Expert workshop on the creation and uses of combined environmental and economic performance datasets at the micro-level - 10-11 July 2018 - OECD, Paris
3. • Policy start in 2010
• Part of Climate Change Act 2008
• First announced in 2007 white paper
• Basic idea: target carbon of firms that are fairly
carbon intensive but not EUETS covered (~10% of UK
emissions)
• Several elements:
• Carbon permit required @ £12/tCO2 from
electricity and gas
• Performance league table
• Senior management sign-off
• Revenue recycling: top performers receive subsidy
back Never implemented
Abolished after 2012
Hence, effectively a carbon tax
4. • UK already had a carbon tax: Climate Change Levy
• Also generous exemption scheme (Climate Change
Agreements, CCA), which extends to CRC
• Hence CRC greatly increases inefficient carbon price
heterogeneity: some firms pay £30 ($50) more than
others per t of CO2
• Looks like bad public policy…..
• ….but great setting for research
5. • Information declarers: firms with at least one half
hourly meter
• Information declarers had to submit data on
electricity consumption for 2008
• Policy eventually imposed on firms with more than 6
GWh of electricity (firm wide) in 2008
6. • We were commissioned by DECC in 2014/15 to figure
out
• Comparing covered firm with comparable non-
covered seems a good start
• However: Comparable in terms of level will be
difficult
• …..or not?
• DECC meter point data: we match meter points
(MP) rather than firms (some firms consist of 100s
of MPs)
We had to delete all data at end
of project
7. Density
0 1 2 3 4
GWh
Info Declarers CRC
Electricity consumption in 2008
(GWh)
Sample Mean p50 p95 Sum Meters
Analysis Sample 1 non CRC 0.64 0.36 2.24 13,530.46 21,256
CRC 1.00 0.30 3.86 75,194.62 74,848
8. • Control group consists of information declarers
• Could be affected by policy:
• Uncertainty pre threshold announcement
• Avoidance of regulation in Phase 2 (2014-2015)
• Should likely lead to downward bias of impact
estimates
9. • Considered regression discontinuity
• Matched Diff-in-Diff
ln 𝐸𝑖𝑡 − ln 𝐸𝑖2008 = 𝛽𝑡𝐶𝑅𝐶 × 𝐶𝑅𝐶𝑖 + 𝛽𝑡 + 𝛽 𝑀𝑡 × 𝑀𝑖
𝑀
+ 𝜖𝑖𝑡
ln 𝐸𝑖𝑡 − ln 𝐸𝑖2008 =
𝛽 𝑃𝑟𝑒,𝐶𝑅𝐶I 𝑡 < 2010 × 𝐶𝑅𝐶𝑖 + 𝛽 𝑃𝑜𝑠𝑡,𝐶𝑅𝐶I 𝑡 ≥ 2010 × 𝐶𝑅𝐶𝑖 + 𝛽𝑡 +
𝛽 𝑀𝑡 × 𝑀𝑖𝑀 + 𝜖𝑖𝑡
• In most basic regression we match on meter point
level energy consumption bands (in 2008) only
• By going to higher aggregation levels we can merge
in other variables.
12. For a subset of the population (about 25% of 2008 energy consumption)
• Matching by building types
• Availability of gas consumption data
13.
14.
15.
16.
17. • Robust Evidence of an impact of CRC on
electricity consumption (between 3 and 7%)
• Impact stronger for very big (and very small)
consumption points
• Evidence of even larger impact for gas >10%
seems less robust, however
• Impact heterogeneous across sectors but strong
and significant impact in most important
sectors
• Some evidence of an impact on intensity
18. • Which aspect of the CRC package is driving the
results?
• No direct econometric evidence but interviews
with stakeholders suggest the financial
component
• Assuming full impact is price effect suggests an
energy price elasticity of 0.3 to 0.52
19. • Regain access to the data
• Get access to more data: Non information
declarers
• The future of CRC: Bleak!
20.
21.
22.
23. Electricity consumption in 2008
(GWh)
Sample Mean p50 p95 Sum Meters
CRC
meters
All data 1.19 0.33 4.36 127,936.68 107,395 84,683
Meeting CRC criteria 1.34 0.32 5.17 113,374.32 84,683 84,683
Analysis Sample 1 0.92 0.32 3.42 88,725.08 96,104 74,848
versus the whole population: i.e. about ¼ of
electricity consumption (probably) slightly
biased towards larger firms
24.
25. Electricity consumption in 2008
(GWh)
Sample Mean p50 p95 Sum Meters
CRC
meters
All data 1.19 0.33 4.36 127,936.68 107,395 84,683
Meeting CRC criteria 1.34 0.32 5.17 113,374.32 84,683 84,683
Analysis Sample 1 0.92 0.32 3.42 88,725.08 96,104 74,848
versus the whole population: i.e. about 5% of
electricity consumption (probably) biased
towards larger firms