Econometric Modelling of Technical Progress, Consumer Tastes: An Application to Petrol Demand in Jap

1
Econometric Modelling of Technical Progress,
Consumer TastesAn Application to Petrol Demand
in Japan

• Yasushi Ninomiya
• Institute for Global Environmental Strategies
• 22 June 2004
• IEW 2004, IEA, Paris

2
Contents of this presentation

• Research questions
• Background of this research
• Model description
• Application to petrol demand in Japan
• Estimated results and findings
• Conclusions

3
Research questions

• How far does technological progress (improvement
  of fuel efficiency) contribute to reduce petrol
  demand?
• Does consumer taste to use petrol reflect recent
  growing concern about global warming and
  establishment of climate change mitigation
  framework?
• What are the implications of these results?
• This research attempts to answer these questions
  numerically

4
Background

• Energy demand is derived demand
• Efficiency improvement of energy appliance
  substantially affect demand for energy
• In addition, change in consumer tastes also
  affect demand for energy
• These create difficulty of energy demand model
• Energy demand model needs to incorporate these
  factors Rather unexplored area

5
Background

• ln E a ln Y b ln P (1)
• E Energy demand Y GDP P Energy price

?
- ? Tmp
Income elasticity around 0.51.0
Temperature
Technical progress
Price elasticity-0.1-0.5
Changes in consumer taste
? Changes in consumer taste ? Unobservable and
data is hardly available often ignored in
energy demand modelling
6
Structural Time Series Model

• Energy demand
• Trend component Seasonal component Exogenous
  variable Residuals
• STSM can easily separate out trend component and
  seasonal component from exogenous variables
  effectively

7
Structural Time Series Model

• Modelling of trend component
• In STSM, a trend component ? is flexibly modelled
  by the stochastic properties ? can shift in
  level and change its direction by a stochastic
  slope component ?
• ?t ?t-1 ?t-1 ?t ?t NID (0, ??)
  (2)

2
2
?t ?t-1 ?t ?t NID (0, ??) (3)
8
Structural Time Series Model

• If the variances of ?t and ?t (hyperparameters)
  are estimated as zero, trend component is not
  stochastic and reverts into the following
  conventional linear time trend
• ?t ? ?t (4)
• This implies that a linear time trend is a
  restricted version of a stochastic trend
• ? This restriction is statistically testable

9
Structural Time Series Model
Linear time trend
Stochastic trend
10
Structural Time Series Model

• Equation (2)(3) and log-linear energy demand
  equation (1) are set in Space State form and the
  Kalman filter can estimates the optimal
  estimators for the unknown parameters
• A(L)lnEt ?t B(L) lnYt C(L)lnPt ?t ?t
  (6)
• ? Observation equation
• ?t ?t-1 ?t-1 ?t
• ?t ?t-1 ?t ?State equation (7)
• S(L)?t ?t

11
Structural Time Series Model
Observation equation
(8)
State equation
(9)
12
Application to petrol demand in Japan

• Why petrol demand is chosen?
• Most rapidly growing energy demand (1990-2003
  35 increased)
• Homogeneous usage (e.g. car) ? technological
  progress can be well approximated by average
  fuel-efficiency data of car stock
• Change in consumer taste to use petrol is of
  particular interest since there is no major
  substitution for passenger car use

13
Data used for estimation (1976q1-2004q1)
Real GDP
Petrol consumption
All in log
Real petrol price
Average fuel efficiency of car stock (Litre per
100Km)
14
Summary of estimated parameters

• LR GDP elasticity 0.3
• LR price elasticity -0.1
• LR fuel efficiency elasticity 0.9
• Large impact of average fuel efficiency of car
  stock on petrol demand
• Relatively small impact of GDP growth on the
  demand new tendency with recent sample data
  period

15
Changes in consumer tastes (after GDP, petrol
price, fuel-efficiency effects are controlled)
Growth rate 1.9 p.a. at 2003
16
Changes in fuel efficiency of car stock
Litre/100Km in log
Growth rate -1.15 p.a. at 2003
17
Findings from the estimated result

• The result shows that changes in consumer tastes
  continuously boosted up the petrol demand by 1.9
  p.a. over the whole sample period
• Despite growing concern about global warming
  recent years and ratification of Kyoto Protocol,
  there has been little impact on consumer taste
• There is no indication that actual consumers
  behaviour becomes to be more environmental
  friendly over the past decade

18
Findings from the estimated result

• It implies that consumers cannot change their
  behaviour within a short period of time
• Therefore, given unchanged consumer behaviour and
  expected GDP growth in the future, petrol demand
  would inevitably increase
• In order to reduce, at least to stabilise, the
  petrol demand, it is absolutely necessary to
  accelerate the improvement rate of fuel
  efficiency of car stock which is currently 1.15
  p.a..

19
Basic assumptions for forecast
Petrol price Remains at the 2004q1 level
GDP2 p.a. growth
Seasonal variations set at the latest pattern
Fuel efficiency improvement 1.15 p.a. (same as
the current rate)
20
Forecast of petrol demand with current rate of
fuel efficiency improvement
18 above the 2003 level at 2012
2003 level
Forecast
Actual
21
Average fuel efficiency of car stock
Actual
Assumptions
Fuel efficiency Improvement by 1.15 p.a
(2004-12)
Litre/100Km
Fuel efficiency Improvement by 1.2 p.a
(2004) 2 p.a.(2005-06) 3 p.a.(2007-12)
22
Forecast of petrol demand with higher rate of
fuel efficiency improvement
9 reduction at 2012
With higher efficiency improvement rate
Actual
Forecast
23
Conclusions (1)

• Technical progress (improvement of fuel
  efficiency of car stock) does significantly
  contribute reduction of petrol demand in Japan
• However, change in consumer taste boosted up the
  petrol demand by 1.9 p.a. which is large enough
  to cancel out the contribution of technical
  progress
• The consumer taste has been little affected by
  recent growing concern about global warming and
  other environmental issues

24
Conclusions (2)

• In order to reduce, or at least stabilise, the
  petrol demand, it is necessary to increase the
  improvement rate of average fuel efficiency of
  car stock which is currently 1.15 p.a..
• Forecasting analysis implies that, even with 2
  real GDP growth and unchanged consumer taste, it
  is possible to stabilise the petrol demand at
  2007 level if fuel efficiency improvement rate is
  set at 2 to 3 p.a. between 2005 and 2012.