Promotion policies for renewable energy and their effects in Taiwan

Journal of Cleaner Production xxx (2016) 1e11

Contents lists available at ScienceDirect

Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro

Promotion policies for renewable energy and their effects in Taiwan Chih-Chun Kung, Liguo Zhang, Meng-Shiuh Chang* Institute of Poyang Lake Eco-economics at Jiangxi University of Finance and Economics, Nanchang, 330013, China

a r t i c l e i n f o

a b s t r a c t

Article history: Received 2 June 2015 Received in revised form 5 May 2016 Accepted 8 May 2016 Available online xxx

Taiwan is an interesting case in bioenergy development, and its government has employed several subsidies to promote bioenergy, such as released land compensation and energy crop subsidies; however, how farmers respond to these economic incentives is not clear. The study employs the concept of “Big Data” by incorporating data on the inputeoutput of crop production, demand elasticities, international trade and tariffs, processing technologies, commodity prices and energy prices for more than 85% of Taiwanese agricultural and forest commodities. The study first utilizes a mathematical programming model to examine the effectiveness of these policies in terms of bioenergy production and GHG (greenhouse gas) emissions reduction and then applies the dynamic structural equation model to analyze the interactions among important input and output factors, such as social welfare, energy prices, bioenergy production, GHG emissions and CO2 trade prices. The results show that (1) the GHG price is more effective than the coal price in the sense of reducing the ethanol production; (2) the gasoline price has a negative impact on contemporary electricity production while the coal and GHG prices have positive impacts; (3) current ethanol production has a negative influence on current GHG emissions reductions; and (4) the gasoline price, coal price, GHG price and GHG emissions reductions have a significant positive impact on contemporary welfare. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Big data Bioenergy Energy crop subsidy Policy analysis

1. Introduction Sustainable development is an important issue because it ensures that future generations will have access to resources for the progression of civilization. The environment is one type of resource, and its degradation is regarded as an inefficient resource allocation. One important factor degrading the environment is the use of fossil fuels, which accelerate the greenhouse gas effects that may make the environment unstable (IPCC, 2007; McCarl and Schneider, 2003), leading to challenges for future generations. Therefore, countries should adopt sustainable and clean production technology to correct resource misallocation and ensure intergenerational fairness. For this reason, finding alternative energy sources that not only reduce the use of fossil fuels but also mitigate climate change is attractive for many countries. Renewable energy is one such attractive possibility because it takes sustainable development and clean production into account simultaneously. Bioenergy is one type of renewable energy that Taiwan has intensively studied (Chen and Chang, 2005; Kung et al., 2013; Tso

* Corresponding author. E-mail address: [email protected] (M.-S. Chang).

and Su, 2009) for several reasons. First, due to global competition in agricultural markets, Taiwan's agricultural industry has been losing its competitiveness, and a substantial amount of cropland has been released, which provides the potential land for planting renewable energy resources. Second, unlike the instability of thermal or hydro power, bioenergy is produced using agricultural commodities, indicating that domestically cultivated bioenergy feedstocks could provide stable sources of renewable energy production. Third, bioenergy offsets greenhouse gas (GHG) emissions. Because Taiwan is vulnerable to the rising sea levels that are the result of climate change, reducing emissions through bioenergy could be an important and attractive program for the Taiwanese government. The production of bioenergy from released land, however, is not a simple issue. The development of the bioenergy industry may bring about various issues, such as (1) government subsidies for both energy crop plantation and bioenergy production: in the beginning stages, economic incentives for farmers may be necessary to encourage them to convert released land or current cultivation patterns into energy crop plantations; (2) GHG emission cutoff targets: because climate change mitigation brings potential benefits to all of society in terms of reduced damage from potential extreme events, some environmental policies may be initiated to reflect the reduction in GHG emissions associated with bioenergy

http://dx.doi.org/10.1016/j.jclepro.2016.05.034 0959-6526/© 2016 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Kung, C.-C., et al., Promotion policies for renewable energy and their effects in Taiwan, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.05.034

2

C.-C. Kung et al. / Journal of Cleaner Production xxx (2016) 1e11

production; (3) efficient resource allocation in terms of land, labor and capital: bioenergy development is quite expensive and involves substantial changes in land competition and labor choice, and the social optimum associated with bioenergy production should therefore be achieved without decreasing the consumer surplus and producer surplus in the agricultural and energy sectors; and (4) market responses: the price and supply of fossil fuels are subject to the influence of international economic and political issues. Domestically produced bioenergy could reduce dependence on uncertain and volatile global energy markets. This study contributes a policy cornerstone by not only providing information on bioenergy development and environmental effects in terms of renewable energy production and climate change mitigation but also by examining the relationships among various influencing factors. Because social welfare is affected by bioenergy production under market operations, this study establishes a dynamic structural equation model to analyze the directional influence (or causal relationships) of multiple variables. The combination of direct and indirect impacts makes the effect of gasoline prices on social welfare debatable. The results are helpful for policy makers in deciding agricultural and environmental policies, including government subsidies, released land payments, environmental protection and energy substitutes. 2. Literature review Many studies regarding renewable energy have been undertaken worldwide (Arvizu, 2008; Fargione et al., 2008; Lehmann, 2007; McCarl, 2008; McCarl et al., 2009; Searchinger et al., 2008). Campiche et al. (2010) show that the cost of cellulosic ethanol in the US agricultural sector will fall in the long run because cellulosic ethanol production will significantly increase as conversion technology improves. Farmers could make more money with sales of corn stover, leading to higher corn production. Bioenergy not only makes energy sustainable but also mitigates global climate change by reducing GHG emissions. McCarl and Schneider (2000) use an economic model to evaluate the carbon displacement potential of agricultural feedstocks. McCarl (2008) also shows that emission offset rates for electricity are higher than other forms of bioenergy because of low transformative energy requirements. In addition to conventional bioenergy technology, one technology called pyrolysis has a higher rate of both electricity conversion and GHG offsetting. If biochar (by-product of pyrolysis) is applied as a soil amendment rather than being used for electricity generation, GHG emissions will be further reduced (Gaunt and Lehmann, 2008; Lehmann, 2007; Lehmann and Joseph, 2009). Among the numerous bioenergy studies, only a few pertain to Taiwanese conditions. Scholars at the Taiwan Institute of Economic Research (TIER) suggest that Taiwan can use multiple energy crops to produce biodiesel to improve Taiwan's current reliance on foreign energy (Tso and Su, 2009). They also examine the energy input/output ratio of corn, sweet sorghum and sugarcane to test the consequent environmental benefits in terms of GHG emissions. Even with positive figures on both energy production and GHG reduction, people question whether Taiwan has sufficient cropland to produce renewable energy. Taiwan had a total of 68,000 ha (hectare) of idle cropland in 2001, which increased to approximately 280,000 ha after Taiwan joined the Word Trade Organization. To ensure farmers' basic living standards in the face of global competition in food markets, the Taiwanese government subsidizes NT$ 90,000 per ha on released land (Chen et al., 2011). Study results show that if released land is fully utilized, approximate 3% of the total gasoline used in Taiwan could be replaced by ethanol. Kung et al. (2013) examine competition between electricity and ethanol and among energy crops. In most cases, sweet potatoes are the

main source of bioenergy production, with switchgrass as a possible feedstock alternative when coal prices are high. To encourage farmers' participation in energy crop plantation, despite the NT$ 90,000 released land payment, an additional NT$ 50,000 subsidy per ha is set by the Taiwanese government. Based on these supporting policies, a bioenergy industry may be established in Taiwan. This study employs several bioenergy technologies, such as ethanol, co-fire and pyrolysis (fast pyrolysis and slow pyrolysis), to examine how these policies affect the net bioenergy production and GHG offset. Structural equation modeling is a general term that is used to describe a family of statistical methods designed to test a conceptual or theoretical model (Goldberger, 1972). Causal modeling, a major application of structural equation modeling, hypothesizes causal relationships among variables and tests the causal models with a linear equation system. Structural equation modeling has been used previously to study renewable energy. For example, Rao et al. (2006) use structural equation modeling to show how environmental indicators correlate with the environmental performance of small and median enterprises in the Philippines. Chien and Hu (2008) employ structural equation modeling to analyze the effect of renewable energy on GDP in 116 economies. Huijts et al. (2014) examine the determinants of the intention to act toward a local hydrogen fuel station using structural equation modeling based on the technology acceptance framework. Additionally, structural vector autocorrelation (VAR), an extension of structural equation modeling, has been widely used. Tiwari (2011) uses the structural VAR to analyze the dynamics of renewable energy consumption, economic growth and CO2 emissions. You (2011) employs structural VAR to study the long-term dynamic relationship between China's energy consumption and economic growth. Silva et al. (2012) analyze how renewable energy sources of electricity generation affect GDP and CO2 emissions. 3. Mathematical programming and analysis 3.1. Data description To properly evaluate this problem, data is the most essential part. This study adopts the idea of big data that are usually used in the commercial and computer industries by including various sources of data for prospective analysis. Bioenergy and sustainable development can also incorporate this concept as well. With various characteristics engaged in agricultural and bioenergy production processes, the concept of big data can be accommodated by including all of the components involved in production, transportation, storage and utilization of commodities, the production of bioenergy, government regulations and environmental consequences, all of which may be adjusted through market operations. To better reflect reality, the dimensionality of the data used in this study includes (a) quantitative variables such as fertilizers, seed, irrigation, chemical applications, mechanics, labor, land, commodity prices, price elasticity and crop outputs; (b) quantitative variables such as cropland classification and forest and pasture use patterns; (c) spatial variables such as Taiwan's fifteen major productive counties, all of which are specified from the first quarter of 2003 to the first quarter of 2015. The study accommodates more than 130 agricultural and forestry commodities and the crop mix data of commodities are specified at the sub-regional level. The availability of total farm labor, cropland, pasture land, forest land and released land is specified at the sub-regional level and eventually mapped to the regional level. The quantity demanded, quantity supplied, elasticity and prices of commodities are also accommodated. Because the import and export of agricultural commodities are important issues in Taiwan, the tariffs of each

Please cite this article in press as: Kung, C.-C., et al., Promotion policies for renewable energy and their effects in Taiwan, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.05.034

C.-C. Kung et al. / Journal of Cleaner Production xxx (2016) 1e11

commodity are included. The transportation costs of commodities and bioenergy production are then aggregated in the mathematical programming to investigate how commodity production, bioenergy production and environmental consequences are affected under market operations. The data are collected from various sources, such as the Taiwan Agricultural Yearbook, Taiwan Agricultural Prices and Costs Monthly and Commodity Price Statistics Monthly. Average quarterly gasoline prices, coal prices and GHG emission trade prices are obtained from CPC Corporation Taiwan, the National Taiwan Power Corporation and Chicago Climate Exchange, respectively. Some of the useful data descriptions are listed in Appendix A. 3.2. Effects of bioenergy on fossil fuel replacement and GHG emissions reduction Due to the potential importance of the bioenergy industry in sustainable development and clean production in terms of energy sustainability and environmental protection, this study examines bioenergy production, GHG emission offset and welfare effects under market operations. To do so, this study employs the framework of Taiwan Agricultural Sector Model originated by Chen and Chang (2005) and utilized and modified by Chen et al. (2011) and Kung et al. (2013). The total value of agricultural commodities accommodated in this framework is more than 85% of Taiwan's total agricultural product value for the past 46 quarters. In addition, because Taiwan began releasing substantial cropland after 2003, the additional land area of approximately 230,000 ha that can be used to produce energy crops is incorporated into the model. Based on this price-endogenous, partial equilibrium mathematical programing model, land and energy crops are competing with each other under given policy and market conditions. The model formulation and descriptions are listed in Appendix B.

3

(through changes in coal, gasoline and GHG prices). The following econometric analysis examines how these factors are interrelated. The factors are defined as follows. (1). Welfare (W): welfare is evaluated by summing the consumer surplus and the producer surplus in the agricultural sector. (2). Gasoline price (GP): the quarterly average gasoline price from 2002 to 2015, measured as NT$/liter, is used in this study (CPC Corporation Taiwan, 2015). (3). Coal price (CP): the quarterly average coal price (NT$/kg) is converted from the global thermal coal prices based on the estimations of the National Taiwan Power Corporation (2015). (4). GHG price (HP): as Taiwan does not have explicit GHG trade prices or a market for emissions trading, the quarterly average GHG price used in this study is obtained from the Chicago Climate Exchange (2015). (5). Ethanol production (TP): ethanol is produced from agricultural feedstocks that reduce both the quantity of imported higher-priced petroleum and the land available for other crops, which potentially affects welfare, in terms of the producer and consumer surpluses, in the total agricultural sector. (6). Electricity production (LP): electricity is generated from agricultural feedstocks that reduce both the quantity of imported coal and land available for other crops, which potentially affects welfare, in terms of producer and consumer surpluses, in the total agricultural sector. (7). GHG emissions reduction (HR): GHG emissions are believed to be the main cause of the global climate shift, and emissions offsets are therefore tradable in the United States and Europe. Thus, gains or losses from agriculturally associated emissions reduction have the potential to influence social welfare.

4. Empirical analysis This study analyzes multiple causal relationships among seven economic and environmental factors, namely, welfare, gasoline prices, coal prices, GHG prices, ethanol production, electricity prices and GHG emissions reduction. This section begins with data analysis and follows with unit root analysis, which are used to verify the stationarity properties of the data. Granger causality is also used to examine the intertemporal correlation among variables. Based on the results of unit root and Granger causality tests, we develop a dynamic structural equation model to construct multiple causal relationships of seven economic and environmental variables. 4.1. Preliminary analysis of variable 4.1.1. Variable analysis In the analysis of welfare related to bioenergy, various economic and environmental factors are associated with the benefits and costs of producing and using bioenergy. The main essentials include gasoline prices, coal prices, GHG prices, ethanol production, electricity production and amounts of GHG reductions. These factors are selected for this study because welfare, in terms of consumer and producer surpluses, can be derived in equilibrium and is influenced by the supply and demand functions. To optimize social welfare in the existence of bioenergy production, energy prices and the production of bioenergy (and associated GHG reduction) play an important role. The mathematical programming model determines the changes in welfare, energy prices, bioenergy production and GHG emissions reductions under market operations

Following the same time span used in the previous section, the quarterly data are collected over the period from October 1, 2003, to March 31, 2015. Table 1 provides the summary statistics of all variables. The average ethanol production is lower than the standard deviations, suggesting relatively high volatility, whereas the average of welfare, gasoline prices, coal prices, GHG prices, electricity production, and GHG emissions reductions are higher than their standard deviations. One possible explanation for this finding is that ethanol production is influenced by the gasoline, coal and GHG prices simultaneously. Therefore, the volatility of ethanol production results from the volatilities of gasoline, coal and GHG prices. Welfare, coal prices, GHG prices, ethanol production and electricity production are skewed to the right, whereas gasoline prices and GHG emissions reductions are skewed to the left. The positive skewness of welfare suggests the slim possibility of extremely good outcomes. Welfare, coal prices and electricity production show excessive kurtosis ranging from 5.23 to 10.28, indicating considerably heavy tails relative to the normal distribution. Pearson's pairwise correlations between the welfare and each factor are reported in the last columns of Table 1. The Pearson's correlations vary from
0.2946 to 0.8535. Five out of six factors (gasoline prices, coal prices, GHG prices, electricity production and GHG emissions reductions) have positive relationships with welfare. The coal price has the highest correlations (0.8535), suggesting the stronger positive dependence of welfare on coal prices than that of welfare on other factors. On the other hand, a negative correlation is detected between welfare and ethanol production, possibly because ethanol production has negative relationships with coal and GHG prices.

Please cite this article in press as: Kung, C.-C., et al., Promotion policies for renewable energy and their effects in Taiwan, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.05.034

4

C.-C. Kung et al. / Journal of Cleaner Production xxx (2016) 1e11

Table 1 Summary statistics.

Welfare Gasoline Price Coal Price GHG Price Ethanol Production Electricity Production GHG Emission Reduction

Mean

SD

Min

Max

Skewness

Kurtosis

Pearson

3248.7753 29.2824

751.6973 4.4508

1822.25 20.05

5936 35.3818

1.0228
0.2287

5.2316 1.8733

1.0000 0.1272

2.6009

0.6262

1.7276

5.3792

2.3315

10.2824

0.8535

45.3782

44.6025

1.5

165.6

0.9592

2.8255

0.4401

20,873.7440

24,520.6590

1300

62,142

0.5741

1.4813


0.2946

270,313.6739

153,908.1288

193,375

684,194.5

2.1587

5.7798

0.0661

561,958.9771

258,592.3104

141,398

800,832.3


0.4535

1.4559

0.3202

Note: ‘SD’ denotes the standard deviation. ‘Pearson’ denotes the Pearson's correlation with respect to the welfare.

In the later econometric analysis, the logarithm of seasonadjusted data is taken to remove the seasonality and mitigate the volatility.

4.1.2. Unit root test Because time series data are used in this study, we must consider the stationarity properties of the data. In our study, we employ the conventional augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) unit root tests. The results of the unit root tests are presented in Table 2. The statistics for both tests confirm that the logarithmic quarterly data are weakly stationary.

4.1.3. Granger causality test In our study, the government plays an important role in altering ethanol and electricity production. To balance the welfare with the GHG emissions reductions, the government must create a policy to influence the current ethanol and electricity production based on the previous welfare levels. This attitude may lead to the intertemporal correlation between ethanol production and welfare as well as between electricity production and welfare. Meanwhile, an intertemporal link exists between ethanol and electricity production because the government's policy of altering the current ethanol and electricity production may rely on historical ethanol and electricity production. Therefore, this study employs the Granger causality test to verify the intertemporal correlation among ethanol production, electricity production, GHG emissions reductions and welfare. Suppose that x and y are two stationary variables observed for T periods. The following model is used:

yt ¼ a0 þ

P X

bk xt
j þ εt :

(1)

k¼1

The null hypothesis of no Granger causality from x to y is supported if b1¼…¼bk ¼ 0. The optimal number of lags is decided by the Bayesian information criterion.

In our study, the optimal lag is 1. The results are reported in Table 3. They suggest a bilateral Granger causality between ethanol production and the GHG emissions reductions as well as between electricity production and GHG emissions reductions. Meanwhile, the hypotheses that GHG emissions reductions Granger cause welfare, welfare Granger causes ethanol production and ethanol production Granger causes electricity production are supported. 4.1.4. Kendall's t test for dynamic correlation In additional to Granger causality test, this study evaluates the Kendall's t to discuss the intertemporal correlations among ethanol production, electricity production, GHG emissions reductions and welfare. The Kendall's tau coefficient (after the Greek letter t), is a statistic used to measure the association between two measured quantities. Suppose (x1,y1), …, (xn,yn) are a set of observations of the joint random variables X and Y respectively. If both xi > xj and yi > yj or if both xi < xj and yi < yj, (xi,yi) and (xj,yj) are said to be concordant. On the other hand, they are said to be discordant, if xi > xj and yi < yj or if xi < xj and yi > yj. The Kendall's t coefficient is defined as:

t¼

ðnumber of concordant pairsÞ
ðnumber of discordant pairsÞ 0:5*n*ðn
1Þ (2)

The null hypothesis of no correlation from lagged x to current y is supported if t¼0. The results, reported in Table 4, are almost Table 3 Granger causality test statistics.

W TP LP GR

W

TP

LP

GR

NA 6.2449 7.0514* 12.444**

11.294*** NA 7.4198* 8.5496***

3.5582 14.862*** NA 21.037***

6.6358* 8.2287** 15.342*** NA

Note: ‘W’, ‘TP’, ‘LP’ and ‘GR’ denote the welfare, ethanol production, electricity production and GHG emission reductions, respectively. The first column and row stand for the Granger cause and the effect. ***, ** and * denote significance at 1%, 5% and 10%, respectively.

Table 2 Test statistics for stationarity analysis.

ADF PP

W

GP

CP

HP

TP

LP

GR


3.305**
3.305**


2.985**
3.023**


4.147***
3.058**


4.131**
4.131**


4.732***
4.732***


4.386***
4.637***


4.465***
4.465***

Note: ‘W’, ‘GP’, ‘CP’, ‘HP’, ‘TP, ‘LP’ and ‘GR’ denote the welfare, gasoline prices, coal prices, GHG prices, ethanol production, electricity production and GHG emission reductions, respectively.***, ** and * denote significance at 1%, 5% and 10%, respectively.

Please cite this article in press as: Kung, C.-C., et al., Promotion policies for renewable energy and their effects in Taiwan, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.05.034

C.-C. Kung et al. / Journal of Cleaner Production xxx (2016) 1e11 Table 4 Kendall's t test.

W TP LP GR

W

TP

LP

GR

NA 0.1408* 0.1866* 0.2123**

0.1310*** NA 0.2677*** 0.5266**

0.0687 0.3283*** NA 0.2475**

0.4566* 0.2535** 0.1505*** NA

identical to those in Table 3. Therefore, the intertemporal correlations among part of variables are supported in our study. 4.2. Dynamic structural equation models 4.2.1. Model development of structural equations In our study, three out of seven factors (ethanol production, electricity production and GHG emissions reduction) could be either dependent variables in one regression or independent variables in another. Thus, this study employs structural equation models to analyze the multiple causal relationships among variables. Structural equation modeling is a confirmatory approach to data analysis requiring the a priori assignment of inter-variable relationships. It statistically tests a hypothesized model to determine the extent to which the proposed model is consistent with the sample data. In our study, we propose a structural equation model, as presented in Fig 1. The developed model relies on the premise that welfare can be explained by several manifest variables. Through this model, it is possible to test a number of hypotheses about the explanatory power and statistical significance between model variables. The underlying causal hypotheses represented from Fig 1 are as follows: (A1) Ethanol production may be influenced by gasoline prices, coal prices and GHG prices. Gasoline prices could influence the demand for gasoline and then influence gasoline's substitute, ethanol. Coal prices could affect the demand for coal and then affect the demand for electricity, which is the competitor of ethanol. GHG prices could impact the emissions reductions that result from the use of ethanol and electricity. (B1) Electricity production may be influenced by gasoline prices, coal prices and GHG prices. Gasoline prices could influence

5

the demand for gasoline and gasoline's substitute, ethanol. They, in turn, influence electricity production. Coal prices could affect the demand for coal and the coal substitute, electricity. GHG prices could impact the emissions reductions that result from the use of ethanol and electricity. (C1) Ethanol production and electricity production may impact GHG emissions reductions. (D1) Social welfare may be influenced by gasoline prices, coal prices, GHG prices, ethanol production, electricity production and GHG emissions reductions.

4.2.2. Dynamic structural equation model The results of Granger causality tests indicate that some explanatory variables are intertemporally correlated. Thus, this study considers the dynamic effects as the control variables when constructing the structural equation models. Based on the results from Table 3, the revised causal hypotheses can be given by: (A2) Ethanol production may be impacted by contemporary gasoline prices, coal prices and GHG prices, as well as the oneperiod lag of ethanol production, GHG emissions reductions and welfare. (B2) In addition to contemporary gasoline prices, coal prices and GHG prices, electricity production may also be influenced by the one-period lag of ethanol production, electricity production and GHG emissions reduction. (C2) Ethanol production, electricity production and the oneperiod lag of ethanol production, electricity production, and GHG emissions reductions may impact contemporary GHG emissions reductions. (D2) Social welfare may be influenced by contemporary gasoline prices, coal prices, GHG prices, ethanol production, electricity production, GHG emissions reductions and the oneperiod lag of GHG emissions reductions. In the estimation, the causal relations (A2eD2) are formulated by Equations (3)e(6), respectively:

TPt ¼ a1 GPt þ a2 CPt þ a3 HPt þ a4 TPt
1 þ a5 GRt
1 þ a6 Wt
1 þ ε1t (3) LPt ¼ b1 GPt þ b2 CPt þ b3 HPt þ b4 TPt
1 þ b5 LPt
1 þ b6 GRt
1 þ ε2t (4) GRt ¼ c1 TPt þ c2 LPt þ c3 TPt
1 þ c4 LPt
1 þ c5 GRt
1 þ ε3t

(5)

Wt ¼ d1 GPt þ d2 CPt þ d3 HPt þ d4 TPt þ d5 LPt þ d6 GRt þ d7 GRt
1 þ ε4t

Fig. 1. Structural equation model for relationship between variables.

(6)

4.2.3. Estimation of dynamic structural equation model Because the Taiwanese government encourages the production of ethanol and electricity through financial subsidies, the constant terms are considered in Equations (3) and (4). The dynamic structural equation model is estimated by the maximum likelihood estimation. This study only reports the contemporary results among all variables because the dynamic terms are used as the control variables. The standardized results in Table 5 can be summarized into four principles, corresponding to Equations (3)e(6).

Please cite this article in press as: Kung, C.-C., et al., Promotion policies for renewable energy and their effects in Taiwan, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.05.034

6

C.-C. Kung et al. / Journal of Cleaner Production xxx (2016) 1e11

Table 5 Estimation of dynamic structural equation model.

constant GP CP HP TP LP GR

Equation (3)

Equation (4)

Equation (5)

Equation (6)

TP

LP

GR

W

0.0000 (0.0036) 0.3481*** (0.0141)
0.1836*** (0.0074)
0.7218*** (0.0290)

0.0000 (0.0003)
0.1861*** (0.0049) 0.0337*** (0.0040) 0.7691*** (0.0128)
0.8027*** (0.1528) 0.1199*** (0.0530)

0.2057*** (0.0093) 0.4588*** (0.0135) 0.3896*** (0.0172)
0.2110 (0.2036) 0.2222 (0.2511) 0.4101** (0.1853)

Note: ‘W’, ‘GP’, ‘CP’, ‘HP’, ‘TP’, ‘LP’ and ‘GR’ denote the welfare, gasoline prices, coal prices, GHG prices, ethanol production, electricity production and GHG reduction, respectively.***, ** and * denote significance at 1%, 5% and 10%, respectively.

Table 6 Equation-level goodness of fit.

2

R mc

Equation (3)

Equation (4)

Equation (5)

Equation (6)

Overall

0.9087 0.9532

0.7204 0.8487

0.8854 0.9409

0.7505 0.8663

0.9906

Note: ‘mc’ denote the Bentler-Raykov multiple correlation coefficient.

(1). Gasoline prices have a positive impact on contemporary ethanol production. The gasoline price elasticity of ethanol production is approximately 0.35, showing a 1% rise in gasoline prices, which can result in a 0.35% increase in ethanol production. On the other hand, the coal and GHG prices have negative impacts on contemporary ethanol production: the coal price and GHG price elasticities of ethanol production are
0.18 and
0.72, respectively. These results indicate that the GHG price is more effective than the coal price in the sense of reducing ethanol production. Among the three prices, the GHG price has the greatest impact on the ethanol production. In sum, the gasoline and GHG prices have the largest positive and negative influences on ethanol production, separately. (2). Different from the results in (1), the gasoline price has a negative impact on contemporary electricity production, whereas coal and GHG prices have positive impacts on contemporary electricity production. The gasoline price, coal price and GHG price elasticities of electricity production are
0.18, 0.03 and 0.77, respectively. Among the three prices, the GHG price dominates the others in its effect on electricity production, implying that affecting the GHG price is the most efficient approach to control electricity production. Among all relative variables, the GHG price and the gasoline price have the greatest positive and negative influences on electricity production, separately. (3). Current ethanol production has a negative influence on current GHG emissions reductions with an elasticity of
0.80. On the other hand, a significant positive effect between electricity production and GHG emissions reductions is observed. The electricity production elasticity of GHG emissions reductions is 0.11. The results imply that controlling ethanol production, rather than electricity production, is a useful method to influence GHG emissions reduction. (4). Gasoline prices, coal prices, GHG prices and GHG emissions reductions all have a significant positive impact on the contemporary welfare. The elasticities are 0.21, 0.45, 0.39 and 0.41, respectively. Among all factors, the coal price plays the most important role in affecting welfare. In sum, increases in the coal price and GHG price results in a decrease in ethanol production and an increase in electricity

production and, finally, results in an increase in GHG emissions reduction. In contrast, an increase in gasoline prices leads to a decline in GHG emissions reductions. The positive effect of coal and GHG prices on welfare can be identified from the results, although the relationship between the gasoline price and welfare is debatable.

4.2.4. Model verification and validation This section first examines the stability of the model. The stability index is computed as the maximum modulus of the eigenvalues for the matrix of coefficients on endogenous variables predicting other endogenous variables. The computed stability index is 8.55e
7, and all eigenvalues lie inside the unit circle. Therefore, our model satisfies the stability condition. Moreover, this study examines the goodness-of-fit for each equation. As the conventional R2 may be negative in the nonrecursive model, Bentler and Raykov (2000) propose an alternative index, the Bentler-Raykov multiple correlation coefficient, to evaluate the goodness-of-fit for each equation. In our study, we report both R2 and the Bentler-Raykov multiple correlation coefficient in the analysis of goodness-of-fit. The results in Table 6 demonstrate that the goodness-of-fit of Equations (3) and (5) is better than that of Equations (4) and (6) in terms of R2 and the Bentler-Raykov multiple correlation coefficient. The overall goodness-of-fit index reaches up to 0.99. These results suggest that the model under analysis could be used to approximate reality. 4.2.5. Consistency check In this section, this study checks the consistency of data. To perform the consistency check, we use a sub-sample of the available data rather than the entire sample. The results are shown in Table 7. It can be observed that the results based on the subsamples are systematically consistent with those based on the entire samples in Table 5, implying that the data in our study can pass the consistency check. 4.2.6. Sensitivity analysis In this section, we test whether the former historical data can influence the contemporary relationships. Although we believe that adding dynamic covariates is more consistent with the nature of data, it is common to use the robust estimator to account for the serial correlation. To verify whether our estimations are affected by the specification of dynamics, we remove the lagged variables in each equation. The results are reported in Table 8. The estimated coefficients in Table 5 are broadly consistent with the results in Table 8, except for Equation (4); however, it still provides qualitatively similar conclusions.

Please cite this article in press as: Kung, C.-C., et al., Promotion policies for renewable energy and their effects in Taiwan, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.05.034

C.-C. Kung et al. / Journal of Cleaner Production xxx (2016) 1e11

7

Table 7 Consistency check.

constant GP CP HP TP LP GR

Equation (3)

Equation (4)

Equation (5)

Equation (5)

TP

LP

GR

W

0.0000 (0.0022) 0.4359*** (0.0126)
0.2125*** (0.0064)
0.7164*** (0.0213)

0.0000 (0.0003)
0.2637*** (0.0048) 0.0529*** (0.0038) 0.6838*** (0.0106)
0.8004*** (0.1468) 0.1253*** (0.0469)

0.3259*** (0.0095) 0.4397*** (0.0110) 0.3524*** (0.0139)
0.3996* (0.2020) 0.2385 (0.2365) 0.4253** (0.1880)

Note: ‘W’, ‘GP’, ‘CP’, ‘HP’, ‘TP’, ‘LP’ and ‘GR’ denote the welfare, gasoline prices, coal prices, GHG prices, ethanol production, electricity production and GHG reduction, respectively.***, ** and * denote significance at 1%, 5% and 10%, respectively.

Table 8 Sensitivity analysis.

constant GP CP HP TP LP GR

Equation (3)

Equation (4)

Equation (5)

Equation (6)

TP

LP

GR

W

0.0000 (0.0022) 0.4025*** (0.0439)
0.1875*** (0.0202)
0.7561*** (0.0972)

0.0000 (0.0003)
0.4383*** (0.0027) 0.1373*** (0.0022) 0.2713*** (0.0029)
0.7903*** (0.1460) 0.1175** (0.0499)

0.2044*** (0.0068) 0.4562*** (0.0068) 0.3876*** (0.0138)
0.2098** (0.0968) 0.2208 (0.1594) 0.4136*** (0.1440)

Note: ‘W’, ‘GP’, ‘CP’, ‘HP’, ‘TP’, ‘LP’ and ‘GR’ denote the welfare, gasoline prices, coal prices, GHG prices, ethanol production, electricity production and GHG reduction, respectively.***, ** and * denote significance at 1%, 5% and 10%, respectively.

4.2.7. Comparison with classic ordinary least square regression To demonstrate the advantages of the proposed model, we compare our proposed model with the conventional ordinary least squared regression (OLS). The results are presented in Table 9. The OLS-estimated coefficients in Equations (2)e(4) are quantitatively the same as those estimated by the proposed model in Table 5, but the estimated standard errors by OLS are larger than those found by the proposed model. On the other hand, the OLS estimation results in Equation (6) deviate from those from the proposed model. In particular, a negative relationship between the GHG price and welfare is found, which is counter-intuitive to the analysis of the welfare function. This property strengthens the importance of systematic estimation in the analysis of social welfare. 5. Concluding remarks The bioenergy industry is important to Taiwan in terms of the domestic energy supply and climate change. Given limited crop land and resources, the development of bioenergy in Taiwan requires intensive investigation before a large scale project can be launched. This study shows that increases in coal prices and GHG prices result in a decrease in ethanol production and increase in

electricity production and, ultimately, in an increase in GHG emissions reductions. The results also indicate that a policy aiming to raise coal and GHG prices is useful to increase GHG emissions reductions and social welfare in the renewable energy and agricultural sectors. On the contrary, if the Taiwanese government employs policies that lead to a rise in gasoline prices, the net GHG emissions reductions will decline. The impact of gasoline prices on social welfare is two-fold: a direct, positive impact and an indirect, negative impact on welfare measures. The GHG price is the most efficient approach to influence the electricity generation, and the control of ethanol production is useful in lowering GHG emissions, implying that the effectiveness of an emissions cutoff policy is highly related to the GHG trade market and gasoline prices. Finally, several policy implications related to Taiwanese sustainable and clean energy production in terms of bioenergy are addressed below: (1) Support for energy crops Because ethanol production is very sensitive to international gasoline prices, higher gasoline prices will push domestic ethanol production to expand. For this reason, it may require some amount

Table 9 Classic ordinary least square estimation.

constant GP CP HP TP LP GR

Equation (3)

Equation (4)

Equation (5)

Equation (6)

TP

LP

GR

W

0.0000 (0.0042) 0.3481*** (0.0781)
0.1836** (0.0709)
0.7218*** (0.0788)

0.0000 (0.0003)
0.1861 (0.1919) 0.0337 (0.1407) 0.7691*** (0.2216)
0.8249*** (0.1261) 0.1200 (0.0785)

0.2216 (0.1539) 0.4945*** (0.1355)
0.4199 (0.3229)
0.2274 (0.3159) 0.2395** (0.0989) 0.4300 (0.2545)

Note: ‘W’, ‘GP’, ‘CP’, ‘HP’, ‘TP’, ‘LP’ and ‘GR’ denote the welfare, gasoline prices, coal prices, GHG prices, ethanol production, electricity production and GHG reduction, respectively.***, ** and * denote significance at 1%, 5% and 10%, respectively.

Please cite this article in press as: Kung, C.-C., et al., Promotion policies for renewable energy and their effects in Taiwan, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.05.034

8

C.-C. Kung et al. / Journal of Cleaner Production xxx (2016) 1e11

of government subsidies for energy crops so that farmers are more willing to convert released land to planting energy crops. Determining the exact amount of subsidies is beyond the scope of this study, however.

sequester more GHG emissions. If the government aims to reduce the impact from OPEC and stabilize regional commodity prices, all energy crops may be used in ethanol production, resulting in a relatively lower GHG emission offset. (5) Maximization of GHG emissions reduction

(2) Support for land acquisition Land is a scarce resource in Taiwan and has been intensively utilized in most regions. To develop bioenergy, multiple bioenergy processing plants may be built and a substantial amount of land may be acquired. Land acquisition may be difficult because a significant portion of Taiwan's land is held by the government and bioenergy development may be a considerable topic in future urban planning. Some forms of government support may therefore be involved. (3) Support for tax or subsidy policy Because the prices of gasoline, coal and GHG emissions are determined by international markets, the Taiwanese government has little control over the prices of gasoline, coal and GHG and must use fiscal policies, such as taxes or subsidies, to interfere with them. If the government's goal is to optimize the reduction of GHG emissions, levying ad valorem or specific taxes on coal and GHG prices and subsidies on gasoline prices will be useful policies to increase coal and GHG market prices and decrease gasoline market prices. These policies can therefore help to decrease ethanol production and increase electricity production, finally resulting in increasing GHG emissions reductions. Policies aiming to encourage bioelectricity production by levying taxes on coal, however, may eventually reduce the welfare of countries where the poor rely on coal-based heat generation. Therefore, although coal is not widely used in households due to climate, geographic and economic reasons, it remains an important source of household heat generation. The failure to consider the differences among countries and regions would limit the usefulness of this study. (4) Maximization of bioenergy production The more bioenergy is produced, the less GHG is emitted, although multiple bioenergy technologies complicate matters. As indicated earlier, feedstocks used in electricity generation could

If the Taiwanese government aims to reduce the potential damage from climate change, its prior goal regarding the development of bioenergy would be to minimize net GHG emissions. Under this consideration, slow pyrolysis plus biochar application will yield the highest GHG emissions offset by the lowest ethanol production and electricity generation. This type of policy slows the trend of climate change and ensures that future generations will not suffer from extreme weather patterns. Due to the small size of the Taiwanese bioenergy industry, however, the effect of GHG emissions reduction may not be significant. Although several advantages of our method are discussed, there are some limitations. First, the sample size is limited so that the estimated variance may not achieve its theoretical minimum. The researchers may overcome the problem of small sample size by analyzing countries such as the United States, Brazil and European Union, which have developed bioenergy for a longer time. Second, the structural equation model only explores the conditional mean of the dependent variable given certain values of the covariate variables. It may also be interesting, however, to estimate either the conditional median or other quantiles of the dependent variable to investigate the conditional quantile effects of the dependent variable. Acknowledgments Chih-Chun Kung would like to thank the financial support from the National Natural Science Foundation of China (#41161087, #71263018, #71303099, #71463022), National Social Science Foundation of China: Major Project (2015YZD16), Postdoctoral Foundation of Jiangxi Province (2013KY56) and China Postdoctoral Foundation (2013M531552, 2015T80685). Appendix A. Data of Taiwanese commodities and energy

A.1 Demand of primary commodity Unit

NT$/kg

Commodity

Average price

JAPONICA CORN CORN1 SORGHUM SOYBEAN PEANUT ADZUKI SWPOTATO POTATO TEA CANEPROC CANEFRESH SESAME RADISH CARROT GINGER SCALLION ONION GARBULB

24.3 4 13.11 8 12 45.94 76.63 12.15 12 200 0.75 10.42 150.4 7.4 7.5 21.5 51.8 9 27.5

Elasticity


0.23
0.4
0.4
0.4
1.56
0.1
1.2
0.4
0.1
1.45
0.1
0.45
0.6
0.58
0.58
0.58
0.28
0.58
1.1

Ton

Unit

NT$/kg

Average quantity

Commodity

Average price

1,329,800 91,653 91,653 8305 213 71,949 11,160 214,068 58,251 18,803 87,545.8 64,754 38,729 132,390 90,930 34,560 91,696 89,748 54,885

LIUCHENG LONGAN JUJUBE LEMON GRAPEFUR MANGO BETEL GUAVA WAXAPPLE GRAPE LOQUAT PLUM PEACH PERSIM APRICOT LICHE CARAM PEAR APPLE

10.2 27.4 32.25 27.63 17.36 40.5 73 15.3 31.8 35.7 102.5 20.5 42.5 30.25 17.72 32.25 21.8 34.5 39.5

Elasticity

Ton Average quantity


0.64
0.5
0.5
0.64
0.64
0.45
0.45
0.85
0.85
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5

212,725 102,790 37,647 17,075 15,642 149,081 137,688 135,887 84,943 104,135 5652 47,287 64,646 27,376 45,358 69,836 16,264 125,018 143,195

Please cite this article in press as: Kung, C.-C., et al., Promotion policies for renewable energy and their effects in Taiwan, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.05.034

C.-C. Kung et al. / Journal of Cleaner Production xxx (2016) 1e11

9

A.1 (continued ) Unit

NT$/kg

Commodity

Average price

LEEK BAMBOO ASPARA WATERBA CABBAGE CAULI CHINESECAB MUSTARD CUCUM BITTER TOMATO PEA VESOY WAMELON CANTA MUSHROOM BANANA PINEAPPLE PONKAN TANKAN WENTAN

23.5 13.8 46 32.5 9 18.75 10.3 9.3 13.5 25.25 12.8 40.5 12 11.5 28.66 54 22 17.5 21.5 20.35 26.5

Elasticity

Ton

Unit

NT$/kg

Average quantity

Commodity

Average price

PAPAYA SUGARAP PASSION COCONUT CHRYSAN GLADIO ROSE BABYS OTHERFLO GOAT MILK GEESE DUCK NATIVE EGG LEG BREAST WING GUT FORBAMBOO

16.8 33.5 21.45 5.3 48 62.4 50.76 50.38 63.11 211.7 19.87 74.56 53.52 52.68 2.03 64 61 70 85.6 7.5


0.58
0.58
0.58
0.58
0.58
0.58
0.5
0.5
0.68
0.68
0.68
0.68
0.68
1.45
1.45
1.25
0.64
1.25
0.64
0.64
0.64

32,316 232,846 13,621 45,801 363,427 89,906 119,102 51,922 52,112 28,378 118,422 13,168 58,924 228,352 60,307 7024 133,481 447,713 93,207 43,656 69,830

A.2 Domestic production

Elasticity

Ton Average quantity


0.5
0.7
0.7
0.7
0.72
0.62
0.35
0.44
0.65
0.81
0.35
0.81
0.93
0.81
0.15
0.93
0.93
0.93
0.93
0.8

88,388 75,554 5035 50,122 20,878 6363 14,252 451 24,210 3234 313,147 23,594 69,760 279,951 6,314,213 178,308 69,899.5 58,731.5 26,371 1725

A.3 Import data of primary commodity

Unit

Ton

Unit

Ton

Commodity

Average production

Commodity

Average production

JAPONICA CORN1 SORGHUM SOYBEAN PEANUT ADZUKI SWPOTATO POTATO TEA CANEPROC CANEFRESH SESAME RADISH CARROT GINGER SCALLION ONION GARBULB LEEK BAMBOO ASPARA WATERBA CABBAGE CAULI CHINESECAB MUSTARD CUCUM BITTER TOMATO PEA VESOY WAMELON CANTA MUSHROOM BANANA PINEAPPLE PONKAN TANKAN

1188381.8 91,653 8325 213 68,000 4850 213,991 41,729 18,803 875,458 64,754 451 132,390 97,934 36,698 91,696 54,000 55,622 32,316 232,856 4635 45,801 316,115 72,548 109,382 73,720 51,363 28,342 118,422 7852 60,629 212,488 60,307 5117 148,715 439,872 93,207 43,656

WENTAN LIUCHENG LONGAN JUJUBE LEMON GRAPEFUR MANGO BETEL GUAVA WAXAPPLE GRAPE LOQUAT PLUM PEACH PERSIM APRICOT LICHE CARAM PEAR APPLE PAPAYA SUGARAP PASSION COCONUT CHRYSAN GLADIO ROSE BABYS OTHERFLO CATTLE HOG GOAT MILK GEESE DUCK NATIVE BROILER EGG

69,830 193,847 102,790 32,844 16,581 8180 149,996 138,095 137,166 84,991 71,105 5652 29,202 25,468 27,111 45,397 70,537 20,928 113,183 5801 88,475 75,554 5035 38,556 21,863 6980 14,253 451 24,210 6048 911,449 3234 303,496 23,594 69,760 279,951 258,110 6,314,213

Unit

NT$/kg

Ton

Unit

NT$/kg

Ton

Commodity

Average price

Average quantity

Commodity

Average price

Average quantity

JAPONICA CORN WHEAT SORGHUM SOYBEAN PEANUT ADZUKI POTATO SESAME CARROT ONION GARBULB BAMBOO ASPARA CABBAGE CAULI CHINESECAB CUCUM PEA WAMELON

11.49 4.63 7.03 6.2 9.31 17.16 23.53 9.66 18.38 12.58 7.52 35.96 10.43 48.47 5.94 20.7 7.59 7.76 21.37 6.25

162,222 4,980,114 1,148,670 75,555 2,446,037 3949 6310 16,522 38,278 3298 36,391 632 278 8986 47,577 17,358 9720 749 5316 15,864

MUSHROOM PINEAPPLE LIUCHENG JUJUBE LEMON GRAPEFUR GRAPE PLUM PEACH PERSIM APRICOT LICHE PEAR APPLE COCONUT MILK LEG BREAST WING GUT

25.7 22.6 13.77 30.87 20.16 15.5 37 26.84 37.67 48.3 45 13.2 48.86 24.05 9.27 38.28 29.5 38.1 29.5 21

1907 9358 19,395 4803 524 8186 33,135 18,085 39,178 265 636 1927 11,835 137,394 11,566 9653 49,253 5372 20,015 560

A.4 Historical fuel and GHG emission prices Year

2015 2014

2013

2012

Unit

NT$/liter

NT$/kg

NT$/Ton

Quarter

95gasoline

Coal

GHG

1 4 3 2 1 4 3 2 1 4 3 2

25.3 29.8 34.4 35.4 35.2 35.1 35.2 34.3 35.4 34.9 34.9 34.1

2.3 2.2 2.4 2.6 2.7 2.5 2.4 2.5 2.6 2.6 2.3 2.5

4.2 4.2 4.2 4.2 4.2 4.2 4.2 4.2 4.2 1.5 3.0 11.7

(continued on next page)

Please cite this article in press as: Kung, C.-C., et al., Promotion policies for renewable energy and their effects in Taiwan, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.05.034

10

C.-C. Kung et al. / Journal of Cleaner Production xxx (2016) 1e11

A.4 (continued ) Year

Unit

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

NT$/liter

NT$/kg

NT$/Ton

Quarter

95gasoline

Coal

GHG

1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1

32.2 31.6 31.7 32.6 32.2 30.7 29.4 29.8 29.6 30.1 28.9 26.7 23.3 23.5 32.7 34.6 32.7 30.3 29.2 27.7 26.5 27.0 27.7 28.1 25.6 24.6 25.4 24.7 23.9 22.9 22.9 22.4 21.9 20.1 20.6 20.2 21.2 20.2 19.4 18.7 18.7

2.6 2.9 3.0 3.2 3.0 2.7 2.6 2.4 2.4 2.3 2.2 2.1 2.0 2.6 3.3 4.0 5.4 4.0 3.3 2.6 1.9 1.8 2.1 2.2 2.5 2.5 2.5 2.5 2.5 2.6 2.6 2.2 2.0 1.7 1.6 1.5 1.4 1.5 1.4 1.3 1.3

3.0 2.1 15.0 37.5 1.5 108.6 41.4 22.5 23.1 20.7 34.2 39.9 57.6 44.4 102.0 165.6 132.6 60.3 99.6 106.5 114.0 122.4 127.5 117.3 60.3 59.4 65.4 40.5 51.6 46.5 28.8 25.5 27.0 29.1 NA NA NA NA NA NA NA

i

X Egik Xik
GHGg  0 cg

XZ XZ XZ Max jðQi ÞdQi þ mðQe ÞdQe
ak ðLk ÞdLk


i

þ

XZ

EXEDðTRQi ÞdTRQi


i

þ

k

XZ

j

Quarter

Ethanol production 1000 L

MWh

Ton

Million NT$

2015 2014

1 4 3 2 1 4 3 2 1 4 3 2 1 4

43,928 52,245 56,126 53,578 53,578 53,578 53,578 56,126 53,578 62,026 55,217 39,000 53,848 52,736

193,375 193,375 193,375 193,375 193,375 193,375 193,375 193,375 193,375 193,375 193,375 250,896 193,375 193,375

324,980 226,682 199,060 226,360 226,360 226,359 226,360 199,060 226,360 141,398 211,803 356,703 211,611 242,506

3116 2673 2827 3087 3394 3036 2798 3020 3087 2645 2197 3031 2636 3550

  ES QiX dQiX

i X Xh X taxi QiM þ outtaxi TRQi þ PiG QiG þ P L ALk i

Year

i

i

X X þ SUBj ALj
Pcarbon  Qghg

where QiG is the government purchase quantity for price supported product i,QiM is the import quantity of product i, QReR is bioenergy technologies, Qi is quantity of commodity i, TRQRiR is the import quantity exceeding the quota for TRQ product i, Qix is export quantity of product i, EDðQiM Þ is the inverse excess import demand curve, ESðQiX Þ is the inverse excess export supply curve, EXED(TRQi) is the inverse excess demand curve of commodities which import quantity is exceeding quota, taxRi Ris import tariff for product i, outtaxRiR is the out of quota tariff of product i, PPLP is the set-aside subsidy, SUBRjR is the subsidy on energy crop j, ALRkR is the released acreage in region k, ECRjkR is the planted acreage of energy crop j in region k, ak(Lk) is the land inverse supply in region k, bk(Rk) is the labor inverse supply in region k, PiG is the government purchase price of commodity i, CRik RisR Rcost of input i in region k, XRikR is the production activity of commodity i in region k, QRghgR is quantity of GHG emissions, YRikR is the total production of commodity i in region k, PRcarbonR is the carbon-equivalent price, and GHGRgR is net emission of greenhouse gas g. The objective function, shown in Equation (7), incorporates Taiwan's agricultural production activities and various domestic and trade policies. Government subsidies on rice purchase, released land and energy crops plantation that are influencing the social welfare in terms of quantity supply and demand are incorporated. All GHG emissions from various sources are converted into carbon equivalent, based on the global warming potential of IPCC. The international carbon trade price is based on the data of Chicago Climate Exchange. The objective assumes that the relationship between social welfare and net GHG emissions is negative. Equation (8) is the demand-supply constraint representing that the demand of commodity plus import should be less than or equal to its supply plus export. Equations (9) and (10) represent the resource endowment constraints by balancing the land and resources usage. Equation (11) balances greenhouse gas components by constraining that the net emissions from agricultural sector should be less than Taiwan's total emissions.

B.1 Bioenergy production and GHG emissions reduction in from 2003 to 2015

k

  XX XZ ED QiM dQiM bk ðRk ÞdRk
Cik Xik þ

k

i

k

2013

ghg

(7) 2012

Subject to

Qi þ QiX þ QiG


X k

  Yik Xik
QiM þ TRQi  0 ci

(11)

ik

The objective function and constraints of the model used are shown as follows:

e

(10)

i

Appendix B. Model formulation and simulation results

i

(9)

j

X fik Xik
Rk  0 ck

Note: Gasoline prices are obtained from CPC corporation Taiwan, coal prices come from National Taiwan Power Corporation and GHG prices are collected from Chicago Climate Exchange.

XZ

X X Xik þ ALk þ ECjk
Lk  0 ck

(8)

2011

Electricity

GHG reduction

Welfare effect

Please cite this article in press as: Kung, C.-C., et al., Promotion policies for renewable energy and their effects in Taiwan, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.05.034

C.-C. Kung et al. / Journal of Cleaner Production xxx (2016) 1e11 B.1 (continued ) Year

2010

2009

2008

2007

2006

2005

2004

2003

Quarter

3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 4

Ethanol production

Electricity

GHG reduction

Welfare effect

1000 L

MWh

Ton

Million NT$

39,000 1300 62,142 1300 1300 16,307 16,307 39,000 14,497 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300 1300

252,225 666,106 193,375 613,380 216,631 193,375 193,375 241,106 193,375 215,965 223,006 223,006 223,013 223,012 223,013 223,006 223,013 223,013 223,012 223,012 223,012 223,013 223,006 223,006 223,007 215,965 223,006 223,006 683,734 656,394 659,299 684,195

357,313 639,254 141,719 800,832 777,922 612,103 612,103 379,354 644,938 775,531 800,808 800,807 800,832 800,832 800,807 800,808 800,832 800,832 800,832 800,832 800,832 800,832 800,808 800,808 800,814 775,531 800,808 800,808 623,706 608,606 611,839 629,829

4143 4335 3387 3747 3501 1822 1863 3104 2804 2858 2762 3501 4485 4424 5936 4424 4485 3501 2653 2516 2803 3008 3350 3446 3446 3405 3405 3501 3555 3049 2762 2366

References Arvizu, D., 2008. Biofuels: too soon to give up. Science 320, 1419e1420. Bentler, P.M., Raykov, T., 2000. On measures of explained variance in nonrecursive structural equation models. J. Appl. Psychol. 85, 125e131. Campiche, J.L., Bryant, H.L., Richardson, J.W., 2010. Long-run effects of falling cellulosic ethanol production costs on the US agricultural economy. Environ. Res. Lett. 5, 14e18. Chen, C.C., McCarl, B.A., Chang, C.C., Tso, C., 2011. Evaluation the potential economic impacts of Taiwanese biomass energy production. Biomass Bioenergy. 35, 1693e1701.

11

Chen, C.C., Chang, C.C., 2005. The impact of weather on crop yield distribution in Taiwan: some new evidence from panel data models and implications for crop insurance. J. Agric. Econ. 33, 503e511. Chien, T., Hu, J.L., 2008. Renewable energy: an efficient mechanism to improve GDP. Energy Policy 36, 3045e3052. Fargione, J., Hill, J., Tilman, D., Polasky, S., Hawthorne, P., 2008. Land clearing and the biofuel carbon debt. Science 319, 1235e1238. Gaunt, J., Lehmann, J., 2008. Energy balance and emissions associated with biochar sequestration and pyrolysis bioenergy production. Environ. Sci. Technol. 42, 4152e4158. Goldberger, A.S., 1972. Structural equation models in the social sciences. Econometrica. 40, 979e1001. Huijts, N.M.A., Molin, E.J.E., Wee, B. van, 2014. Hydrogen fuel station acceptance: a structural equation model based on the technology acceptance framework. J. Environ. Psychol. 38, 153e166. Intergovernmental Panel on Climate Change, 2007. Guidelines for National Greenhouse Gas Inventories, Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge. Kung, C.C., McCarl, B.A., Cao, X.Y., Xie, H.L., 2013. Bioenergy prospects in Taiwan using set-aside landean economic evaluation. China Agric. Econ. Rev. 5, 489e511. Lehmann, J., 2007. A handful of carbon. Nature 447, 143e144. Lehmann, J., Joseph, S., 2009. Biochar for Environmental Management: Science and Technology. Earthscan Publisher, London UK. McCarl, B.A., 2008. Food, biofuel, global agriculture, and environment: discussion. Rev. Agric. Econ. 30, 530e532. McCarl, B.A., Schneider, U.A., 2000. US agriculture's role in a greenhouse gas emission mitigation world: an economic perspective. Rev. Agric. Econ. 22, 134e159. McCarl, B.A., Schneider, U.A., 2003. Economic potential of biomass based fuels for greenhouse gas emission mitigation. Environ. Resour. Econ. 24, 291e312. McCarl, B.A., Peacocke, C., Chrisman, R., Kung, C.C., Ronald, D., 2009. Economics of biochar production, utilization, and GHG offsets. In: Lehmann, J., Joseph, S. (Eds.), Biochar for Environmental Management: Science and Technology. Earthscan Publisher, London, UK, pp. 341e357. Rao, P., Castillo, O., Intal, P.S., Saji, A., 2006. Environmental indicators for small and medium enterprises in the Philippines: an empirical research. J. Clean. Prod. 14, 505e515. Searchinger, T., Heimlich, R., Houghton, R.A., Dong, F., Elobeid, A., Fabiosa, J., Tokgoz, S., Hayes, D., Yu, T., 2008. Use of US croplands for biofuels increases greenhouse gases through emissions from land-use change. Science 319, 1238e1240. Silva, S., Soares, I., Pinho, C., 2012. The impact of renewable energy sources on economic growth and CO2 emissions e a SVAR approach. Eur. Res. Stud. 15, 133e144. Tiwari, A.K., 2011. A structural VAR analysis of renewable energy consumption, real GDP and CO2 emissions: evidence from India. Econ. Bull. 31, 1793e1806. Tso, C., Su, M., 2009. Domestic bio-ethanol sources productivity and energy and economic indicators research. Taiwan J. Agric. Econ. 30, 47e64. You, J., 2011. China's energy consumption and sustainable development: comparative evidence from GDP and genuine savings. Renew. Sustain. Energy Rev. 15, 2984e2989.

Please cite this article in press as: Kung, C.-C., et al., Promotion policies for renewable energy and their effects in Taiwan, Journal of Cleaner Production (2016), http://dx.doi.org/10.1016/j.jclepro.2016.05.034