E nergy Proc edia 61 ( 2014 )1912 – 1916
Available online at
ScienceDirect
1876-6102 © 2014 The Authors. Published by Elvier Ltd. This is an open access article under the CC BY-NC-ND licen (creativecommons/licens/by-nc-nd/3.0/).
Peer-review under responsibility of the Organizing Committee of ICAE2014doi: 10.pro.2014.12.240
The 6th International Conference on Applied Energy – ICAE 2014
Elasticity of substitution between energy and non-energy factors
explaining from energy efficiency by threshold effect
Donglan ZHA a,b 1 Dequn ZHOU a,b
a
College of Economics and Management, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China b
Rearch Centre for Soft Energy Science, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China
Abstract
The source of discrepancies in the empirical study of complementarity or substitutability between energy and non-energy inputs remains controversial. The prent study eks to motivate and initiate exploration of alternative ways of explaining the disparity. A translog cost function is estimated by time-ries data for three inputs: capital, labour and energy of 29 provinces in China. Employing the threshold effect model, we obtain the overwhelming evidences of a threshold point which parates the substitution of energy and non-energy by energy efficiency. The findings imply that the energy-saving technologies are encouraged to be applied or invested in regions with comparative higher energy intensity, which would receive better effects.
Key words: Elasticity of substitution; Threshold effect; Energy efficiency;
1. Introduction
Although many efforts have been attempted to calculate the elasticities, there has been a substantial controversy in the literature on the issue of complementariy or substitutability between energy and non-energy. No connsus has emerged on either the sign or the magnitude of substitution between energy and non-energy inputs, especially for the interplay between energy and capital. The heterogeneity can be explained by the differences in model specification, data characteristics, regions and time periods. But the influencing factors are not limited to the, Pindyck [1] obtain the elasticity of developing countries tend to be somewhat lower than tho of developed countries. Comparing various industries, Kim and Labys [2] find the possibility of capital and labour for energy intensive industries are much larger than tho for less energy intensive industries. It ems many types of energy efficiency improvement may be understood as the ‘substitution’ of capital for energy inputs [3]. In their tup, we assume that energy efficiency may have effects on the elasticities of substitution. And we attempt to shed further light on explaining the divergences of elasticity of substitution from the perspective of energy efficiency. The final aim is to detect an effective way to conrve energy and make protecting the environment obligatory.
2. Data description
1
Corresponding author: Donglan ZHA; Tel.: +86-25-84893190-1010; fax: +86-25 -84892751 E-mail address: zdl@nuaa.edu;
D onglan Zha and Dequn Zhou /
E nergy Procedia 61 ( 2014 ) 1912 – 1916 1913
The datat ud in this study was asmbled from (1) various years of China Statistical Yearbook p
ublished by China’s National Bureau of Statist ics [4], and (2) China Energy Yearbook published by China’s National Bureau of Statistics [5]. Our panel data study is bad on three obrvations, capital stock (K), labor (L) and energy (E) for 29 provinces over the period 1993 to 2010, which leaves us with 493 obrvations 2. The total cost is aggregated by three inputs multiplying by their respective prices.
We obtained capital stock by applying the perpetual inventory method where K=K t-1(1-δ)+I t [6], which has been compared and widely ud in capital stock calculation. Here, K t-1 is capital stock for ba year t-1 (t=1). K 0 is about the capital in the ba year and is derived from K 0=I 0/(g 0+δ), where I 0 is denoted by Total Investment in Fixed Asts by region and g 0 is the growth rate of added value. δ is the depreciate rate assigned 9.6% following [7], which gives the detailed calculation progress. I t means the additional capital investment for year t and it is reprented by the first order difference of Total Investment in Fixed Asts in year t th and t-1th . δ is the price of capital measured by the weighted one year official interest rates of loans of financial institutions.
The cost of L is compod of two parts, the urban and the rural. The former is obtained from multiplying Number of Employed Persons at Year-end in Urban multiplies by Average Wage of Employed Persons in Urban Units Per Capita. And the later is derived from multiplying Number of E
mployed Persons at Year-end in Rural Areas by Annual Net Income of Rural Houholds. The aggregated wage ud to prent the cost of L is inflated by consumer price index. The price of L is measured by the aggregated wage divided by the labor input.
E is each ctor ’s energy consumption with unit of standard coal. Therefore, we u the coal price to reprent the energy price. The ba year of coal price is from China Energy Databook v. 6.0 and we u the coal Production Price Index (PPI) from CSY to estimate the coal price for each year. 3. Panel threshold regression model and results
Intuitively, differences of estimates of substitution across regions may be partly explained from the point of energy efficiency as illustrated in the previous ction. And, it is expected that the influencing factor is not be a constant. To address this issue, we apply a class of panel threshold models recently developed by [8]. With respect to our interest, the corresponding single threshold model is defined as follows:
1it 2it ()()+it it it it it Y X I e X I e D E J E J H d ! (1) where the dependent variable Y is a scalar, α is the fixed effect, X is a vector of regressors, I (·) is an indicator function and e are scalar. The subscripts i and t index the individual and time. γ is the threshold value. The threshold variable ε is al
so a scalar and follows ii d N(0,σ2). The slope coefficients βi can be obtained by ordinary least squares (OLS). If the model exists double threshold effects, the model (1) could be re-written as:
直立猿
1121232()()()it it it it it it it it it it it it it X I e Y X I e X I e D E J H D E J J H D E J H d ½
°°
d ®¾°°
联系英语
! ¯¿
(2)
To investigate the relationship between energy efficiency and energy substitution for 29 regions during 1993 to 2010, the specific threshold regression model is re-written as,
平均工资
01222131+()(<)+()+it it t Et it Et it Et it it S u P P I e P I e P I e E E J E J J E J H d d ! (3)
In our model, S notes cost share of energy or capital. The vector P E is the price of energy. P means the price of capital (P K ) or labor (P L ). e i is the energy efficiency. The subscript indexes i and t denote the industry ctors and the time
2
The municipality of Chongqing parated from Sichuan province in 1997 and become an independently administered municipal city. To keep the
consistency of the classification, we combine Chongqing and Sichuan together, which is the common way to deal with their relationship for the purpo of time-ries rearch. Thus we get 29 cross-ction obrvations, which we name province for short.
1914 D onglan Zha and Dequn Zhou / E nergy Procedia 61 ( 2014 ) 1912 – 1916
dimensions of the panel. Here i is 29, t is 18. The summary statistics of the variables are showed in Table 1.
Table 1. Summary statistics
Minimum 25% quantile
Average 75% quantile
Maximum S K (%) 0.04 0.29 0.33 0.38 0.57 S L (%) 0.23 0.40 0.45 0.51 0.88 S E (%) 0.04 0.16 0.21 0.26 0.44 P K (%) 5.31 5.43 7.21 9.81 10.98 P L (Yuan) 593 2015 4226 5042 27461 P E (Yuan)
248 379 498 671 829 e (10000/Ton of Standard Coal Equivalent)
0.01
0.28
0.46
0.64
1.36
Table 2 prents the results for both S K and S L specifications together with the statistic indicator Fs, the p-values and the critical values. From the table, we find that the test for a single threshold F 1 is significant both for S K and S L with p-values of 0.07 and 0.01 respectively. On the other hand, the tests for double and triple threshold F 2 and F 3 are not statistically significant even under the crit
ical values of 10%. Therefore, we conclude that there is a single threshold effect between energy price and substitution elasticities for energy and non-energy inputs.
Table 2. Test for threshold effects
Number of thresholds
S K S L Test for single threshold
F 1
61.02* 76.33*** p-value
0.07
0.01
Critical values (10%, 5%, 1%)
(53.28, 66.01, 85.84)
(48.71, 55.40, 75.68)
Test for double threshold
F 2
19.87 27.57 p-value
0.63
0.26
Critical values (10%, 5%, 1%)
(45.12, 51.84, 70.35)
(41.85, 51.75, 69.34)
Test for triple threshold
F 3
12.56 13.10 p-value
0.66 0.63
Critical values (10%, 5%, 1%)
(27.75, 32, 44.88)
(33.33, 42.37, 64.22)
八年级语文试卷
Furthermore, we estimate the thresholds for S K and S L under their asymptotic 95% and 99% confidence intervals respectively, e Table 3. The arch procedure is to scrutinize the potential threshold values when giving the smallest residual variance. The SK J and SL J are the least-squares estimates, which occur at 0.6738 both for S K and S L . The value is a bit larger than 75% pecentile of energy efficiency shown in Table 1.
Table 3. Threshold estimates for the single threshold model
Threshold Estimate 95% confidence interval 99% confidence interval SK J 0.6738 ˄0.6694ˈ0.6866˅ ˄0.6634ˈ0.6892˅ SL J
0.6738
仄怎么读什么意思(0.6649ˈ0.6866)
(0.6605ˈ0.6866)
The regression parameters are displayed in Table 4 together with the OLS standard errors and White-corrected standard errors. There are veral points to note about the results. The price of energy has a positive and significant effect on the share of capital, which reflects that energy and capital are substitutes. This can demonstrate that policies taken to introduce energy-saving technologies have positive effect on reducing energy consumption.
责令限期改正通知书
Table 4. Regression estimates: single threshold model
Regressor
Coefficient estimate
OLS SE
White SE
D onglan Zha and Dequn Zhou /
E nergy Procedia 61 ( 2014 )1912 – 1916 1915
S K
P K-0.0858 0.001 0.001 P E I(D≤0.6738)0.0844 0.034 0.033 P E I(D>0.6738) 0.0791 0.037 0.035
S L
P L0.0915 0.002 0.002 P E I(D≤0.6738)-0.2238 0.147 0.144 P E I(D>0.6738) -0.2145 0.148 0.142
Looking the absolute values of coefficient estimates of P E for capital share equation, it clearly shows when the energy efficiency of the provinces are lower than the threshold point, the substitution elasticities of them are larger than the opposite ca. That means, investing energy-saving technology in the regions with relative low energy efficiency can receive more positive effects. It is possible the areas are energy intensive development model and have more energy-saving potential.鼠和龙合不合
For the labour share equation, it is obrved that the coefficient of P E is negative which indicates the complementarity relationship between the energy and labour. Also we can find when the energy efficiency of the provinces are lower than the threshold point, the elasticities of complement between
energy and labour prent stronger than the opposite.
4. Discussion and conclusion
There are considerable differences between the assumptions ud by empirical studies of elasticities of substitution and by tho employed within energy or associated environment models. Using the panel threshold model, we explore the effects of energy efficiency on energy and non-energy substitution. The results indicate that there exists a single threshold effect both between energy and capital, and energy and labor, which mean energy efficiency does affect the elasticities in a significant way. The smaller are the energy efficiency of the regions, the larger energy and non-energy elasticities perform. In this light, we suggest substitutes should be estimated individually for each region when being applied to the energy and environment economic rearch and policy issues. Turning to policy implication, it is commonly believed that improved energy efficiency is still an important energy development policy. From our study, we suggest, investing energy-saving technologies in regions with comparative higher energy intensity can receive better effects.
Acknowledgements
The authors are grateful to the financial support provided by the China Natural Science Funding NO.
71203092, and the Fundamental Rearch Funds for the Central Universities NO. NJ20140029.
References
[1]Pindyck RS. The structure of world energy demand. MIT Press, Cambridge, MA, 1979.
猪年大吉[2]Kim BC, Labys WC. Application of the translog model of energy substitution to developing countries the ca of Korea. Energy Economics, 1988;10(4):313-23.
[3]UKERC. Review of evidence for the rebound effect technical report 3: elasticity of substitution studies, Working Paper; 2007.
[4]CNBSa (China’s National Bureau of Statistics), China Statistic Yearbook 1998-2010. China Statistical Press, Beijing; 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011.
[5]CNBSb (China’s National Bureau of Statistics), China Statistic Yearbook 1998-2010. China Statistical Press, Beijing; 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011.
[6]Goldsmith R. A perpetual inventory of national wealth. National bureau of economic rearch, New York; 1951.
[7]Zhang J, Wu G. The estimation of China’s provincial capital stock: 1952-2000. Economic Rearch Journal, 2004;10:35-44 (In Chine).
1916D onglan Zha and Dequn Zhou / E nergy Procedia 61 ( 2014 )1912 – 1916
[8]Hann BE. Threshold effects in non-dynamic panels: estimation, testing, and inference. Journal of Econometrics, 1999;93(2):345-68.
