protection fee levied on the ship but also is included in the government income. Note that social welfare is only counted once and does not accumulate repeatedly.

The use of shore power is abandoned after the ship converts to shore power owing to the high price. In the worst case, all ships consume fuel for power supply. In this way:

$$\begin{cases} \mathsf{B}\_{\mathrm{s},\rho\mathrm{il}} = \mathsf{C}\_{2}\mathsf{a}\_{2} \\ \mathsf{C}\_{\mathrm{s},\rho\mathrm{il}} = \mathsf{C}\_{\mathrm{oil},\rho\mathrm{il}} + \mathsf{C}\_{\mathrm{oil},\mathrm{sp}} + T\_{\mathrm{sp},\rho\mathrm{il}} + T\_{\mathrm{ep},\mathrm{sp}} + \mathsf{C}\_{2} \\ \quad F\_{\mathrm{s},\rho\mathrm{il}} = B\_{\mathrm{s},\rho\mathrm{il}} - \mathsf{C}\_{\mathrm{s},\rho\mathrm{il}} \end{cases} \tag{15}$$

where *Bs*,*oil* is the ship income; *Cs*,*oil* is the ship cost; *Fs*,*oil* is the ship profit.

When some ships use shore power, the benefits of the entire ship are improved:

$$F\_{\rm s,save} = F\_{\rm s} - F\_{\rm s,oil} \tag{16}$$

When the ships use all fuel oil, the costs come from the fuel costs and the environmental protection taxes paid. The income comes from freight and is unincluded in the scope of the three-party game, so the ship's benefit is negative, which is similar to social welfare.

#### *3.2. Game Researches on the Use of Shore Power under the Carbon Tax Policy*

Figure 3 illustrates the three-way game model in the case of carbon tax policy. The government sets the carbon tax rate, and ports must offer the government a carbon tax for every metric ton of carbon dioxide they emit.

**Figure 3.** Tripartite benefit analysis under carbon tax policy.

1. Government Benefit Analysis Model

$$\begin{cases} \mathcal{B}\_{\mathcal{S}} = T\_{\varepsilon} + T\_{\text{s}} + T\_{c2} + T\_{ep,sp} + T\_{ep,oil} \\\\ \mathcal{C}\_{\mathcal{S}} = \mathcal{C}\_{1}\mathfrak{a}\_{1} + \mathcal{C}\_{2}\mathfrak{a}\_{2} + T\_{ep,oil} \\\\ F\_{\mathcal{S}} = B\_{\mathcal{S}} - \mathcal{C}\_{\mathcal{S}} \end{cases} \tag{17}$$

where *B<sup>g</sup>* is the government income; *C<sup>g</sup>* is the government cost; *F<sup>g</sup>* is the government profit.

2. Port Benefit Analysis Model

 *B<sup>p</sup>* = *Csp* + *C*1*α*<sup>1</sup> *C<sup>p</sup>* = *C*<sup>1</sup> + *C<sup>e</sup>* + *T<sup>s</sup>* + *Tc*<sup>2</sup> *F<sup>p</sup>* = *B<sup>p</sup>* − *C<sup>p</sup>* (18)

where *B<sup>p</sup>* is the port income; *C<sup>p</sup>* is the port cost; *F<sup>p</sup>* is the port profit.

3. Ship Benefit Analysis Model

$$\begin{cases} \text{ } B\_s = \text{C}\_2\text{a}\_2 + \text{C}\_{oil,sp} + T\_{ep,sp} \\\\ \text{ } \text{C}\_s = \text{C}\_{sp} + \text{C}\_2 + \text{C}\_{oil,oil} + T\_{ep,oil} \\\\ F\_s = B\_s - \text{C}\_s \end{cases} \tag{19}$$

where *B<sup>s</sup>* is the ship income; *C<sup>s</sup>* is the ship cost; *F<sup>s</sup>* is the ship profit.

Social welfare is given by:

$$SW\_2 = F\_p + F\_s + T\_\varepsilon + T\_s \tag{20}$$

If all types of ships are powered by fuel oil, the benefit analysis of the ship is as follows:

$$\begin{cases} \mathsf{B}\_{\mathrm{s},\rho\mathrm{il}} = \mathsf{C}\_{2}\mathsf{a}\_{2} \\ \mathsf{C}\_{\mathrm{s},\rho\mathrm{il}} = \mathsf{C}\_{\mathrm{oil},\rho\mathrm{il}} + \mathsf{C}\_{\mathrm{oil},\mathrm{sp}} + T\_{\mathrm{ep},\rho\mathrm{il}} + T\_{\mathrm{ep},\mathrm{sp}} + \mathsf{C}\_{2} \\ \quad \mathsf{F}\_{\mathrm{s},\rho\mathrm{il}} = \mathsf{B}\_{\mathrm{s},\rho\mathrm{il}} - \mathsf{C}\_{\mathrm{s},\rho\mathrm{il}} \end{cases} \tag{21}$$

where *Bs*,*oil* is the ship income; *Cs*,*oil* is the ship cost; *Fs*,*oil* is the ship profit.

Some ships are converted from oil to electricity, and the overall benefit of the ship is improved:

$$F\_{\rm s,save} = F\_{\rm s} - F\_{\rm s,oil} \tag{22}$$

#### **4. Data Selection**

#### *4.1. Government Data Acquisition*

According to the "Interim Measures for the Management of Subsidy Funds for Ports, Ship Shore Power Facilities and Marine Low Sulfur Oil Subsidy Funds in Shenzhen", the subsidy will be provided for the reconstruction of port shore power facilities, which will not exceed 30% of the project construction costs [32].

#### *4.2. Port Data Acquisition*

According to the survey, a port has built 14 sets of shore power systems, covering a total of 23 berths. The installed capacity of the shore power system has reached 11,600 kVA. Including the equipment purchase fee and construction installation fee, the investment and renovation costs of the power equipment are CMY55,144,134. Taking the service life as 30 years, the interest rate of the annualized cost is 8%, which is equivalent to the annual renovation cost:

$$\mathcal{C}\_1 = \frac{8\% \times (1 + 8\%)^{30}}{(1 + 8\%)^{30} - 1} \times 55, 144, 134 = 4,898, 312 \text{ CMY}$$

#### *4.3. Ship Data Acquisition*

The investment and transformation cost of onboard electrical equipment, referring to the report of the European Commission Environment Directorate (ECDGE) [33], converted into unit power is 1530 CMY/kW.

The berthing time of the ship in the port, the length of the use of shore power in a certain year, and the electricity consumed by the use of shore power are shown in Figure 4.

**Figure 4.** Ship information.

#### *4.4. Electricity Price Acquisition*

According to the "Notice on Clarifying the Electricity Price and Service Price of Ship's Shore-based Power Supply Facilities" of the price bureau of Jiangsu province, *C k grid* takes the electricity price of large industrial electricity at 0.6601 CMY/kWh. The maximum price of shore power used by ships in Taizhou is 1.20 CMY/kWh [34].

In order to standardize the accounting of carbon dioxide emissions implied by electricity consumption by regions, industries, enterprises, and other units, and to ensure comparability of results, the government organized a study to determine the average carbon dioxide emission factor for regional power grids in China. It refers to the carbon emissions generated by one unit of electricity used in the grid, and is obtained by dividing the total emissions of the entire grid by the total electricity generation. As the port study is in the southern region, the average CO<sup>2</sup> emission factor for the southern regional grid was used. *F<sup>e</sup>* is the annual average power supply emission factor of the regional power grid, taking 527.1 g/kWh.

#### *4.5. Pollutant Data Acquisition*

According to the literature [33], *E*<sup>1</sup> is taken as 213 g/kWh. According to the "2020 Implementation Plan for the Global Sulfur Restriction Order for Marine Fuel Oil" issued by the China Maritime Safety Administration, those entering the country's inland river ships' air pollutant emission control areas should use fuel oil with a sulfur content of no more than 0.10% [35]. *Cper*−*oil* was taken as 3800 CMY/Mt.

According to the "Decision of the Standing Committee of the Jiangsu Provincial People's Congress on the Applicable Tax Amount of Environmental Protection Tax for Air Pollutants and Water Pollutants", the tax rate in Nanjing is CMY8.4 per pollution equivalent of air pollutants [36].

The annual emission factors and pollution equivalent values of various pollutants caused by marine auxiliary engine fuel are shown in Table 1 [37,38].


**Table 1.** Annual emission factors and pollution equivalent values of pollutants.

#### *4.6. Carbon Price Data Acquisition*

The carbon price data from 2020 to 2050 comes from the "2020 China Carbon Price Survey" [39]. Figure 5 reveals that there has been a steady increase in the carbon price in China since 2020.

**Figure 5.** The expected price of the national carbon emissions trading market in 2021–2050.

#### **5. The Impact of Various Factors on the Benefits of Each Party**

*5.1. Comparison of the Impact of Subsidy Rates*

Based on the maximization of social welfare, we explore the impact of subsidy rates under two carbon mechanisms on the optimal price and the benefits to all parties. Restrictions:

$$\begin{cases} \ 0 \le a\_1 \le 30\% \quad 0 \le a\_2 \le 30\% \quad \mathcal{C}\_{\text{service}} \ge 0 \quad 0 \le \mathcal{C}\_{d\text{j}} \le 1.2\\\ \quad F\_{\mathcal{S}} \ge 0 \quad F\_p \ge 0 \quad F\_{\text{s.save}} \ge 0 \end{cases} \tag{23}$$

Objective function:

$$\text{max.}SW\tag{24}$$

Tables 2 and 3 state the impact of the subsidy rate on the economic and environmental benefits of each party in the three-way game model under the two carbon regimes, respectively.


**Table 2.** Comparison of the impact of subsidy rates on economic benefits to all parties under the two carbon regimes.

**Table 3.** Comparison of the impact of subsidy rates on environmental benefits to all parties under the two carbon regimes.


As can be seen from Figure 6a, whether under the cap-and-trade or the carbon tax policy, social welfare decreases when the electricity service price rises with roughly a linear negative correlation between the two. The cost of power supply services is rising, which does not promote social welfare, and there is clear resistance to the use of shore power. In addition, the lowest value of social welfare under cap and trade is much higher than the maximum value under a carbon tax policy. As a result, cap and trade offers a significant benefit in this case and merits consideration. From Figure 6b–e, we can observe that the subsidy rate a1 affects social welfare, port benefit, and ship benefit to a larger extent than a2 affects all three. The social welfare and ship benefits rise as the subsidy rate a1 rises. The effects of a1 and a2 on government benefits are identical. Government benefits rise with an increase in a2, while they decline with an increase in a1. A comparison of the two results reveals that the cap-and-trade group reported far more social welfare and ship benefits than the other one. On the contrary, the government benefit and port benefit under cap and trade are in every case short of what they are under the other system.

**Figure 6.** The relationship between subsidies and the benefits to all parties. (**a**) The relationship between electricity service price and social welfare; (**b**) The relationship between subsidy and social **Figure 6.** The relationship between subsidies and the benefits to all parties. (**a**) The relationship between electricity service price and social welfare; (**b**) The relationship between subsidy and social welfare; (**c**) The relationship between subsidy and government benefit; (**d**) The relationship between subsidy and port benefit; (**e**) The relationship between subsidy and ship benefit.

than the maximum value under a carbon tax policy. As a result, cap and trade offers a significant benefit in this case and merits consideration. From Figure 6b–e, we can observe that the subsidy rate a1 affects social welfare, port benefit, and ship benefit to a larger extent than a2 affects all three. The social welfare and ship benefits rise as the subsidy rate a1 rises. The effects of a1 and a2 on government benefits are identical. Government benefits rise with an increase in a2, while they decline with an increase in a1. A comparison of the two results reveals that the cap-and-trade group reported far more social welfare and ship benefits than the other one. On the contrary, the government benefit and port benefit under cap and trade are in every case short of what they are under the other system.

#### *5.2. Carbon Price Impact Comparisons*

We evaluated how the price of carbon affects the best price and the gains for all parties under two carbon mechanisms based on the maximization of social welfare.

Restrictions:

$$\begin{cases} \begin{array}{ll} 20 \le P\_{\mathcal{C}} \le 100 & \mathcal{C}\_{\text{server}} \ge 0 & 0 \le \mathcal{C}\_{dj} \le 1.2 \end{array} \\\\ F\_{\mathcal{S}} \ge 0 & F\_{p} \ge 0 & F\_{\text{s.save}} \ge 0 \end{cases} \tag{25}$$

Objective function:

#### max.*SW* (26)

Tables 4 and 5 indicate the impact of the carbon price on the economic and environmental benefits of each party in the three-way game model under the two carbon regimes, respectively.

**Table 4.** Comparison of the impact of carbon price on economic benefits to all parties under the two carbon regimes.


**Table 5.** Comparison of the impact of carbon price on environmental benefits to all parties under the two carbon regimes.


Figure 7 compares the outcomes gained from the analysis of the two carbon mechanisms. Under both the cap-and-trade and carbon tax policies, social welfare decreases as electricity service prices rise. The social welfare under cap and trade is always greater than that under the carbon tax policy. Figure 7b–e shows that the price of carbon has a significant impact on social welfare, government benefits, port benefits, and ship benefits under the carbon tax policy. The four are not significantly impacted by the carbon price under cap and trade. According to the cap-and-trade model, while government benefits decline as the price of carbon rises, social welfare and ship benefits increase. The carbon tax policy states that when carbon prices rise, the government gains more advantages whereas social welfare and ship benefits decline. Social welfare and ship benefits under cap and trade tend to be invariably greater than those under the carbon tax approach. In comparison to a carbon tax, the government's benefit under cap and trade is always less. The port efficiency fluctuates positively around 0.

**Figure 7.** The relationship between carbon price and the benefits to all parties. (**a**) The relationship between electricity service price and social welfare; (**b**) The relationship between carbon price and social welfare; (**c**) The relationship between carbon price and government benefit; (**d**) The relationship between carbon price and port benefit; (**e**) The relationship between carbon price and ship benefit.

#### *5.3. Comparison of Time-Proportional Effects of Using Shore Power*

This study investigates the impact of the ratio of time spent using shore power to total docking time on the best pricing and the gains for all parties under two carbon mechanisms, based on the maximization of total social welfare.

$$T\_{\mu}^{k} = \lambda \sum\_{i=1}^{N\_{\bar{i}}} T^{k,i} P^{k,i} / S\omega \cos \varphi \tag{27}$$

where *λ* is the ratio of using shore power to the berthing time. As the total port call time is constant, *T k <sup>u</sup>* and *λ* are proportional. *T k <sup>u</sup>* decreases, so *λ* decreases, which affects all aspects of electricity prices, social welfare, government benefits, port benefits, ship benefits, etc. Restrictions:

*Cserve* ≥ 0 0 ≤ *Cdj* ≤ 1.2 *T<sup>u</sup>* ≤ 8760 (28)

Objective function:

$$\max.SW\tag{29}$$

Tables 6 and 7 explicate the impact of the time-proportional on the economic and environmental benefits of each party in the three-way game model under the two carbon regimes, respectively.

**Table 6.** Comparison of the impact of time-proportional on economic benefits to all parties under the two carbon regimes.


**Table 7.** Comparison of the impact of time-proportional on environmental benefits to all parties under the two carbon regimes.


What stands out in Figure 8a is the price of electricity supply service changes regularly between 0 and 0.5399 CMY/kWh, and social welfare is also changing. The service price must be 0.5399/kWh under cap and trade in order to reach its maximum value, and it must be 0 under the carbon tax policy in order to reach its minimum value. In Figure 8b–d, the similarity between the two carbon mechanisms is highlighted above. The overall level of social welfare rises as the ratio of time spent utilizing shore power to docking time grows. Under the two carbon mechanisms, the change curves for overall social welfare, government benefits, and port benefits are comparable and can be attained by moving up and down. The curve of ship benefit with the time proportion of using shore power is exactly the same in Figure 8e. With the share of time spent utilizing shore power increasing, the overall level of social welfare rises monotonically. However, government benefits, port benefits, and ship benefits show an upward trend with the increase in the proportion of time when shore power is used.

**Figure 8.** The relationship between time proportion and the benefits to all parties. (**a**) The relationship between electricity service price and social welfare; (**b**) The relationship between time proportion and social welfare; (**c**) The relationship between time proportion and government benefit; (**d**) The relationship between time proportion and port benefit; (**e**) The relationship between time proportion and ship benefit.

#### **6. Conclusions**

The two systems of cap and trade and carbon tax are clear and easy to understand, but too many similar parameters appear in the modeling process, leading to easy confusion. In both cap and trade and carbon tax, the specific components of tripartite benefits and social welfare are different and are simply represented by the same parameters, and the specific values change as the electricity service price, carbon tax, and the ratio of using shore power to the berthing time change. Also note that the government is happy to see and promote the use of shore power, which is in line with the plan to reduce carbon emissions; the port is to take the responsibility of a state-owned enterprise, respond to the national policy, take responsibility for emission reduction and establish a good image among the public. Ships have the least public pressure and social responsibility, and they are oriented by economic interests, so they are more unstable, and they need government guidance and support because they have to face technical and economic difficulties in the "oil-to-electricity" conversion. In the model, we should focus on understanding the cost saving of some ships after "oil-to-electricity" conversion, which is the key to decide whether ships should insist on using shore power. There is a difference between the CO<sup>2</sup> emitted from using oil to meet the power and the CO<sup>2</sup> emitted from switching to shore power to replace this power, and the focus is on calculating the difference and the accompanying economic and environmental benefits.

Depending on the port's yearly business volume and historical carbon emission intensity base, the government calculates the annual carbon cap for the port. The comprehensive CO<sup>2</sup> emission factor of ships in the first three years is greater than the current year's CO<sup>2</sup> emission factor. Therefore, the carbon cap is frequently higher than the actual CO<sup>2</sup> emission, according to calculations of the port's real energy usage. Currently, compared to other businesses, the government's criteria for reducing carbon emissions in the maritime sector are not stringent enough.

In summary, these results indicate that the annual hours of shore power facility use, the amount of environmental protection tax saved by utilizing shore power, and the reduction in pollutants and carbon dioxide are all equivalent as long as the ships' energy consumption is constant. The social welfare, governmental benefits, port benefits, and ship benefits appear to vary depending on the subsidy rate, carbon price, and percentage of time using shore power.

It is reasonable for the port to set the shore electricity price within the range of 0.6601 to 1.2 CMY/kWh, taking into consideration both its own transformation costs and government subsidies. As long as ships use electricity instead of fuel, the economic benefits will be significantly improved, no matter what changes in various carbon mechanisms and influencing factors. The primary source of improvement may partly be related to reduced fuel costs. The government's benefits are always greater than zero, mostly due to the environmental benefits of decreasing pollutant emissions. Low economic advantages are expected for the port, as a result of high costs of self-renovation and government limits on the actual electricity price.

The major limitation of this study is that ships are assumed to be equivalent to a virtual ship although they have distinct types and numbers when in berth. It is a macro analysis for multiagent games; hence, it cannot accurately reflect the benefit to an individual ship. The three-party game is dynamic and played repeatedly, and ships are allowed to choose whether to utilize shore power or not, which makes it challenging to calculate the best electricity price. By examining the status of the tripartite game in electricity substitution, this study offers a particular reference value for the cycle planning and benefit distribution of electric energy replacement projects in the future.

**Author Contributions:** Conceptualization, Y.H. and Y.Z.; methodology, Y.H.; software, Y.H.; validation, Y.H.; formal analysis, Y.H.; investigation, Y.H.; resources, Y.Z.; data curation, Y.H.; writing—original draft preparation, Y.H.; writing—review and editing, Y.Z.; visualization, Y.H.; supervision, Y.Z.; project

administration, Y.Z.; funding acquisition, Y.Z. All authors have read and agreed to the published version of the manuscript.

**Funding:** Guangxi Special Fund for Innovation-Driven Development (AA19254034).

**Data Availability Statement:** Publicly available datasets were analyzed in this study. These data can be found in the References section.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
