Modelling of Distributional Impacts of Energy Subsidy Reforms: An Illustration with Indonesia - Presentation by Olivier Durand-Lasserve, OECD

1. MODELLING OF
DISTRIBUTIONAL IMPACTS OF
ENERGY SUBSIDY REFORMS:
AN ILLUSTRATION WITH
INDONESIA
Olivier Durand-Lasserve, Economist/Policy Analyst,
OECD Environment Directorate – IEA Directorate of Global
Energy Economics
Expert seminar organised by the OECD and Indonesia’s
Ministry of Finance
4-5 November 2015

2. Adding distributional aspects to the
economic and environmental analysis
2
Modelling assessment of impacts of FFS reforms in Indonesia
• Macroeconomic & environmental: IEA (1999), Magné et al (2014)
• Macroeconomic & distributional: Clements et al (2007), Yusuf and
Resosudarmo (2008), Widod et al. (2012)
• Distributional: World Bank (2006, 2011), Dartanto (2013)
Construction of a dedicated framework to deal
simultaneously with all 3 dimensions energy subsidy reforms
• Environmental
• Macroeconomic
• Distributional

3. Macro model Microsimulation models
The analytical framework
Economy -wide effects
• Market prices
• Wages and rents
• Taxes and transfers
Household-level effects
• Income
• Expenditures
• Demand adjustment
Fossil fuel consumption subsidy phase out
Economic
Efficiency
Environmental
effectiveness
Distributional
effect 3

4. The energy consumption subsidy phase
out scenarios
Phase out of energy consumption subsidies in Indonesia from
2012 to 2020
• Electricity, gasoline, diesel, kerosene, LPG
• For firms and households
3 different redistribution schemes:
1. Cash transfer: all the households receive identical payment
2. Food subsidies: food & agricultural products are subsidized
3. Labour support: households receive payments proportional
to their labour income
4

5. Strong emissions reduction: ~8% for GHG for all scenarios
Results: GHG impacts
5
Note: The emission reductions exclude emissions from deforestation, which are large in
Indonesia, but highly uncertain and for which the model cannot make reliable projections
-9
-8
-7
-6
-5
-4
-3
-2
-1
0
2012 2013 2014 2015 2016 2017 2018 2019 2020
GHG emissions (% of baseline)
Cash transfers
Labour support
Food subsidies
Source: ENV-Linkages

6. Results: macroeconomic impacts
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
Real GDP (% of baseline)
Cash transfers
Labour support
Food subsidies
Source: ENV-Linkages
6

7. Results: distributional impacts
-2
0
2
4
6
8
10
12
14
1 11 21 31 41 51 61 71 81 91
Income centile groups
Real income variation in 2020 (% baseline)
Cash transfers
Labour tax
Food subsidies
Source: ENV-Linkages
7

8. Results: cash transfers are efficient and
equitable,... But the scheme represented is very
stylized
Questions
• How to target beneficiary?
• Development of infrastructure, education,…
• Improved access to energy
• Renewable energy and energy efficiency
What we learnt,….
8

9. THANK YOU
For further information:
Olivier.durand-lasserve@oecd.org
Jean.chateau@OECD.ORG
www.oecd.org/environment/modelling

10. APPENDIX
10

11. Cash transfer requirements to
compensate households
Only a fraction of gov. expenditures avoided by the subsidy reform
is redistributed with cash transfers and the recipients are targeted
What fraction needs to be redistributed to keep constant the real
income of households until to a given decile?
Notes: (1) the cash transfer payments are adjusted for each decile group,
(2) here the feedback effects on growth are ignored 11

12. Transfers: the revenues generated by the phase out can be
redistributed to the households based on eligibility criteria.
Households may be impacted differently
by an energy consumption subsidy reform
Expenditures: consumer prices change and the impacts on
household groups depend their consumption structure and on
how they behave
Income: the policy can affect wage rates as well as non-wage
factor income and the impact on households’ income depend on
their sources of income
12

13. Microsimulation models, based on household budget surveys,
are often needed for quantifying the differences in impacts between
household groups
Combining micro and macro approaches
But for a energy subsidy phase out has economy-wide effects,
and a microsimulation model has to be linked with a
macroeconomic model
• The effects on the phase on final prices and households’ income
depend on the whole structure of the economy
• Redistribution schemes cannot be featured without representing
how governments’ budget adjusts to the policy
• Distributional, environmental, economic effects interplay
13

14. The decomposition algorithm at iteration
k and period t
CGE model
with 1 representative
household
Microsimulation model:
1-Household-level revenues computation
2-Household-level optimal consumption
and savings
Final good
prices
Total households’
consumptions and
savings
Production
factors’ prices
Recalibration of the
representative
household’s
utility function
calibrated representative
household’s utility function
iteration k+1
iteration k
14

15. • Low budget shares of energy
• Increasing share of food consumption
• The share of formal labour in revenue decrease with income
Elements on households heterogeneity
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 2 3 4 5 6 7 8 9 10
deciles
Labour (formal) Labour (informal)
Capital Land
Natural resource Transfers
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 2 3 4 5 6 7 8 9 10
deciles
Electricity Gasoline and Diesel
Kerosene and LPG Transportation services
Agricultural products Food products
Manufactured goods Services
15

16. Based on the IFLS4 household budget survey data, >10000 Indonesian
household groups are explicitly represented in the multi-period,
multi-region CGE model
For the models’ base year, household-specific parameter are
calibrated on the HS data
• Parameters of the extended linear demand system
• Endowments in production factors
The baseline projection is generated assuming that the households’
preferences and endowment parameters are static: no structural
change
Resolution of the integrated model using a decomposition algorithm
(Rutherford et al. 2006)
Representative households groups for
Indonesia in ENV-Linkages
16

17. Remarks on the imputation process
• Gasoline and diesel expenditures are not explicitly represented in the HS.
They are proxied based on information about expenditures for transportation
and vehicle ownership
• Profits from own business assigned to labour and capital revenue categories
using values shares from the corresponding sector in the CGE.
The process for reconciling the HS and the CGE data ensures:
• Micro-macro accounting consistency: for each expenditure and income
category, the total over households adds-up to the CGE base-year value
• Balance budget for individual households: for each household, the total
disposable income is equal to the sum of expenditures and savings
• “Steady state savings”: for all the households, the ratio between savings and
capital revenues is the same. This requirement is used because of the multi-
period setting in order to avoid a massive reallocation of capital income
between the households in the baseline scenario
Imputation and reconciliation of survey
data with the CGE
17

18. Driving mechanisms of distributional
impacts in the cash transfer scenario
The redistribution scheme makes the reform progressive
• The direct effect through final price changes is slightly regressive
• The cash transfers effect is very progressive and dominates the
regressive effects of final price changes
-5
-3
-1
1
3
5
7
9
11
13
15
1 11 21 31 41 51 61 71 81 91income centile groups
Total effect
Final price effect
Transfer effect
Source: ENV-Linkages
18

19. Driving mechanisms of distributional
impacts in the labour support scenario
The effect through labour income is regressive as labour
income from formal activities represents a lower proportion
of total income for higher-income than for lower-income
households
-5
-4
-3
-2
-1
0
1
2
3
4
5
1 11 21 31 41 51 61 71 81 91
income centile groups
Total effect Final price effect Labour income effect
Source: ENV-Linkages
19

20. Conclusions (1/2)
The energy consumption subsidy phase out contributes to
emissions reduction: ~8% for GHG and ~12% for CO2, if
excluding emission from tropical forest fires
Positive impacts on GDP (+0.4 to +.7% in 2020 w.r.t. baseline)
Direct price effect of subsidy removal is regressive but the
compensation schemes can make the reform progressive
Cash transfers are the best policies in terms of GDP growth, and
inequality reduction
Impacts of the redistribution schemes
• Cash transfers: very progressive
• Food subsidies: slightly progressive
• Labour support: regressive
20