Advertisement

International Journal of Economic Policy Studies

, Volume 13, Issue 1, pp 173–191 | Cite as

Estimating price elasticity of demand for electricity: the case of Japanese manufacturing industry

  • Yasunobu WakashiroEmail author
Research article
  • 28 Downloads

Abstract

Many papers have estimated the residential and/or industrial price elasticity of demand for electricity. Most papers that study industrial elasticities analyze the elasticity for the whole industrial sector. Only a few studies have estimated elasticities for individual sectors, but even then, sectors are classified by broad divisions (alphabetical-letter industrial classification) such as agriculture, manufacturing, and services. Studies that classify sectors by major groups (two-digit industrial classification) such as food, chemicals or iron are rare. Companies that require large amounts of electricity will likely be influenced by an increase in the electricity price. After the Great East Japan Earthquake in 2011, activities at all nuclear power plants were halted. Electric power companies switched to generating electric power using thermal power plants instead of nuclear plants. This increased the electricity price because thermal power plants use expensive fossil fuels such as coal, petroleum, and liquefied natural gas (LNG). The increase in the electricity price imposed a heavy burden on manufacturing companies that consume a large amount of electricity. Some papers have discussed the fact that certain domestic manufacturing companies faced disadvantages and accelerated off-shoring when electricity prices increased. Hosoe (Appl Econ 46(17):2010–2020, 2014) simulated the effects of the power crisis on Japanese industrial sectors using a CGE (Computable General Equilibrium) model. The simulation indicated that the power crisis would decrease domestic output of the wood, paper and printing, pottery, steel and nonferrous metal, and food sectors in Japan, and would accelerate their foreign direct investment. For this paper we estimated the price elasticity of the electricity demand for each industry (major groups) in manufacturing, using the partial adjustment model and the Kalman filter model. In the partial adjustment model, the elasticity of electricity demand of manufacturing in aggregate is − 0.400. Other studies showed that the elasticities of electricity demand including different industrial sectors range from − 0.034 to − 0.300. We found that demand in the manufacturing sector is more elastic than in the aggregate of industrial sectors. They also found that elasticities differ greatly between sectors (major groups) in manufacturing. Sectors with more elastic electricity demand than the aggregate of manufacturing include textile mill products (− 0.775) followed by plastic, rubber and leather products (− 0.701), ceramic, stone and clay products (− 0.701), pulp, paper and paper products (− 0.570), printing and allied industries (− 0.530), machinery (− 0.485), food, beverages, tobacco and feed (− 0.468), miscellaneous manufacturing industries (− 0.413), and lumber and wood products (− 0.403). On the other hand, the less elastic sector is iron, steel, non-ferrous metals and products (− 0.251). The chemical and allied products (− 0.147) sector is not statistically significant at 5% level. In general, less elastic industries need electricity more. In other words, electricity is a necessary good for inelastic industries. The low elasticity implies that these industries cannot reduce electricity consumption even when electricity prices increase. This implies that a high electricity price is a heavy burden on these companies. Inelastic industries can move their operations overseas to access cheaper electricity or they can stop their operations when the price of electricity increases. We believe that policy makers should consider the elasticity of electricity demand because an increase in electricity price has the real possibility of aggravating de-industrialization and/or raising the unemployment rate.

Keywords

Price elasticity of industrial electricity demand Regulatory reform of Japanese electric power industry Partial adjustment Kalman filter 

JEL Classification

C32 C33 Q41 Q48 

Introduction

During the Great East Japan Earthquake of 2011, nuclear power plants at the Fukushima Daiichi Power Station suffered severe damages from the tsunami caused by the earthquake. This incident increased distrust of nuclear power generation among the Japanese population, which resulted in policy makers deciding to halt operations at all nuclear power plants in Japan. Figure 1 illustrates the electricity generation of Japanese incumbent electric companies. Note that electricity generated by nuclear power plants decreased to zero in 2014.
Fig. 1

Electricity generation in Japan.

Source: Agency for Natural Resources and Energy [1]

Figure 1 also indicates that electric power companies had to use thermal power plants to generate electric power that was previously generated by nuclear power plants. This increased the electricity price because thermal power plants use expensive fossil fuels such as coal, petroleum and liquefied natural gas (LNG). Figure 2 shows that the electricity price was tending downwards until 2010, but began rising thereafter.
Fig. 2

Electricity price of electric power companies (Yen, Fiscal Year).

Source: Federation of Electric Power Companies of Japan [10]

The increase in electricity price imposed a heavy burden on manufacturing companies that consume a large amount of electricity. The Ministry of Economy, Trade and Industry [23] estimated that the cost of electricity generation would increase by 3 trillion yen if all nuclear power plants ceased operations, and all the substituted electricity was generated by thermal generation power plants. In such a case, the decline in profits was estimated to be over 50% in the plastics industry, and over 30% in the non-ferrous metal, fibers and transport equipment industries.

The Ministry of Economy, Trade and Industry [23] also reported the influence of an increase in electricity price using the following examples. An electric furnace company expressed concerns that its competitiveness decreased because of the increase of imported steel from Korea, where the electricity price is lower. A chemical manufacturer reported that their manufacturing cost increased by one billion yen for each yen/kWh increase in electricity price. This manufacturer reported that they had to shift investment to factories abroad.

Hosoe [14] showed that some domestic manufacturing companies were facing disadvantages and accelerated off-shoring to avoid the soaring electricity price. The paper simulated the effects of the power crisis on Japanese industrial sectors using a CGE (Computable General Equilibrium) model. The simulation indicated that the power crisis would decrease the domestic output of the wood, paper and printing, pottery, steel and nonferrous metal, and food sectors, and would accelerate foreign direct investments in these sectors.

A power crisis like the one in 2011 rarely happens. However, electricity prices can rise due to other factors such as an increase in the costs and tariffs of Feed-in-tariffs (FIT: Fig. 3), and fluctuations in fuel prices (Fig. 4).
Fig. 3

Costs and tariffs of FIT.

Source: Agency for Natural Resources and Energy [2]

Fig. 4

Prices of fuels (yen/calorie).

Source: Institute of Energy Economics, Japan [20]

We estimate the sectoral elasticities of manufacturing industries. We believe this kind of analysis is essential in discussing the effects of industrial policies because policy makers should understand which sectors are affected the most by an increase in electricity prices when they formulate industrial policies including tariffs, grants, and other industry-specific policies.

Literature review

Many past studies have estimated residential and industrial elasticities of electricity demand. Our paper studies Japanese industrial elasticities using several models. Therefore, we categorize past studies in the following terms: (1) studies of Japanese electricity demand, (2) studies about industrial sectors, and (3) studies about estimation models.

Studies of Japanese electricity demand

Several studies have estimated the Japanese electricity demand function in residential, industrial, and commercial sectors. Most studies have focused on the residential sector (e.g., [25, 28, 29, 33, 34]), and reported that the elasticity of the residential sector ranges from − 0.26 to − 1.204.

Wang and Mogi [35] estimated the elasticity of electricity demand in the residential and industrial sectors, and reported that the industrial sector is much more inelastic than the residential sector (industrial: − 0.16, residential: − 0.51).1 Otsuka [29] estimated elasticity for the industrial sector and found it to be rather inelastic (− 0.034). Hosoe and Akiyama [15] reported that industrial elasticity ranges from − 0.105 to − 0.300. Studies that showed that the elasticity of the industrial sector is less elastic (− 0.034 to − 0.300) employed partial adjustment and Kalman filter models.

Studies of industrial sectors

Most studies using data from foreign countries analyzed the residential sector (e.g., [3, 9, 11, 12, 13, 22]; [26, 27, 37]). Elasticities estimated in these papers ranged from − 0.08 to − 0.41, which are higher than those for Japan.

Some studies examined the aggregate industrial sector [5, 7, 8]. Zachariadis and Pashourtidou [36] estimated the price elasticity of electricity demand for the commercial sector. Inglesi-Lotz and Blignaut [19] estimated the sectoral price elasticity of electricity demand in South Africa from 1993 to 2006. Blignaut et al. [6] estimated the price elasticity of electricity demand for various industrial sectors in South Africa from 2002 to 2011. They focused on showing that a majority of industrial sectors became much more sensitive to electricity price change after the sharp rise of electricity tariffs in 2007/2008. In their study, the estimated sectors are agriculture, mining, iron and steel, liquid fuels, non-ferrous metals, chemicals (other), manufacturing (other), transport, and commercial. The price elasticity was estimated using SUR (seemingly unrelated regression). However, most of their estimation results were statistically insignificant.

There are only a few studies that have estimated the elasticity of electricity demand at a detailed sectoral level. To the best of our knowledge, we do not know of any peer-reviewed paper that has estimated the elasticity of electricity demand in Japan at a detailed sectoral level.

Studies of estimation models

Panel data and time series analyses have been used to estimate the price elasticity of electricity demand. The main model used in the panel data analysis to estimate the elasticity of electricity demand is a partial adjustment model. This can estimate stable parameters even for data with only a short time-length. A time series analysis contains two main models, an autoregressive distributed lag (e.g. [4]) and a Kalman filter (e.g. Inglesi-Lotz [18]). Some papers have adopted the autoregressive distributed lag model, but this model requires a long-time series of more than 20 years, as Okajima and Okajima [28] have pointed out. A Kalman filter model is a kind of state-space model that can estimate a non-stationary model, while an autoregressive distributed lag model can estimate only a stationary model. As indicated in Table 1, we do not see any agreed model in estimating elasticities of electricity demand.
Table 1

Literature review

Author

Country or region

Model

Period

Category

Short run elasticity

Hosoe and Akiyama [15]

Japan (regional)

PA

1976–2006

Industrial

− 0.105 to − 0.300

Commercial

Otsuka [29]

Japan (regional)

PA

1990–2010

Industrial

− 0.034

Commercial

Wang and Mogi [35]

Japan

KF

1989–2014

Residential

− 0.511

Industrial

− 0.16

Tamechika [34]

Japan (prefectural)

PA

1996–2009

Residential

− 0.26 to − 0.35

Okajima and Okajima [28]

Japan (regional)

PA

1990–2007

Residential

− 0.397

Tanishita [33]

Japan (regional)

PA

1986–2006

Residential

− 0.60 to − 0.92

Nakajima [25]

Japan

ADL

1975–2005

Residential

− 1.13 to − 1.2

Chang et al. [7]

Korea

TVC

1995.01–2012.12

Residential

− 0.07

1985.01–2012.12

Industrial

0.12

Commercial

− 0.22

Arisoya and Ozturk [5]

Turkey

KF

1960–2008

Residential

− 0.014

Industrial

− 0.023

Dilaver and Hunt [8]

Turkey

ADL

1960–2008

Industrial

− 0.161

Blignaut et al. [6]

South Africa

PA

2002–2011

Agriculture

− 0.235

Coal Mining

− 0.291

Commercial

− 0.19

Gold Mining

− 0.417

Iron and Steel

− 0.279

Liquid Fuels

− 0.418

Non-ferrous Metals

− 0.342

Rest of Chemicals

− 0.24

Rest of Manufacturing

− 0.251

Rest of Mining

− 0.465

Transport

− 0.346

Inglesi-Lotz and Blignaut [19]

South Africa

Panel data

1993–2006

Industrial

− 0.869

Agriculture

0.152

Transport

− 1.22

Commercial

0.677

Mining

0.204

Zachariadis and Pashourtidou [36]

Cyprus

VEC

1960–2004

Residential

− 0.103

Commercial

− 0.009

Inglesi-Lotz [18]

South Africa

KF

1986–2005

Aggregate

− 0.075

Amusa et al. [4]

South Africa

ADL

1960–2007

Aggregate

0.0387

Kamerschen and Porter [22]

The United State

PA

1973–1998

Residential

0.13

Alberini and Filippini [3]

The United State

PA

1995–2007

Residential

− 0.08 to − 0.15

Narayan and Smyth [26]

Australia

ADL

1959–1972

Residential

− 0.26

Halicioglu [11]

Turkey

ADL

1968–2005

Residential

− 0.33

Ziramba [37]

South Africa

ADL

1978–2005

Residential

− 0.02

Dilaver and Hunt [9]

Turkey

ADL

1960–2008

Residential

− 0.092

Holtedahl and Joutz [12]

Taiwan

VEC

1955–1995

Residential

− 0.154172

Hondroyiannis [13]

Greece

VEC

1986–1999

Residential

− 0.41

Narayan et al. [27]

G7

PC

1978–2003

Residential

− 0.0001

PA partial adjustment, KF Kalman filter, ADL autoregressive distributed lag, VEC vector error-correction model, TVC time-varying cointegrating vector, PC panel cointegration

The model

The partial adjustment model

In this paper, the elasticity of electricity demand is estimated by employing partial adjustment because our data has a small T and a large N (T = 24 and N = 47). A first difference estimator was used to control for individual effects because it is well known that the correlation of individual effects and independent variables causes a dynamic panel bias.

The estimation model is formulated as below.

$$\Delta \ln ELE_{i,t} = \alpha + \beta_{1} \Delta \ln p_{i,t}^{ELE} + \beta_{2} \Delta \ln EMP_{i,t} + \beta_{3} \Delta \ln ELE_{i,t - 1} + \Delta \mu$$
(1)
where \(\Delta\) denotes the first difference operator, \(ln\) represents the natural logarithm, i (i = 1, 2, …, N) stands for the prefecture, and t (t = 1, 2, …, T) means time. The dependent variable, \({ELE}_{{{i},{t}}}\), is the electricity consumption in each industry. Independent variables are defined as follows. \(p_{{i,{\text{t}}}}^{ELE}\) is the real electricity price (yen/kWh)\(,\; {EMP}_{{i,t}}\) is the number of employees, and \(ELE_{i,t - 1}\) is lagged electricity consumption. One of the independent variables is electricity consumption in the previous period, which indicates that electricity demand depends not only on electricity use in the present period but also on use in the previous period. This is because facilities that use electricity cannot be replaced in a single time period but, instead, can only be replaced gradually. The number of employees is a control variable. This represents the scale of an industry. Tanishita [33] estimates the elasticity of the electricity price by using a partial adjustment model based on OLS (ordinary least squares) estimation. However, the paper indicates that a lagged dependent variable has the possibility of endogeneity [16]. In other words, a lagged dependent variable may correlate with the error term. This dynamic panel bias would make the estimated long-run price elasticity higher than the true value.

Electricity demand is affected by other factors beyond those captured by the independent variables, as seen in the relationship between the electricity price and the error term. To avoid these biases, we employ an additional lag of a lagged dependent variable, \(\Delta ELE_{i,t - 2}\) and a lagged electricity price, \(\Delta p_{{i,{\text{t}} - 1}}^{ELE}\) as instrumental variables, and estimate by using the first difference generalized method of moments (FD GMM). \(\beta_{1}\) is the short-run price elasticity of electricity demand and \(\beta_{1} / (1 - \beta_{3} )\) is the long-run price elasticity.

The Kalman filter model

Some papers adopt the autoregressive distributed lag model, which requires a long time-series. Okajima and Okajima [28] pointed out that such a model would require data over more than 20 years. A Kalman filter model is a kind of state-space model, and can estimate a non-stationary model, while an autoregressive distributed lag model can only estimate a stationary model. The advantage of a Kalman filter is that this model does not need a large sample size. In our estimation, the only required data are electricity consumption and electricity price.

The model is expressed as:
$$Y_{i,t} = \beta_{0t} + \mathop \sum \limits_{j = 1}^{k} \beta_{jt} x_{jt} + \varepsilon_{t} ,$$
(2)
$$\beta_{j,t + 1} = \beta_{jt} + \mu_{jt} ,$$
(3)
where \(\mu_{1t} , \ldots ,\mu_{kt}\) are independent of each other, and the regression coefficients, \(\beta_{jt}\) vary over time, and are distributed as a random walk. We can fix the regression coefficients by \(\sigma_{\mu ,0}^{2} = \sigma_{\mu ,1}^{2} = \cdots = \sigma_{\mu ,k}^{2}\). Coefficients \(\beta_{jt}\) in equation [2] are updated by equation [3]. The Kalman filter model employed in this study is described below.
$$\ln ELE_{i,t} = \alpha + \beta \ln p_{t}^{ELE} + Z^{\left( \mu \right)} \mu_{t} + \varepsilon_{t} ,$$
(4)
$$\beta_{t + 1} = \beta_{t} + \upsilon_{t} ,$$
(5)
$$\mu_{t + 1} = \mu_{t} + \eta_{t} ,$$
(6)
where \(ln\) represents the natural logarithm, i (i = 1, 2, …, N) stands for the prefecture, and t (t = 1, 2, …, T) means time. The dependent variable, \({ELE}_{i,t}\), is industrial electricity consumption, the independent variables are \(p_{i,t}^{ELE}\), which is real electricity price (yen/kWh) and \(Z\), which is the trend variable.

Data

To estimate the elasticity of electricity demand, we use data on electricity consumption, electricity price and other control variables. To obtain a correct estimation, the period of the data should be long enough and the sample size should be large enough.

Industrial categories are listed below.

0. Manufacturing

1. Food, beverages, tobacco, and feed

2. Textile mill products

3. Lumber and wood products

4. Pulp, paper, and paper products

5. Printing and related industries

6. Chemical and related products

7. Plastic, rubber, and leather products

8. Ceramic, stone, and clay products

9. Iron, steel, non-ferrous metals and products

10. Machinery

11. Miscellaneous manufacturing industries

Electricity consumption

Electricity consumption data is obtained from the Prefectural Energy Consuming Statistics [1]. This is not primary data; however, it is used to evaluate CO2 emissions and the energy balance of allocated electricity use data. This is done by using the proportion of employees in each industrial sector of electricity consumption for each prefecture. As far as we know, the prefectural energy consumption statistics are the only sectoral electricity consumption data aggregated by prefectures over a long period of time. The observation periods are from 1990 to 2014 (in fiscal years) and the number of samples per year is 47 (which is the number of prefectures).

Electricity price

The electricity price (yen/kwh) is calculated from the electricity sales revenues of the 10 existing electric power companies (Hokkaido, Tohoku, Tokyo, Hokuriku, Chubu, Kansai, Chugoku, Shikoku, Kyushu, and Okinawa) divided by their gross electricity generation, where the revenue includes sales from the commercial sector. The data is obtained from the Federation of Electric Power Companies of Japan [10]. The electricity price in each prefecture is derived from the corresponding electric power companies.2 As of 1999, new electric companies can enter the electricity market. However, we calculated the prices only for the existing companies because the prices of the new companies are not available.3

Control variables

The Ministry of Economy, Trade and Industry [24] surveys manufacturers using questionnaires. This survey contains data on electricity consumption, numbers of employees, salary payments, material uses, outputs, added value and other information. The period covered by these surveys is from 1990 to 2014 (in fiscal years) and the sample size is 47 per year.

In estimating the price elasticity of electricity, we chose the number of employees as an independent variable. Although outputs and added values can be independent variables, those variables have endogeneity with electricity use. Therefore we employed the number of employees as a control variable.

Empirical Results

Cross-sectional dependency test and panel unit root test

There are two panel unit root tests: first generation and second generation. The first-generation panel unit root test requires cross-sectional independency. To test the cross-sectional dependency in the panel data, Pesaran’s cross-sectional dependency test is employed [31]. The results of the tests are presented in Table 2. The estimation model is as below, where the instrumental variables are \(\Delta lnELE_{i,t - 2}\) and \(\Delta lnp_{i,t - 1}^{ELE}\).
$$\Delta \ln ELE_{i,t} = \alpha + \beta_{1} \Delta \ln p_{i,t}^{ELE} + \beta_{2} \Delta \ln EMP_{i,t} + \beta_{3} \Delta \ln ELE_{i,t - 1} + \Delta \mu$$
(7)
Table 2

Pesaran’s test of cross-sectional dependence in panels

 

z statictic

p value

Manufacturing

70.4203

0.0000

Food, beverages, tobacco and feed

64.8231

0.0000

Textile mill products

63.2691

0.0000

Lumber and wood products

62.1269

0.0000

Pulp, paper and paper products

17.0303

0.0000

Printing and allied industries

64.4233

0.0000

Chemical and allied products

40.4038

0.0000

Plastic, rubber and leather products

55.0399

0.0000

Ceramic, stone and clay products

39.7333

0.0000

Iron, steel, non-ferrous metals and products

35.1928

0.0000

Machineries

51.1143

0.0000

Miscellaneous manufacturing industries

50.1331

0.0000

The results reject the null hypothesis that there is no cross-sectional dependence in the data, and as such, the second-generation unit root test is needed.

We employed an augmented Im, Pesaran, and Shin (IPS) test [17, 30] to test the panel unit root. As indicated in Table 3 below, for the electricity price and electricity consumption, the test shows that there is no unit root.
Table 3

Cross-sectionally augmented Im, Pesaran, and Shin (IPS) test

 

Employee

Electricity

Electricity price

Manufacturing

− 2.1035

− 2.0592

− 1.6313

Food, beverages, tobacco and feed

− 1.8941

− 1.8091

− 1.6313

Textile mill products

− 2.2724

− 1.8782

− 1.6313

Lumber and wood products

− 2.3546

− 1.9089

− 1.6313

Pulp, paper and paper products

− 1.9976

− 2.2359

− 1.6313

Printing and allied industries

− 2.4459

− 1.7549

− 1.6313

Chemical and allied products

− 2.3112

− 2.2626

− 1.6313

Plastic, rubber and leather products

− 1.7704

− 1.8105

− 1.6313

Ceramic, stone and clay products

− 2.3369

− 2.2169

− 1.6313

Iron, steel, non-ferrous metals and products

− 2.6437**

− 1.9938

− 1.6313

Machineries

− 1.7818

− 1.8716

− 1.6313

Miscellaneous manufacturing industries

− 3.3085***

− 1.9496

− 1.6313

**Rejection of the null of a unit root at 5% level

***Rejection of the null of a unit root at 1% level

Partial adjustment estimation

Table 4 below illustrates the estimation results of the partial adjustment model. The price elasticity of electricity demand in aggregate manufacturing is − 0.400.
Table 4

Estimation results of the partial adjustment model

 

∆ln (employee)

∆ln (pele)

∆ln (elet_1)

Manufacturing

− 0.07732

− 0.39667***

0.59478***

0.165

0.000

0.000

Food, beverages, tobacco and feed

0.03792

− 0.46817***

0.74899***

0.827

0.000

0.000

Textile mill products

0.04368

− 0.77529***

0.50475***

0.278

0.000

0.000

Lumber and wood products

− 0.02587

− 0.40256***

0.68423***

0.444

0.000

0.000

Pulp, paper and paper products

0.4518***

− 0.56992***

0.53058***

0.002

0.000

0.000

Printing and allied industries

0.22984***

− 0.52982***

0.70556***

0.001

0.000

0.000

Chemical and allied products

− 0.76611**

− 0.14663*

0.72031***

0.010

0.059

0.000

Plastic, rubber and leather products

0.63497**

− 0.70124***

0.64126***

0.012

0.000

0.000

Ceramic, stone and clay products

0.36655***

− 0.70058***

0.41742***

0.007

0.001

0.000

Iron, steel, non-ferrous metals and products

− 0.09363

− 0.25065**

0.40993***

0.606

0.011

0.000

Machineries

− 0.02195

− 0.4846***

0.6859***

0.693

0.000

0.000

Miscellaneous manufacturing industries

0.08995**

− 0.41272***

0.86954***

0.022

0.000

0.000

The individual coefficient is statistically significant at 1% level (***), 5% level (**), 10% level (*)

The more elastic sectors than aggregate manufacturing are textile mill products (− 0.775) followed by plastics, rubber and leather products (− 0.701), ceramic, stone and clay products (− 0.701), pulp, paper and paper products (− 0.570), printing and allied industries (− 0.530), machinery (− 0.485), food, beverages, tobacco and feed (− 0.468), miscellaneous manufacturing industries (− 0.413), and lumber and wood products (− 0.403). On the other hand, the less elastic sector is iron, steel, non-ferrous metals and products (− 0.251). The chemical and allied products (− 0.147) sector is not statistically significant at 5% level.

Table 5 presents the short-run and long-run price elasticities of electricity demand.
Table 5

Short-run and long-run price elasticities of electricity demand

 

Short-run elasticity

Long-run elasticity

Manufacturing

− 0.397

− 0.979

Food, beverages, tobacco and feed

− 0.468

− 1.865

Textile mill products

− 0.775

− 1.565

Lumber and wood products

− 0.403

− 1.275

Pulp, paper and paper products

− 0.570

− 1.214

Printing and allied industries

− 0.530

− 1.799

Chemical and allied products

− 0.147

− 0.524

Plastic, rubber and leather products

− 0.701

− 1.955

Ceramic, stone and clay products

− 0.701

− 1.203

Iron, steel, non-ferrous metals and products

− 0.251

− 0.425

Machineries

− 0.485

− 1.543

Miscellaneous manufacturing industries

− 0.413

− 3.164

Kalman filter estimation

Table 6 presents estimation results of the Kalman filter estimation. We do not find any major differences between the “fluctuate estimation” and “constant estimation”. We refer to the coefficients from the “fluctuate estimation” below. The price elasticity of electricity demand in manufacturing is − 0.28.
Table 6

Estimation results of the Kalman filter model

 

Fluctuate estimation

Constant estimation

Estimate

Std. error

Estimate

Std. error

I1200

Manufacturing

− 0.2778*

0.1921

− 0.2778*

0.1921

I1201

Food, beverages, tobacco and feed

− 0.5305*

0.2781

− 0.5547*

0.3096

I1202

Textile mill products

− 0.0954

0.3155

− 0.0946

0.3156

I1203

Lumber and wood products

− 0.3673*

0.3567

− 0.3674*

0.3568

I1204

Pulp, paper and paper products

− 0.2804*

0.2045

− 0.2804*

0.2045

I1205

Printing and allied industries

− 0.07816

0.4112

− 0.0782

0.4112

I1206

Chemical and allied products

− 0.3577**

0.1593

− 0.3577*

0.1593

I1207

Plastic, rubber and leather products

− 0.6449*

0.3236

− 0.6448*

0.3237

I1208

Ceramic, stone and clay products

0.04937

0.2447

0.0493

0.2447

I1209

Iron, steel, non-ferrous metals and products

− 0.07477

0.0773

− 0.0749

0.0773

I1210

Machineries

− 0.2971*

0.2441

− 0.2971*

0.2441

I1211

Miscellaneous manufacturing industries

− 0.7343**

0.3577

− 0.8458**

0.3950

The individual coefficient is statistically significant at 5% level (**), 10% level (*)

The more elastic sectors than aggregate manufacturing are miscellaneous manufacturing industries (− 0.734) followed by plastic, rubber and leather products (− 0.645), food, beverages, tobacco and feed (− 0.531), lumber and wood products (− 0.367), chemical and allied products (− 0.358), machinery (− 0.297), pulp, paper and paper products (− 0.280). On the other hand, the less elastic sectors are ceramic, stone and clay products (0.049), iron, steel, non-ferrous metals and products (− 0.075), printing and allied industries (− 0.078), textile mill products (− 0.095).

In the Kalman filter model, however, coefficients of sectors other than miscellaneous manufacturing industries and chemical and allied products are statistically insignificant at 5% level.

In the next section, we refer to the results of the partial adjustment model.

Conclusions

In this paper we estimated the price elasticity of electricity demand for each manufacturing industry (major groups) using the partial adjustment and the Kalman filter models.

In general, less price-elastic industries need electricity more. In other words, electricity is a necessary good for inelastic industries. The low elasticity implies that these industries cannot reduce electricity consumption even when electricity prices increase. This implies that a high electricity price is a heavy burden on these companies. Figure 5 illustrates the relationship between electricity consumption per unit of output and price elasticity.
Fig. 5

Electricity consumption per output and price elasticity (partial adjustment)

As stated before, we are not aware of any studies that have calculated Japanese sectoral price-elasticities of electricity demand in peer-reviewed papers, and thus it is difficult to directly compare our results with other econometric studies. We refer to three studies to compare results.

In Blignaut et al. [6], the estimated sectors in the manufacturing industry are iron and steel (− 0.79), non-ferrous metals (− 0.34), chemicals (other) (− 0.24), and manufacturing (other) (− 0.251). These results are consistent with ours in that the iron, steel, non-ferrous metals and products sector is more elastic than the chemical and allied products sector.

In the Ministry of Economy, Trade and Industry [23], sectors that have decreased profits due to increased electricity prices are identified, and their likely profit decrease estimated. However, because the sectoral definition differs from ours, the consistency with our results is ambiguous.

Hosoe [14] simulates the effects of the power crisis on Japanese industrial sectors using a CGE model. The simulation indicated that the power crisis would decrease domestic outputs of the wood, paper and printing, pottery, steel and nonferrous metal and food industries in Japan, and would accelerate foreign direct investment in these sectors. In our estimation of the partial adjustment model, the price-elasticity in the iron, steel, non-ferrous metals and products is low. Because it means it’s impossible for this sector to adjust electricity consumption when electricity price increases, this sector has to decrease their output or increase foreign direct investment. Then their result is consistent with our result in iron, steel, non-ferrous metals and products sector.

There are three studies that estimate Japanese price-elasticities of electricity demand including industrial sectors. Hosoe and Akiyama [15] and Otsuka [29] estimate the industrial and commercial elasticity, the results are − 0.105 to − 0.300 and − 0.034 each. Wang and Mogi [35] estimate the industrial elasticity: the result is − 0.16. We find that the manufacturing sector is more elastic than total industry, and also find that many sectors within manufacturing are more elastic.

Our results showed that price-elasticities vary greatly between different sectors. Policy makers need to understand which sectors are most affected by an increase in electricity prices in order to formulate industrial policies including tariffs, grants, and other industry-specific policies, because an increase in electricity price has the real possibility of accelerating de-industrialization and/or raising the unemployment rate.

Finally, we discuss possible extensions of this study. First, we can straightforwardly extend this study to all industrial categories beyond manufacturing (e.g., construction, services). We are certain that such an exercise will yield many useful findings. This paper focuses on manufacturing industry, because we assumed that this industry is sensitive to electricity prices, and because of the often controversial relationship between electricity prices and global competitiveness.

Second, we can simulate the influence of an increase in electricity prices on each industry’s global competitiveness. We are currently constructing a CGE model to account for this effect. Finally, the cross-elasticity of demand can be examined. In the short-run, an increase in electricity price may increase the use of alternative energy resources, and in the long-run, it may lead to acquiring energy-saving machines, and also investing in private power generation. Since companies which own private power generations can switch to private power generation when the electricity price increases, industrial categories in which many companies introduce private power generation reduce electricity consumption more than actual electricity use. We recognize that we need to examine the cross-elasticity of electricity and alternative energy resources.

In the future, we are planning to study a simulation model which uses the elasticities estimated in this paper to draw more definite conclusions, while we also recognize that it is effective to research the reasons why elasticities are different among industrial sectors.

Footnotes

  1. 1.

    Some studies state that whether the industrial sector is more inelastic than the household sector is ambiguous. Sonoda et al. [32] estimated that the elasticity of the household sector is − 0.219 to − 1.368, the commercial sector is − 0.268 to − 0.943. Kaino [21] estimated long-run elasticities, and found the household sector is − 0.121, and the industrial sector is − 0.033 to − 0.157.

  2. 2.

    Each electric power company covers the prefectures as listed below.

    Hokkaido Electric Power Company: Hokkaido

    Tohoku Electric Power Company: Aomori, Iwate, Miyagi, Akita, and Yamagata

    Tokyo Electric Power Company: Tokyo, Kanagawa, Saitama, Chiba, Tochigi, Ibaragi, Yamanashi, and Shizuoka

    Hokuriku Electric Power Company: Toyama, Ishikawa, Fukui, and Gifu

    Chubu Electric Power Company: Aichi, Nagano, Gifu, Mie, and Shizuoka

    Kansai Electric Power Company: Shiga, Kyoto, Osaka, Hyogo, Nara, and Wakayama

    Chugoku Electric Power Company: Hiroshima, Yamaguchi, Shimane, Tottori, and Okayama

    Shikoku Electric Power Company: Kagawa, Tokushima, Ehime, and Kochi

    Kyushu Electric Power Company: Fukuoka, Nagasaki, Oita, Saga, Miyazaki, Kumamoto, and Kagoshima

    Okinawa Electric Power Company: Okinawa

    *Shizuoka prefecture is covered by both Tokyo and Chubu Electric Power Companies. Therefore, the electricity price of Shizuoka is obtained by taking the average of the prices from Tokyo and Chubu.

  3. 3.

    We should note that company–facing electricity prices are different from the accounting data, because the electricity price which each company faces depends on each company’s electricity consumption volume, load facility, and load factor.

Notes

Acknowledgements

I would like to thank Professor Takashi Yanagawa for dedicated mentoring. I also thank Mr.Teizo Anayama for insightful comments during the 16th International Conference of the Japan Economic Policy Association. Of course, all remaining errors are the author’s responsibility.

References

  1. 1.
    Agency for Natural Resources and Energy. (2016). Electric power statistics. http://www.enecho.meti.go.jp/statistics/electric_power/ep002/results.html (in Japanese).
  2. 2.
    Agency for Natural Resources and Energy (2017) Kaisei FIT hou shikou ni mukete. http://www.meti.go.jp/committee/sougouenergy/shoene_shinene/shin_ene/pdf/017_01_00.pdf (in Japanese).
  3. 3.
    Alberini, M., & Filippini, M. (2011). Response of residential electricity demand to price: The effect of measurement error. Energy Economics, 33, 889–895.CrossRefGoogle Scholar
  4. 4.
    Amusa, H., Amusa, K., & Mabugu, R. (2009). Aggregate demand for electricity in South Africa: An analysis using the bounds testing approach to co-integration. Energy Policy, 37, 4167–4175.CrossRefGoogle Scholar
  5. 5.
    Arisoya, I., & Ozturk, J. (2014). Estimating industrial and residential electricity demand in Turkey: A time varying parameter approach. Energy, 66, 959–964.CrossRefGoogle Scholar
  6. 6.
    Blignaut, J., Inglesi-Lotz, R., & Weideman, J. P. (2015). Sectoral electricity elasticities in South Africa: Before and after the supply crisis of 2008. South African Journal of Science, 111(9/10), 01.CrossRefGoogle Scholar
  7. 7.
    Chang, Y., Kim, S. K., Miller, J. I., Park, J. Y., & Park, S. (2014). Time-varying long-run income and output elasticities of electricity demand with an application to Korea. Energy Economics, 46, 334–347.CrossRefGoogle Scholar
  8. 8.
    Dilaver, Z., & Hunt, L. C. (2011). Industrial electricity demand for Turkey: A structural time series analysis. Energy Economics, 33, 426–436.CrossRefGoogle Scholar
  9. 9.
    Dilaver, Z., & Hunt, L. C. (2011). Modelling and forecasting Turkish residential electricity demand. Energy Policy, 39, 3117–3127.CrossRefGoogle Scholar
  10. 10.
    Federation of Electric Power Companies of Japan. (2017) Statistical information of electric power. http://www.fepc.or.jp/library/data/tokei/index.html. Accessed 30 June 2017.
  11. 11.
    Halicioglu, F. (2007). Residential electricity demand dynamics in Turkey. Energy Economics, 29, 199–210.CrossRefGoogle Scholar
  12. 12.
    Holtedahl, P., & Joutz, F. L. (2004). Residential electricity demand in Taiwan. Energy Economics, 26, 201–224.CrossRefGoogle Scholar
  13. 13.
    Hondroyiannis, G. (2004). Estimating residential demand for electricity in Greece. Energy Economics, 26, 319–334.CrossRefGoogle Scholar
  14. 14.
    Hosoe, N. (2014). Japanese manufacturing facing post-Fukushima power crisis: a dynamic computable general equilibrium analysis with foreign direct investment. Applied Economics, 46(17), 2010–2020.CrossRefGoogle Scholar
  15. 15.
    Hosoe, N., & Akiyama, S. (2009). Regional electric power demand elasticities of Japan’s industrial and commercial sectors. Energy Policy, 37, 4313–4319.CrossRefGoogle Scholar
  16. 16.
    Hsiao, C. (2002). Analysis of panel data, Edition 2. UK: Cambridge University Press.Google Scholar
  17. 17.
    Ima, K. S., Pesaran, M. H., & Shin, Y. (2003). Testing for unit roots in heterogeneous panels. Journal of Econometrics, 115, 53–74.CrossRefGoogle Scholar
  18. 18.
    Inglesi-Lotz, R. (2011). The evolution of price elasticity of electricity demand in South Africa: A Kalman filter application. Energy Policy, 39, 3690–3696.CrossRefGoogle Scholar
  19. 19.
    Inglesi-Lotz, R., & Blignaut, J. N. (2011). Estimating the price elasticity of demand for electricity by sector in South Africa. SAJEMS NS, 14(4), 449–465.CrossRefGoogle Scholar
  20. 20.
    Institute of Energy Economics, Japan. (2017). EDMC handbook of Japan’s and world energy and economic statistics 2017. The Energy Conservation Center, Japan.Google Scholar
  21. 21.
    Kaino, K. (2002). Development of energy policies. In 3rd Technical Committee on Economic Analysis of Environmental Tax in General Assembly and Global Environment Committee on Central Environment Council (in Japanese). Google Scholar
  22. 22.
    Kamerschen, D. R., & Porter, D. V. (2004). The demand for residential, industrial and total electricity, 1973–1998. Energy Economics, 26, 87–100.CrossRefGoogle Scholar
  23. 23.
    Ministry of Economy, Trade and Industry (METI). (2011) White Paper on Manufacturing Industries (Monodzukuri) 2011. Ch. 2—Sec. 2. http://www.meti.go.jp/report/whitepaper/mono/2011/pdf/honbun02_02_01.pdf (in Japanese).
  24. 24.
    Ministry of Economy, Trade and Industry (METI). (2016) Economic Census for business activity. http://www.meti.go.jp/english/statistics/tyo/census/index.html
  25. 25.
    Nakajima, T. (2010). The residential demand for electricity in Japan: An examination using empirical panel analysis techniques. Journal of Asian Economics, 21, 412–420.CrossRefGoogle Scholar
  26. 26.
    Narayan, P. K., & Smyth, R. (2005). Electricity consumption, employment and real income in Australia evidence from multivariate Granger causality tests. Energy Policy, 33, 1109–1116.CrossRefGoogle Scholar
  27. 27.
    Narayan, P. K., Smyth, R., & Prasad, A. (2007). Electricity consumption in G7 countries: A panel co-integration analysis of residential demand elasticities. Energy Policy, 35, 4485–4494.CrossRefGoogle Scholar
  28. 28.
    Okajima, S., & Okajima, H. (2013). Estimation of Japanese price elasticities of residential electricity demand, 1990-2007. Energy Economics, 40, 433–440.CrossRefGoogle Scholar
  29. 29.
    Otsuka, A. (2015). Demand for industrial and commercial electricity: evidence from Japan. Journal of Economic Structures, 4, 9.CrossRefGoogle Scholar
  30. 30.
    Pesaran, M. H. (2007). A simple panel unit root test in the presence of cross-section dependence. Journal of Applied Econometrics, 22(2), 265–312.CrossRefGoogle Scholar
  31. 31.
    Pesaran, M. H. (2015). Testing weak cross-sectional dependence in large panels. Econometric Review, 34(6–10), 1089–1117.CrossRefGoogle Scholar
  32. 32.
    Sonoda, K et al. (1999). Estimating price elasticity in considering the stagnation of energy price. Essays form the 18th Energy system, Economy, Environment conference (in Japanese). Google Scholar
  33. 33.
    Tanishita, M. (2009). Estimation of regional price elasticities of household’s electricity demand (in Japanese). Journal of Japan Society of Energy and Resources, 30, 1–7.Google Scholar
  34. 34.
    Tamechika, H. (2014). Residential electricity demand in Japan. Istanbul, Turkey: Proceeding of Fifth World Congress of Environmental and Resource Economists.Google Scholar
  35. 35.
    Wang, N., & Mogi, G. (2017). Industrial and residential electricity demand dynamics in Japan: How did price and income elasticities evolve from 1989 to 2014? Energy Policy, 106, 233–243.CrossRefGoogle Scholar
  36. 36.
    Zachariadis, T., & Pashourtidou, N. (2007). An empirical analysis of electricity consumption in Cyprus. Energy Economics, 29, 183–198.CrossRefGoogle Scholar
  37. 37.
    Ziramba, E. (2008). The demand for residential electricity in South Africa. Energy Policy, 36, 3460–3466.CrossRefGoogle Scholar

Copyright information

© Japan Economic Policy Association (JEPA) 2018

Authors and Affiliations

  1. 1.Graduate School of EconomicsKobe UniversityKobeJapan

Personalised recommendations