The Efficiency of Economic Performance, Electricity Consumption, and Environmental Pollutants in Taiwan

Finding the balance between economic development and environmental protection is a major problem for many countries around the world. Air pollution caused by economic growth has caused serious damage to humans’ living environment, and as improving energy and resource eﬃciencies is the ﬁrst priority, many countries are targeting to move towards a sustainable environment and economic development. This study uses the modiﬁed dynamic SBM (slack-based measure) model to explore the economic eﬃciency and air pollutants emission eﬃciency in Taiwan’s counties and cities from 2012 to 2015 by taking labor, motor vehicles, and electricity consumption as inputs and average disposable income as output. Particulate matter (PM 2.5 ), nitrogen oxide emissions (NO 2 ), and sulfur oxide emissions (SO 2 ) are undesirable outputs, whereas factory ﬁxed assets are a carry-over variable, and the results show the following: (1) the regions with the best overall eﬃciency between 2012 and 2015 include Taipei City, Keelung City, Hsinchu City, Chiayi City, and Taitung County; (2) in counties and cities with poor overall eﬃciency performance, the average disposable income per household has no signiﬁcant relationship with air pollutant emissions; (3) in counties and cities where overall eﬃciency is poor, the average eﬃciency of each household’s disposable income is small; and (4) except for the ﬁve counties and cities with the best overall performance, the three air pollutants in the other fourteen counties and cities are high. Overall, the air pollution of most areas needs improvement.


Introduction
Taiwan, one of the Asian four dragons, has high energy (electricity) consumption, population, and vehicle density and severe air pollution. is study is going to explore the economic performance efficiency, energy consumption (electricity), and air pollutant emission efficiency of Taiwan.
From the World Health Organization's [1] national ranking of PM 2.5 concentrations in September 2011, Taiwan ranks 32nd among 38 survey countries. Among nearly 600 cities worldwide, Chiayi and Kaohsiung made it among the top ten. From the average concentration of PM 2.5 in 2013, the risk of lung cancer and asthma in children increased to 15%, with the risk from stroke, heart disease, and chronic respiratory disease increasing by 25%. In 2014, more than 6,000 deaths in Taiwan were caused by exposure to PM 2.5 .
Indeed, PM 2.5 causes damage in Taiwan. e impact of CO 2 , SO 2 , and PM 2.5 cannot be overlooked. Most studies in the literature explore the effects of energy and environmental efficiencies on CO 2 , SO 2 , and NO 2 emissions. Many researches analyze the energy efficiency of China. Wu et al. [2] use two-stage network DEA (data envelopment analysis) to assess China's energy conservation and emission reduction efficiency during 2006-2010. Energy saving and emission reduction in the eastern region are better than in the central and western regions. Lin and Du [3] employ the new nonradial directional distance function to assess regional energy and carbon dioxide emissions efficiency in China from 1997 to 2009. e results show that most of China's performances in energy use and carbon dioxide emissions are poor. Industrial sector expansion is negatively correlated with China's regional energy and CO 2 emissions performance. Wang et al. [4] utilize multidirectional efficiency analysis (MEA) to look at regional energy and emissions efficiencies in China from 1997 to 2010. e eastern region is more efficient than the central and western regions. Hebei, Shanxi, Inner Mongolia, Shandong, Henan, and Hubei have higher potentials for energy conservation and emission reduction. Li et al. [5] collect energy data from 2000 to 2009 in China and analyze the impact of three internal factors (economic structure, energy consumption structure, and technological progress) on energy intensity in China using the DEA-based Malmquist method. ey convert technology into three components to see the different impacts in various regions. Other researches such as [6][7][8][9][10][11][12][13][14][15][16][17][18][19] also focus on the energy efficiency of China.
Some in the literature analyze the impacts of energy and environmental efficiencies on PM 2.5 emissions, such as [20][21][22][23][24][25][26][27][28][29][30]. Martínez [20] uses two-stage DEA to assess the energy efficiency of non-energy-intensive industries (NEISs) in Germany and Colombia from 1998 to 2005. e highest energy efficiency in non-energy-intensive industries (NEISs) in Colombia is derived from the cost minimization model, showing that energy prices are not the key to improving energy efficiency. Sueyoshi and Yuan [21] utilize the DEA model to explore regional environmental efficiency performance in China from 2013 to 2014. e Chinese government should allocate economic resources to cities located in the northwestern region (including Lanzhou, Xining, Yinchuan, and Urumqi) and strengthen stricter regulation of energy consumption in major urban environments (such as Beijing, Tianjin, Shanghai, and Chongqing). Ma et al. [22] use the spatial autoregressive model to analyze the spatial diffusion effects of PM 2.5 in 152 cities in China. PM 2.5 is significantly affected by geospatial and regional economies. Li et al. [23] utilize the multilevel frontiers DEA model to explore the environmental efficiency of 49 cities in China.
eir results present that PM 2.5 and SO 2 emissions are significantly related to urban population and energy technologies.
ere are two contributions of this study. First, we use small economy as the research sample. As can be seen from the above literature, most of the research on air pollutants is based on large economies, such as China.
ese large economies have rich natural resources and focus on industrial and manufacturing development. However, the problem of air pollution is not limited to large economies, and it cannot be overlooked in some non-industrial-oriented small economies. For example, according to the Taiwan Environmental Protection Agency's 2016 and 2017 Air Quality Monitoring Report [31], the annual air quality indicators (AQIs) hit 39.34% and 42.1%, respectively, or out of reach from a good grade of 50%.
e other contribution is model modification.
Most past models are still dominated by radial (Charnes, Cooper, and Rhodes model, abbreviation as CCR model; Banker, Charnes, and Cooper model, abbreviation as BCC model), nonradial (slack-based measure, abbreviation as SBM), two-stage DEA analysis, and directional distance function. However, these models employ static analysis, lack dynamic considerations, and cannot understand the changes in efficiency of energy and environmental pollutants. us, this study employs the modified dynamic SBM to evaluate the situation for each county and city. We utilize 19 counties and cities in Taiwan from 2012 to 2015 with data on the number of employed people, motor vehicles, and electricity consumption and take the average disposable income per household as output, PM 2.5 , nitrogen oxide emissions (NO 2 ), and sulfur oxide emissions (SO 2 ) as undesirable outputs (recently, the problem of air pollution has drawn the attention of many scholars; because the issue of CO 2 emission has been analyzed by many researches, this study focuses on the other air pollutants (SO 2 , NO 2 , and PM 2.5 in Taiwan), and fixed assets as the carry-over variable). By above input and output variables, this study evaluates the economic performance, electricity consumption efficiency, and air pollutant emission efficiency of Taiwan.  [37] then offer a new analysis of the dynamic impact of consecutive activities. Chen [38] and K. S. Park and K. Park [39] subsequently present SBM studies of several dynamics, with the dynamic analysis model extended into a slack-based measure by Tone and Tsutsui [40]. In order to carry-over activities as a form of connectivity, they propose the SBM (slack-based measures) dynamic DEA model. Tone and Tsutsui [40] develop the model into SBM dynamic analysis, with carry-over activities as a link, and the existence of activities divided into a four-model analysis: (1) desirable (Z good ); (2) undesirable (Z bad ); (3) discretionary (Z free ); (4) nondiscretionary (Z fix ), with carry-over variables from period t to period t + 1.

Research Methods
e following is the nonoriented model: Equation (2) is the connection equation between t and t + 1.
e most efficient solution is where Y g is the desirable output, Y b is the undesirable output, and Z good is carried over from period t to period t + 1. e following is the nonoriented model: e following six equations show the connection equation between t and t + 1: e most efficient solution is In equation (5) Hu and Wang [19] total-factor energy efficiency index is used to overcome any possible bias in the traditional energy efficiency indicator. For each specific evaluated country, we calculate the number of motor vehicles, electricity consumption, average disposable income, and NO 2 , SO 2 , and PM 2.5 efficiencies from the following equations: the number of motor vehicles' efficiency � target motor vehicles' efficiency input (i, t) actual motor vehicles' efficiency input (i, t) , electricity consumption efficiency � target electricity consumption input (i, t) actual electricity consumption input (i, t) , average disposable efficiency � actual average disposible income output(i, t) target average disposible output(i, t) , e efficiency index indicates the ratio of target value and actual value. e target value indicates the most efficient value. us, the efficiency index denotes the difference of actual vale and target value. e index (ratio) equals to 1 when the actual value reaches the target value, and the actual value is most efficient. e index is more efficient when the value is close to 1.
If the target motor vehicle number and electricity consumption input equal the actual inputs and the NO 2 , PM 2.5 , and SO 2 outputs equal the actual undesirable outputs, then the motor vehicle number, electricity consumption, and NO 2 , PM 2.5 , and SO 2 efficiencies equal 1, indicating no room for improvement on their efficiency. e actual value reaches the target. If the target motor vehicle number and electricity consumption inputs are less than the actual input and the NO 2 , PM 2.5 , and SO 2 outputs are less than the actual undesirable outputs, then the motor vehicle number, electricity consumption ,, and NO 2 , PM 2.5 , and SO 2 efficiencies are less than 1, indicating the actual value is inefficiency. ere is room for improvement on actual value.
If the target average disposable income output is equal to the actual average disposable income output, then the average disposable income efficiency equals 1, indicating overall efficiency. If the actual average disposable income output is less than the target average disposable income output, then the average disposable income efficiency is less than 1, indicating overall inefficiency. e software used by this research is MaxDEA. is software is a benefit for DEA analysis, especially for model with undesirable output.

Variable and the Structure of Model.
is is a dynamic model with several periods, such as period t and period t + 1. e inputs are labor, motor vehicle number, and electricity consumption. Labor and electricity consumption are used for economic development. Electricity is the main energy consumption of Taiwan. A large amount of air pollutants are generated during the production of electricity (ex: thermal power). Vehicles are a source of air pollutants in daily life.
ere are two kinds of output. e desirable output is average disposable income which is an indicator of economic performance. e undesirable outputs are air pollutant which is generated by economic development and citizen's daily life. e carry-over factor which continues to each period (ex: period t to period t + 1) is fixed assets. e linkage of variables is shown in Figure 1. Table 2  According to Table 2, regardless of the amount of labor, motor vehicles, and electricity consumption, the maximum values are mainly concentrated in New Taipei City and Tainan City. e minimum values are mainly concentrated in Taitung County and Chiayi City. For average disposable income, the maximum value in the 4 years is in Taipei City.

Statistics of Input and Output Variables.
e minimum values are mainly concentrated in Taitung County and Chiayi City. e maximum emission of suspended particulates is mainly in Kaohsiung City, and the minimum emissions are in Chiayi City.
e maximum emissions of sulfur oxides are in Kaohsiung City, and the minimum emissions are in Chiayi City. e maximum fixed asset investment for a factory is in Tainan City, and the minimum value is in Taitung County.

Empirical Analysis.
is study explore the overall efficiency of Taiwan's counties and cities from 2012 to 2015. Table 3, the overall efficiency average is 0.8215, and the average efficiency for each year from 2012 to 2015 is, respectively, 0.8360, 0.7984, 0.8199, and 0.8370. e room for improvement is still between 16  Output:

Overall Efficiency. As shown in
Period t + 1 Output: Motor vehicle Electricity consumption   From Table 5             In order to further understand the counties' and cities' electricity consumption, motor vehicles, aerosol emissions, sulfur oxide emissions, nitrogen oxide emissions, and average disposable income per household, this research offers Table 10 for illustration. From Table 10, Taipei City, Taitung County, Hsinchu City, Chiayi City, and Keelung City maintain the highest efficiency values in the 4 years, regardless of overall score, electricity consumption, motor vehicles, fine aerosol emissions, nitrogen oxide emissions, and average per efficiency analysis of household disposable income. From the efficiency analysis of sulfur oxide emissions, Taipei City, Taitung County, Hsinchu City, and Chiayi City still rank first in 4 years. Keelung City also maintains first place in 2012, 2013, and 2015, but in 2014, it ranks 19 th (last). In 2014, the Keelung Port Art Exhibition attracted a large number of people from other counties and cities, and the air pollution was serious. e 19 th overall score (last place) is Tainan City. e electricity efficiency value of Tainan City is also in second to last place or third from last place. Its value for motor vehicle efficiency remains in 15 th place in the 4 years (5 th from last). For its efficiencies of suspended particulate emissions and sulfur oxide emissions, it maintains 11 th or 12 th place in the 4 years. e average efficiency of each household's disposable income is last place in 2012, 2014, and 2015 (19 th place). In 2013, it improves slightly to 17 th (third from last place).
Among the six municipalities, only Taipei City and Taoyuan City rank first and eighth. e remaining 4 municipalities have poor overall scores. In last place (19 th ) is Tainan City, 15 th is Kaohsiung City, 14 th is Taichung City, and 13 th is New Taipei City.
Among the 13 nonmunicipalities with low overall scores, 18 th is Changhua County (second lowest), 17 th is Yunlin County (third lowest), 16 th is Pingtung County, and 12 th is Chiayi County. For nonmunicipalities with middle overall scores, 6 th is Hsinchu County, 7 th is Yilan County, 9 th is Miaoli County, 10 th is Hualien County, and 11 th is Nantou County.
For the efficiency of electricity consumption, in last place is Taoyuan City for all 4 years. For the efficiency of motor vehicles, in last place is New Taipei City and in 18 th is Kaohsiung City for the 4 years (second to last place). For the efficiency of suspended matter emissions, in last place is Hualien County, in 18 th is Kaohsiung City (second lowest), in 17 th is Yilan County (third from last), and Pingtung County is in 16 th place in the 4 years. e undesirable output and desirable output of each county and city have different degrees of progress or regression in the 4 years.
We note that Taipei City, Taitung County, Hsinchu City, Chiayi City, and Keelung City have the best performances. e rest of the counties and cities have a lot of room for improvement. Taiwan's local governments thus should formulate strong policy interventions in air pollution.

Conclusions
During the current situation of global warming and deteriorating environmental conditions, countries around the world are thinking about how to balance economic growth, reduce environmental pollution, and move forward in the direction of sustainable development. erefore, this article collects data from 19 counties and cities in Taiwan from 2012 to 2015, using the modified dynamic SBM model to explore the change in efficiency of air pollutants in various regions of the country from the economic perspective. e results are as follows: (1) From 2012 to 2015, the counties with the best overall efficiency performance are Taipei City, Taitung County, Keelung City, Hsinchu City, and Chiayi City. eir average overall efficiency value is 1. e average overall efficiency performances are poor in Tainan City, Changhua County, Yunlin County, Pingtung County, and Kaohsiung City, with efficiency values of 0.6488, 0.6716, 0.6756, 0.7032, and 0.7113. In counties and cities with the best overall performance and poor performance, the average disposable income per household has no significant relationship with air pollutant emissions. County and need to improve their efficiency. Due to the high demand for industrial electricity, Taoyuan City accounts for more than half of all electricity in Taiwan. e authority must increase the utilization rate of energy use.
Compared with advanced countries, Taiwan has fallen far behind in air pollution control. e prevention of air pollution still only focuses on propaganda and should change to enforcement as soon as possible. More detailed regulations of pollution reduction in different industries are also necessary. is study has pointed out that the efficiency values of the three air pollutants in fourteen counties and cities are far below 0.5. Among the six municipalities directly under the central government with relatively high financial autonomy, only Taipei City and Taoyuan City have higher overall scores. e remaining 4 municipalities of Tainan City, Kaohsiung City, Taichung City, and New Taipei City have poor overall scores. Taiwan must pay attention to the future adjustment of its energy structure such as the use of coal and petrochemical energy and renewable energy development policies. Lastly, Taiwan should face the problem of air pollution without dividing the political parties in order to achieve a steady economy and sustainable development for all involved.
By above results, this research provides the following policy recommendation: (1) Air pollutants move with air flow; thus, the issue of air pollution should be jointly treated with surrounding countries (2) Local governments reduce coal use to generate electricity to reduce air pollution (3) Public sector and private sector should replace petrochemical energy with renewable energy to reduce air pollution (4) e government encourages the public sector and the private sector to use electric vehicles to reduce air pollutant emission by vehicle

Data Availability
e data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare that there are no conflicts of interest.