Structural Constraints on Public Goods Provision: Evidence from Pakistan by D...
Macro-economic Implications of Electricity Subsidies by Paul Dorosh, IFPRI
1. IFPRI
Macro-economic Implications of
Electricity Subsidies
Dario Debowicz, Paul Dorosh and Angga Pradesha
Productivity, Growth and Poverty Reduction in Rural Pakistan
Pakistan Strategy Support Program (PSSP)
December 13-14, 2012
Islamabad, Pakistan
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
2. Resolving Pakistan’s Electricity Crisis:
IFPRI
Macro-economic Implications of
Alternative Policy Options
Dario Debowicz, Paul Dorosh and Angga Pradesha
International Food Policy Research Institute (IFPRI)
First presented in Islamabad, Pakistan
25 July, 2012
This presentation is based on a note written by Dario Debowicz, Paul Dorosh and Angga
Pradesha. Sherman Robinson, Haroon Sarwar Awan, Wajiha Saeed and Syed Hamza Haider also
provided valuable inputs to the disaggregation of the Social Accounting Matrix and analysis of
Pakistan’s energy sector. This work was funded by the United States Agency for International
Development (USAID) through IFPRI’s Pakistan Strategy Support Program.
INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
3. Plan of Presentation
Introduction: Electricity Shortages and Subsidies
Methodology: Data and Model
Simulations
• Additional load shedding with no change in the subsidy;
• No additional load shedding; increased subsidy financed
through domestic borrowing and thus reduced investment
• No additional load shedding financed through an increase
in money supply (monetizing the fiscal deficit).
Concluding Observations: Growth, Employment and
Inflation Tradeoffs
4. Pakistan’s Energy Sector: Major
Problems
Frequent electricity shortages (load-shedding)
• Load shedding as a percentage of demand (actual
sales plus load shedding) increased from 2.8% to
22.9% from 2007 to 2010.
Arrears in payments
Massive fiscal subsidies which are often
seriously underestimated in the budget
• In 2007-08, 2009-10 and 2010-11, the realized
(actual) value of the subsidy was two to three time
the budgeted amount.
7. Methodology
Dynamic economy-wide real-side CGE model plus a simple
equation determining aggregate price level
• Data base: 2007-08 Pakistan Social Accounting Matrix
(SAM)
• 62 Activities; 27 factors of production; regional disaggregation
of agriculture; 19 household groups
Key model features:
– Disaggregation of agricultural sectors
– Links to upstream (processing) activities
– Inclusion of non-agricultural industries and services
– Regionalization by province
– differentiation among household with different resources and
sources of income
8. Main sources of data for the SAM
• 2007-08 National Accounts
• Value added by 15 sectors
• Macroeconomic Aggregates
• 1990-91 Input-Output Table (97 sectors)
• 2007-08 Agricultural Statistics of Pakistan
• 2007-08 Pakistan Integrated Household Survey (consumption
disaggregation)
• Commodity level trade data from the Ministry of Finance
• 2000-2001 SAM for Pakistan which uses the 2001 Pakistan
Rural Household Survey (for household income disaggregation)
9. Rules for Supply-Demand Balances
(Model Closure)
• Commodity markets
• Prices adjust to equate supply and demand
• Capital is fixed by sector
• Total livestock capital is fixed but is mobile across region
• Land is fixed by region, but is mobile across crops
• Labor markets
• Urban skilled labor: Nominal wages are fixed; levels of
employment (and unemployment) are endogenous
• Other labor: Total supply of labor of each skill type is fixed
(and fully employed); real wages of other labor adjusts so
that demand for labor is equal to supply
10. Design of Simulations
• Compare three alternative policies to address the
electricity crisis:
• 1) additional load shedding with no change in the subsidy; (the
base scenario assumes 25 percent load shedding in 2011-12).
• 2) no additional load shedding with an increased subsidy
resulting in a larger budget deficit (implicitly financed through
domestic borrowing and thus reduced investment); and
• 3) no additional load shedding financed through an increase in
money supply (monetizing the fiscal deficit).
• Each simulation is run twice: once assuming flexible real
wages and full employment and a second time with open
unemployment.
11. Productivity Effects of Electricity Shortages
Total factor productivity in each sector is reduced
according to the following formula:
ECOSTa
αa
PLa = ES
TCOSTa
PLa : % decline in output (productivity loss)
ES: Electricity Shortage (% load shedding),
ECOSTa : cost of electricity (intermediate input) in sector a;
TCOSTa : total cost of production of sector a,
αa : substitutability factor in sector a.
12. Simulation 1: Additional Load Shedding
We model additional load shedding (an increase of 50
percent, i.e. from 25 percent to 37.5 percent of the time)
with no change in the subsidy.
Assuming flexible real wages, reduced sectoral
productivity results in a 1.3 percent decline in real GDP.
• Under an assumption of open unemployment (Simulation 4),
real GDP falls further (1.5 percent) and employment declines
by 0.6 percent.
Investment also falls (by 1.2 percent with flexible real
wages and 1.6 percent with open unemployment) because
the reduction in real GDP implies lower incomes and lower
total savings in the economy, thereby limiting finances
available for investment.
13. Simulation Results: Macro-economic
Effects of Alternative Electricity Policies
Sim 1 Sim 2 Sim 3 Sim 4 Sim 5 Sim 6
No load shedding;
Load shedding; No load shedding; monetized
no increase in increase in subsidy increase in Sim 1 with open Sim 2 with open Sim 3 with open
Base subsidy and budget deficit subsidy unemployment unemployment unemployment
(% change) (% change) (% change) (% change) (% change) (% change)
Real GDP 17,238.0 -1.3 0.0 0.0 -1.5 -0.1 0.0
Fiscal balance -683.0 0.3 19.7 1.0 0.8 19.7 1.0
Investment 1,983.5 -1.2 -3.8 0.4 -1.6 -4.0 0.4
Inflation 12.0 0.0 0.0 1.6 0.0 0.0 1.6
Unemployment 5.9 0.0 0.0 0.0 0.6 0.3 -0.1
Base data: Real GDP and fiscal balance are in billions of Pakistan rupees 2010/2011; inflation and
unemployment are percentages.
Sources: Model simulations; Pakistan SAM 2010/11.
14. Simulation Results: Macro-economic Effects of
Alternative Electricity Policies
2.0
1.0
0.0
Real GDP Investment Inflation Unemployment
-1.0
Percent
-2.0
-3.0
-4.0
-5.0
Load Shedding Increased Subsidy Increased Subsidy Monetized
Note: Change in inflation and unemployment are in percentage points.
Source: Model simulations
15. Simulation 2: Increased Electricity Subsidy
If the electricity subsidy is increased (by an assumed 50
percent) so that load shedding is unchanged from the base
scenario…
The fiscal deficit increases by 19.7 percent (by 0.7 percent
of real GDP).
• Under an assumption of open unemployment, real GDP
declines by 0.1 percent and unemployment increases by 0.3
percent, but the change in the fiscal deficit is the same: 19.7
percent.
In both simulations, the decline in total savings in the
economy causes a reduction in investment (by 3.8 to 4.0
percent).
16. Simulation 3: Additional Load Shedding
Financing an increased energy subsidy through expansion in
money supply prevents a short-run decline in real GDP but could
potentially increase money supply and the price level by about 2
percent.
Real GDP is essentially unchanged under both labor market
closures: fixed real wage and open unemployment.
The electricity subsidy is essentially financed through an inflation
tax (a loss in the value of money held by households and
enterprises) equivalent to about 0.8 percent of total household
expenditures.
A full analysis of these policy options would require a fully
specified financial sector. These simulations were done using a
simple monetary sector framework assuming the subsidy leads to
an equivalent increase in M2 with a constant velocity of money.
17. Household Income Effects
In Simulation 1, the urban poor have smaller declines in
incomes than the rural poor or the national average as the
decline in total factor productivity in non-agricultural
sectors leads to an increase in returns to informal sector
capital (a major source of income of the urban poor).
In Simulation 2, by contrast, the large decline in
investment reduces demand for unskilled non-agricultural
labor in construction and other activities, reducing the
incomes of the urban poor by more than other groups.
Finally, in simulation 3, both rural and urban poor suffer
larger declines in real incomes than the urban non-poor
since this latter group enjoys a gain in returns to formal
capital.
19. Tradeoffs Between Output/Employment,
Growth and Price Stability
• If there is no increase in the electricity subsidy and
total load shedding increased by 50 percent relative
to 2010/11, real GDP could decline by 1.3 to 1.5
percent with up to a 0.6 percentage point increase in
unemployment.
• An increase in the electricity subsidy financed
through government borrowing in the domestic
economy could prevent a significant decline in real
GDP and employment, but reduce investment by 3.8
to 4.0 percent of GDP and hamper future growth.
20. Tradeoffs Between Output/Employment,
Growth and Price Stability
• Financing an increase in the electricity subsidy
through increases in the money supply could
likewise prevent a fall in real GDP or employment,
but would raise money supply and the aggregate
price level by about 2 percent.
21. Next Steps
• Further refinement is needed on:
Productivity effects of electricity load shedding across
sectors
Behavior of the labor markets in Pakistan to enable
better estimates of the effects of policy on employment.
A fuller specification of the monetary and financial
sectors to better estimate the effects of budget deficits
on inflation.
• Other approaches to the modeling the macro-
linkages between real output, money supply and
price level are also being planned:
Full specification of financial sector in a CGE model
Dynamic stochastic general equilibrium model