Transcript Carbon tax scenarios - Centre International de Recherche

IMACLIM-SA
Insights from version 2
J. Schers, F. Ghersi, F. Lecocq
AfD, 4 December 2014
1
Photo : Mairie Paris
Objectives of the presentation
 Describe status of part II (modeling) of AfD research
convention RCH262-2011
• “Green growth and its implications for public policy - The
case of South Africa”
 Obtain feedback on version 2 of IMACLIM-SA model
and results
 Prepare final report (in the “Recherche” AfD
publication series)
4 December 2014
2
Outline
 Context and objectives
 Existing literature
 The IMACLIM-SA_v2 Model
 Scenarios
 Results
4 December 2014
3
CONTEXT AND OBJECTIVES
4 December 2014
4
A coal-based economy generating high
CO2 emissions
 South Africa is a rapidly growing middle-income economy
• World 25th largest economy (in PPP GDP)
• World 72nd GDP per capita in 2012
 South Africa per capita CO2 emissions, however, are on
par with richest economies
• World 33rd Emissions per capita in 2010
 Economic development driven by abundant coal
resources, long subsidized electricity from coal, and coalto-liquids technology
• Energy-intensive mining and industry historic drivers of SA
economy
4 December 2014
Source: World Bank World Development Indicators
5
Structure of current (2011) GHG
emissions
South Africa
France
51.6
65.4
11 848
36 248
GHG emissions (MtCO2e)
456
487
CO2 emissions (MtCO2)
374
338
CO2 emissions from electricity
generation (MtCO2)
229
60
Population (millions)
GDP per capita (USD PPP)
Sources: World Bank, World Resources Institute CAIT 2.0
4 December 2014
6
Ambitious decarbonization objectives
 Mitigation pledge announced at COP15 (2009)
• Reduce GHG emissions by 34% in 2020 and by 42% in
2025 relative to baseline
• Conditional on finance and technological support from
international community
 2011: Climate Change Response white paper
• Outlines strategy
• Including carbon pricing, possibly carbon tax
 2013: Revised carbon tax plan for consultation
“To allow for further consultation, implementation of the carbon
tax is postponed by a year to 2016”
4 December 2014
7
Interacting with energy sector policies
 Long-Term Electricity Investment Plan
• Urgent need to ease the tension between supply and demand
• 2030 targets:
• 35% energy efficiency improvements in demand
• Renewables largest contribution to additional capacity.
• Coal share would fall from 93% to 46% of installed capacity
 Electricity prices rising sharply to recover system costs and
maintain ESKOM financial viability
• From ca. R0.25/kWh in 2005 to R0.65/kWh in 2013
• Further increases approved by NERSA
 Price-competitive procurement for volume of RE
 Energy efficiency and demand side management initiatives
Sources: IEA, 2013 ; Baker et al. 2014
4 December 2014
8
Within a rapidly transforming society
facing important challenges
 South Africa is experiencing rapid economic growth
• Real GDP has doubled over the past 30 years
 Population is also growing rapidly
• 51M in 2010 (+14.6M relative to 1990)
 Unemployment remains large
• 25% unemployment rate (39% if “discouraged workseekers”
are included)
• Particularly for low-skilled individuals
 With persisting inequalities in the society
• South Africa has one of the highest Gini indices in the World
Sources: SA Treasury; UN population data
4 December 2014
9
Objectives of the project
 To provide numerical insights on articulation between
decarbonization and other key development
objectives
• In the context of the ‘green growth’ conversation
 Interactions have been mostly with academics
• Energy Research Center (Univ. Cape Town): 3 missions,
including MAPs workshop Oct./Nov. 2014
• June 13, 1st meeting w/SA Treasury modeling team
• Further exchanges since (e-mail and de visu in Cape Town)
4 December 2014
10
EXISTING LITERATURE
4 December 2014
11
An already deep spectrum of analysis
 Carbon taxes already much analyzed
• Van Heerden et al. 2006
• De Pauw 2007
• Devarajan et al. 2009, 2010
• Alton et al. 2014
• Musango et al. 2014
 To our knowledge few countries, OECD included,
have such depth of analysis
• 4 models in France
4 December 2014
12
van Heerden et al. 2006

Question: Can revenue recycling from a C tax generate double dividend in
SA?

Model: Static CGEM of SA
•
High-skill labor supply inelastic to real wages
•
Low-skill labor supply perfectly elastic to real wages

Timeframe: 1998

Policy scenarios:

•
4 instruments: C tax; fuel tax; electricity tax ; energy tax
•
3 recycling targets: direct taxes, indirect taxes, food taxes
Outcome:
•
Recycling policy critical for dividends (over policy instrument)
•
Food tax break can induce triple dividend on GHG emissions, GDP, poverty reduction
4 December 2014
13
De Pauw 2007
 Question: what impact of various mitigation scenarios on SA
economy?
 Model: Dynamic CGEM of SA
 Timeframe: 2005-2015
 Policy scenarios:
• 3 mitigation scenarios, one of them w/ C tax
• Informed by MARKAL energy system modeling
 Outcome:
• C tax scenario yields -2% GDP in 2015
4 December 2014
14
Devarajan et al. 2009, 2011
 Question: What role of prior market distortions on C tax impacts
in SA?
 Model: Static CGE of SA economy
 Time: 2003
 Policy scenarios:
• -15% CO2 emissions achieved through C tax, energy input tax or
energy-intensive goods tax
• Lump-sum rebate to households
• Sensitivity analysis on economic rigidities and labor market distortions
 Outcome:
• Market rigidities critical in choice of instrument and impact on welfare
and equity
4 December 2014
15
Alton et al. 2012, 2014
 Question: What impact of a C tax on SA?
 Model: Dynamic CGEM of SA linked to energy sector model
 Time: 2005 to 2025
 Policy scenarios:
• Domestic C tax ; C tax + border tax adjustment; Unilateral C tax + BTA
in SA trading partners
• 3 recycling options: sales tax; corporate tax; transfers
 Outcome:
• Phased-in C tax reaching $30/tCO2 in 2022 achieves 2025 SA
emissions target
• Recycling option strongly impacts trade-off between distribution and
growth
4 December 2014
16
Musango et al. 2014

Objective: Multicriteria analysis of “green economy” policies

Model: South Africa Green Economy Model
•
Differential equations
•
Economic and natural resource dynamics

Timeframe: 2001 to 2030

Policy scenarios:

•
BAU (BAU2%): continuation (slight increase) of investment in natural resources,
energy, agriculture and transport
•
GETS (GE2%): additional 2% investment in above mentioned sectors, targeted to
energy (evenly distributed)
Outcome:
•

GE policies do not impact real GDP, increase employment and reduce environmental
pressures (including CO2 emissions)
Insignificant economic impact of additional investment questionable
4 December 2014
17
THE IMACLIM-SA_V2 MODEL
4 December 2014
18
IMACLIM-SA_v2 Overview
 Recursive, exogenous growth, “GE” model of a small,
open economy
IMACLIM approach
 2 periods: 2005 (calibration year) to 2035
 Calibrated on “hybrid” data
• National accounts hybridized with energy data
• Agent-specific net-of-tax prices
 Second-best features in equilibrium structure
• Markup prices
• Rigid real wages
 Hybrid top-down/bottom-up vision of technical change
• Part nested CES, part BU-informed Leontief
4 December 2014
19
Rationale of IMACLIM approach
 Experience with France, Brazil suggests that
reconciling economic and energy calibration data
significantly impacts model outcome
 “Second-best” market features and explicit integration
of technical constraints allow testing broad range of
views/beliefs about the functioning of the economy
4 December 2014
20
Modeling scope
 4 primary factors
• Capital (FCC), 3 labour classes
 10 goods/production sectors:
• 5 E sectors, E-intensive, Manufacture, High-skill services,
Low-skill sectors, Transport services
 5 household (expenditure) classes
 Secondary income distribution
• Taxes and transfers between agents
• Debt positions (under constant rate assumption)
4 December 2014
21
Specific development for IMACLIM_SA
 Hybridization of national accounts and energy tables
for South Africa

 Skill level differentiation with varying degrees of
rigidities and limited substitutability

 Integrating bottom-up information on the SA energy
sector from SATIM energy model

4 December 2014
22
10 Production sectors
Sector
Name
Content
COA
Coal
Coal
OIL
Oil
Oil resources
GAS
Gas
Gas resources, Gas distribution
REF
Refineries
Refineries
ELC
Electricity
Electricity
EIN
Energy
Intensive
Industries
Gold, Other mining, Petrochemical industry, Other NMM products,
Basic iron/steel, Non-ferrous metals
MAN
Manufacturing
Food, Footwear, Metals basic manufacturing, Electrical
machinery, Radio, Transport equipment, Other manufacturing
TRS
Transport
Transport services
LSS
Low-Skill
Sectors
Agriculture, Construction, Trade, Hotels and restaurants,
Domestic and other services
HSS
High-Skill
Services
Water, Communications, Financial intermediation, Real estate,
Business activities, General government, Health and social work,
Education
Sectors in IMACLIM_SA model (left) matched with South Africa Social Accounting Matrix sectors
(w/energy sectors split up during hybridization process) – Automatic aggregation / disaggregation
23
Factor trade-offs in production
Y
EIN
MAN
COA
Mat
KLE
OIL
GAS
TRS
LSS
KL
E
REF
L12
L1
4 December 2014
L
K
ELC
HSS
L3
L2
Constant Elasticity
of Substitution
(CES)
Fixed production
coefficients
(Leontieff)
24
Producer costs
 Net wage in each sector varies proportionally to
average wage
 Labour costs pL1,2,3 are net wages + labour taxes
 Capital cost pK is weighted average price of
investment goods (homogeneous capital good)
 Production prices are factor costs + output tax τY +
markup τNOS
10
𝑝𝑌,𝑖 =
𝑝𝑗,𝑖 𝛼𝑗,𝑖 +
𝑗=1
4 December 2014
3
𝑝𝐿𝑗,𝑖 𝑙𝑗,𝑖 + 𝑝𝐾 𝑘𝑖 + 𝜏𝑌,𝑖 𝑝𝑌,𝑖 + 𝜏𝑁𝑂𝑆,𝑖 𝑝𝑌,𝑖
𝑗=1
25
Agent-specific consumer prices
 Consumption prices pile up trade, transport and
agent-specific margins, then taxes
𝑝𝐶,𝑖
= 𝑝𝑌,𝑖 1 + 𝜏𝑀𝐶,𝑖 + 𝜏𝑀𝑇,𝑖 + 𝜏𝑀𝑆𝐶,𝑖 + 𝑡𝐶1,𝑖 + 𝑡𝐶2,𝑖
 Negative trade and transport margins adjust to
balance out positive (calibrated) ones
4 December 2014
26
Households behaviour
 GDI of class h is wages + share of operating surplus
GOS + pensions RP + unemployment benefits RU+
residual social transfers RA + residual transfers AT –
income tax TIT – interest payments ih Dh
𝐺𝐷𝐼ℎ =
𝑤𝑗,𝑖 𝐿𝑗ℎ,𝑖 + 𝜔𝐾ℎ 𝐺𝑂𝑆 +
𝑖,𝑗
𝑅𝑖ℎ + 𝜔𝐴𝑇ℎ 𝐴𝑇 − 𝑇𝐼𝑇 − 𝑖ℎ 𝐷ℎ
𝑃,𝑈,𝐴
 Classes consume, save and invest proportionally to
GDI ; their self-financing capacity SFC is the
difference between their savings S and investments
GFCF
𝑆𝐹𝐶ℎ = 𝑆ℎ − 𝐺𝐹𝐶𝐹ℎ = 𝜏𝑆ℎ 𝐺𝐷𝐼ℎ − 𝜏𝐺𝐹𝐶𝐹ℎ 𝐺𝐷𝐼ℎ
4 December 2014
27
Households consumption trade-offs
CONS + basic needs of:
ELC, REF, COA, TRS
COMP
TRS EIN MAT
4 December 2014
LSS HSS
EAG
REF
ELC
28
Agents’ positions: Firms
 Firms’ GDI is share of operating surplus GOS +
pension levies TACS + residual transfers AT –
business tax TBT – pensions RP – debt service iS
DS
𝐺𝐷𝐼𝑆 = 𝜔𝐾𝑆 𝐺𝑂𝑆 + 𝑇𝐴𝐶𝑆 + 𝜔𝐴𝑇𝑆 𝐴 𝑇 − 𝑇𝐵𝑇 − 𝑅𝑃 − 𝑖𝑆 𝐷𝑆
 Firms invest constant proportion of own GDI
𝐹𝐵𝐶𝐹𝑆 = 𝜏𝐹𝐵𝐶𝐹𝑆 𝐺𝐷𝐼𝑆
 Self-financing capacity SFCS is GDIS remainder
𝑆𝐹𝐶𝑆 = 𝑅𝐷𝐵𝑆 − 𝐹𝐵𝐶𝐹𝑆 = (1 − 𝜏𝐹𝐵𝐶𝐹𝑆 )𝑅𝐷𝐵𝑆
4 December 2014
29
Agents’ positions: public admin.
 Government gross disposable income is computed
as households’ and firms’:
GDIG = Tax revenues + share of gross operating surplus –
unemployment benefits - other social benefits + share of
residual transfers
 Public expenditures pG G and public investment
GFCFG are fixed shares of GDP
 Public budget deficit SFCG is remainder of GDIG
10
𝑆𝐹𝐶𝐺 = 𝐺𝐷𝐼𝐺 −
𝑝𝐺,𝑖 𝐺𝑖 − 𝐺𝐹𝐶𝐹𝐺
𝑖=1
4 December 2014
30
Trade and the Rest-of-the-World
 Domestic and international goods assumed
homogeneous in each sector
 Imports market shares and absolute exports
(quantities) elastic to terms-of-trade + exogenous
trend for exports
𝑀𝑖 𝑀𝑖2005 𝑝𝑀𝑖2005 𝑝𝑌𝑖
=
𝑌𝑖
𝑌𝑖2005 𝑝𝑌𝑖2005 𝑝𝑀𝑖
𝜎𝑀𝑝𝑖
𝑋𝑖
𝑋𝑖2005
= 1+𝜃
30
𝑝𝑋𝑖2005 𝑝𝑀𝑖
𝑝𝑀𝑖2005 𝑝𝑋𝑖
𝜎𝑋𝑝𝑖
 ROW agent balances current account
𝑆𝐹𝐶𝑅𝑂𝑊 =
𝑝𝑀𝑖 𝑀𝑖 −
𝑝𝑋𝑖 𝑋𝑖 +
𝑖𝑖 𝐷𝑖 − 𝐴𝑇
𝐻,𝑆,𝐺
4 December 2014
31
Debt accumulation
 Self-financing capacities of 8 agents’ net of debt
interest payments are assumed to linearly transit
from their 2005 to their 2035 values
• Agents are: firms, public administration, 5 household classes, and
the rest-of-the-world
 This allows computing compounded interest
payments as functions of 2005 debt levels and 2005
observed agent-specific interest rates (ratios of
payments to net debt), which are assumed constant
to the horizon
4 December 2014
32
Goods & investment market balances
 Mass conservation equation: domestic output
and imports balance household and public
consumptions, immobilization and exports
𝑌𝑖 + 𝑀𝑖 = 𝐶𝑖 + 𝐺𝑖 + 𝐼𝑖 + 𝑋𝑖
 Total GFCF of agents matches immobilization by
adjustment of the GOS share accruing to firms
𝐺𝐹𝐶𝐹𝑖 =
𝑝𝐼𝑖 𝐼𝑖
𝐻,𝑆,𝐺
 Immobilized capital is homogeneous; investment
ratio to consumption of fixed capital is fixed
𝐼𝑖
=
𝑘𝑗 𝑌𝑗
4 December 2014
𝐼𝑖2005
𝑘𝑗2005 𝑌𝑗2005
33
Market Balance - Employment
 Labor endowment 𝐿 proportional to total active
population, split in 3 skill levels 𝐿1 , 𝐿2 , 𝐿3
 For each skill j unemployment uj is endogenous
1 − 𝑢𝑗 𝐿𝑗 =
𝑙𝑗,𝑖 𝑌𝑖
𝑖
 … and real sector-average wage wj /CPI corrected for
share ωLPj of labor productivity gains Φj accruing to
workers is elastic to unemployment (wage curve)
1 − 𝜔𝐿𝑃𝑗
4 December 2014
𝑤𝑗
𝑢𝑗2005
= 𝑤𝑗2005
𝐶𝑃𝐼 + 𝜔𝐿𝑃𝑗 𝜙𝑗 𝐶𝑃𝐼
𝑢𝑗
𝜎𝑤𝑢𝑗
34
Model implementation
4 December 2014
35
Model implementation (cont.)
 Programmed under Excel and VBA
 Advantage:
• Ease of use and of communication
• Easy visualization of data/result in accounting matrix term
• Link to input data
 Limitations:
• Equation programming less efficient
• Excel solver constraints
 About 5 mn/run on a recent computer
4 December 2014
36
CALIBRATION & SCENARIOS
4 December 2014
37
Overview
 Base year:
2005
• Data availability
 Selected Horizon:
2035
• Consistent with planning horizon
Step #1: Build image of 2005 economy
Step #2: Build projected image of 2035 economy w/o
carbon tax (RP)
Step #3: Introduce C-tax at constant budget position,
with no recycling and 5 recycling options
4 December 2014
38
Data hybridization
 Reconciling material and monetary balances
Material Flows
Energy balances
+ energy price stats
Monetary Flows
I/O Tables
Data hybridization: Methodology
Sources
statistiques
disponibles
Étape 1
Matrice en volume
(unité physique)
CF
CI
MAT
MAT
CI
Matrice en prix
(monnaie/unité physique)
Qij
Pij
M
M
Étape 2
CF
Factures des achats de matière
MAT
CI
CF
M
Vij = Pij . Qij
AUTRE
MAT
CF
M
VA
MAT
AUTRE
Étape 3
Vij
Insertion des factures des achats (Vij)
& réarrangement des nomenclatures
Data hybridization: Application to SA
 An example: SAM vs. "electricity bills"
Supplied by
electricity sector
to…
Refineries
SAM, incl gas
distr. (MZAR)
Energy balance
and prices (MZAR)
Difference
(to EIN)
222
1 112
+ 890
5 484
1 451
- 4 033
646
1 137
+ 491
Basic Iron & Steel
2 563
4 278
+ 1 715
Non-ferrous
1 562
2 919
+ 1 357
889
685
- 204
Other sectors
23 265
18 551
- 3 777
All sectors
33 704
30 133
- 3 571
Petro- & chemical
industry
NMM
Basic metals
manufacturing
4 December 2014
41
Population growth assumptions
15 287
14 407
None
1 589
287
Primary
7 530
3 764
Lower secondary
13 248
15 075
Upper secondary
9 701
18 989
Post secondary
1 407
2 241
33 477
40 356
65+
2 689
4 765
All
51 453
59 528
0-14
15-64
Total
Skill 1
2035
Skill 3 Skill 2
2010
Population per age group and education in 2010 and 2035;
Source: UN projections, distributed by education class following IIASA projections
4 December 2014
42
Integration of bottom-up information
Energy Investment Plan
 Technical coefficients for inputs to electricity
based on outcomes of SATIM, ERC of Univ
of Cape Town’s bottom-up (TIMES) model
• PJ fuel per PJ of ELC
• Units of capital per PJ of ELC
• Interpretation of O&M costs for labour, and goods &
services inputs per PJ of ELC
 SATIM simulations include South Africa’s Energy
Investment Plan (IRP policy updated) and different
carbon tax levels
4 December 2014
43
Economic growth assumptions
 Generating real per capita growth and reducing
unemployment at the same time turns out to be difficult
• Base year unemployment is already high
• Population growth is rapid
 We achieve both in RP by combining:
• Autonomous labor productivity increase of +0.375% (skill 1) to
+1.25%/year (skill 3)
• Increase in demand for exports (+2%/year)
• Only part of productivity gains passed on to wages (percentage
skill-class dependent)
• Capital productivity gains (+1.5%/year)
• Material & services intensity decreases (-0.25%/year)
4 December 2014
44
Key features of RP
2005
2035 - RP
Population (millions)
47.6
59.5
GDP (billion 2005 Rands)
1 571
5 757
4.4% annual
growth rate
GDP per capita (2005 USD PPP)
9 448
27 701
Equivalent to
Spain in 2005
Unemployment (%)
38.8%
24.6%
443
1017
CO2 emissions (MtCO2)
note
+129%
 Though economically plausible, this outcome is
strongly dependent on parameters
4 December 2014
45
Discussion of RP – Structural
transformation of labor market
 According to projections, skill 3 population more than
doubles relative to 2005, while skill 1 + skill 2
population increases by 8%.
Total working age
population (thousands)
2005
2035
Skill 1
7 748
3 530
Skill 2
16 161
22 662
Skill 3
6 182
14 164
 Correlated to this, tensions on skill 3 jobs ease, while
tensions on skills 1-2 appear.
Broad unemployment rate
4 December 2014
2005
2035 (RP)
Skill 1
41%
10%
Skill 2
44%
23%
Skill 3
26%
31%
46
Carbon Taxes
 Economy-wide Carbon tax (Ctax), applied to
intermediate consumption of firms and on household
final direct consumption of energy
 Given (long-term) production trade-offs embedded in
the model, the Ctax is implicitly assumed to be
imposed progressively over time
 Purely domestic scheme:
• No levy on imports
• No rebate for exports
 C tax in 2035: R100/tCO2 to R500/tCO2
4 December 2014
47
Recycling options
 R0. No recycling: C revenues are added to government
resources (no budget position constraint)
 R1. VAT rate is adjusted downward, budget deficit held identical
to RP (-2%)
• Homogeneous reduction in VAT rate across all products
 R2. Income tax rate is adjusted downward (same budget rule)
 R3. Combines R1 and R2
 R4. Proceeds from carbon tax add up to government
expenditures on “public administration” (salaries and
consumption)
 R5. Proceeds from carbon tax directed to an across the board
increase in “other transfers”
• Transfers increase by same amount for all households (lumpsum)
4 December 2014
48
RESULTS
4 December 2014
49
Environmental outcome
RP (IRP)
RP
R0
R1
R2
R3
R4
Carbon tax scenarios (ZAR/tCO2)
100
300
500
853
(-16%)
896
(-12%)
869
(-15%)
878
(-14%)
867
(-15%)
656
(-36%)
768
(-25%)
659
(-35%)
721
(-29%)
693
(-32%)
596
(-41%)
559
(-45%)
1 017
R5
Total South Africa CO2 emissions in 2035 with IRP investment
plan for electricity, for different scenarios in IMACLIM-SA-v2
(MtCO2, and difference to RP in percentage)
4 December 2014
 Emissions increase
significantly relative to
2005 (+129%)
 Rapid but decreasing
response to Ctax
• Calibration of
degrees of flexibility
• Basic consumption
levels
• Break-even point of
other technologies
with coal lies above
ZAR 100/tCO2
50
Environmental outcome w/o IRP
Ref w/o IRP Carbon tax scenarios (ZAR/tCO2)
100
Ref w/o
300
500
1 148
877
655
(-24%)
(-43%)
921
757
669
R1
(-20%)
(-34%)
(-42%)
894
R2
(-22%)
903
715
610
R3
(-21%)
(-38%)
(-47%)
892
689
573
R4
(-22%)
(-40%)
-50%
898
702
593
R5
(-22%)
(-39%)
(-48%)
Total South Africa CO2 emissions in 2035 for different scenarios
in IMACLIM-SA-v2 w/o IRP
(MtCO2, and difference to RP in percentage)
R0
4 December 2014
 Reference
emissions
higher than RP
• CES
underestimates
planned
investments in
low-carbon
technologies
 But is also more
responsive to
Ctax
51
Economic outcome
 All C tax scenarios
have growth prospects
of higher than 2,8%
average real growth per
year

RP growth rate
sensitive to
assumptions on
drivers of growth
 Still, Ctax recycling
schemes have lower
real growth forecasts
than RP
RP (IRP)
Carbon tax scenarios (ZAR/tCO2)
100
300
R0
3,9%
3,0%
R1
4,2%
3,9%
3,5%
R2
4,0%
R3
4,0%
3,5%
2,9%
R4
4,0%
3,4%
2,8%
R5
4,0%
3,3%
2,8%
RP
500
4,4%
Average real GDP growth per year (for 2005 – 2035) for
different scenarios in IMACLIM-SA-v2
 R1, recycling through VAT reduction has significantly higher
economic growth rates than other scenarios (but higher emissions
as well)
 Scenarios have slightly different results for public debt
4 December 2014
52
Economic outcome (absolute terms)
RP
RP
R0
R1
R2
R3
R4
R5
Carbon tax scenarios (ZAR/tCO2)
100
300
500
315%
-14%
344%
-6%
321%
-12%
329%
-10%
324%
-12%
322%
-12%
245%
-33%
314%
-14%
278%
-24%
280%
237%
271%
-26%
268%
-27%
228%
-38%
228%
-38%
366,5%
Average real GDP growth (for 2005 – 2035) for different
scenarios in IMACLIM-SA-v2 (and difference with RP)
4 December 2014
53
w/o IRP, CES would yield slightly
higher growth
Ref w/o IRP
Carbon tax scenarios (ZAR/tCO2)
100
300
R0
4,1%
3,3%
R1
4,4%
4,1%
3,9%
R2
4,1%
R3
4,2%
3,7%
3,3%
R4
4,2%
3,6%
3,2%
R5
4,2%
3,6%
3,1%
Ref w/o
500
4,6%
Average real GDP growth per year (for 2005 – 2035) for
different scenarios in IMACLIM-SA-v2 w/o IRP
4 December 2014
54
Economic outcome w/IRP
Economic outcome: average annual real GDP growth
IRP ref
100
300
R0
3,9%
3,0%
R1
4,2%
3,9%
3,5%
R2
4,0%
R3
4,0%
3,5%
2,9%
R4
4,0%
3,4%
2,8%
R5
4,0%
3,3%
2,8%
RP
4 December 2014
Carbon tax scenarios (ZAR/tCO2)
500
4,4%
55
Employment outcome
IRP ref
Carbon tax scenarios (ZAR/tCO2)
100
300
R0
31,0%
41,0%
R1
26,9%
29,1%
32,7%
R2
29,3%
R3
28,5%
33,6%
39,5%
R4
28,7%
34,4%
40,8%
R5
28,9%
34,8%
40,7%
RP
500
25,4%
Broad unemployment levels in 2035 (% of active population)
for different scenarios in IMACLIM-SA-v2
4 December 2014
 Employment
outcomes follow
pattern for real
GDP growth:
 Most scenarios
improve versus
base year
unemployment
(38,8%)
 Significantly better
outcome of R1
recycling option
56
Employment outcome w/o IRP
RP
Carbon tax scenarios (ZAR/tCO2)
100
300
R0
29,7%
38,5%
R1
25,6%
27,2%
28,9%
R2
28,0%
R3
27,2%
31,5%
35,4%
R4
27,4%
32,2%
36,8%
R5
27,6%
32,8%
37,3%
RP
500
24,6%
Broad unemployment levels in 2035 (% of active population)
for different scenarios in IMACLIM-SA-v2
4 December 2014
 Employment
outcomes follow
pattern for real
GDP growth:
 All scenarios
improve versus
base year
unemployment
(38,8%)
 Significantly better
outcome of R1
recycling option
57
Distributional implications
 Large decrease of inequality
in RP and all Ctax scenarios
• GDI per capita ratio
between class 5 and class 1
decreases from 41 to 21
 Decrease in inequality follows
from relaxation of constraint
on high-skilled labour market
 Results are sensitive to labour
market hypotheses
 Lumpsum transfers of carbon
tax revenues (R5) have a
large impact, due to still low
income levels in classes 1 to 3
4 December 2014
IRP ref
Carbon tax scenarios (ZAR/tCO2)
100
300
R0
20,2
19,8
R1
20,5
20,6
20,5
R2
20,8
R3
20,7
21,1
21,3
R4
20,3
19,9
19,5
R5
15,5
11,2
9,1
RP
500
20,5
Ratio of class 5 GDI per capita over class 1 GDI
per capita for different scenarios in IMACLIM-SAv2 (higher means larger income inequality)
58
Distributional implications w/o IRP are
slightly worse
 Large decrease of inequality
in RP and all Ctax scenarios
RP
Carbon tax scenarios (ZAR/tCO2)
100
300
R0
20,4
20,1
 Decrease in inequality follows
from relaxation of constraint
on high-skilled labour market
R1
20,7
20,8
20,9
R2
21,1
R3
20,9
21,3
21,6
 Results are sensitive to labour
market hypotheses
R4
20,5
20,2
19,9
R5
15,6
11,5
9,5
• GDI per capita ratio
between class 5 and class 1
decreases from 41 to 21
 Lumpsum transfers of carbon
tax revenues (R5) have a
large impact, due to still low
income levels in classes 1 to 3
4 December 2014
RP
500
20,7
Ratio of class 5 GDI per capita over class 1 GDI
per capita for different scenarios in IMACLIM-SAv2 (higher means larger income inequality)
59
CARBON TAX FOR INVESTING IN
EDUCATION?
4 December 2014
60
Labour market and education (1/2)
 By assumption of constant
allocation of degrees over skill
levels,
 South Africa changes from a
country with a lot of people
(63%) with only a high school
degree working in skill 3 jobs,
2005
RP
38,8%
25,4%
- skill 3
26,5%
31,0%
- skill 2
43,7%
23,2%
- skill 1
41,0%
10,2%
Overall (broad)
unemployment level
2005 and RP 2035 unemployment levels
disaggregated by skill level
 into a country with over-supply
of skill 3
 Labour market model is segmented, meaning no spill-over
of high-skilled people to low-skilled jobs
4 December 2014
61
Does RP justify progress in education?
 If…
• we follow IIASA’s projection of educational attainment, and
• we follow international data on expenditure by student by type of
education for South Africa (UNESCO, World Bank, StatsSA)
• and if we assume education keeps the same share of public
expenditures (20% in 2005), and given:
• (i) relative stagnation in number of pupils,
• (ii) gov expenditures increase as GDP
• (iii) wage moderation in HSS
 Then: there is ample room for largely decreasing pupils to
teacher ratio – continuing current trends (currently 29, against
32 in 2005)
 If… share of education in public expenditure drops to 10%
(France in 2013), there is room for keeping pupil to teacher ratio
constant
• In that case answer is: yes, RP and education prospects are aligned.
4 December 2014
62
Labour market and education (2/2)
 What if companies stop
hiring (most) high-school
leavers for high skilled
jobs?
 When assuming "degree
inflation" disaggregated
labour market results
change radically,
 and economic growth is
negatively affected by
"degree rigidity"
4 December 2014
RP
Degree
inflation
25,4%
26,1%
- skill 3
31,0%
7,3%
- skill 2
23,2%
35,7%
- skill 1
10,2%
13,3%
4,4%
4,1%
Overall broad
unemployment level
Real avg GDP growth
per year
RP 2035 unemployment levels disaggregated by
skill level compared to with degree inflation
63
What model for investment in education?
 We want to keep technological coefficients for labour
factor by skill level in the production function
 We do not model "degree inflation" with
unemployment level endogenously
 Therefore:
• in our segmented labour market model for South Africa
investing in having more people with a degrees for highskilled jobs doesn’t make sense
 What we can model is additional impact of
investment in education through productivity gains
4 December 2014
64
Investment in education scenario
 For projection year the education budget increases 10%,
compared to 20% of government expenses
• Meaning: government expenses in 2035 are 2% higher
compared to constant expenses over PIB assumption
 Forced social return:
• The benefits from the investment in productivity terms is a
reduction of costs on labour (for domestic production)
• These should be 10% bigger than the increase in expenses on
education in the same year (2035)
 Model outcomes are:
• A required (for social return) additional labour productivity gain
• Other macro-economic outcomes
4 December 2014
65
Results for IRP plus R0, R1, and R2
Educational investment with…
R0 (no)
R1 (VAT)
R2 (income
tax)
Carbon tax level: 100 ZAR/tCO2
Resultsing changes for period of 2005 to 2035:
Additional labour productivity
+3,23%
+3,18%
+3,23%
GDP growth (2005 - 2035)
+1,59%
-0,29%
+1,25%
Overall 2035 unemployment
-0,97%
+1,48%
+1,03%
Comparison of results of carbon tax recycling scenarios with additional investment in
education vs. Recycling scenarios without additional investment in education
 Impacts of investment in education under our
assumptions slightly improves either slightly
deteriorates outcomes of recycling
4 December 2014
66
CONCLUSION AND DISCUSSION
4 December 2014
67
Reference projection sensitivity
 A large number of runs allow us to stress…
 The conflicting impacts of labour productivity on growth
and employment in our second-best setting: imperfect
labour markets, markup pricing, partially Leontief input
structures require adjustments of other inputs
productivities to balance growth
 The impact on unemployment and growth of the share of
productivity gains accruing to workers vs. to cost reduction
 The importance of trade: the exogenous upward trend of
export markets (with terms-of-trade corrections) accounts
for more than a third of growth of Reference projection
after feedbacks
• What impact of carbon taxes abroad on international prices?
4 December 2014
68
Labour market segmentation
 Labour market segmentation is tricky in our modelling
timeframe of one long leap into the future
• Unchanged segmentation rules over 30 years questionable;
changing rules strongly impacts unemployment distribution
• Risk of double counting by simultaneously programming LP
increases and massive skill shifts
• Uncertainty about the costs of the shifts envisioned by
IIASA: built in the reference or requiring extra investments?
 Further work will look into
• Dedicated literature stressing the differentiated
substitutabilities among skill levels/aggregates and K
4 December 2014
69
Recycling options
 Clear differences in environmental and economic
impacts of C tax recycling options
• Recycling through VAT reduction (Sc. R1) better option by
focusing on consumption incentive
• Whereas income tax and lump-sum recycling partly
absorbed by savings
• Lumpsum recycling (scen. R5) outperforms RP on reducing
inequality
 … but no free lunch
• R1 still costs 0.2pts average annual growth and increases
unemployment by 1pt per ZAR100 tax
4 December 2014
70
Technical change and CO2 emissions
 Technical assumptions are central to CO2 emissions
in RP and to responses to Ctax
• Strong contrast between nested CES and BU-informed runs
illustrates biases attached to poorly-backed elasticity
estimates
4 December 2014
71
Thank you
4 December 2014
72