Primary energy demand in Japan: an empirical analysis of long-term trends and future CO2 emissions
Introduction
For the past one hundred years, Japan has experienced dramatic changes in economic and social structure. People’s life styles, as well as in its political system, have changed beyond all recognition. Between 1900 and 2000 real per capita GNP in Japan increased by over 15 times, from a million to 3.9 million Yen (at 1990 prices). This rapid economic growth is sometimes referred to as a ‘miracle’ in comparison to other OECD countries (Burda and Wyplosz, 1993, p. 152). For example, over the same period, real per capita GNP increased by only four to five times in the UK and US, and by around seven times in Germany and France (Maddison, 1989, EDMC, 2003). This indicates that Japan passed through many stages, from an economically developing country to an advanced country within a relatively short period of time.
Energy consumption in Japan has also risen drastically over the past hundred years. Between 1900 and 2000, primary energy consumption per capita expanded by about 18 times, from 2.4 million kcal to 44.0 million kcal.1 This growth was also much faster than other OECD countries. Primary energy consumption in the UK, for instance, increased by just 2.3 times over a similar period (Fouquet and Pearson, 1998).
This fast growth in energy consumption has been accompanied by a significant increase in CO2 emissions contributing to the environmental problem of global warming. Under the Kyoto Protocol, Japan committed to reducing its greenhouse gas (GHG) emissions (of which 85% is CO2) by 6% on average over the first commitment period (2008–2012) compared to its 1990 emissions level. To achieve this overall GHG reduction target, the Japanese Government's target is to reduce CO2 emissions generated by energy consumption to the same level as that in 1990 during the commitment period.2
A substantial number of econometric studies on energy demand have been conducted using historical data; see Atkinson and Manning (1995) for a survey of developed countries and Dahl (1994) for a survey of developing countries. However, very few of them use long-term data covering more than 50 years. For example, Stern (2000) uses US annual energy data for 1948–1994, and Bentzén and Engsted (1993) uses Danish annual data for 1948–1990; both use considerably longer data than other studies but still less than 50 years. The likely reason for the majority of previous studies not using longer samples is the lack of reliable and consistent energy data for long periods.
Fortunately such a long data set, in excess of 100 years, is available for Japan. It is therefore of interest to investigate the relationships between energy demand, output/income, and the energy price over this period. If a stable long-run relationship can be found between these variables, it will provide a useful analytical tool for considering the past and the future, in particular: (a) to provide interesting insights into the evolution of Japanese energy demand as the economy grew from a developing to an advanced stage; (b) to analyse whether the relationship between energy, GNP and the real energy price has been affected by the dramatic change in the mode of energy usage over the past century such as the shift in the main energy resource from wood and coal to petroleum (as illustrated in Fig. 1); and (c) to provide reliable forecasts of future Japanese energy demand (and hence CO2 emissions) which is crucial given Japan's commitment to meeting the Kyoto target.
This paper therefore estimates the long-run relationship between energy demand, GNP3 and the real energy price in Japan using data for the period 1887–2001 to allow a historical analysis of the past and to create future scenarios for energy consumption and CO2 emissions and is organised as follows. The next section describes the data and Section 3 introduces the model employed in this study. Section 4 reports the estimated results with the historical analysis followed by Section 5 that presents the forecasts of results based on the preferred model in the previous section. The final section concludes the paper.
Section snippets
Data
The annual data for primary energy consumption per capita (1000 kcal) in Japan between 1887 and 2001 are taken from EDMC (2003). Fig. 2 shows a plot of this series in logarithms. World War II and the subsequent economic disruption caused the huge fall during the 1940s. In addition, stagnation during the late 1970s and the early 1980s occurred after the oil crisis, becoming stable at the end of the sample period. Except for these periods, energy consumption increased almost constantly for over
Model
It is assumed that aggregated energy demand is defined by the following function:where E is the aggregate energy demand, Y is GNP, P is the aggregate real energy price and μ is the underlying energy demand trend (UEDT).
Furthermore, following Hunt et al (2003a), Hunt et al (2003b), the function is specified as a combination of a log linear autoregressive distributed lag (ARDL) model and Harvey’s structural time series model within a space state form framework (Charemza and Deadman,
Results
Table 1 reports a summary of the estimated results and diagnostics of each model.
Model 1, with the stochastic trend, passes all the diagnostic tests comfortably and there is no sign of mis-specification. The estimated long-run income elasticity is just over unity indicating the existence of a long-run steady relationship between energy demand and GNP over the past hundred years. In contrast, the estimated price elasticity is −0.18 suggesting that the impact of a price change on the demand is
Energy consumption
As stated above, the estimated model in the final column of Table 2 was used to drive the future scenarios. Since GNP is the predominant driving factor of Japanese energy demand, four cases are considered as follows:
- (1)
a low GNP growth case (denoted by ‘Low’);
- (2)
a medium GNP growth case (denoted by ‘Medium’);
- (3)
a high GNP growth case (denoted by ‘High’); and
- (4)
the Japanese Government plan case (denoted by ‘Gov’).
The growth rates in each scenario are summarised in Table 4. The Japanese Government plan
Summary and conclusion
This paper has examined the long-run linkage between energy demand, GNP, and energy price for Japan using annual data between 1887 and 2001. It is found that, if an Underlying Energy Demand Trend (UEDT) is appropriately incorporated, the econometric model produces a long-run income elasticity of about unity and a lower (in absolute term) long-run price elasticity of –0.2; with no signs of model mis-specification. The UEDT is found to be stochastic with an inverse U-shape. The estimated UEDT
Acknowledgements
We would like to thank an anonymous referee for very helpful comments on an earlier draft. The authors are, of course, responsible for all errors and omissions.
References (30)
- et al.
Short- and long-run elasticities in energy demanda cointegration approach
Energy Economics
(1993) - et al.
An empirical analysis of economic fluctuations in Japan1885–1940
Japan and the World Economy
(2000) - et al.
Underlying trends and seasonality in UK energy demanda sectoral analysis
Energy Economics
(2003) - et al.
Production and high energy pricesa case of some pan-pacific countries
Energy Economics
(1990) - et al.
Econometric modelling of international carbon tax regimes
Energy Economics
(1995) A multivariate cointegration analysis of the role of energy in the US macroeconomy
Energy Economics
(2000)Is GNP-energy model logical?
Energy Policy
(2001)The reliability of aggregated energy demand functions
Energy Economics
(1989)- et al.
A Survey of international energy elasticities
The Practice of EconometricsClassic and Contemporary
(1991)