QUANTIFYING ECONOMY-WIDE IMPACTS OF REDUCING SUBSIDY TO THE ELECTRICITY SECTOR IN KUWAIT
Paper to be presented at International Conference on Economic Modeling, 2014
Bali – Indonesia, July 16–18, 2014
Ayele GelanEconomic Public Policy Research Programme (EPP)
Techno-Economics Division (TED)Kuwait Institute for Scientific Research (KISR)
P.O Box 24885Safat 13109
KuwaitTel.: +9652489495 (office)
+96560677415 (mobile)
Abstract
This paper examined economy-wide impacts of reducing electricity subsidy in Kuwait. A social accounting matrix (SAM) was constructed with 2010 as a base year and a computable general equilibrium (CGE) model was developed. The CGE model was used to simulate effects of a 30% reduction the subsidy. This policy shock was applied to the model in two scenarios. In the first scenario, the subsidy reduction was applied and results were compared with the baseline scenario given in the SAM. This yielded contractionary effects across a range of endogenous variables. Electricity tariff increased by three fold, rising from 2.0 to 6.1 fils/kWh. GDP fell by 1.2% and aggregate household welfare declined by 1.6%. Sectoral value-added declined with substantial variations across the sectors, notably the crude oil extraction as well as the petroleum production subsector of manufacturing being among those that experience the largest declines. The electricity sector out itself experienced a 21% decline in output. In scenario 2, the 30% subsidy reduction was accompanied with cash transfers to households by the amount of subsidy reduction to compensate for welfare loss. The results indicated that such transfers would reduce the adverse effects of the policy reform. The effects on GDP and household welfare were reversed, each rising by 0.3% and 0.2% respectively. Finally, sensitivity of the model results were examined by varying the elasticity of substitution of electricity use with factor uses in the process of production. It was found out that greater possibility of substitution causes the subsidy reform to have much larger positive effects compared to scenario 2 (or smaller contractionary effects compared with scenario 1). The simulation experiments suggest that subsidy reform would need to be accompanied by demand side stimulus or supply side initiatives to encourage electricity conservation.
Key worlds: Electricity subsidy reform, welfare, cross-price elasticity, Kuwait, SAM, CGE.JEL: E61, E64, H20, L94, Q43, Q48.
1
1. Introduction
Kuwait’s electricity is the most highly subsidized sector in the country - the rate of subsidy is
such that it amounts to generating electricity and almost freely distributing to users. The
extremely generous subsidy is intended to serve as a means of allocating welfare transfers to
resident households and businesses. The government owns a vertically integrated monopoly
and manages the entire supply chain all along electricity generation to retails (Wood and
Alsayegh 2014, Burney, 1998).
In recent years, however, it has become increasingly clear that the welfare oriented electricity
production and distribution have had adverse economy-wide effects, specifically conflicting
with other policy priorities such as environmental protections and misguiding resource
allocation (BuShehri and Wohlggenant 2012). There is a growing awareness among policy-
makers and researchers that the existing policy is not sustainable (Alotabi 2011, Darwish, et
al 2008; Darwish and Darwish 2008). The necessity for economic reforms in wider areas of
public utilities management is rooted in recent shifts in economic development strategies as
well as initiatives related to regional integration among GCC member countries (EIU 2010).
Despite the consensus on the necessity to reform electricity tariff, there are very few
analytical studies providing insights into options for optimal mix of policy instruments.
Burney and Al-Matrouk (1996) developed an econometric model and examined factor
substitution possibilities in the Kuwait electricity generation. Their findings implied that
“using factor prices, particularly fuel prices, as instruments will not only help with energy
conservation, but will also induce more demand for capital.” (p. 78). Since there can be many
options to apply fuel price as instrument of reform, it is important to go further and specify
alternatives ways to put the price policy into practice. For instance, the most straightforward
way to cause changes in fuel prices is to withdraw or reduce their subsidies.
BuShehri and Wohlggenant (2012) examined possible conflicts in objectives of reforming the
Kuwaiti electricity sector. The specifically drew attention to the inevitable welfare loss when
fuel prices are used as instruments of policy reform. They suggested that “the government
may, in the short run adopt a cash payment plan (i.e., credit balance) not exceeding the loss in
user welfare for the target group. This plan would allow consumers to enjoy the same level of
satisfaction as before the price increase, offsetting any political backlash…” (p. 425).
However, it should be noted that the rationale for compensating households for welfare loss
2
goes wider than equity considerations and gaining popular support for the reform. Given the
extent of the current subsidy, its withdrawal or reduction is bound to cause substantial
increases in electricity tariff. This will inevitably cause adverse demand side shock and hence
reductions in economic activities. Therefore, compensation schemes can also be seen as a
means of mitigating short to medium run adverse effects on economic activities.
This objective of this paper is to quantify economy-wide effects of reducing subsidy to the
Kuwait electricity sector and shed some light on the likely impacts of the envisaged reform.
A computable general equilibrium (CGE) model is applied to conduct simulation
experiments1. The simulation results indicate that reforming the electricity sector is likely to
have adverse economy-wide consequences unless it is accompanied by demand or supply
side measures to offset adverse consequences of the price increases. The former may include
compensating households for welfare loss through cash transfers to households while the
latter may include policy interventions to introduce and promote energy saving technologies.
The remaining part of this paper is classified into four sections. The next section highlights
study context. This is followed by description of approaches and methods. Results of
simulation results are discussed in the fourth section. The final section provides concluding
remarks.
2. Study context
2.1. Electricity tariff
Electricity users in Kuwait pay a nominal tariff of 2 fils (about 0.7 cents) per KWh2. In fact,
all GCC member countries are known for having extremely low electricity tariff rates
compared to the rest of the world. Qatar has relatively low tariffs among GCC member
countries other than Kuwait. Its residential tariff rate is about 2.2 cents which is about three
times the corresponding figure for Kuwait. This means Kuwait’s electricity tariff is the
lowest even by the standard of GCC member countries. The rest of GCC have differential
tariff rates for residential and business premises but their tariff levels are much higher
compared to corresponding figures for Qatar and Kuwait. For instance, tariffs in Saudi 1 An early and abridged version of this paper was published in: Gelan, A. 2014, Simulating impacts of reducing subsidy to Kuwait’s electricity sector. Oxford Energy Forum, 96: 32-35
2 This section heavily relied on data in a project report: TED-KISR 2014 (forthcoming). Household Conservation Behavior: A Case Study of Electricity and Water Demand in Kuwait, project progress report.
3
Arabia are close to 4.1 cents/KWh. This means that perhaps Kuwait’s tariff rate is the lowest
in the world.
Kuwait’s 2 fils (0.7 cents)/KWh was introduced in 1966 and has been retained at that level
ever since. On the other hand, the cost of generating electricity has sharping risen over the
years. For instance, cost of production per KWh escalated from 20 fils (7 cents) in 2000 to
about 40 fils (14 cents) in 2010, nearly doubling during that decade. This suggests that rate
of Kuwait’s electricity subsidy was 95% in 2010.
2.2. Patterns of consumption
The extremely generous subsidy has given rise to a pattern of unsustainable behaviour in
electricity use. Krane (2013) described the situation as a ‘dichotomy between energy value
and price’ to say that excessively low energy pricing induced ‘wanton consumption’ whereby
‘low pricing encourages consumption at rates above those warranted by the opportunity cost
of these fuels on global energy markets. Low prices also distort energy allocation preferences
while undercutting upstream investment and efficiency incentives’ Krane (2013).
These are reflected in a number of key aggregate indicators. In terms of economic efficiency
in use of electricity, measured in terms of GDP generated per unit of KWh used, Kuwait does
not only stand among the lowest in the world but also the situation has gotten worse over the
years. In 1990 GDP/KWh was USD 1.4 but this fell to USD1.2 in 2005. This contrasts with
experiences of other countries in the world; for instance, USA which was doing already much
better in 1990 (about USD 2.2/KWh) but also electricity use efficiency improved and reached
about USD2.7 in 2005.
In terms of electricity consumption per capita in 2005 Kuwait was the second highest in the
world (after Norway). This figure doubled between 1985 and 2005, rising from about 8
thousands KWh to 17 thousand KWh. It should be noted that Kuwait’s electricity is
generated entirely by using fossil fuels but other countries like Norway mostly use
renewables mainly hydroelectric power.
2.3. Prospects for reform
The pressure to reduce subsidy comes from various sources that can be classified into
domestic and regional policy environments. The first one is related to the latest medium term
development plan which expressed the government’s commitment to implement far reaching
4
liberalization of Kuwait’s economy (General Secretariat of Higher Council for Planning and
Development 2010). These are planned to be implemented through two firmly interrelated
strategies: (a) diversifying the structure of the economy by reducing the dominance of the oil
sector and encouraging growth of the non-oil sectors; and (b) promoting private sector
development and reducing the dominance of the public sector. Liberalization of public
utilities including electricity and water are prime targets in achieving these goals.
The regional policy environment is related to the GCC electricity grid interconnection.
Tabors (2009) provide interesting summary and economic analysis of the GCC grid
interconnection. The primary goal of this initiative is to provide power supply stability and
reliability by integrating high voltage transmission systems of all GCC member countries.
The economic rationale for this lies in the need to improve competitiveness in generation and
distribution capacity which is badly needed in the medium to the long-run in each country.
The system encourages countries to engage in trading electricity among each other based on
their comparative marginal costs. Cross border electrical energy trading has already been
taking place starting from summer 2010 although information on quantity traded is not yet
made available.
The relevance of GCC electricity grid connection to reform and regulation by each member
country lies in the pressure the interconnection is expected to cause in each country to
improve its efficiency so that its marginal cost of production and distribution would be
competitive relative its neighbours. In this regard, Kuwait is already at a relatively
disadvantageous position since its marginal cost is relatively high compared to other GCC
member countries. For instance, Qatar’s marginal cost of electricity production at peak is
less than half of that of Kuwait, $88/MWh and $188/MWh respectively. These differences
are largely explained by types of turbine or types of fuel used to fire electricity generating
plants – mostly natural gas in Qatar and heavy oil in Kuwait.
3. Approaches and methods
The structure of the Kuwaiti economy in the baseline year (2010) is displayed in a social
accounting matrix (SAM) developed for that year. Simulation experiments were conducted
by formulating a CGE model and utilizing the baseline SAM. These are discussed in detail
below.
5
3.1. The Social Accounting Matrix
A SAM was constructed for Kuwait with 2010 as a base year. The choice of the base year is
influenced by the existence of Kuwait Input-output Table produced for that year (CSB 2011).
The micro-SAM, the detailed SAM with its full dimension, has 85 accounts. The condensed
SAM (Table 1) has 13 accounts. The latter is obtained mainly by aggregating 37 activities
into a aggregate activity (labelled as ACT row 1 and column 1) and 37 commodities accounts
into a aggregate commodity account (labelled as COM in headings of row 2 and column 2).
Complete list of the activities and commodities is presented in Appendix A-1.
Table 1 – Condensed SAM for Kuwait 2010 (million KD)
ACT
1
COM
2
LAB-K
3
LAB-N
4
CAP
5
PTAX
6
SUBS
7
ITAX
8
HHD
9
GVT
10
S-I
11
ROW
12
Total
13
ACT 1 - 57,876 - - - - - - - - - - 57,876
COM 2 21,646 1,822 - - - - - - 9,916 5,203 6,878 22,054 67,520
LAB-K 3 4,469 - - - - - - - - - - - 4,469
LAB-N 4 4,795 - - - - - - - - - - - 4,795
CAP 5 26,879 - - - - - - - - - - 3,044 29,923
PTAX 6 87 - - - - - - - - - - - 87
SUBS 7 - -2,447 - - - - - - - - - - -2,447
ITAX 8 - 228 - - - - - - - - - - 228
HHD 9 - - 4,469 4,771 1,929 - - - - 5,786 - - 16,954
GVT 10 - - - - 27,675 87 -2,447 228 - - - - 25,543
S-I 11 - - - - - - - - 5,167 12,684 - - 17,851
ROW 12 - 10,040 - 24 320 - - - 1,870 1,871 10,973 - 25,098
Total 13 57,876 67,520 4,469 4,795 29,923 87 -2,447 228 16,954 25,543 17,851 25,098ACT : activities or industries COM : commodities or productsLAB-K : Kuwaiti labourLAB-N : Non-Kuwaiti labourCAP : capitalPTAX : production tax
ITAX : import taxesHHD : householdsGVT : governmentS-I : saving-investment balanceROW : rest of the worldTotal : control totals of each account
SUBS : product subsidies
Details of the remaining 11 accounts in the condensed SAM appear as they are in the micro-
SAM. Descriptions for the abbreviated account headings in Table 1 are presented below the
table.
In each SAM account, entries in columns denote out-goings or payments and entries in the
corresponding row represent in-coming or receipts. A particular cell in the matrix represents
a payment by the account in the column heading and a receipt by the account in the row
heading.
6
Column 1 represents cost of production which consists of intermediate inputs worth 21,646
million (in row 2) and 36,143 million value-added (sums of intersections with rows 3, 4, and
5). The CGE model is implemented with three factors of production: Kuwaiti labour, Non-
Kuwaiti labour, and producer surplus. This represents GDP at factor cost during 2010, whose
details are presented in Appendix-2. Intersection with Row 4 represents production tax
(PTAX) which is given as 87 million. Row 1 has only one entry, domestic commodity output
supply, which is estimated 57,876 million, intersection with column 2. Although it is a single
entry in the condensed version of the SAM, the activity-commodity linkage, the make matrix
has a dimension of 37 rows (number of activities) by 37 columns (number of commodities) in
the full SAM
Total commodity supplied to the domestic economy is presented in column 2. This consists
of commodity output from domestic activities at producer prices (in row 1) and aggregate
imports at world prices (in row 12). The remaining entries in this column are net commodity
tax (intersection with rows 7 and 8) and trade margins (intersection with row 2). Aggregate
commodity supply at market price (inclusive of tax and trade margins) is 67,520 million (in
row 13). Row 2 allocates total commodity supply to different destinations: intermediate
demand, transaction demand, final demand, investment demand, and exports.
Column 3, 4 and 5 distribute national income, generated in the process of production and
factor income earned from abroad to domestic institutions (rows 9 and 10). This model has
only one aggregate household group whose receipts (in row 9) consists of labour income,
capital income, and transfers from the government. The main item in household expenditure
(column 9) is final consumption expenditure (intersection with row 2); other outgoings from
this account are transfer to the ROW (in row 12) and private household savings (in row 11).
The bulk of government revenue comes from the oil sector, which in this SAM framework
comes as a transfer from the capital account (through column 5). The other two sources with
negligible contributions are production tax (in column 6) and import taxes (in column 8).
The other figure in the government revenues row denotes net commodity taxes, which is a
negative number (-2,447), representing product subsidy. When expanded into 37 separate
cells, it denotes subsidy by commodity group. It is the particular cell against the electricity
sector (IO class 40, see Appendix A-3) which is the focus of this study – economy-wide
impacts of reducing electricity subsidy. While final expenditure on goods and services and
transfers to households and transfer to the rest of the world represent major items of
7
government expenditure, the balance between the sum of these items and total government
revenue gives government surplus (in row 11).
In the Kuwaiti SAM, substantial surplus in the external sector is transferred to the ROW
through the capital account (in row 12). It should be noted that domestic capital formation
6,878 million (in row 2) constitutes only 40% of the total savings made available by domestic
institutions (household savings plus government savings).
In addition to SAM data, the implementation of the CGE model requires supplementary
satellite accounts such as employment are separately estimated in line with flow variables in
the SAM. In appendix – 2, columns 1 and 2, come from such satellite data while the other
columns provide value-added related micro SAM version the concisely presented macro
SAM discussed in this section. The satellite accounts are compiled by making use of data
obtained from CSB (2011) and reconciling the latter with other sources (PACI 2014, CSB
2011, CSB 2010).
3.2. Description of the CGE Model
The model used for this study was previously adapted from IFPRI standard CGE model
(Lofgren et al 2002) to the Kuwait Economy and used for policy analysis (TED-KISR 2004).
In this study we have undertaken further reformulation and adaptations of the previous model
to conduct the simulation experiments. The focus in this version of the model is on
application of key elasticity values obtained through previous econometric studies using
Kuwaiti data. Furthermore, the model is calibrated with the recently constructed Kuwaiti
SAM which is discussed in the preceding section.
Figure 1 displays the structure of the model. Interactions between its components are
classified into three blocks (see Panels 1, 2 and 3). In Panel 1, starting from activity level
(QAA), the process of production is modelled as a nested multi-level structure. The first level
of the nesting structure determines sectoral output (QAA) as aggregation of intermediate
inputs (QINTAA) and value-added and electricity (QVEA) using a Leontief functional form.
This means substitution between material inputs is not allowed at this level (subscripts A and
C denote activities and commodities respectively). At the second nest of the production
function, the value-added and electricity bundle QVEa is split into demand for aggregate
factor input (QVAA) and electricity (QINTEA). At this stage, the constant elasticity of
substitution (CES) is employed to allow for substitution in input uses. In other words, this
specification allows electricity saving in the process of production. Burney and Al-Matrouk
8
(1996, p.77) have used Kuwaiti data and estimated cross-price elasticity between energy and
labour 0.1691 with labour and 0.3597 with capital. In this study electricity-Gross value-
added substitution elasticity of (ve) of 0.3 in such a way that it would fall within the range of
the previous estimates. During the simulation experiment, a sensitivity analysis is conducted
to examine the role of variations in this parameter value on endogenous variables in the
model.
9
Figure 1 – Structure of the Kuwaiti CGE Model
10
CD+CD
PANEL 3
PANEL 2
PANEL 1
Investment demand
(QINVC | PQC)
Government final demand(QGC | PQC)
Household final demand(QHC | PQC)
LEONTIEF
LEONTIEF
CET
Imports(QMC | PMC)
Composite commodity(QQC | PQC)
Exports(QEC | PEC)
Domestic demand(QDC | PDC)
CD
LEONTIEF
Aggregate commodity output(QXC | PXC)
CES
>>> Commodity output(QXACA,C | PXACA,C)
Commodity output(QXAC1,1 | PXAC1,1)
Activity output(QAA | PAA)
>>>Other int. inputs(QINT36,A | PQ36,A)
Other int. inputs(QINT1,A | PQ1,A)
Other composite int inputs(QINTAA | PINTAA)
VA & Elec composite(QVEA | PVEA )
CES (0.3)
Electricity(QINTEE,A | PQE)
Value-added(QVAA | PVAA)
Capital(QFCAP,A | PFCAP,A)
Labour (QFLAB,A|PFLAB,A)
CD
Non-Kuwaiti labour (QFLABN,A| PFLABN,A)
Kuwaiti labour (QFLABK,A| PFLABK,A)
The value-added and other intermediate composites are split into their component parts.
Using the Leontief functional form, the composite quantity of intermediate demand by each
producing sector is disaggregated into demand for different commodities (QINTC,A). The
value-added composite is disaggregated into aggregate labour and capital using a Cobb-
Douglass functional form. Finally, the aggregate labour is disaggregated into demand for
Kuwaiti labour and non-Kuwaiti labour, again using Cobb-Douglass specification.
Panel 2 displays specifications and functional forms used in transforming activity output to
commodity outputs. This corresponds with the make matrix part of the SAM that provides
linkages between commodity supply and activity outputs. Totals of domestic commodity
outputs (QXC) are obtained by aggregating over their activity origins of QXACA,C
(commodity C produced by Activity A). On the other hand, the sum of commodities C
produced by an activity A should equal the level of activity output (QAA). This model
follows the functional form applied by Lofgren et al (2002) - CES for the former and Leontief
for the latter.
Panel 3 displays flows of commodity supply. The upper part shows a constant elasticity of
substitution (CET) function that allocates domestic commodity output (QXC) to exports to the
rest of the world (QEC) and domestic supply (QDC). The lower part of the diagram shows
determination of domestic demand for a composite commodity (QQC) which is obtained
through a two-way aggregation. The first one is the Armington assumption implies that
commodities from different geographical origins are treated as imperfect substitutes
(Armington, 1969). This is implemented by applying a CES aggregation of commodities by
source of supply – domestic output (QDC) and imports from the rest of the world (QMC). The
second one is summation of commodity destinations by domestic institutions – intermediate
demand by producing sectors (QINTC,A), household final consumption (QHC,H), government
final consumption (QGC), and capital formation or investment demand (QINVC).
4. Simulation results
4.1. Scenarios
The model was run three rounds to conduct simulation experiments. The first run replicates
the base year situation or the baseline scenario. This represents the status quo whereby the
subsidy system remains as in the baseline SAM. The second and third runs impose policy
shock under different scenarios. These yield results which are different from the base run.
11
The differences between the base scenario and the policy shocks yield the effects of the
policy change.
A 30% reduction in electricity subsidy was applied and economy-wide impact of this change
was evaluated. This rate of subsidy reduction was arbitrarily chosen, it can be set at lower or
higher rate of changes. However, it is not feasible to implement much larger shocks such as
abolishing electricity subsidy altogether in a CGE modelling context, particularly when the
rate of subsidy at the base is as high as 91% (see appendix 3). In such cases, policy shocks
can be evaluated by applying relatively small changes to get a sense magnitude regarding
impacts.
For this policy experiment, the 30% subsidy reduction is simulated in two scenarios.
Scenario 1 was run without compensating households for any welfare loss resulting from
inevitable rises in electricity tariffs due to the partial withdrawal of subsidy. Scenario 2 is
inspired by the recommendation BuShehri and Wohlgenant (2012) that the Kuwait
government would need to compensate target household groups for their welfare loss.
4.2. Intra-sectoral effects
Sectoral effects are discussed in the subsequent section but it is important to examine impacts
on the electricity sector itself. Table 1 displays changes in selected indicators. Clearly, the
intra-sectoral effects of the policy shocks are about equal in both scenarios. The reason is that
scenario 2 is an increase in transfer payments which is not likely to cause the amount of
electricity consumed by households, given the functional forms applied to specify the
household consumption function.
The most immediate effect is about three fold increase in electricity tariff in both scenarios.
This means a 30% reduction in electricity subsidy in Kuwait is likely cause tariff to rise from
2.0 to 6.1 fills / KWh. The amount of electricity generated and gross value added in the
sector is expected to fall by 21% and 24% respectively.
Table 2 - Effects on the electricity sector
Scenario 1 Scenario 2Tariff 307.58 306.02Output -20.90 -20.44Value added -24.16 -23.74Employment:
Total -3.73 -3.66Kuwaiti 0.00 0.00Non-Kuwaiti -24.06 -23.58
12
Demand for electricity:Final demand -5.00 -4.96Intermediate demand -35.10 -34.31
Employment in the sector declines from the baseline level by 3.7%. It should be noted that
Kuwaitis constitute the bulk of the work force in the electricity and water production and
distribution sectors, accounting 84% of total employees (see appendix 2). Given that the
number of Kuwaiti employees is not allowed to vary (by assumption). It follows that the
brunt of changes in employment falls on the non-Kuwaiti labour force that decline by about
24% in both scenarios.
The last two rows of Table 2 indicate the extent to which electricity users will reduce their
consumption as a result of subsidy reduction and hence increase in tariff rates. In the
framework of this model, household final consumption will fall by about 5% from the base
line level while total intermediate demand for electricity declines by 35%. It should be noted
that the difference in the rate of energy saving between the final and intermediate demand is
explained by the functional forms employed in specifying the two categories of demand for
electricity. Household demand specification followed the LES system, where demand for
each community group cannot fall below a certain minimum. Expenditure elasticity are
borrowed from Burney and Al-Mutairi (1993). As far as are aware, this is the only
econometric study on Kuwait economy on the relationship between income and expenditure.
On the other hand, intermediate demand by the production sector is specified using the CES
functional form which allows substitution between electricity and factor inputs.
4.3. Macroeconomic effects
Now we turn to effects on key macroeconomic indicators. Figure 1 displays percentages
changes for the two scenarios from the baseline levels for selected macroeconomic variables.
Figure 1 - Effects on key macroeconomic variables
13
GDP
BOP
Govt Saving
Welfare
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
-1.2
-2.1
-2.1
-1.6
0.3
0.9
-0.5
0.2
Scenario 2Scenario 1
Change from the base line (%)
Scenario 1 results indicate electricity subsidy withdrawals without compensating households
for welfare loss is likely to lead to adverse macroeconomic outcomes. GDP, measured in
gross value-added, terms decline by 1.2%. The government budget surplus and balance of
payments (GSV and BOP) decline by equal proportions, 2.1%. Aggregate welfare indicator
declines by 1.6%.
Scenario 2 results show that relocating subsidy from producer to consumer, at least in the
context of Kuwait, is likely to lead to noticeable positive macroeconomic effects. GDP
increases by 0.3% when a subsidy reduction is accompanied by household transfer payment
of equivalent magnitude with the quantity of subsidy withdrawn. The BOP surplus increases
from the baseline level by 0.9% but government savings declines by 0.5%, which is a smaller
proportion compared to corresponding results for scenario 1.
Aggregate household welfare marginally rises, by 0.2%, from the baseline level. The fact
that scenario 2 has a consistently positive effects on most macroeconomic variables indicates
that shifting subsidy from producer to consumers leads to positive economy-wide effects. On
the one hand, the reduction of producer subsidy improves resource allocation across sectors,
resources moving out of less productive to more productive sectors. On the other hand, the
relocation of subsidies to households causes consumer demand, which in turn stimulate more
production activities.
14
Figure 2 presents effects of the policy shock on aggregate expenditure components. GDP,
expenditure measure, declines by 1.2% in scenario 1 and rises by 0.3% in scenario 2. As we
expect, these are equal to the corresponding changes in GDP net of taxes, production
approach, reported in Figure 1. It is interesting to note that the effects on household
consumption in the two scenarios are about equal but in opposite directions. Without
compensation, household consumption declines by about 0.9% but with compensation
household expenditure it increases by about the same proportion.
Figure 2 - Effects on domestic expenditure
Gross domestic expenditure
Household demand
Government demand
Investment demand
Exports
Import
-2.50 -2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50
-1.17
-0.91
0.07
-1.13
-2.33
-1.28
0.25
0.93
0.14
-1.70
-0.89
-0.73
Scenario 2Scenario 1
Change from the base line (%)
The model is implemented with fixed government expenditure and flexible government
budget surpluses. As a result, the effect on government expenditure is minimal. On the other
hand, the saving-investment balance was obtained by allowing investment to vary but fixing
private savings. Consequently, investment declined from the baseline level in both scenarios.
Both exports and imports decline in each scenario but they declined in scenario 2 by
relatively smaller proportion than in scenario 2.
4.4. Labour market effects
The Kuwaiti labour market is highly segmented by nationality (Kuwaiti and non-Kuwaiti) as
well as in pay structure. One of the indicators for labour market segmentation is the huge
15
wage differentials between Kuwaiti and non-Kuwaiti employees. Additionally, there are
differences in degree of flexibilities in employment contracts as well as sectoral mobility. The
Kuwait government guarantees full employment of nationals (Kuwaitis), which means
employers are not free to lay off Kuwaiti employees regardless of the economic conditions.
Also, Kuwaiti employees are highly concentrated in certain sectors with limited possibility of
their relocation or mobility across sectors regardless of differentials in sectoral employment
opportunities. On the contrary, employment contracts for non-Kuwaitis is highly competitive
in that they can be employed with high degree of flexibilities in employment contracts both in
terms of in mobility between sectors and payment conditions.
Table 2 – Labour market effects
Scenario 1 Scenario 2Employment level Total -0.89 -0.12
Kuwaiti 0.00 0.00Non-Kuwaiti -1.09 -0.15
Labour income Total -1.05 -0.30Kuwaiti -0.86 -0.23Non-Kuwaiti -1.21 -0.37
The model specifications have taken into account the underlying labour market conditions
outlined above. Accordingly, the labour market closures were specified in such a way that
wage and employment of Kuwaitis were fixed, which means they are not allowed to vary
during the model run. However, non-Kuwaiti labour is characterized by full flexibility in
wage levels and mobility across sectors.
Table 2 presents aggregate labour market effects that taken into account details of the labour
market closures outlined above. In scenario 1, total employment income decline by 1.05%,
with the Kuwaiti and non-Kuwaiti labour income declining by 0.86% and 1.21%
respectively. In terms of levels of employment, overall employment level falls by 0.89%.
This is mostly explained by declines in non-Kuwaiti labour (by 1.09%) but Kuwaiti labour
remains at the baseline level as expected under fixed employment closure for this category of
labour. Scenario 2 shows much less adverse labour market effects. Kuwaiti labour income
declines by only 0.23 and non-Kuwaiti labour income by 0.37%. Total employment declines
by 0.12%.
16
4.5. Sectoral value-added effects
As noted earlier, the reduction of subsidy causes electricity tariff to rise by around 307% in
both scenarios. The marginal differences between the two scenarios in their price effects can
be explained by the role in demand side stimulus on commodity prices in scenario 2. Now
we turn our attention to explaining how the three-fold increase in electricity tariff triggers
further increases in prices of other commodities and economic activities measured by sectoral
gross value added.
Figure 3 – Effects on sectoral gross value-added
Base metalsReal estate activitiesPetroleum products
Crude oil and natural gasThe furniture and related products
Repair of goodsAgriculture and livestock
Hotels and restaurantsLeather products
Machinery and equipmentFin. Insur. and related services
Food products and beveragesPrecision instruments
Business services ActivitiesWater production and distribution
Computer and related activitiesTextiles
Other non-metallic mineral productsWholesale Trade
Clothing and fur productsRoad transport services
Transport and storage servicesRetail Trade
Printing and publishingWood and wood products exc furn.
Fabricated metal products Construction
Other servicesPaper and paper productsEquipment rental services
Waste and scrapsPlastics and rubber products
Postal and Telecom. servicesEducation and health services
Public administration and defenseChemicals products
-5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00
Scenario 2Scenario 1
Figure 3 displays changes in sectoral value added in both scenarios, the data is sorted in
increasing order of scenario 1 effects. Clearly, the highest declines in value-added happen in
manufacturing sectors that are use electricity more intensively. It should be noted that the oil
sector, both primary crude oil extraction and petroleum products sub-sector of manufacturing,
are among the most adversely affected production sectors. Chemical products subsector is an
exception, its output increased in both scenarios. Service sectors encounter contractions but
only by a relatively small percentage points compared to the industrial sectors.
17
4.6. Sensitivity to elasticity values
It is useful to examine responsiveness of the simulation results to variations in elasticity
values. Energy demand elasticity play in play a critical role in quantifying effects of energy
sector reforms, particularly reduction of energy subsidies (He et al 2011; 2010, Burney 1996).
Table 3 presents variations in simulation results in response to changes in the elasticity value
for substitutions between electricity and aggregate factors of production. The simulation
results discussed so far were obtained with the electricity -value-added aggregate elasticity of
substitution (ve) being held at 0.30. The corresponding results for the selected variables are
displayed in the middle (shaded) column in Table 3. Now we conduct sensitivity of results
by varying the values of ve and then examining how simulation results change from those
reported in the preceding section.
Table 3 - Sensitivity of simulation results to variations in demand elasticity (ve)
(ve =0.25 (ve =0.30 (ve =0.35Tariff 307.51 306.02 304.63Output -17.98 -20.44 -22.72Value added -20.95 -23.74 -26.32Employment (Total) -3.20 -3.66 -4.08Intermediate demand for electricity -29.58 -34.31 -38.70Output, savings and welfare: Gross Domestic Product (GDP) 0.22 0.25 0.28 Balance of payments 0.74 0.92 1.08 Government saving -0.53 -0.51 -0.49 Household welfare 0.19 0.22 0.25Expenditure: Gross domestic expenditure (GDE) 0.22 0.25 0.28 Household demand 0.91 0.93 0.96 Government demand 0.06 0.14 0.22 Investment demand -1.48 -1.70 -1.91 Exports -0.82 -0.89 -0.97 Import -0.65 -0.73 -0.81
It proves useful to start with changes in intermediate demand for electricity. It is
straightforward that higher price elasticity means greater change in quantity demanded.
Accordingly, change in intermediate demand for electricity is directly related to the sizes of
the elasticity values. When the value of ve reduced to 0.25, then electricity use fell by less
percentage point, about 30% compared to the 34% decline with ve = 0.30. On the contrary,
when the value of ve increased to 0.35 then we get a larger fall in demand for electricity,
39%. The rises in tariff rates go in the opposite directions, i.e., greater increase in electricity
18
tariff with smaller elasticity values and vice versa. It should be noted that “substitution
elasticity” in the context of this simulation analysis can only mean savings in electricity use,
change in quantity of electricity demand. The latter imply presence of excess use of
electricity over and above the essential requirement in technical relationships in the process
of production.
The variations with parameter values in the price and quantity demanded for electricity
causes changes in the rest of the economy. Within the electricity production sector,
employment value-added and total output rise and fall with possible extents of electricity
savings. Similarly, much greater possibility of electricity savings will cause subsidy
reductions to have greater expansionary effects on most macroeconomic variables.
Electricity subsidy withdrawals can have expansionary effects or positive stimulus on the
economy only when electricity users are in a position to reduce electricity use. This may
indicate the importance of accompanying measures to enable consumers to conserve
electricity.
5. Conclusion
The simulation experiments indicate that subsidy reduction does not necessarily cause any
substantial contractions in economic activities or declines in household welfare. The
differences between the two policy scenarios indicated that the adverse demand side effect of
the subsidy reform dominates unless the reform is accompanied with other measures.
Specifically, when households are compensated for their welfare loss, then this changes the
effects of the policy reform to become positive and hence aggregate GDP and household
welfare effects became positive as well. However, the results of the simulation experiments
reported in this study should be interpreted with caution. The model used for this analysis is
highly aggregate, and hence it does not account for distributional effects, particularly
differential impacts on households in different income brackets.
Policy reforms, such as reduction of subsidy to Kuwait’s electricity, can realize positive and
desired results by implementing accompanying measures in addition to the actual change to
the policy instruments. The accompanying measures may include organizational or
technological changes, both of which mean innovations which are not quantifiable in a
modelling framework. This means depending on whether or not these innovations
accompany the policy change, the results reported in this study can overestimate or
19
underestimate the effects of reducing subsidy. Similarly, the applications of packages of
reform measures can influence the speed with which the economy will realize potential
benefits from reducing or abolishing subsidies to public utilities.
For instance, the 21% contraction in the electricity sector was likely to overstate the adverse
effects since the reform package will not be confined to just reducing or removing the
subsidy but also partial or full privatization of the public utility which in turn would lead to
substantial efficiency gains through organizational changes and introduction of latest
technology in the process of generating electricity. If this is the case, then the economy-wide
positive stimulus of the reform can be much greater or adverse effect much smaller than the
reported simulation results.
References
Armington, P. S. 1969. The geographic pattern of trade and the effects of price changes. IMF Staff Papers 16: 176 – 199.
Al-Hasan, A.Y.; A.A. Ghoneim; and A.H. Abdullah. 2004. Optimizing electrical load pattern in Kuwait using grid connected photovoltaic systems. Energy Conversion and Management 45 (4):483-494.
Alotaibi, Sorour. 2011. Energy consumption in Kuwait: Prospects and future approaches. Energy Policy 39(2): 637-643.
Ayyash, S.; M. Salman; and N. Al-Hafi. 1985. Modelling the impact of temperature on summer electricity consumption in Kuwait. Energy 10(8): 941-949.
Burney, Nadeem A. 1998. Economies of scale and utilization in electricity generation in Kuwait. Applied Economics 30: 815-819.
Burney, Nadeem A. and Faisal T. Al-Matrouk. 1996. Energy conservation in electricity generation: A case study of the electricity and water industry in Kuwait. Energy Economics. 18(1–2): 69-79.
Burney, Nadeem A., and Naief Al-Mutairi. 1993. Household consumption patterns in Kuwait. The Middle Eastern Business and Economic Review 5(2): 1-12.
BuShehri, Mahmoud A.M. and Michael K. Wohlgenant. 2012. Measuring the welfare effects of reducing a subsidy on a commodity using micro-models: An application to Kuwait's residential demand for electricity. Energy Economics 34(2): 419-425.
CSB. 2010. National Accounts Statistics 2011. Central Statistical Bureau, Ministry of Planning Kuwait. http://www.csb.gov.kw/Socan_Statistic_EN.aspx?ID=26. (last accessed on 07/05/2014)
20
CSB. 2011. Annual Survey of Establishments. Central Statistical Bureau, Ministry of Planning Kuwait. http://www.csb.gov.kw/Socan_Statistic_EN.aspx?ID=26. (last accessed on 07/05/2014)
CSB. 2014. National Accounts Statistics Input & Output Tables 2005-2010. Central Statistical Bureau, Ministry of Planning Kuwait. http://www.csb.gov.kw/Socan_Statistic_EN.aspx?ID=26 . (last accessed on 07/05/2014).
Darwish, M.A ; F.M. Al-Awadhi; and A.M. Darwish. 2008. Energy and water in Kuwait Part I. A sustainability view point. Desalination 225(1–3): 341-355.
Darwish, M.A. 2001. On electric power and desalted water production in Kuwait. Desalination 138 (1–3): 183-190.
Darwish, M.A. and A.M. Darwish. 2008. Energy and water in Kuwait: A sustainability viewpoint Part II. Desalination 230(1–3): 140-152.
EIU. 2010. The GCC In 2020: Resources for the Future – debating how the GCC states will manage key natural resources (oil and gas, minerals, water and food) over the next decade. A report from the Economist Intelligence Unit Sponsored by the Qatar Financial Centre Authority, London. www.eiu.com/gulfin2020
Gelan, A. 2014. Simulating impacts of reducing subsidy to Kuwait’s electricity sector. Oxford Energy Forum 96: 32-35
General Secretariat of Higher Council for Planning and Development. 2010. General Frame of Development Plan For the State of Kuwait 2010/2011 – 2013/2014 . State of Kuwait, February 2010
He, Y. X.; L. F.Yang; H. Y. He; T. Luo; and Y. J. Wang. 2011. Electricity demand price elasticity in China based on computable general equilibrium model analysis. Energy 36(2): 1115-1123.
He, Y.X.; S.L. Zhang, L.Y. Yang; Y.J. Wang; and J. Wang. 2010. Economic analysis of coal price–electricity price adjustment in China based on the CGE model. Energy Policy. 38(11): 6629–6637.
Koesler, Simon and Michael Schymura. 2012. Substitution Elasticities in a CES Production Framework An Empirical Analysis on the Basis of Non-Linear Least Squares Estimations. ZEW Discussion Paper No. 12-007. Centre for European Economic Research, ftp://ftp.zew.de/pub/zew-docs/dp/dp12007.pdf
Krane, Jim. 2013. Stability versus Sustainability: Energy Policy in the Gulf Monarchies. EPRG Working Paper 1302, Cambridge Working Paper in Economics 1304.
Lofgren, H.; R. L. Harris; and S. Robinson. 2002. A Standard Computable General Equilibrium (CGE) Model in GAMS. International Food Policy Research Institute. Microcomputers in Policy Research, Volume 5, Washington, D.C.
PACI. 2014. Population: Statistical reports. The Public Authorities for Civil Information (PACI).
21
http://stat.paci.gov.kw/englishreports/#DataTabPlace:view1ArcGISRegionMap. (last accessed on 12/06/2014)
Tabors, Richard D. 2009. Interconnection in the GCC Grid: The economics of change. Presented at System Sciences, HICSS 2009: 42nd Hawaii International Conference, Big Island, Hawaii, 5-8 Jan. http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=4755521.
TED-KISR. 2014(forthcoming). Household conservation behavior: a case study of electricity and water demand in Kuwait. Project progress report, Kuwait.
TED-KISR. 2004. A detailed specification of the standard static computable general equilibrium model for Kuwait developing quantitative planning tools. Technical report No. 3. Submitted to Ministry of Planning State of Kuwait
Wood, Michael and Osamah A. Alsayegh. 2014. Impact of oil prices, economic diversification policies and energy conservation programs on the electricity and water demands in Kuwait. Energy Policy 66: 144-156
22
Appendix – A: Baseline Data
Appendix – A-1: Compensation of employees by sector
Input-output categoriesCodes Descriptions
01 Agriculture and Livestock11 Extraction of crude oil and natural gas15 Food products and beverages17 Textiles18 Clothing and the making and dyeing of fur19 Leather preparation and tanning plus manufacture handbags and leisure bags plus saddles20 Wood and wood products and cork excluding furniture21 Paper and paper products22 Printing publishing and copying of recorded media science23 Coke and refined petroleum products and nuclear fuel24 Chemicals and chemical products25 Plastics and rubber products26 Other non-metallic mineral products27 Base metals28 Fabricated metal products excluding machinery and equipment29 Machinery and equipment that were excluded in previous category31 Precision instruments36 The furniture industry and furniture upholstery and paint37 Waste and scrap that is of non-ferrous metals40 Electricity and gas41 Water production and distribution45 Construction51 Wholesale Trade52 Retail Trade526 Repair of goods55 Hotels and restaurants60 Road transport services
61-63 Transport and storage services64 Postal and Telecommunications services
65-67 Financial insurance and related services70 Real estate activities71 Machinery and equipment rental and personal goods72 Computer and related activities74 Business services Activities75 Public administration and defense
80-85 Education and health services90+ Other services
23
Appendix – A-2: Sectoral distribution of value-added (2010, Thousands KD)
No. personsemployed
Compensationof Employees
Othervalue-added
Grossvalue-added
%share
Codes Kuwaiti Non-Kuwaiti
Kuwaiti Non-Kuwaiti
Total
01 705 30,891 347 3,486 3,833 121,448 125,281 0.311 8,115 12,193 255,665 164,667 420,332 18,032,054 18,452,386 51.115 487 21,942 7,771 66,542 74,313 120,448 194,761 0.517 18 1,972 127 3,921 4,048 4,203 8,251 0.018 112 14,877 522 25,559 26,081 22,671 48,752 0.119 2 227 27 592 619 596 1,215 0.020 6 1,194 47 2,353 2,400 2,682 5,082 0.021 26 2,268 323 7,308 7,631 13,180 20,811 0.122 288 5,427 4,590 31,093 35,683 15,347 51,030 0.123 4,087 1,163 185,315 37,432 222,747 1,613,573 1,836,320 5.124 1,401 4,090 48,428 29,227 77,655 247,644 325,299 0.925 77 4,235 971 12,255 13,226 24,732 37,958 0.126 145 10,206 2,032 35,313 37,345 84,290 121,635 0.327 11 1,172 90 5,574 5,664 6,484 12,148 0.028 177 11,775 1,790 31,871 33,661 32,363 66,024 0.229 32 4,490 286 10,870 11,156 4,231 15,387 0.031 370 14,519 7,654 58,472 66,126 47,610 113,736 0.336 48 6,506 278 14,665 14,943 17,084 32,027 0.137 11 763 80 1,918 1,998 1,841 3,839 0.040 6,568 1,206 38,300 1,614 39,914 224,704 264,619 0.741 4,863 893 28,360 1,195 29,555 414,567 444,123 1.245 2,611 138,318 16,977 422,803 439,780 287,705 727,485 2.051 1,393 35,912 7,085 113,142 120,227 282,943 403,170 1.152 4,309 120,752 24,563 347,865 372,428 515,390 887,818 2.5526 425 17,563 3,364 36,092 39,456 40,353 79,809 0.255 1,346 62,198 17,087 174,894 191,981 43,378 235,359 0.760 1,659 21,921 12,736 54,907 67,643 59,881 127,524 0.4
61-63 6,356 55,541 140,170 382,586 522,756 144,445 667,201 1.864 955 17,243 23,676 261,465 285,141 750,646 1,035,787 2.9
65-67 22,701 39,522 687,288 659,391 1,346,679 1,527,006 2,873,685 8.070 1,431 9,778 14,592 65,862 80,454 1,714,585 1,795,039 5.071 156 4,218 1,628 15,089 16,717 62,532 79,249 0.272 189 6,401 3,487 56,914 60,401 15,062 75,463 0.274 2,001 65,510 15,730 149,548 165,278 53,607 218,885 0.675 187,002 67,863 1,399,870 116,593 1,516,463 140,408 1,656,871 4.6
80-85 68,670 53,047 1,476,981 681,720 2,158,700 104,877 2,263,578 6.390+ 8,139 619,007 40,682 710,111 750,793 84,439 835,231 2.3
Totals 336,891 1,486,802 4,468,917 4,794,910 9,263,828 26,879,012 36,142,839 100.0
24
Appendix – A-3: Sectoral distribution of subsidy payments (2010, Thousands KD)
Codes Value ofcommodity output Amount of subsidy Subsidy rate (%)
01 182,995 -38,537 -21.111 18,930,170 - -15 583,682 -64,514 -11.117 23,593 - -18 79,437 - -19 2,447 - -20 20,120 - -21 76,108 - -22 73,066 -543 -0.723 11,673,717 -710,777 -6.124 1,109,152 - -25 121,378 - -26 341,584 -4,244 -1.227 87,591 - -28 214,163 - -29 38,690 - -31 260,594 - -36 95,572 - -37 12,627 - -40 663,905 -606,541 -91.441 1,056,271 -1,007,717 -95.445 2,324,698 - -51 617,465 - -52 1,204,622 - -526 112,757 - -55 567,137 - -60 217,624 - -
61-63 2,227,697 - -64 2,046,854 - -
65-67 3,483,180 - -70 2,230,729 - -71 197,745 - -72 153,802 - -74 462,335 - -75 2,629,140 - -
80-85 2,679,173 - -90+ 1,074,265 -14,226 -1.3
Totals 57,876,085 -2,447,100 -4.2
25
26
27
Top Related