Expert workshop on the creation and uses of combined environmental and economic performance datasets at the micro-level - 10-11 July 2018 - OECD, Paris
2. • Concerns that there is an energy efficiency paradox
or gap
• Or perhaps not? Maybe costs and benefits are wrong
in bottom up paradox calculations?
• Actual performance data should give us a better idea
➞ Benchmarking
3. • Compare energy intensity of different firms
• Energy (Expenditure) per
• Output
• Revenue
• Employee
• Measures of paradox:
• Spread of energy intensity
• Counterfactual improvement: Leader
Laggard
8. • Interesting for policy makers: win win potential
• Various existing policies focus explicitly on this; e.g.:
• UK Carbon Trust, French ADEME: Advice
businesses on how to reduce costs and energy
consumption
• UK Carbon Reduction Commitment (CRC):
benchmarking exercise with financial transfers
from laggards to leaders (at least initially)
• EU ETS: Permit allocation on the basis of
performance benchmarks.
To motivate and evaluate such policies, measurement of the
severity of the energy paradox is key
9. Cobb-Douglas production function
with firm specific energy intensity
Firm specific TFP shifter
Log linear demand with firms specific demand
shifter (quality, consumer valuation)
10. Most datasets only have revenue data
Markup parameter μ=1/(1-1/η)
rit = qit + pit =
1
m
qit + lit( )
Using demand
function
=
aEit
m
eit - kit( )+
aMit
m
mit - kit( )+
aLit
m
Lit - kit( )+
g
m
kit +
1
m
wit
Using production function
Composite demand &
technology shock a+λ
11. From short run profit maximisationsEit =
aEit
m
=
EitWEit
Rit
cit = rit - sEit eit - kit( )- sMit mit - kit( )- sLit Lit - kit( )
Computable from data
12. Below we compute this holding
parameters fixed at the 3 digit sector
level
sEit =
aEit
m
=
EitWEit
Rit
We compare the distribution of
economic efficiency to the distribution of
energy intensity at the sectoral level
13. Average energy intensity of firms with
above median productivity and below
median energy intensity
Sector specific
14. By how much does energy consumption reduce if below
median productive and above median energy intensive firms
had the energy intensity of the benchmark – ceteris paribus
15. • Production data for French manufacturing firms:
Enquete Annuelle des Entreprise (EAE) & Enquete
Annuelle sur les Consommations d'Energie dans
l'Industrie (EACEI)
• We focus (initially) on firms regulated by the EU ETS
• Gives an idea of the relevance of benchmarking
within the ETS
• Ensures we are dealing with comparable firms
• Unbalanced sample of ~300 firms, 1996-2007
24. • New approach to measure energy gap from common
firm level data
• Suggests there is a reduction potential of about 10%
within ETS manufacturing firms
• Wide variation between sectors (0 to 25%)
• Naïve approach would yield vastly larger estimates
and different ranking of industries
25. • Compute for more datasets, samples & policies
• Compute benchmark and counterfactual in
alternative ways
• More sophisticated model of firms; e.g.
• Allowing for firm specific variations in market
power
• Allow for capital stock age, vintages
• Size/Rebound effects?
• Market structure?
Thanks – r.martin@imperial.ac.uk