1. Email: okagawa.azusa@nies.go.jp
Econometric analysis of key factors contributing to energy intensity changes
Azusa OKAGAWA
National Institute for Environmental Studies
Good afternoon, everyone. I am Azusa Okagawa from NIES, the National Institute for Environmental Studies in Japan.
Today I’d like to talk about our econometric analysis of the reason for energy intensity changes of Japan.

2. Outline Background Purpose of the study Econometric model Result
Conclusion
This is the outline of my talk.
First, I will brabrabrabra…

3. Background
Japan’s mid-term target is a reduction in GHG of 15% from 2005 level by 2020.
Prime Minister Taro ASO (10 June, 2009)
“The target will be achieved through energy conservation measures and not carbon emission credits.”
The key factor will be technical innovation and diffusion.
Last year Japanese government committed Japan to reduce GHG of 60-80% by 2050.
Last week Prime Minister Aso announced that Japan’s mid-term target is a reduction in GHG of 15% from 2005 level by This target will be achieved through energy conservation measures and not carbon emission credit.
So, the key issue is technical innovation and diffusion.

4. Energy intensity improvement
Three channels
Inputs substitution
Innovation
Autonomous energy efficiency improvement (AEEI)
Induced technical change (ITC)
Change in the mix of industries

5. Energy intensity and factor prices
This figure shows the relationship between energy intensity and energy price of a whole economy.
Energy intensity improved by energy price increasing in oil crisis, but recently it is getting worse besides energy price compared to labor and capital price were increasing.

6. Energy intensity and capital
Energy can be substituted by capital in a economic model, but is it true? I thought I should distinguish capitals because one might be energy using, but others are energy-saving, it might depends on industries.

7. Purpose of our study
To investigate key factors contributing to energy efficiency change of Japanese economy
Three factors:
Substitution between factors
Technological innovation
Autonomous energy efficiency improvement
Induced by energy cost increase
Character of assets

8. Econometric model
The model to be estimated is simultaneous equation like this.
These equations are derived from producers’ cost minimization problem.
I assumed short-run restricted variable cost function with dynamic adjustment of quasi-fixed assets, using a quadratic approximation. And then, by Shepherd’s Lemma, I derived the optimal conditional short-run input demand functions for energy, materials, services and labor.
Using alpha eei and …., short- and long-term elasticities can be calculated.
E: energy Y: output M: materials S: services L: labor
t: time trend p: input price normalized by wage
x: quasi-fixed assets normalized by output

9. Knowledge capital Accumulation of energy price increase : Decay rate
: Diffusion rate

10. Econometric model Short-run elasticity Long-run elasticity
: Input substitution
: AEEI
: Induced tech change
of knowledge capital
Long-run elasticity
where
Short-term elasticity and long-term elasticity can be calculated using estimate results.
Epsilon evi indicates how much increase/decrease energy intensity when variable price increases.
Epsilon Eti is change of energy intensity when time goes by, that is exogenous technical progress as a time trend.
Mu Eti is elasticity for ITC of knowledge capital caused by contemporaneous energy price shocks.
Using equilibrium condition, optimal quasi-fixed assets are expressed by estimated parameters and original data.
This xi is the difference between short-run and long-run elasticities, which might make the production process more flexible, that is, input substitution easier.
The last is average long-term effect through the change of knowledge and other capital accumulation.
where

11. Data EU-KLEMS EU committee Time series data of 71 industries
Factor inputs and its prices
Outputs and its prices
Capital stocks
Aggregated into 28 industries
European countries, US, Korea and Japan
Quasi fixed assets
ICT, transport equipment, other machinery and other consumption
I used the EU-KLEMS dataset formed by the EU committee.
This is the time series data of 71 industries for OECD countries. I estimated 28 industries which covers Japanese economy.
I used capital stock data of ICT, transport eq, other machinery and other consumption as quasi-fixed assets.

12. Knowledge capital accumulation
Decay and diffusion rate
Using the value Sue Wing (2008) estimated
This is a graph of knowledge capital data for total industries. I used the value of delay and diffusion rate in Sue Wing (2008) for each industry.

13. Estimation results
This is the estimation results. I picked up 8 of 28 industries.
We can find significant relation between energy intensity and factor prices. I expected negative elasticities of energy prices for energy intensive industries, but in chemical, basic metals and electricity, gas and water industries, I could not get significant relations.

14. Long- and short-run elasticity
Higher intensiveness of ICT brings higher intensiveness of energy in a half of industries.
Transport eq. brings energy efficiency improvement in more than a half of industries.
Other consumption capital are energy-using in many industries.
Long-run elasticities of factor prices are smaller than short-run elasticities in many industries.
Knowledge capital is energy-using …
This is the using estimation results.
Capitals have the variety of impacts between industries and we cannot say which capital is energy- saving or using.

15. Median and summary of results
Carbon pricing improved energy efficiency for these 35 years.
Materials and energy were substitutable in a short-run, but it decreased energy intensity in a long-term.
AEEI was positive and did not help energy efficiency improvement.
Except for some industries, we might not able to expect tech progress so much in short- and long-term.

16. Conclusion The factor of energy intensity improvement
Energy cost increasing
Effective to improve energy intensity
Production plants and equipments
Variety
Transport eq. improved energy efficiency in many industries.
AEEI
Averagely positive
Knowledge capital

17. Future work Estimation of delta 1 and delta 2
Patent data as a proxy variable of knowledge stock
おまけ（見せない）

18. Summary of result (1) Positive Negative EE
CKREFOIL, MANUF, HOTELRES, FIN, EDU (5)
AGR, MIN, FOOD, WOOD, PULP, RUBBER, O_MIN, TRANS, POSTEL, PUB, HEALTH (11) EM
AGR, CONST, O_MIN, BMET, MANUF, EGW, HOTELRES, POSTEL, HEALTH, PSERV (10)
WOOD, PULP, CKREFOIL, RUBBER, MACH, ELEQ. TREQ, BSERV (8) ES
CKREFOIL, MANUF, EGW, TRADE, EDU (5)
AGR, WOOD, BMET, POSTEL, PUBSERV (5) ET
AGR, TXT, BMET, TRANS, POSTEL, ESTATE, EDU, HEALTH, PSERV (9)
FOOD, PULP, CHEM, MANUF (4)
If energy price increases, energy intensity will decrease in energy intensive industries.

19. Summary of result (2) Positive Negative E, ICT
FOOD, PULP, CHEMI, BMET, MACH, EGW, HTLRES, BSERV, EDU (9)
MIN, TXT, ELEQ, TRANS, POSTEL, FIN, PUBSERV (7)
E, Transport Equipment
AGR, RUBBER, MACH, TRADE, HEALTH, PSERV (6)
CONST, FOOD, TRANS, HTLRES, POSTEL, ESTATE, BSERV, PUBSERV, PSERV (9)
E, Other Machinery
MIN, FOOD, CKREFOIL, MACH, MANUF, TRANS, POSTEL, FIN, ESTATE, BSERV, EDU, HEALTH (12)
PULP, RUBBER, OMIN, PUBSERV (4)
E, Other Consumption
FOOD, PULP, RUBBER, OMIN, BMET, MACH, ESTATE, BSERV, HEALTH (9)
MIN, CONST, MANUF, TRANS, FIN, BSERV, EDU (7)
E, Other
TXT, RUBBER, OMIN, MACH, MANUF, TRANS, TRADE, POSTEL, FIN, PUBSERV (10)
FOOD, PULP, CKREFOIL, BMET, ESTATE, HEALTH (6)
E, Knowledge
CONST, CKREFOIL, MACH, MANUF, TRANS, FIN (6)
PULP, OMIN,TREQ, EGW, ESTATE (5)