各類進口能源消費與台灣貿易表現之關聯性分析

全文

(1) . . . . . . . .

(2) . The Linkage between ImportedEnergy and Trade in Taiwan. . . . . . . #. $. . . !. ". !. . . . . .

(3) . .

(4). . . . . . . . . The Linkage between Imported Energy and Trade in Taiwan.

(5) . Student: Tzu-Cheng Yang Advisor: Jin-Li Hu. . . . . . . . . . . A Thesis Submitted to Institute of Business and Management College of Management National Chiao Tung University in Partial Fulfillment of the Requirements for the Degree of Master of Business Administration. June 2010 Taipei, Taiwan, Republic of China . . . . !. " # $.

(6) %. . :. . &. '. (. ). *. +. ,. -. .. /. 0. 1. 2. 3. 4. 5. 6. ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;. . . . . . . . . . . . . <. 7. 8. 9. . :

(7). . . ;. ;. =;;;;;;;;>. . .

(8)

(9) ! " # $ % & ' ( ) * + , ( - . / 01 2 3 4 5 678 9 :;< = > ? @ A67 B 9 :;< = > C @ A: < > C D @ AE F < G D @ AH < D D @ AI J K < L D @ A67M N O P B 9 ;< C = ? @ A67Q M N O P B 9 ;< C = R ? @ S67 T Q M N B 9 ;< C = R ? @ UV W ' ( ) * + , ( - . / 01 2 S = X Y Z [ \ ] ^ 1 _ à . b c : 67B 9 :;A67M N O P B 9 ;A67Q M N O P B 9 ;d e f g ' ( ) * + , ( - . / 0h i j g [ \ UE F k l m d 3 n 5 678 9 :;A67B 9 :;A67M N O P B 9 ;A6 7Q M N O P B 9 ;S67T Q M N B 9 ;o d f g ' ( ) * + , ( - . / 0pq r 678 9 :;S67T Q M N B 9 ;i s g [ \ q t u i j g [ \ Uv w x yH k l z{ m d f g [ \ q r | 5 67Q M N O P B 9 ;S67T Q M N B 9 ;l s g [ \ } q t u l j g [ \ UI J K ~ Z | 5 z{ 67B 9 :;A67M N O P B 9 ;A67Q M N O P B 9 ; s g [ \ } p b c ~ T Q M N B 9 ;SI J K ! ~ e / 0~ Z z s g [ \ U [ \ o z e ! 2 U. . i.

(10) The Linkage between Imported Energy and Trade in Taiwan. Student: Tzu-Cheng Yang. Advisor: Dr. Jin-Li Hu. Institute of Business and Management National Chiao Tung University; ;. ABSTRACT. This study examines co-integration and Granger causality between Taiwan’s energy consumptions and trade variables, including real value of total imports (VTI), real value of total exports (VTE), total energy consumption (TEC), oil and petroleum products consumption (OC), coal and coal products consumption (CC), natural gas consumption (NC), real exports value of industrial sector (EVI), real exports value of heavy-chemical industrial products (EVHI), and real exports value of non-heavy-chemical industrial products (EVNHI) with monthly data during 1998-2009. Via applying Hsiao' s version of the Granger causality method, causality running from TEC to VTE, EVI, and EVHI are found. The impulse-response simulations show that the above relations have positive responses at the initial period. OC will Granger cause all trade variables including VTI, VTE, EVI, EVHI, and EVNHI. The impulse directions to VTI and EVNHI are negative; and others are positive. On the other hand, CC will respond to impulses in all trade variables. The impulse-response simulations show that these relations have positive responses at the initial period except the causality running from EVHI and EVNHI to CC. VTE, NC negatively responds to impulses in EVI and EVHI at the initial period. The bi-directional Granger causality between NC and EVNHI are found, and both sides in this relationship negatively influence each other in short-run. In response to these shocks, all response curves will converge to the pre-shock level gradually and become stable after 20 months.. ii.

(11) ? A. "B. RHS l. T. . U. B. |. C. . . ¨. © ·. Å. r Hâ B ö . . . w. Æ. p. `. r. . . . . Æ. w. M. . B. N. d. O. e. u . . . f. . g. P. h. v. ] Hw. H_. . B. B. Q. i. P . B. @. ´. j. k H. xy >. z{ @. . p. Ï. b. . °. w. ]. . ê. ]. þ. . ãz. . Ð Þ. à. . B. í. ô. î. õ. ^ ãä. P. H. zB. ^. ß. ì. ) Ä. Ï Ý. H. µ Q. Î. ë. g. Ã. Ü. }ì. §. B. pÂ. .

(12). . . ). ^ ì. p iii. 4. . ¦. Û. é. ¥ ´. Q. ê Hx B. ê. ¤ ³. Á 1. Hzý. ãä Ë. ó. £ v. Í. é. . ¢. Ú. è. é. zB ¯. V. ç . ü B. { P. . û. ì Ô. s. zÀ. Ù. -. ² Hu. pÌ. æ. ú. A. . ± y. × HØ. . ¡ Hy. £ . Ë. zB. . ¢ ¿. Ê. @. . ¾. Ö. ò. . °. É |. >. . ½. È. ñ. . x Í. L. tB . . ¯ ¼. ë. . ) [. B. ®. Õ. . ù. . . . Ç. å. . . "K. b Hc. . HJ. B. . ø. . (. zB ÷. A. a. . . . ããä. {. J. s. ^. P. H w. . Ô. Hâ . . Ó. . »zB. -. @. . ] ^_. ^). º ^». > B. HI. ^. zB. H. k. . -. @. (. ð. . . ¹. £ Á. ï. zì. ¸. ¢. \. G. Hx. > Ò. . ¬. . b. HÑ. «. «. B. ). F. pq. ~. . E. [. o. -. . ª. tp¶. á. . D. Z. '. }]. B. . Y. V. . _. C. X. n. zB. . W. m. B {. V. [. b . . @. . B . ° w. . Hz> . . . .

(13) w. @. . Hã.

(14) ]. zì. . . < «. . B. . . O. s. y. \. B -. ö. . ¢. b. £. { l. |. 6. ¢. B. <. O H. ã¢. £. @. . . }r c. £ B. V _. p Hì &. B. ó. x Hy. H. . ã. }5. B. @. 2. B B M. «. 2. ^. -. Y. *. z. «. L. '. f ^c. 1. A K. X. . . @. «. N. B. C. . D B. . §. Z. H [. ^. ¢. ¯. £ . `. h. ) §. z-. _. h. i §. '. z5. ó. B. . . F. 4 y. . @. b. r. . . B .. . ^. . B. .. H l. . . k. B z. . . j. H. ^ Q. B i. ). . iv. ½ 0. ?. . }P. . "B. . . f Hg. Q. ". x>. ãzé. A. 9. k. &. . e. l. `. q. %. @. ~. e pf d. $. >. W. ö. . . ãä. . #. . P. . . 8. }. À. ". / H Q. £. B. . .. J. «. Ô . 7. ¢. ¢. A. l. A. -. Ù I. . £. o. B. è. . . =. Hy. c. ç. @. ,. z6. £. ^. ~ ^ . +. U Hx. . n. . H". . r. ! H. Hy. X p.. } . B. q. HzL. ñ. 1. *. °. ¢. ]. v Hw _. V. >~ . u. ;. T. ³. « 5. . S. . £ -. t. R. ó Ha ¢. n. . HG. :. l. s. F. Q Hw . m. :. B ). B. . . «. "B B. P. î. p. . z. 4. . -. }. @. 3. A. . Á. . o. E. .

(15). p . H(. x> ì. . . . !. . . ? . "#$.

(16) Table of Contents ............................................................................................................................... i. Abstract ................................................................................................................................ ii . . .................................................................................................................................... iii. Table of Contents .................................................................................................................. v List of Tables ....................................................................................................................... vi List of Figures ..................................................................................................................... vii Chapter 1.. Introduction ..................................................................................................... 1. Chapter 2.. Literature Review and Hypotheses ................................................................... 5. Chapter 3.. Model and Econometric Methodology ............................................................ 12. 3.1. Unit Root Tests .................................................................................................. 12. 3.2. Co-integration .................................................................................................... 13. 3.3. Granger Causality Test ....................................................................................... 16. 3.4. The Vector Autoregression Model (VAR) .......................................................... 17. 3.5. Impulse-Response Analysis with VAR ............................................................... 18. Chapter 4.. Data and Empirical Analysis .......................................................................... 20. 4.1. Data Sources and Descriptive Statistics .............................................................. 20. 4.2. Unit Root Test .................................................................................................... 22. 4.3. Co-integration Test ............................................................................................. 24. 4.4. Granger Causality Test ....................................................................................... 25. 4.5. Impulse-Response Simulations ........................................................................... 27. Chapter 5.. Conclusion ..................................................................................................... 33. References .......................................................................................................................... 36 . v.

(17) List of Tables Table 1.. Total domestic energy consumption quantity and growth rate................................ 2. Table 2.. Summary of descriptive statistics for each variable ............................................. 22. Table 3.. Results of unit root tests ...................................................................................... 24. Table 4.. Results of Granger causality tests ........................................................................ 26. Table 5.. Lag length determination for VAR model ........................................................... 28. . vi.

(18) List of Figures Figure 1.. Energy consumption structure in Taiwan ............................................................. 3. Figure 2.. Energy consumption structure in Taiwan (by sectors) .......................................... 4. Figure 3.. Research framework ............................................................................................ 8. Figure 4. Trends of energy consumption and trade variables in Taiwan............................. 21 Figure 5.. Impulses-responses of variables with Granger causal relations ........................... 29. . vii.

(19) Chapter 1. Introduction After the 1973 energy crisis, countries all over the world have noticed the importance of energy to national security as well as economic development. Moreover, greenhouse gases produced by energy consumption have drastic impacts on global climate change, which attracts attention and efforts under the UN Framework Convention on Climate Change. Since domestic energy sources in Taiwan is finite, Taiwan’s dependence on imported energy rated up to 99.34% in 2008 (Bureau of Energy, 2009a).. The cost of. imported energy accounts for 10.02% of Taiwan’s GDP (Bureau of Energy, 2009b). Total energy consumption has grown greatly over the past two decades, going from 48.04 million kiloliters of oil equivalent in 1989 to 113.09million kiloliters in 2009, which is an average annual growth of 4.41% (Table 1).. The growing energy demand. is causing several energy related issues, including cost competitiveness (Rourke et al., 2009) and customer welfare.. In the face of constant rise of energy prices, the heavy. cost of imported energy impacts not only household expenditures, but also the island’s overall economic growth (Chien et al., 2007). As the international energy prices skyrocket, the rising energy prices increase the production cost higher for energy-consuming industries, and hence directly reduce their profit margins and production.. 1.

(20) Table 1.. Total domestic energy consumption quantity and growth rate Year. Quantity (103KLOE). Growth Rate (%). 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009. 48,035.8 50,986.7 54,554.7 57,952.6 60,745.1 65,021.4 68,475.5 71,754.8 75,357.3 80,291.0 84,645.1 91,736.5 97,055.2 100,495.0 104,371.5 108,766.3 111,143.5 113,738.6 119,175.8 115,701.2 113,085.2. 6.14 7.00 6.23 4.82 7.04 5.31 4.79 5.02 6.55 5.42 8.38 5.80 3.54 3.86 4.21 2.19 2.33 4.78 -2.92 -2.26. Average. 85,385.18. 4.4115. Data source: 2009 Energy Statistics Handbook. In Taiwan primary imported energy resources are petroleum, coal, and natural gas, accounting for 49.45%, 32.42% and 9.42%, respectively, of the value of energy imports in 2009.. For energy demands, the biggest share in energy consumption. belongs to electricity (49.28%) followed by oil products (41.35%), coal products (6.81%), natural gas (2.20%), solar energy (0.10%), and others (0.26%) (Figure 1). Since electricity price is fixed and the components of it are not clear, this study does not use it to be an observation variable. 2. In addition, energy consumption in.

(21) industrial sector accounts for 52.48% of total (Figure 2), and up to 98% in the value of total exports.. The relations between international trade variables are discussed with. economic growth in some articles, such as Ghartey (1993), Kwan et al. (1996), and Shirazi and Manap (2005). Furthermore, international trade is an important factor in shaping the industrial structure of a country and, consequently, in affecting a country' s energy use and CO2 emissions (Machado et al., 2001).. In the other words, the. imported energy prices may not only slow down the development of the industrial sector of Taiwan, but also influence the trade surplus in Taiwan.. Trade surplus is an. index to measure the difference in value between the total exports and total imports of an economy during a specific period of time and it will be used as one of research variables in this work.. Figure 1.. Energy consumption structure in Taiwan. 3.

(22) Figure 2.. Energy consumption structure in Taiwan (by sectors). 4.

(23) Chapter 2. Literature Review and Hypotheses Kraft and Kraft (1978) address the idea of causal relationship between energy consumption and economic growth for USA, and it was the earlier paper examining the relationship between these variables.. Recently, considerable attention still has. been focused on this kind of relationship.. In general, these related studies can be. divided into four types:. (1) Some of these studies find the causality running from. energy consumption to economic growth, and it is referred to as ‘the growth hypothesis’, for example, Stern (1993), Wolde-Rufael (2004), Lee and Chang (2005), Ho and Siu (2007), Ang (2007), and Bowden and Payne (2009).. (2) Some find. causal relationship running from economic growth to energy consumption, for example, Kraft an Kraft (1978), Abosedra and Baghestani (1989), Cheng and Lai (1997), Cheng (1998), Aqeel and Butt (2001), Ang (2008), and Zhang and Cheng (2009). It is also called ‘the conservation hypothesis’.. (3) Some find that there are. no causality between energy consumption and economic growth; people also call them ‘the neutrality hypothesis’, for example, Akarca and Long (1980), Yu and Hwang (1984), Yu and Jin (1992), Cheng (1995), Fatai et al. (2002), and Payne (2009).. (4) Others find that there are bi-directional causality between energy. consumption and economic growth; people call them ‘the feedback hypothesis’, for example, Hwang and Gum (1991), Hondroyiannis et al. (2002), Glasure (2002), Ghali and El-Sakka (2004), Paul and Bhattacharya (2004), Erdal et al. (2008), and Belloumi (2009).. This study will consult these four categories to make a brief literature. review. The first type literature on the causal linkage running from energy consumption to economic growth includes: Stern (1993) uses multivariate VAR model to explore the causal relation running 5.

(24) from energy consumption to GDP in USA during 1947-1990, and obtains the same result in 2000 through use the data during 1948-1994 with co-integration and Granger causality approach.. Wolde-Rufael (2004), and Bowden and Payne (2009) through a. modified version of Granger causality approach which called Toda-Yamamoto causality test and find the causal relationship running from energy consumption to GDP in Shanghai during 1952-1999 and in USA during 1949-2006, respectively. Lee and Chang (2005), Ho and Siu (2007) use co-integration approach and VEC model to explore the outcome of Taiwan during 1954-2003 and Hong Kong during 1966-2002, respectively.. Ang (2007) observe the causal relation running from. energy use to GDP would appear in the short run in France with co-integration approach and VEC model. The second type literature on the causal linkage running from economic growth to energy consumption includes: Kraft an Kraft (1978), Cheng and Lai (1997), Cheng (1998) and Zhang and Cheng (2009) follow Hsiao’s (1981) version of the Granger causality approach to find a causal linkage running from GDP to energy consumption in USA during 1947-1974, in Taiwan during 1955-1993, in Japan during 1952-1995, and in China during 1960-2007, respectively.. Abosedra and Baghestani (1989) and Aqeel and Butt. (2001) explore the results of this type in USA during 1947-1987 and in Pakistan during 1955-1996.. Ang (2008) also uses co-integration approach and VEC model to. observe the causal relations running from GDP to energy consumption in Malaysia during 1971-1999. Type 3 literature finding no causality between energy consumption and economic growth contains the following: Fatai et al. (2002) establish no causal relations between energy consumption and GDP in New Zealand during 1960-1999.. Meanwhile, Akarca and Long (1980), Yu 6.

(25) and Hwang (1984), Yu and Jin (1992), Cheng (1995), and Payne (2009) all observe no causal linkage between energy consumption and economic growth in USA under different approaches and periods. Type 4 literature finding bi-directional causality between energy consumption and economic growth includes the following: Hondroyiannis et al. (2002) explore the linkage between energy consumption, economic growth, and the consumer price index (CPI) for Greece during 1960-1996 and find a long-run relationship among them and a bi-directional causality between energy consumption and economic growth.. Glasure (2002) establish a bi-directional. causality between these two variables in Korea during 1961-1990.. Hwang and Gum. (1991), Ghali and El-Sakka (2004), Paul and Bhattacharya (2004), Erdal et al. (2008), and Belloumi (2009) observe a bi-directional causality between energy consumption and GDP. The relations between international trade and economic variables are also discussed widely.. Ghartey (1993) find that exports growth causes economic growth. in Taiwan in short-run.. Kwan et al. (1996) address the result of their research is. conflict with theexport-led growth hypothesis in Taiwan.. Shirazi and Manap (2005). also discuss the export-led growth hypothesis in South Asia, and find a strong support for a long-run relationship among exports, imports, and real output for all the countries except Sri Lanka in South Asia.. Furthermore, feedback effects between. exports and GDP for Bangladesh and Nepal and unidirectional causality from exports to output in the case of Pakistan were found; a feedback effect between imports and GDP was also documented for Pakistan, Bangladesh, and Nepal, as well as unidirectional causality from imports to Output growth for Sri Lanka. Mongelli et al. (2006) argue that there are strong dynamic inter-relationships between output, energy consumption, environmental pollutants and foreign trade, 7.

(26) which should be investigated in the same multivariate framework. Therefore, the main objective of this study is to discuss whether high dependence of energy consumption has apparent influence to the trade activities performance in Taiwan. This study will examine the co-integration and Hsiao’s version of Granger causality between trade and variety of imported energy consumption of Taiwan in bivariate vector autoregression (VAR) framework. Figure 3 is the framework of this study.. It prepares a discussion of the relations. and directions between variety trade and energy consumption variables.. Trade. variables include real value of total imports (VTI), real value of total exports (VTE), real export value of industrial sector (EVI), heavy-chemical industrial products (EVHI),. and. non-heavy-chemical. industrial. products. (EVNHI).. Energy. consumption variables include total energy consumption (TEC), oil and petroleum products consumption (OC), coal and coal products consumption (CC), and natural gas consumption (NC).. Total energy consumption (TEC) H1. Oil and petroleum products consumption (OC). H2. Trade variables a. b. c. d.. Real value of total imports (VTI) Real value of total exports (VTE) Real export value of industrial sector (EVI) Real export value of heavy-chemical industrial products (EVHI) e. Real export value of non-heavy-chemical industrial products (EVNHI). H3 H4. Coal and coal products consumption (CC) Natural gas consumption (NC). Figure 3.. Research framework. 8.

(27) After a brief literature review, this study provides Hypothesis 1a to Hypothesis 1e. Hypothesis 1a: There is a negative Granger causal relation between total energy consumption and real value of total imports. Hypotheses 1b ~ 1e: There are positive Granger causal relations between total energy consumption and real value of export variables. Recent studies research individual energy consumption as well, most of them discuss the ideas with electricity and nuclear, for example, Ouedraogo (2010), Yoo and Kwak (2010), Yoo and Lee (2010), Pao (2009), Akinlo AE (2009), Odhiambo (2009), Balat (2009), Abosedra et al. (2009), Apergis and Payne (2010a), Wolde-Rufael and Menyah (2010), Wolde-Rufael (2010a), and Yoo and Ku (2009). Therefore, it is rare to find studies which discuss the ideas with petroleum, coal, and natural gas.. Zou and Chau (2006) examine both the equilibrium relationship and the. predictability between oil consumption and economic growth in China.. They find. that these two variables tend to move together in the long run, and oil consumption could be a useful factor that forecasts changes in the economy in the short run as well as in the long run.. Therefore, the oil consumption is found to have great effects on. the economy. Yoo (2006b) investigates the short- and long-run causality issues between oil consumption and economic growth in Korea.. The results show that bi-directional. causality runs from oil consumption to economic growth in Korea, implying that an increase in oil consumption directly affects economic growth and that economic growth also stimulates further oil consumption.. Apergis and Payne (2010b; 2010c). examine the relationship between coal consumption and economic growth for 25 OECD countries within a multivariate panel framework over period 1980-2005. The results reveal bi-directional effects between coal consumption and economic growth in both the short- and long-run; however, the bi-directional effects in the short-run are 9.

(28) both negative.. Wolde-Rufael (2010b) finds a unidirectional causality running from. coal consumption to economic growth in India and Japan while the opposite causality running from economic growth to coal consumption was found in China and South Korea.. In contrast, there was a bi-directional causality running between economic. growth and coal consumption in South Africa and the United States, implying that to mitigate the adverse effects of coal consumption may be taken without harming economic growth in China and South Korea, but in India, Japan, South Africa, and the United States. Yoo (2006a) investigates the short- and long-run causality issues between oil consumption and economic growth in Korea as well, and the overall results show that there exists bi-directional causality running from coal consumption to economic growth with feedback.. Yang (2000) uses Granger causality test using 1954-1997. data of Taiwan to test the causality issue between coal consumption and economic growth.. However, there is no literature discussing the causal relations between. natural gas consumption and economic variables.. Hence this study provides. Hypothesis 2a to Hypothesis 4e which are listed as follows: Hypothesis 2a: There is a negative Granger causal relation between oil and petroleum products consumption and real value of total imports. Hypotheses 2b ~ 2e: There are positive Granger causal relations between oil and petroleum products consumption and real value of export variables. Hypothesis 3a: There is a negative Granger causal relation between coal and coal products consumption and real value of total imports. Hypotheses 3b ~ 3e: There are positive Granger causal relations between coal and coal products consumption and real value of export variables. Hypothesis 4a: There is a negative Granger causal relation between natural gas consumption and real value of total imports. 10.

(29) Hypotheses 4b ~ 4e: There are positive Granger causal relations between natural gas consumption and real value of export variables. The structure of this study is organized as follows: the econometric method applied in this study. statistics, and empirical results.. Chapter 3 briefly describes. Chapter 4 presents data, descriptive. Concluding remarks are given in the final chapter.. 11.

(30) Chapter 3. Model and Econometric Methodology 3.1 Unit Root Tests If the mean, variance and covariance of a time series variable were independent with time, and any extraneous impact only make a short influence, then it is a ‘stationary time series’; otherwise, it is a ‘non-stationary time series’.. Prior to the. co-integration analysis, the stationarity of time series should be tested; this is also known as a unit-root test. This research will test the stationarity of time series by two conventional unit root test techniques:. the augmented Dickey-Fuller (ADF) (1981) and the. Phillips-Perron (PP) (1988) test. Dickey and Fuller proposed Dickey-Fuller test method in 1979. Since DF test did not consider the first order autoregression model (AR(1)) and the existence of autocorrelation in residual terms.. Dickey and Fuller modified their equations with. lag length in the process of producing data, so that the residual terms would turn into white noise.. It is known as the augmented Dickey-Fuller test.. The three differencing AR models of ADF are expressed as follows: Model 1: ∆Yt = γYt − 1 + ∑i =1 δi∆Yt − i + εt. (1). Model 2: ∆Yt = α + γYt − 1 + ∑ i =1 δi∆Yt − i + εt. (2). k. k. Model 3: ∆Yt = α + βt + γYt − 1 + ∑i =1 δi∆Yt − i + εt k. (3). where ∆ is the difference operator; t denotes time t ; and γYt − 1 + ∑ i =1 δi∆Yt − i is k. 12.

(31) an augmented part of ADF test.. Therefore, considerable time has been spent. selecting the lag structure using the minimizing Akaike’ s information criterion (AIC) (Akaike, 1974) or Schwarz’s Bayesian information criterion (SBC) (Schwartz, 1978). Model 1 is defined as a pure random walk with lag terms; Model 2 is a random walk. with drift and Model 3 is a random walk with drift around a stochastic trend.. The. null hypothesis and alternative hypothesis for the ADF test are H0: γ = 0 and H1: − 2 < γ < 0.. Although the ADF test with long lag terms is thought to be superior to the others (Schwert, 1989), this research also apply the PP test which is robust in the presence of serial correlation and heteroscedasticity.. Phillips (1987) and Phillips and Perron. (1988) relax the assumption that residual terms must be iid and allow residual terms to have serially correction and heteroscedasticity.. It is known as the PP test.. If the series are tested to be non-stationary in levels and become stationary when dth-order differenced, then they are said to be integrated of order d; i.e., I(d), and co-integration approaches are supposed to be applied to observe the possible long-run relationships among the variables.. 3.2 Co-integration According to Engle and Granger (1987) a linear combination of two or more non-stationary series (with the same order of integration) may be stationary.. If such. a stationary linear combination exists, the series are considered to be co-integrated and long-run equilibrium relationships exist (Erdal et al., 2008).. The ‘spurious. problem’ (Granger and Newbold, 1974) is the first task to be ruled out and with more 13.

(32) powerful multivariate maximum likelihood co-integration test which is conducted by means of the method developed by Johansen (1988) and Johansen and Juselius (1990), is scheduled to investigate the common stochastic trend among variables. Following Johansen (1988) and let Yt be a k×1 matrix, the VAR representation of general form can be set up:. Yt = A0 + A1Yt − 1 + A2∆Yt − 2 + ... + Ap∆Yt − p + εt. (4). εt ~ i.i.d .N (0, σ 2 ). In order to apply the Johansen test, the VAR model above needs to be rewritten in error correction form:. ∆Yt = A0 + ΠYt − 1 + Γ1∆Yt − 2 + Γ 2∆Yt − 3 + ... + Γt − p∆Yt − p − 1 + εt. (5). where ∆ is the first-order difference operator, ΠYt − 1 is termed as the equilibrium error or error correction, and define that Π = ∑i =1 Ai − I , Γi = ∑ j =1 Aj − I . p. i. The parameter matrix r ( Π ) will be further marked, that the rank r of this matrix r ( Π ) , where 0 < r < p , will determine the number of co-integrating vectors in the VAR system. According to the property of the matrix r ( Π ) , there are three cases as follows: 1. Rank (Π ) = p 2. Rank (Π) = 0 3. Rank (Π) = r < p In case 1, Π is full rank and Yt is a stationary series; i.e., Yt ~ I(0), and we 14.

(33) may directly estimate the VAR model with Yt .. In case 2, none is stationary; i.e.,. Yt ~ I(1), there is no co-integration, and estimating the VAR model using ∆Yt is suggested. Under the condition of last case, which is called ‘reduces rank’ , the matrix. Π can be decomposed as Π = αβ , where α is known as the speed of adjustment vector; β is the co-integrating vector, and both α and β are p × r matrices. The number of co-integrating vectors can be judged by determining the significance of the characteristic roots of Π .. There are two test statistics for. co-integration under the Johansen approach, which are formulated as: n λtrace( r ) = −2 ln(θ ) = −T ∑ i = r +1 ln(1 − λî ). (6). and. λnax(r , r + 1) = −2 ln(θ , r | r + 1) = −T ln(1 − λˆr + 1). (7). where r is the number of co-integrating vectors under the null hypothesis; T is the total number of observations; and λî is the estimated value for the ith ordered. eigenvalue from the Π matrix.. If both test statistics are greater than the critical. value, we reject the null hypothesis that there are r co-integrating vectors in favor of the alternative that there are r + 1 (for λtrace ) or more than r (for λ max ) stationary relationships between the relevant variables.. The Schwartz Bayesian information. criterion (SBC) is again used to select the optimal number of lag length.. 15.

(34) 3.3 Granger Causality Test If the predication of the current value of Yt is improved by including past values of X t , then it is supposed that the variable X t Granger causes Yt .. The main. concept to use Granger test is ‘predictability’ the causal relations between variables. The Granger causality mentioned goes in statistics, not completely the ‘causation relations’ .. It can be deemed the lead and backward relation between variables.. The causal relationships between energy consumption and trade variables are detected by Hsiao’ s (1981) version of Granger causality test.. If two variables are. tested to be stationary, then the standard form of Granger causality approach can be expressed as follows: ∆Yt = α 0 + ∑ i =1α i∆Yt − i + u 1t P. (8). Hsiao' s procedure involves two steps to determine the optimum number of own and cross-lagged terms and the direction of causality of variables using the final prediction error (FPE).. The first step calculates the sum of squared errors (SSE) for. Equation (8) where i = 1, 2,..., P.. The FPE(p) considering the lag terms which will. be obtained in the following equation:. FPE( p) =. SSE p +1 (1 + ) T − p −1 T. (9). where T is the total number of observations, p is the order of lags altering from 1 to P, and SSE is the sum of squared errors.. The minimum FPE is decided by the. corresponding SSE and p* , which is expressed as FPE(p*) to make a comparison in the next step.. The second step shifts focus to the following equation: 16.

(35) ∆Yt = α 0 + ∑i =1αi∆Yt − i + ∑ j =1 βj∆Xt − j +u 2t P. Q. (10). From the above equation, Yt is defined as a controlled variable, with the order of lags set at p* from Equation (2), and X t as a manipulated variable.. According to. Equation (3), we estimate the SSE of Yt by altering the lag order of X t from 1 to Q and decide the order producing the smallest FPE, which is denoted as q*.. Finally,. the corresponding two-dimensional FPE is of the form: FPE( p * , q) =. SSE( p * , q ) p* + q + 1 ( 1 + ) T T − p* − q −1. (11). where q is known as the lag order of series X t altering from 1 to Q; and p* is the optimum number of lags estimated in the preceding step.. Summing up the above,. we may draw the conclusion that series X t Granger causes series Yt , if FPE(p*, q*) is smaller than FPE(p*).. 3.4 The Vector Autoregression Model (VAR) Since many economic empirical works are traditionally established according to the prior knowledge, it is hard to settle the proper causality and the endogenous-exogenous relationship between variables until the vector autoregression model (VAR) unfolded by Sims (1980).. The VAR model treats every variable as. been endogenous and expresses their interaction relationship with multiple regression equations rather than one regression equation. The general mathematic form of VAR model is given by: 17.

(36) Yt = α + ∑i =1 βiYt − i + εt m. (12). E (εtεs ) = 0 ; E (εtε ′t ) = ∑ ≠ 0 where Yt is a n × 1 vector of variables; β i is n × n matrices of coefficients; α is n × 1 vector of intercept terms; and εt is a n × 1 vector of disturbances, i.e., the process of one-step-ahead forecast errors. To make sure that the residual terms are all white noise, AIC still will be the optimal lag length selection approach.. If the common stochastic trend among. non-stationary variables does not exist, the VAR model in first-order differences will be applied to carry out the analysis, which is as follows: ∆Yt = α + ∑i =1 βi∆Yt − i + εt m. (13). 3.5 Impulse-Response Analysis with VAR To better comprehend the dynamic response pattern in the VAR model, we further employ the impulse responses to trace out the responsiveness of the dependent variables to shocks to each of the variables.. In other words, impulse response. analysis makes it possible to examine how the variables can be destabilized by shocks that arise with other variables.. By utilizing the Wold’s decomposition theorem. (1954), the VAR model can be transformed into the form of moving average (MA), that is, each variable can be expressed as a linear combination of current value and previous values of a white noise error term. The process is as follows:. 18.

(37) Yt − ∑i =1 βiYt − i = α + εt m. (1 − β 1L − β 2 L2 − ... − βmLm )Yt = α + εt. Yt = (1 − β 1L − β 2 L2 − ... − βmLm ) −1α + (1 − β 1L − β 2 L2 − ... − βmLm ) −1 εt. (14). Yt = α ′ + ∑i =0 Ciεt − i ∞. where L is the lag operator; α ′ is n × 1 vector of constants; Ci is n × n matrices; and C 0 = I is a unit matrix. Typically, the estimated VAR residuals are deemed contemporaneously correlated. For dissecting the effects of innovations in one variable uncontaminated by contemporaneous innovations in other variables, one feasible implementation is to apply the Choleski decomposition to generate triangular orthogonalization matrices. Equation (12) ensures the following equation: Yt = α t + ∑i =0 CiKK −1εt − i ∞. Let Ci * = CiK and et − i = K −1εt − i . The above equation can be re-expressed as: Yt = α t + ∑ i =0 Ci * et − i ∞. (15). where C i * is an impact multiplier and et − i s are neither autocorrelated nor contemporaneously correlated. In this way, each variables can be turned into the function of innovations and these matrices multiplied by the estimated VAR model create uncorrelated residuals, which aid to observe how the coefficients change when the objective variable receive spontaneous shocks from other variables.. 19.

(38) Chapter 4. Data and Empirical Analysis 4.1 Data Sources and Descriptive Statistics Time series variables over the periods from 1998 to 2009 are employed in empirical tests. Monthly imported energy consumption quantity including total energy consumption (TEC), oil and petroleum products consumption (OC), coal and coal products consumption (CC), and natural gas consumption (NC), are defined in KL oil equivalents and obtained from the energy monthly report by Bureau of Energy (2009c) , Ministry of Economic Affairs, Taiwan, covering a period which extends from 1998:1 to 2009:12.. Since the oil, coal, and natural gas used in Taiwan are. almost 100% imported, this study adopts individual total energy consumption quantity to substitute individual imported energy consumption quantity.. Trade data including. real value of total imports (VTI), real value of total exports (VTE), real exports value of industrial sector (EVI), real exports value of heavy-chemical industrial products (EVHI), and real exports value of non-heavy-chemical industrial products (EVNHI), are obtained from the National Statistics Database by the Directorate-General of Budget, Accounting and Statistics, Executive Yuan, Taiwan and deflated with base period of 2005.. These data are defined in million US dollars.. The trends and descriptive statistics for all imported energy consumption and macroeconomic variables are shown in Figure 4 and Table 2.. 20.

(39) 20,000. 24,000. 18,000. 20,000. 16,000. 16,000. 14,000 12,000. 12,000. 8,000. 10,000 8,000. 1998. 2000. 2002. 2004. 2006. 4,000. 2008. 1998. 2000. 2002. 2004. 2006. 2008. Value of total exports. Value of total imports. 11,000,000. 4,800,000. 10,000,000. 4,400,000. 9,000,000. 4,000,000. 8,000,000. 3,600,000. 7,000,000. 3,200,000. 6,000,000. 2,800,000. 5,000,000. 2,400,000 1998. 2000. 2002. 2004. 2006. Total energy consumption. 2008. 1998. 2000. 2002. 2004. 2006. 2008. Oil and petroleum products consumption. 280,000. 900,000. 260,000. 800,000. 240,000. 700,000. 220,000. 600,000. 200,000. 500,000. 180,000 160,000. 400,000 1998. 2000. 2002. 2004. 2006. 1998. 2008. 2000. Coal and coal products consumption. 24,000. 20,000. 20,000. 16,000. 16,000. 12,000. 12,000. 8,000. 8,000. 4,000. 4,000. 1998. 2000. 2002. 2004. 2006. 0. 2008. Exports value of industrial sector. Figure 4.. 1998. 2000. 2002. 2004. 2006. 2008. Natural gas consumption. 2002. 2004. 2006. 2008. Exports value of heavy-chemical industrial products. Trends of energy consumption and trade variables in Taiwan 21.

(40) 4,500 4,000 3,500 3,000 2,500 2,000 1,500. 1998. 2000. 2002. 2004. 2006. 2008. Exports value of non-heavy-chemical industrial products. Figure 4.. Table 2.. Continued. Summary of descriptive statistics for each variable. Variables. VTI. VTE. TEC. OC. CC. Mean. 14,344.07. 14,697.77. 8,612,447.00. 3,534,980.00. 608,598.60. Median. 14,162.30. 14,047.77. 8,732,205.00. 3,597,263.00. 610,022.50. Maximum. 19,304.83. 23,346.08. 10,767,180.00. 4,520,889.00. 817,602.00. Minimum. 8,873.38. 6,682.42. 5,835,672.00. 2,496,122.00. 405,642.00. Std. Dev.. 2,438.75. 4,183.41. 1,165,938.00. 470,136.10. 89,032.28. Skewness. -0.11. 0.17. -0.34. -0.24. 0.13. Kurtosis. 2.06. 1.86. 2.27. 2.14. 2.48. Jarque-Bera. 5.61. 8.45. 5.93. 5.82. 2.02. Variables. NC. EVI. EVHI. EVNHI. Mean. 196,386.70. 14,477.83. 11,331.66. 3,145.34. Median. 193,052.50. 13,807.70. 11,277.06. 3,228.32. Maximum. 265,803.00. 23,160.29. 19,793.93. 4,061.47. Minimum. 161,029.00. 6,453.43. 3,894.20. 1,933.31. Std. Dev.. 18,147.55. 4,211.12. 4,225.56. 350.29. Skewness. 0.67. 0.17. 0.16. -0.90. Kurtosis. 3.52. 1.86. 1.78. 3.99. Jarque-Bera. 12.53. 8.45. 9.56. 25.09. 4.2 Unit Root Test The time series properties of the variables are checked through augmented Dickey-Fuller test (ADF) (1981) and Phillips-Perron’s (PP) (1988) unit root-testing 22.

(41) procedures.. As shown in Table 3, under testing equation computed with intercept,. natural gas consumption is stationary; under testing equation computed with linear trend and intercept, real value of total exports, total energy consumption, coal and coal products consumption, oil and petroleum products consumption, natural gas consumption, real export value of industrial sector, and real export value of heavy-chemical industrial products are stationary; and no variable is stationary without intercept and linear trend in levels. The. series. of real. value. of total imports,. real exports. value. of. non-heavy-chemical industrial products appear to contain unit roots in their levels but stationary after first-order differences; i.e., they are I(1) variables.. Therefore, the. co-integration test can be prepared to examine the long-run equilibrium relationship between these I(1) series.. 23.

(42) Table 3.. Results of unit root tests. Variable. VTI. VTE. TEC. CC. OC. NC. EVI. EVHI. EVNHI. ADF Tu. -2.23. -1.65. -1.58. -2.28. -2.62. -5.43***. -1.63. -1.23. -2.34. Tt. -2.79. -3.84**. -3.76**. -4.58***. -4.14***. -5.55***. -3.80**. -3.56**. -2.97. T. 0.12. 0.63. 1.32. 0.46. 1.00. 0.03. 0.65. 0.90. -0.55. Tu. -4.92***. -1.99. -2.94**. -3.89***. -2.94**. -6.02***. -1.94. -1.39. -8.22***. Tt. -6.18***. -5.85***. -5.23***. -7.51***. -5.71***. -6.10***. -5.79***. -5.29***. -8.64***. T. 0.28. 1.23. 0.64. 0.58. 1.36. -0.06. 1.30. 1.62. -0.24. Level. PP. First Difference ADF Tu. -14.54***. -18.64***. -3.20**. -13.07***. -3.84***. -17.86***. -18.67***. -18.21***. -3.32**. Tt. -14.49***. -18.58***. -3.21. -13.03***. -3.86**. -17.89***. -18.60***. -18.15***. -3.41. T. -14.58***. -18.59***. -2.82***. -13.07***. -17.47***. -17.92***. -18.61***. -18.10***. -3.31***. Tu. -29.08***. -26.19***. -17.03***. -28.54***. -23.22***. -18.02***. -25.57***. -22.45***. -60.55***. Tt. -29.57***. -26.33***. -16.90***. -28.70***. -23.51***. -18.09***. -25.69***. -22.38***. -84.14***. T. -28.76***. -22.43***. -17.01*. -27.23***. -22.20***. -18.08***. -22.19***. -19.39***. -61.41***. PP. Note: 1. *** and ** represent significance at the 1% and 5% levels, respectively. 2. The numbers showed in this table represent t value.. Tu, Tt, and T respectively denote test equation computed. with intercept, with linear trend and intercept, and without intercept and linear trend.. ADF and PP stand for. augmented Dickey-Fuller and Phillips and Perron unit root tests with the same critical values at 5% are -2.86, -3.14, -1.94 and at 1% are -3.43, -3.96, -2.57, respectively.. 4.3 Co-integration Test This study using the Johansen multivariate maximum likelihood procedure (Johansen, 1988), which has been shown to be superior to Engle and Granger’ s residual-based approach (Engle and Granger, 1987) to test the co-integration relationship between variables.. Although there are two non-stationary variables, this. study only discusses the pairwise comparisons with energy consumption and trade variables. Hence, this study does not use co-integration test to observe the long-run relations between variables. 24.

(43) The next step is to apply Granger causality test to examine all pairwise comparisons which include stationary series and non-stationary series after first-order differences under research framework.. 4.4 Granger Causality Test The precondition to apply the standard Granger’s is that the series of variables need to be stationary.. According to the results of unit root test, real value of total. exports, total energy consumption, coal and coal products consumption, oil and petroleum products consumption, natural gas consumption, real export value of industrial sector, and real export value of heavy-chemical industrial products are stationary; while real value of total import, and real exports value of non-heavy-chemical industrial products are appear to contain a unit root in their levels. Therefore, we need to transform these non-stationary variables into difference form to compare with stationary series in Granger causality test.. According to frameworks,. test results are built in Table 4. The results of Granger causality test between total energy consumption and trade variables reveal that total energy consumption has a unidirectional Granger causal linkage running to real value of total exports, real export value of industrial sector, and real export value of heavy-chemical industrial products. After examining the causal relationships between oil and petroleum products consumption and trade variables, this study observe that oil and petroleum products consumption would Granger cause real value of total exports and all real export values under industrial sector.. On the other hand, coal and coal products 25.

(44) consumption is Granger caused by all trade variables.. The natural gas consumption. is Granger caused by real value of total exports, real exports value of industrial sector, and real exports value of heavy-chemical industrial products.. Furthermore, there is a. bi-directional causality between natural gas consumption and real exports value of non-heavy-chemical industrial products, implying that there exists a feedback relationship between these two variables.. Table 4.. Results of Granger causality tests. Null Hypothesis. F-Statistic. Probability. Result. TEC does not Granger cause VTI. 3.01185. 0.0525. Accepted. VTI does not Granger cause TEC. 0.90554. 0.4067. Accepted. TEC does not Granger cause VTE. 4.67160. 0.0109. Rejected**. VTE does not Granger cause TEC. 0.87260. 0.4202. Accepted. TEC does not Granger cause EVI. 4.55679. 0.0121. Rejected**. EVI does not Granger cause TEC. 0.92758. 0.3980. Accepted. TEC does not Granger cause EVHI. 4.24986. 0.0162. Rejected**. EVHI does not Granger cause TEC. 1.48400. 0.2304. Accepted. TEC does not Granger cause EVNHI. 1.66973. 0.1766. Accepted. EVNHI does not Granger cause TEC. 0.38545. 0.7637. Accepted. OC does not Granger cause VTI. 3.61486. 0.0295. Rejected**. VTI does not Granger cause OC. 0.21409. 0.8075. Accepted. OC does not Granger cause VTE. 9.13576. 0.0002. Rejected***. VTE does not Granger cause OC. 1.31452. 0.2720. Accepted. OC does not Granger cause EVI. 9.11179. 0.0002. Rejected***. EVI does not Granger cause OC. 1.42001. 0.2452. Accepted. OC does not Granger cause EVHI. 8.40103. 0.0004. Rejected***. EVHI does not Granger cause OC. 2.07448. 0.1296. Accepted. OC does not Granger cause EVNHI. 4.45012. 0.0052. Rejected***. EVNHI does not Granger cause OC. 1.26568. 0.2888. Accepted. Note: 1. *** and ** represent significance at the 1% and 5% levels, respectively.. 2. denote the non-stationary series variable after first-order differences.. 26.

(45) Table 4.. Continued. Null Hypothesis. F-Statistic. Probability. Result. CC does not Granger cause VTI. 1.49041. 0.2289. Accepted. VTI does not Granger cause CC. 7.77420. 0.0006. Rejected***. CC does not Granger cause VTE. 1.30278. 0.2762. Accepted. VTE does not Granger cause CC. 7.48613. 0.0001. Rejected***. CC does not Granger cause EVI. 1.29781. 0.2779. Accepted. EVI does not Granger cause CC. 7.67668. <0.001. Rejected***. CC does not Granger cause EVHI. 2.28715. 0.1054. Accepted. EVHI does not Granger cause CC. 9.02300. 0.0002. Rejected***. CC does not Granger cause EVNHI. 0.98330. 0.4028. Accepted. EVNHI does not Granger cause CC. 3.38181. 0.0202. Rejected**. NC does not Granger cause VTI. 2.32756. 0.1014. Accepted. VTI does not Granger cause NC. 1.33712. 0.2660. Accepted. NC does not Granger cause VTE. 2.17586. 0.1174. Accepted. VTE does not Granger cause NC. 4.70585. 0.0106. Rejected**. NC does not Granger cause EVI. 2.12137. 0.1238. Accepted. EVI does not Granger cause NC. 4.73500. 0.0103. Rejected**. NC does not Granger cause EVHI. 2.37697. 0.0966. Accepted. EVHI does not Granger cause NC. 4.41751. 0.0138. Rejected**. NC does not Granger cause EVNHI. 4.23376. 0.0068. Rejected***. EVNHI does not Granger cause NC. 4.39757. 0.0055. Rejected***. Note: 1. *** and ** represent significance at the 1% and 5% levels, respectively.. 2. denote the non-stationary series variable after first-order differences.. 4.5 Impulse-Response Simulations In order to obtain additional insight into how the volatility of energy consumption variables to trade performance variables which have causal relations from one to the other, we conduct impulse response analysis.. Before establishing. impulse-response simulations, we have to choose an optimal lag order of the VAR model.. The optimal lag order of the VAR model is selected as 1 on the basis of SBC. as in Table 5.. Impulse response analysis makes it possible to examine how the 27.

(46) variables can be destabilized by shocks that arise with other variables.. Figure 5. presents the results of variables which are observed with causal relationships in Granger causality test from the impulse-response analysis based on VAR (1). It shows the impulse-response paths of variables up to 20 months after a one standard deviation shock stimulated from others.. Table 5.. Lag length determination for VAR model. Model. SBC. Model. SBC. VAR(1). 167.5419*. VAR(6). 175.5312. VAR(2) VAR(3) VAR(4) VAR(5). 168.9623 170.5082 172.5123 174.0371. VAR(7) VAR(8) VAR(9) VAR(10). 176.6639 177.7346 178.7649 179.5172. Note: SBC is the Schwarz information criterion.. 28.

(47) Response to Cholesky One S.D. Innovations ± 2 S.E. 1,400. 1,400. 1,200. 1,200. 1,000. 1,000. 800. 800. 600. 600. 400. 400. 200. 200. 0. 0. -200. 2. 4. 6. 8. 10. 12. 14. 16. 18. -200. 20. 2. 4. 2. 4. 2. 4. Response of VTE to TEC. 1,200. 6. 8. 10. 12. 14. 16. 18. 20. 6. 8. 10. 12. 14. 16. 18. 20. 6. 8. 10. 12. 14. 16. 18. 20. Response of EVI to TEC. 1,200. 1,000. 800. 800 400. 600 400. 0. 200 -400. 0 -200. 2. 4. 6. 8. 10. 12. 14. 16. Response of EVHI to TEC. 18. -800. 20. 1,400. 1,400. 1,200. 1,200. 1,000. 1,000. 800. 800. 600. 600. 400. 400. 200. 200. 0. 0. -200. 2. 4. 6. 8. 10. 12. 14. 16. Response of VTE to OC. Figure 5.. 18. -200. 20. Response of VTI to OC. Response of EVI to OC. Impulses-responses of variables with Granger causal relations. 29.

(48) 300. 1,200 1,000. 200. 800. 100. 600 400. 0. 200. -100 0 -200. 2. 4. 6. 8. 10. 12. 14. 16. Response of EVHI to OC. 18. 20. -200. 2. 35,000. 35,000. 30,000. 30,000. 25,000. 25,000. 20,000. 20,000. 15,000. 15,000. 10,000. 10,000. 5,000. 5,000. 0. 0. -5,000. 2. 4. 6. 8. 10. 12. 14. 16. Response of CC to VTI. 18. 20. 35,000. -5,000. 4. 6. 8. 10. 12. 14. 16. 18. 20. Response of EVNHI to OC. 2. 4. 2. 4. 6. 8. 10. 12. 14. 16. 18. 20. 6. 8. 10. 12. 14. 16. 18. 20. 18. 20. Response of CC to VTE. 30,000. 30,000. 25,000. 25,000 20,000. 20,000 15,000. 15,000. 10,000. 10,000. 5,000 5,000. 0 -5,000. 2. 4. 6. 8. 10. 12. 14. 16. 18. 0. 20. Response of CC to EVI. Response of CC to EVHI. 35,000. 8,000. 30,000. 7,000. 25,000. 6,000 5,000. 20,000. 4,000. 15,000. 3,000 10,000. 2,000. 5,000. 1,000. 0. 0. -5,000. -1,000. 2. 4. 6. 8. 10. 12. 14. 16. Response of CC to EVNHI. 18. 20. Figure 5.. 2. 4. Continued. 30. 6. 8. 10. 12. 14. 16. Response of NC to VTE.

(49) 8,000. 8,000. 7,000. 7,000. 6,000. 6,000. 5,000. 5,000. 4,000. 4,000. 3,000. 3,000. 2,000. 2,000. 1,000. 1,000. 0. 0. -1,000. 2. 4. 6. 8. 10. 12. 14. 16. Response of NC to EVI. 18. -1,000. 20. 15,000. 400. 10,000. 200. 5,000. 0. 0. -200. -5,000. 2. 4. 6. 8. 10. 12. 14. 16. Response of NC to EVNHI. 18. -400. 20. Figure 5.. 2. 4. 2. 4. 6. 8. 10. 12. 14. 6. 8. 10. 12. 14. 16. Response of NC to EVHI. 16. 18. 18. Response of EVNHI to NC. 20. 20. Continued. From the test results, real value of total exports, real exports value of industrial sector, and real exports value of heavy-chemical industrial products tend to have positive responses to the shock of total energy consumption at initial period, and then move downward to the pre-shock level smoothly. Real value of total imports and real exports value of non-heavy-chemical industrial products tend to have negative responses to the shock of oil and petroleum products consumption and fluctuates up and down around the pre-shock level and become stable gradually.. Furthermore, real value of total exports, real exports value. of industrial sector, and real exports value of heavy-chemical industrial products have positive responses to oil and petroleum products consumption at initial period, and the 31.

(50) trend to approach the pre-shock level is downward smoothly. To discuss the responses of energy consumption which is shocked by all trade variables, coal and coal products consumption tends to have positive responses to the shock of real value of total imports, real value of total exports; real exports value of industrial sector at initial period, and have negative responses to the shock of real exports value of heavy-chemical industrial products and real exports value of non-heavy-chemical industrial products at initial period.. These curves also move. downward to the pre-shock level smoothly. On the other hand, real value of total exports, real export value of industrial sector and real value of heavy-chemical industrial impulses natural gas consumption with negative response directions and converge to the pre-shock levels.. Furthermore,. natural gas consumption and real value of non-heavy-chemical industrial products would impulse each other negatively, and fluctuates up and down around the pre-shock level gradually and become stable during the period.. 32.

(51) Chapter 5. Conclusion As has been demonstrated, this study provides an outlook that we can enhance our exports value such as real value of total exports, real export value of industrial sector, and real export value of heavy-chemical through total energy use and oil consumption.. Ghartey (1993) follows the Granger causality tests of Hsiao’s (1979). version to confirm that exports growth causes economic growth in Taiwan. Following this literature, increasing total energy consumption and oil consumption quantity may causes economic growth.. However, Kwan et al. (1996) address the. result of their research is conflict with the export-led growth hypothesis in Taiwan. Furthermore, this study finds oil consumption may Granger causes real value of total imports and real export value of non-heavy-chemical industrial products negatively. It implies oil conservation may enhance our imports value and reduce our trade surplus.. Lee and Chang (2005) show unanimously in the long run that energy acts as. an engine of economic growth in Taiwan, and it implies that energy conservation may harm economic in the long-run. Yang (2000) uses Granger causality test based on 1954-1997 in Taiwan to test the causality issue between coal consumption and economic growth.. The co-integration. and Granger causality test are applied to investigate the relationship between the two economic series.. Results of the co-integration show a unidirectional causality from. economic growth to coal consumption, and support the neutrality hypothesis of coal consumption with respect to economic growth.. Even though this study does not find. the neutrality relationship between coal consumption and export variables, but the conclusion is the same.. That is, to make a suitable policy decision in the area of. macroeconomic planning, coal conservation is a feasible policy with no damaging repercussions on economic growth. 33.

(52) Yoo (2006a) shows that there exists bi-directional causality running from coal consumption to economic growth with feedback.. Our findings only observe Granger. causal relations running from trade variables to coal consumption.. It is likely that. the greater value industrial sector produce is, the greater coal demand in the next period will be, implying that coal is one of the major raw materials of industrial sector. Once we separating industrial sector to two parts- heavy-chemical industrial sector and non-heavy-chemical industrial sector, this study finds both of their value of export products will Granger cause coal consumption negatively.. These results coincide. with Hypothesis 3d and Hypothesis 3e of this study. The bi-directional causality between natural gas consumption and real value of non-heavy-chemical industrial products is observed in this study. directions in this Granger causal relationship are negatively.. Both shock. It seems that the. development of non-heavy chemical industry contains close relation with natural gas consumption.. Since there is no research to discuss this before, this study infers that. it is a kind of index to predict the development of non-heavy-chemical industry. For example, if the consumption of natural gas arises, the development of non-heavy-chemical industry will probably enhance. However, Taiwan is an extreme example that its dependence degree of imported energy is approaching 100%, and it is really a serious problem while energy resources get exhausted.. Under the circumstances that economic development and trade. performance enhance stability, energy demand will still increase continuously. increasing. energy consumption will. development, and carbon dioxide.. influence national security,. The. economic. Therefore, we should be more positive to develop. renewable energy, even the process of expanding renewable energy has been started for a long time.. However, public subsidies could be a measure, but it also increases 34.

(53) government expense and budget deficit. measure to promote the use of clean energy.. The preferential loan is also a feasible According to the Second-Stage Rules of. Preferential Loan for Purchasing Clean Energy Equipment (Bureau of Energy, 1999), there have been preferential loans to enterprises and households to adopt clean energy. Compared to direct subsidies, preferential loans make the users pay more but still provide incentive and can ease the government’s deficit.. Moreover, the development. of clean energy industry in Taiwan, such as wind energy, solar energy, biomass energy, hydro energy, and geothermal energy, can also provide employment opportunity and promote export.. 35.

(54) References [1] Abosedra, S. and Baghestani, H. “New evidence on the causal relationship between United States energy consumption and gross national product”, Journal of Energy Development, 14, pp. 285–292, 1989. [2] Abosedra, S., Dah, A. and Ghosh, S. “Electricity consumption and economic growth, the case of Lebanon”, Applied Energy, 86, pp. 429-432, 2009. [3] Akaike, H. “A new look at the statistical model identification”, IEEE Transaction on Automatic Control, 1, pp. 716–723, 1974. [4] Akarca, A.T. and Long, T.V. “On the relationship between energy and GNP: a reexamination”, Journal of Energy Development, 5, pp. 326–331, 1980. [5] Akinlo, A.E. “Electricity consumption and economic growth in Nigeria: Evidence from cointegration and co-feature analysis”, Journal of Policy Modeling, 31, pp. 681-693, 2009. [6] Ang, J.B. “CO2 emissions, energy consumption, and output in France”, Energy Policy, 35, pp. 4772–4778, 2007. [7]. , “Economic development, pollutant emissions and energy consumption in Malaysia”, Journal of Policy Modeling, 30, pp. 271–278, 2008.. [8] Apergis, N. and Payne, J.E. “A panel study of nuclear energy consumption and economic growth”, Energy Economics, 32, pp. 545-549, 2010a. [9]. , “Coal consumption and economic growth: Evidence from a panel of OECD countries”, Energy Policy, 38, 1353-1359, 2010b.. [10]. , “Renewable energy consumption and economic growth: Evidence from a panel of OECD countries”, Energy Policy, 38, pp. 656–660, 2010c.. [11] Aqeel, A. and Butt, M.S. “The relationship between energy consumption and 36.

(55) economic growth in Pakistan”, Asia Pacific Development Journal, 8, pp. 101–110, 2001. [12] Balat, M. “Electricity consumption and economic growth in turkey: A case study”, Energy Sources- Part B, 4, pp. 155-165, 2009. [13] Belloumi, M. “Energy consumption and GDP in Tunisia: cointegration and causality analysis”, Energy Policy, 37, pp. 2745–2753, 2009. [14] Bowden, N. and Payne, J.E. “The causal relationship between US energy consumption and real output: a disaggregated analysis”, Journal of Policy Modeling, 31, pp. 180–188, 2009. [15] Bureau of Energy, Revised Second-Stage Rules of Preferential Loan for Purchasing. Clean. Energy. Equipment,. 1999.. Available. from:. http://www.moeaboe.gov.tw/opengovinfo/Laws/saveenergy/LSaveMain.aspx?Pa geId=l_save_10. [16] Bureau of Energy, 2009a. Website: [17]. http://www.moeaec.gov.tw.. , 2009 Energy Statistics Handbook, 2009b. Available from: http://www.moeaboe.gov.tw/Download/opengovinfo/plan/all/files/EnergyStatisti calDataBook.pdf.. [18]. , Energy Stat. Monthly Report, 2009c. Available from: http://www.moeaboe.gov.tw/opengovinfo/Plan/all/energy_mthreport/main.. [19] Chen, T.Y., et al. “Renewable energy technology portfolio planning with scenario analysis: A case study for Taiwan”, Energy Policy, 37, pp. 2900–2906, 2009. [20] Cheng, B. “An investigation of cointegration and causality between energy consumption and economic growth”, Journal of Energy Development, 21, pp. 73–84, 1995. [21] Cheng, B.S. and Lai, T.W. “An investigation of co-integration and causality between energy consumption and economic activity in Taiwan”, Energy 37.

(56) Economics, 19, pp. 435–444, 1997. [22] Cheng, B.S. “Energy consumption, employment and causality in Japan: a multivariate approach”, Indian Economic Review, 33, pp. 19–29, 1998. [23] Chien, T.C. and Hu, J.L. “Renewable energy and macroeconomic efficiency of OECD and non-OECD economies”, Energy Policy, 35, pp. 3606-3615, 2007. [24] Chow, L.C.H. “Changes in energy intensiveness of Hong Kong economy, 1995–2007”, Energy Policy, 38, pp. 2076–2085, 2010. [25] Dickey, D.A. and Fuller, W.A. “Distribution of the estimators for autoregressive time series with a unit root”, Journal of the American Statistical Association, 74, pp. 427–431, 1979. [26]. , “Likelihood ratio statistics for autoregressive time series with a unit root”, Econometrica, 49, pp. 1057-1072, 1981.. [27] Directorate-General of Budget, National Statistics Database, 2009. Available from: http://ebas1.ebas.gov.tw/pxweb/Dialog/statfile9L.asp. [28] Engle, R.F. and Granger, C.W.J. “Co-integration and error correction: representation, estimation and testing”, Econometrica, 55, pp. 251-276, 1987. [29] Erdal, G., Erdal, H. and Esengün, K. “The causality between energy consumption and economic growth in Turkey”, Energy Policy, 36, pp. 3838–3842, 2008. [30] Fatai, K., Oxley, L. and Scrimgeour, F. “Energy consumption and employment in New Zealand: searching for causality”, Paper presented at NZAE Conference, Wellington, New Zealand, 2002. [31] Feng, T.W., Sun, L.Y. and Zhang, Y. “The relationship between energy consumption structure, economic structure and energy intensity in China”, Energy Policy, 37, pp. 5475–5483, 2009. [32] Ghali, K.H. and El-Sakka, M.I.T. “Energy use and output growth in Canada: a multivariate cointegration analysis”, Energy Economics, 26, pp. 225–238, 2004. 38.

(57) [33] Ghartey, E.E. “Causal relationship between exports and economic-growth - some empirical-evidence in Taiwan, Japan and the United-States”, Applied Economics, 25, pp. 1145-1152, 1993. [34] Glasure, Y.U. “Energy and national income in Korea: further evidence on the role of omitted variables”, Energy Economics, 24, pp. 355–365, 2002. [35] Granger, C. and Newbold, P. “Spurious regressions in econometrics”, Journal of Econometrics, 2, pp. 111-120, 1974. [36] Hondroyiannis, G. and Lolos, S., Papapetrou, E. “Energy consumption and economic growth: assessing the evidence from Greece”, Energy Economics, 24, pp. 319–336, 2002. [37] Hsiao, C. “Causality tests in econometrics”, Journal of Economic Dynamics and Control, 1, pp. 321-346, 1979. [38]. , “Autoregressive modeling and money income causality detection”, Journal Monetary Economics, 7, pp. 85–106, 1981.. [39] Ho, C.Y. and Siu, K.W. “A dynamic equilibrium of electricity consumption and GDP in Hong Kong: an empirical investigation”, Energy Policy, 35, pp. 2507–2513, 2007. [40] Hu, J.L., Fang, C.Y. and Kuo, T.H. “The linkage of global raw material prices and Taiwan’ s CPI level”, Empirical Economics Letters, 8, pp. 261-270, 2009. [41] Hu, J.L., et al. “The subsidy-based solar energy policy in Taiwan”, Open Environmental Sciences, 4, pp. 8-13, 2010. [42] Huang, B.N., Hwang, M.J. and Yang, C.W. “Does more energy consumption bolster economic growth? An application of the nonlinear threshold regression model”, Energy Policy, 36, pp. 755–767, 2008. [43] Hwang, D. and Gum, B. “The causal relationship between energy and GNP: the case of Taiwan”, Journal of Energy Development, 16, pp. 219–226, 1991. 39.

(58) [44] Johansen, S. “Statistical analysis of cointegration vectors”, Journal of Economic Dynamics and Control, 12, pp. 231-254, 1988. [45] Johansen, S. and Juselius, K. “Maximum likelihood estimation and inference on cointegration with applications to the demand for money”, Oxford Bulletin of Economics and Statistics, 52, pp. 169-209, 1990. [46] Kraft, J. and Kraft, A. “On the relationship between energy and GNP”, Journal of Energy and Development, 3, pp. 401–403, 1978. [47] Kwan, A.C.C., Cotsomitis, J.A. and Kwok, B. “Exports, economic growth and exogeneity. Taiwan 1953-88”, Applied Economics, 28, pp. 467-471, 1996. [48] Lee, C.C. and Chang, C.P. “Structural breaks, energy consumption, and economic growth revisited: evidence from Taiwan”, Energy Economics, 27, pp. 857–872, 2005. [49] Lee C.C. “The causality relationship between energy consumption and GDP in G-11 countries revisited”, Energy Policy, 34, pp. 1086–1093, 2006. [50] Machado, G., Schaeffer, R. and Worrell, E. “Energy and carbon embodied in the international trade of Brazil: an input–output approach”, Ecological Economics, 39, pp. 409-424, 2001. [51] Mongelli, I., Tassielli, G. and Notarnicola, B. “Global warming agreements, international trade and energy/carbon embodiments: an input-output approach to the Italian case”, Energy Policy, 34, pp. 88–100, 2006. [52] Nieh, C.C. and Lee, C.F. “Dynamic relationship between stock prices and exchange rates for G-7 countries”, Quarterly Review of Economics and Finance, 41, pp. 477–490, 2001. [53] Odhiambo, N.M. “Electricity consumption and economic growth in South Africa: A trivariate causality test”, Energy Economics, 31, pp. 635-640, 2009. [54] Ouedraogo, I.M. “Electricity consumption and economic growth in Burkina Faso: 40.

(59) A cointegration analysis”, Energy Economics, 32, pp. 524-531, 2010. [55] Ozturk, I. “A literature survey on energy–growth nexus”, Energy Policy, 38, pp. 340–349, 2010. [56] Pao, H.T. “Forecast of electricity consumption and economic growth in Taiwan by state space modeling”, Energy, 34, pp. 1779–1791, 2009. [57] Paul, S. and Bhattacharya, R.N. “Causality between energy consumption and economic growth in India: a note on conflicting results”, Energy Economics, 26, pp. 977–983, 2004. [58] Payne, J.E. “On the dynamics of energy consumption and output in the US”, Applied Energy, 86, pp. 575–577, 2009. [59] Phillips, P.C.B. “Time series regression with a unit root”, Econometrica, 55, pp. 277-301, 1987. [60] Phillips, P.C.B. and Perron, P. “Testing for a unit root in time series regression”, Biometrika, 75, pp. 335-346, 1988. [61] Rourke, F.O., Boyle, F. and Reynolds, A. “Renewable energy resources and technologies applicable to Ireland”, Renewable and Sustainable Energy Reviews, 13, pp. 1975-1984, 2009. [62] Sadorsky, P. “Renewable energy consumption and income in emerging economies”, Energy Policy, 37, pp. 4021–4028, 2009. [63]. , “The impact of financial development on energy consumption in emerging Economies”, Energy Policy, 38, pp. 2528–2535, 2010.. [64] Sari, R. and Soytas, U. “The growth of income and energy consumption in six developing countries”, Energy Policy, 35, pp. 889–898, 2007. [65] Schwert, G.W. “Tests for unit roots: a Monte Carlo investigation”, Journal of Business and Economic Statistics, 7, pp. 147-159, 1989. [66] Schwartz, G. “Estimating the dimension of a model”, Annals of Statistics, 6, pp. 41.

(60) 461-464, 1978. [67] Shirazi, N.S. and Manap, T.A.A. “Export-led growth hypothesis: Further econometric evidence from South Asia”, Developing Economies, 43, pp. 472-488, 2005. [68] Sims, C.A. “Macroeconomics and reality”, Econometrica, 48, pp. 1-47, 1980. [69] Stern, D.I. “Energy and economic growth in the USA. A multivariate approach” Energy Economics, 15, pp. 137–150, 1993. [70] Wold, H. “A study in the analysis of stationary time series”, Almqvist and Wiksell Book Co., Uppsala, 1954. [71] Wolde-Rufael, Y. “Disaggregated industrial energy consumption and GDP: the case of Shanghai”, Energy Economics, 26, pp. 69–75, 2004. [72]. , “Bounds test approach to cointegration and causality between nuclear energy consumption and economic growth in India”, Energy Policy, 38, pp. 52-58, 2010a.. [73]. , “Coal consumption and economic growth revisited”, Applied Energy, 87, pp. 160-167, 2010b.. [74] Wolde-Rufael, Y. and Menyah, K. “Nuclear energy consumption and economic growth in nine developed countries”, Energy Economics, 32, pp. 550-556, 2010. [75] Yang, H.Y. “Coal consumption and economic growth in Taiwan”, Energy Sources, 22, pp. 109-115, 2000. [76] Yoo, S.H. “Causal relationship between coal consumption and economic growth in Korea”, Applied Energy, 83, pp. 1181-1189, 2006a. [77]. , “Oil consumption and economic growth: Evidence from Korea”, Energy Sources- Part B, 1, pp. 235-243, 2006b.. [78] Yoo, S.H. and Ku, S.J. “Causal relationship between nuclear energy consumption and economic growth: A multi-country analysis”, Energy Policy, 37, pp. 42.

(61) 1905-1913, 2009. [79] Yoo, S.H. and Kwak, S.Y. “Electricity consumption and economic growth in seven South American countries”, Energy Policy, 38, pp. 181-188, 2010. [80] Yoo, S.H. and Lee, J.S. “Electricity consumption and economic growth: A cross-country analysis”, Energy Policy, 38, pp. 622-625, 2010. [81] Yu, E.S.H. and Hwang, B.K. “The relationship between energy and GNP: further results”, Energy Economics, 6, pp. 186–190, 1984. [82] Yu, E.S.H. and Jin, J.C. “Cointegration tests of energy consumption, income, and employment”, Resources and Energy, 14, pp. 259–266, 1992. [83] Yuan, C.Q., Liu, S.F. and Wu, J.L. “The relationship among energy prices and energy consumption in China”, Energy Policy, 38, pp. 197-207, 2010. [84] Zhang, X.P. and Cheng, X.M. “Energy consumption, carbon emissions, and economic growth in China”, Ecological Economics, 68, pp. 2706–2712, 2009. [85] Zou, G.L. and Chau, K.W. “Short- and long-run effects between oil consumption and economic growth in China”, Energy Policy, 34, pp. 3644-3655, 2006.. 43.

(62)