I. Introduction

Sound fiscal policy is crucial to promote price stability and sustainable growth in output and employment. Thus, understanding the relationship between government spending and revenue is important to evaluate how to address fiscal imbalances. Several hypotheses have resulted from the causal link of this relationship. A bi-directional causality between revenue and expenditure supports “the fiscal synchronization hypothesis,” i.e., over time, expenditure decisions are not made in isolation from revenue decisions. In contrast, if causality runs from revenue to expenditure, “the revenue-spend hypothesis,” the spending level adjusts to changes in revenue, while “the spend-revenue hypothesis” suggests the opposite in that changes in spending induce changes in revenue. Although this relationship has been extensively studied in industrial and some developing countries, with the exception of a paper by Ghamdi (1991), it has not been examined for the oil-dependent Gulf Cooperation Council (GCC) countries—Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and the United Arab Emirates.² Some GCC countries have consistently recorded overall fiscal deficits since the early 1980s after oil prices peaked in 1979–81. In addition, with oil revenue accounting for about three-quarters of government revenue in most of these countries, fluctuations in crude oil prices have led to volatile revenue and swings in spending.

The focus of the paper is, therefore, to test the validity of these various hypotheses in the case of the GCC countries for the period from 1975–2000—a common available period of observations for all these countries. We examine the direction of causality between total government expenditure and revenue in these countries by utilizing a cointegration and error-correction modeling framework, which provides a more comprehensive test of causality than the standard Granger causality test, and by calculating a variance decomposition analysis to determine the stronger causality if results show a bi-directional causality (Appendix I). In addition, we present impulse responses to shed light on the dynamic relation of expenditure to a revenue shock.

The paper proceeds as follows. First, a brief review is presented of developments in government revenue and spending in GCC countries over the past three decades. Then, we report our empirical results of the testing of the causality between revenue and expenditure. The last section summarizes our findings and the policy implications.

II. Revenue and Spending Developments in GCC Countries

Over the past three decades, GCC countries have generally followed procyclical fiscal policies to changes in oil revenue (Table 1). Following the sharp increase in global oil prices in the 1970s and early 1980s, government spending in all these countries rose as fast as oil revenue through a massive public investment program in infrastructure, fiscal incentives to develop the industrial sector, and the adoption of a generous welfare system. Notwithstanding the increase in spending, sizable overall fiscal surpluses were recorded in all GCC countries during those years, leading to a sharp accumulation of official assets (Figures 1–6).

Table 1.

GCC Countries: Change in Total Government Spending and Revenue in Selected Periods, 1975-2000 ^1/

(In percent)

Source: Authors’ calculations based on national statistics.

^1/Total expenditure includes net lending, and revenue, grants and investment income.

^2/Consolidated fiscal accounts (federal plus the three largest emirate governments).

^3/Includes investment income from 1991 onwards.

^4/Average nominal price of UK Brent, Dubai, and West Texas Intermediate.

Table 1.

GCC Countries: Change in Total Government Spending and Revenue in Selected Periods, 1975-2000 ^1/

(In percent)

			1975-80	1980-86	1986-90	1990-94	1994-96	1996-98	1998-2000
Bahrain
	Change in expenditure		201.4	35.5	23.1	−8.8	8.8	3.6	16.2
		Current	…	85.0	17.2	27.4	−1.0	6.9	18.5
		Capital	…	11.9	−13.7	5.3	−19.1	39.8	−14.8
	Change in revenue		236.1	−4.6	17.1	5.8	20.2	−12.5	89.1
	Ratio change in expenditure/revenue		0.9	−7.6	1.3	−1.5	0.4	−0.3	0.2
Kuwait
	Change in expenditure		108.6	28.0	−3.7	46.3	4.8	−9.6	0.7
		Current	…	37.5	19.2	57.4	7.0	−11.7	5.0
		Capital	…	38.3	−43.3	−3.0	−10.5	8.3	−29.8
	Change in revenue		153.9	−31.6	0.2	−23.2	29.7	13.1	31.8
	Ratio change in expenditure/revenue		0.7	−0.9	−15.8	−2.0	0.2	−0.7	0.0
Oman
	Change in expenditure		107.5	116.4	−2.0	19.4	0.4	−1.4	18.7
		Current	…	112.5	25.3	13.2	3.1	−3.6	18.5
		Capital	…	139.0	−60.1	66.7	−11.6	9.7	10.9
	Change in revenue		201.2	16.2	46.5	−11.5	32.4	−22.1	87.5
	Ratio change in expenditure/revenue		0.5	7.2	0.0	−1.7	0.0	0.1	0.2
Qatar
	Change in expenditure		106.8	−4.7	9.3	15.0	26.9	1.1	8.7
		Current	…	13.9	15.5	6.4	27.1	6.6	9.2
		Capital	…	−51.9	−27.6	66.9	35.5	−19.9	5.0
	Change in revenue		166.4	−69.0	106.2	−15.9	31.9	−5.1	83.4
	Ratio change in expenditure/revenue		0.6	0.1	0.1	−0.9	0.8	−0.2	0.1
Saudi Arabia
	Change in expenditure		223.9	−40.4	47.9	−17.9	15.7	1.1	23.9
		Current	…	−7.7	90.6	−26.2	16.4	−0.1	21.7
		Capital	…	−70.4	−19.5	−18.0	12.0	−29.8	43.8
	Change in revenue		261.4	−77.0	88.6	−14.7	38.7	−21.0	64.4
	Ratio change in expenditure/revenue		0.9	0.5	0.5	1.2	0.4	0.2	0.4
United Arab Emirates 2/
	Change in expenditure		788.1	−9.0	49.8	6.3	40.2	−3.2	13.8
		Current	…	21.8	72.9	−3.1	61.2	−11.9	33.4
		Capital	…	−19.6	−5.6	103.5	−12.6	36.3	−19.7
	Change in revenue 3/		821.5	−59.2	101.7	36.0	18.1	−2.2	54.5
	Ratio change in expenditure/revenue		1.0	0.2	0.5	0.2	2.2	1.5	0.3
Memorandum item:
	Change in average crude oil prices 4/		234.8	−61.4	62.2	−30.6	27.7	−35.8	116.0

Source: Authors’ calculations based on national statistics.

^1/Total expenditure includes net lending, and revenue, grants and investment income.

^2/Consolidated fiscal accounts (federal plus the three largest emirate governments).

^3/Includes investment income from 1991 onwards.

^4/Average nominal price of UK Brent, Dubai, and West Texas Intermediate.

View Table

Table 1.

GCC Countries: Change in Total Government Spending and Revenue in Selected Periods, 1975-2000 ^1/

(In percent)

			1975-80	1980-86	1986-90	1990-94	1994-96	1996-98	1998-2000
Bahrain
	Change in expenditure		201.4	35.5	23.1	−8.8	8.8	3.6	16.2
		Current	…	85.0	17.2	27.4	−1.0	6.9	18.5
		Capital	…	11.9	−13.7	5.3	−19.1	39.8	−14.8
	Change in revenue		236.1	−4.6	17.1	5.8	20.2	−12.5	89.1
	Ratio change in expenditure/revenue		0.9	−7.6	1.3	−1.5	0.4	−0.3	0.2
Kuwait
	Change in expenditure		108.6	28.0	−3.7	46.3	4.8	−9.6	0.7
		Current	…	37.5	19.2	57.4	7.0	−11.7	5.0
		Capital	…	38.3	−43.3	−3.0	−10.5	8.3	−29.8
	Change in revenue		153.9	−31.6	0.2	−23.2	29.7	13.1	31.8
	Ratio change in expenditure/revenue		0.7	−0.9	−15.8	−2.0	0.2	−0.7	0.0
Oman
	Change in expenditure		107.5	116.4	−2.0	19.4	0.4	−1.4	18.7
		Current	…	112.5	25.3	13.2	3.1	−3.6	18.5
		Capital	…	139.0	−60.1	66.7	−11.6	9.7	10.9
	Change in revenue		201.2	16.2	46.5	−11.5	32.4	−22.1	87.5
	Ratio change in expenditure/revenue		0.5	7.2	0.0	−1.7	0.0	0.1	0.2
Qatar
	Change in expenditure		106.8	−4.7	9.3	15.0	26.9	1.1	8.7
		Current	…	13.9	15.5	6.4	27.1	6.6	9.2
		Capital	…	−51.9	−27.6	66.9	35.5	−19.9	5.0
	Change in revenue		166.4	−69.0	106.2	−15.9	31.9	−5.1	83.4
	Ratio change in expenditure/revenue		0.6	0.1	0.1	−0.9	0.8	−0.2	0.1
Saudi Arabia
	Change in expenditure		223.9	−40.4	47.9	−17.9	15.7	1.1	23.9
		Current	…	−7.7	90.6	−26.2	16.4	−0.1	21.7
		Capital	…	−70.4	−19.5	−18.0	12.0	−29.8	43.8
	Change in revenue		261.4	−77.0	88.6	−14.7	38.7	−21.0	64.4
	Ratio change in expenditure/revenue		0.9	0.5	0.5	1.2	0.4	0.2	0.4
United Arab Emirates 2/
	Change in expenditure		788.1	−9.0	49.8	6.3	40.2	−3.2	13.8
		Current	…	21.8	72.9	−3.1	61.2	−11.9	33.4
		Capital	…	−19.6	−5.6	103.5	−12.6	36.3	−19.7
	Change in revenue 3/		821.5	−59.2	101.7	36.0	18.1	−2.2	54.5
	Ratio change in expenditure/revenue		1.0	0.2	0.5	0.2	2.2	1.5	0.3
Memorandum item:
	Change in average crude oil prices 4/		234.8	−61.4	62.2	−30.6	27.7	−35.8	116.0

Source: Authors’ calculations based on national statistics.

^1/Total expenditure includes net lending, and revenue, grants and investment income.

^2/Consolidated fiscal accounts (federal plus the three largest emirate governments).

^3/Includes investment income from 1991 onwards.

^4/Average nominal price of UK Brent, Dubai, and West Texas Intermediate.

View Full Size

Bahrain: Central Government Fiscal Indicators, 1980–2000

(In millions of Bahraini Dinars)

Citation: IMF Working Papers 2002, 201; 10.5089/9781451874365.001.A001

Sources: National authorities; and IMF staff estimates.1/ Non-oil fiscal balance defined as overall fiscal balance excluding oil revenue.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Kuwait: Central Government Fiscal Indicators, 1980–2000

(In millions of Kuwaiti dinars)

Citation: IMF Working Papers 2002, 201; 10.5089/9781451874365.001.A001

Sources: National authorities; and IMF staff estimates.1/ Non-oil fiscal balance defined as overall fiscal balance excluding oil revenue.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Oman: Central Government Fiscal Indicators, 1980–2000

(In millions of rials Omani)

Citation: IMF Working Papers 2002, 201; 10.5089/9781451874365.001.A001

Sources: National authorities; and IMF staff estimates.1/ Non-oil fiscal balance defined as overall fiscal balance excluding oil revenue.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Qatar: Central Government Fiscal Indicators, 1980–2000

(In millions of Qatari riyals)

Citation: IMF Working Papers 2002, 201; 10.5089/9781451874365.001.A001

Sources: National authorities; and IMF staff estimates.1/ Non-oil fiscal balance defined as overall fiscal balance excluding oil revenue.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Saudi Arabia: Central Government Fiscal Indicators, 1980–2000

(In millions of Saudi riyals)

Citation: IMF Working Papers 2002, 201; 10.5089/9781451874365.001.A001

Sources: National authorities; and IMF staff estimates.1/ Non-oil fiscal balance defined as overall fiscal balance excluding oil revenue.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

United Arab Emirates: Central Government Fiscal Indicators, 1980–2000 ^1/

(In millions of U.A.E. dirhams)

Citation: IMF Working Papers 2002, 201; 10.5089/9781451874365.001.A001

Sources: National authorities; and IMF staff estimates.^1/ Consolidated fiscal accounts of the federal government and the three largest emirates.2/ Non-oil fiscal balance defined as overall fiscal balance excluding oil revenue.

• Download Figure
• Download figure as PowerPoint slide

The existence of large foreign official assets facilitated a relatively low level of adjustment in spending in the period 1980–86, when crude oil prices declined significantly. Concerns for sustaining domestic demand in order to stave off a sharp reduction in non-oil growth has usually militated against significant fiscal adjustment in the face of falling oil prices in GCC countries.³ Spending was only cut by the equivalent to about half the fall in total revenue in Saudi Arabia, 20 percent in the United Arab Emirates, and 10 percent in Qatar. Facilitated by the completion of major infrastructure investments, the cutbacks fell mostly on outlays for projects, while current expenditure rose in all these countries, except in Saudi Arabia. In Bahrain and Kuwait, spending continued to rise across the board. In contrast, in Oman, lower oil revenue was more than offset by higher investment income and fees and charges, leading to a further increase in expenditure in the period.

The boost in expenditure towards the end of the 1980s, particularly in Saudi Arabia and the United Arab Emirates, mainly reflected the emergency and higher external assistance requirements created by the 1990/91 regional conflict. However, this increase mainly resulted from higher current spending, since capital expenditure was cut further across the GCC area, particularly in Oman. Following the end of the conflict, oil prices declined and increased attention was devoted to fiscal adjustment, particularly in Saudi Arabia, where both current and capital expenditure was cut, with the overall fiscal deficit in nominal terms returning to the pre-conflict level by the mid-1990s. In the other GCC countries, except Bahrain, spending continued to rise during the first half of the 1990s, even though revenue declined, but the sharp rise in spending in Kuwait reflected the reconstruction period.

The second half of the 1990s witnessed renewed volatility in government expenditure in some GCC countries. The recovery in oil prices during 1994–96 was accompanied by an increase in spending, particularly in Qatar, though this was related to the country’s liquefied natural gas projects, and in the United Arab Emirates. In Oman, spending remained stable despite the increase in revenue. In response to the 1998 oil price collapse, all GCC countries, except Bahrain and Qatar, cut spending—albeit modestly. However, for the first time, in some GCC countries (Kuwait, Oman, and the United Arab Emirates), current spending bore much of the adjustment. Some GCC countries also mobilized non-oil revenue to dampen the decline in oil prices. More recently, most GCC countries made progress in decoupling public expenditure decisions from the evolution of oil revenue. In fact, total spending rose moderately in most of these countries despite the sharp recovery in global oil prices through end-2000.

III. Econometric Methodology and Empirical Results

Our empirical analysis employs annual total real government expenditures and revenues for the GCC countries. The choice is for total revenue since oil accounts for the majority of government revenue in these countries.⁴ The variables are in logarithmic form (let R_t and E_t denote the natural logarithm of real government revenues and government expenditures, respectively), and are converted to real terms using the consumer price index. The cointegration and error-correction modeling approach is well suited to study the causality between government revenue and government expenditure. It is quite conceivable that the levels of revenue and expenditure are cointegrated, namely, they move on a common trend as long as the intertemporal budget constraint is binding over the long run. This important information is incorporated through the error-correction model in examining the causality between the changes in government expenditure and revenue.

A. Empirical Results

Test results for unit root

Each time series in our sample are first tested for their orders of integration by using Augmented Dickey-Fuller (ADF) test (Table 2). The test shows that R_t and E_t are integrated of order one, or I (1), for all GCC countries in our sample.⁵

Table 2.

ADF Test Results for Unit Roots

Note: * and ** denotes significance at 5 and 1 percent, respectively.

Table 2.

ADF Test Results for Unit Roots

Variables	Bahrain	Oman	Qatar	Kuwait	Saudi Arabia	U.A.E.
Rt	I(1)**	I(1)**	I(1)**	I(1)**	I(1)**	I(1)**
Et	I(1) **	I(1) *	I(1) **	I(1) **	I(1) *	I(1) *

Note: * and ** denotes significance at 5 and 1 percent, respectively.

View Table

Table 2.

ADF Test Results for Unit Roots

Variables	Bahrain	Oman	Qatar	Kuwait	Saudi Arabia	U.A.E.
Rt	I(1)**	I(1)**	I(1)**	I(1)**	I(1)**	I(1)**
Et	I(1) **	I(1) *	I(1) **	I(1) **	I(1) *	I(1) *

Note: * and ** denotes significance at 5 and 1 percent, respectively.

Test results for cointegration

Since the time series of R_t and E_t are found to be integrated of the same order (i.e, order one), a cointegration test can be conducted to determine whether a long-run equilibriating relationship exists between them. Johanson cointegration test is performed, assuming a cointegrating relationship as specified by equation (1):

$\begin{array}{cc}{E}_t+a*R_t+c={\epsilon }_t& \left(1\right)\end{array}$

Cointegration relationships are found for all GCC countries in our sample (Table 3).⁶ The results indicate that, not surprisingly, government expenditure and government revenue are subject to an equilibrating relationship and positively related to each other over the long run.⁷,⁸ However, the degree of their association differed across countries. To maintain the long-run equilibrium, for Bahrain, Oman, and Kuwait, when government revenue increased by 1 percent, government expenditure also increased by about 1 percent during the period under analysis; for the United Arab Emirates and Saudi Arabia, when government revenue increased by 1 percent, government expenditure rose by less than 0.5 percent; for Qatar, when government revenue increased by 1 percent, government expenditure increased by more than 1 percent. It should be noted that these different degrees of association are negatively related to the levels of fiscal deficits for each country, i.e., starting from a position of a higher deficit, when government revenue increases, maintaining long-run equilibrium would require government expenditure to increase by less in percentage term.

Table 3.

Johanson Cointegration Test: Estimated Cointegration Vectors

Note: 1. The cointegration coefficients are normalized on E_t. 2. t-ratios are in parentheses. 3. *(**) indicates the significance level at which the number of cointegration equation(s) (CE) is identified is 5% (1%). 4. The length of the lag is estimated with Schwartz’s criterion and chosen, with an view to balancing between ensuring approximately white-noise errors and allowing for enough degrees of freedom in estimation.

Table 3.

Johanson Cointegration Test: Estimated Cointegration Vectors

Variables	Bahrain	Oman	Qatar	Kuwait	Saudi Arabia	U.A.E.
Et	1	1	1	1	1	1
Rt	−0.90 (-13.08)	−1.02 (-23.59)	−1.36 (-3.89)	−1.24 (-5.36)	−0.42 (-4.73)	−0.42 (-1.79)
Constant	−0.01	−0.004	0.34	0.23	−1.35	−0.95
Likelihood Ratio	20.60	18.80	20.56	27.66	21.14	22.17
Eigenvalue	0.48	0.50	0.45	0.70	0.52	0.39
No. of CE(s)	1**, 2*	1*	1**, 2*	1**, 2**	1**	1**,2*
Dummy	No	No	Yes	Yes	No	No

Note: 1. The cointegration coefficients are normalized on E_t. 2. t-ratios are in parentheses. 3. *(**) indicates the significance level at which the number of cointegration equation(s) (CE) is identified is 5% (1%). 4. The length of the lag is estimated with Schwartz’s criterion and chosen, with an view to balancing between ensuring approximately white-noise errors and allowing for enough degrees of freedom in estimation.

View Table

Table 3.

Johanson Cointegration Test: Estimated Cointegration Vectors

Variables	Bahrain	Oman	Qatar	Kuwait	Saudi Arabia	U.A.E.
Et	1	1	1	1	1	1
Rt	−0.90 (-13.08)	−1.02 (-23.59)	−1.36 (-3.89)	−1.24 (-5.36)	−0.42 (-4.73)	−0.42 (-1.79)
Constant	−0.01	−0.004	0.34	0.23	−1.35	−0.95
Likelihood Ratio	20.60	18.80	20.56	27.66	21.14	22.17
Eigenvalue	0.48	0.50	0.45	0.70	0.52	0.39
No. of CE(s)	1**, 2*	1*	1**, 2*	1**, 2**	1**	1**,2*
Dummy	No	No	Yes	Yes	No	No

Note: 1. The cointegration coefficients are normalized on E_t. 2. t-ratios are in parentheses. 3. *(**) indicates the significance level at which the number of cointegration equation(s) (CE) is identified is 5% (1%). 4. The length of the lag is estimated with Schwartz’s criterion and chosen, with an view to balancing between ensuring approximately white-noise errors and allowing for enough degrees of freedom in estimation.

Correspondingly, during the period under analysis, Saudi Arabia and the United Arab Emirates registered the highest average level of overall fiscal deficits.

Test results for causality

Since the series for all the countries in our sample are found to be cointegrated, the Granger causality test is conducted in the context of an error correction model by estimating a vector auto regressive (VAR) model as follows:

$\begin{array}{cc}\mathrm{\Delta }{E}_t={\vartheta }_0+{\alpha }_0\mathrm{\Delta }{E}_{t-1}+{\beta }_0\mathrm{\Delta }{R}_{t-1}+{\lambda }_0{\mathit{ECT}}_{t-1}+{\mu }_t& \left(2\right)\end{array}$

$\begin{array}{cc}\mathrm{\Delta }{R}_t={\vartheta }_1+{\alpha }_1\mathrm{\Delta }{E}_{t-1}+{\beta }_1\mathrm{\Delta }{R}_{t-1}+{\lambda }_1{\mathit{ECT}}_{t-1}+v_t& \left(3\right)\end{array}$

where, θ, α, β, and λ are the coefficients, and μ and ν are the error terms. ECT_t is equivalent to ε_t in equation (1), representing the disequilibrium residuals of the cointegration equation identified above.

As discussed in more detail in Appendix I, by examining the t-statistics on β₀ and λ₀ in equation (2) and β₁ and λ₁ in equation (3), one can infer the direction of causality either from revenue to expenditure or from expenditure to revenue, respectively.⁹ For instance, from equation (2), a statistically significant coefficient on either ∆R_t-1 or ECT_t-1 or both, suggests that revenues cause expenditures thereby supporting the spent-revenue hypothesis. Likewise, from equation (3), a statistically significant coefficient on either ∆R_t-1 or ECT_t-1 or both suggests that expenditure causes revenue thereby supporting the revenue-spend hypothesis. If the relevant coffecients from equations (2) and (3) are both statistically significant, then the fiscal synchronization hypothesis is supported.

The results from estimating the error-correction model show a unidirectional causality from revenue to expenditure for Bahrain, the United Arab Emirates, and Oman, supporting the revenue-spend hypothesis (Table 4). A bi-directional causality between revenue and expenditure was found for Qatar, Saudi Arabia, and Kuwait, supporting the fiscal synchronization hypothesis. Nevertheless, for Kuwait and Saudi Arabia, the causality from revenue to expenditure shows higher significance than the reversal direction.

Table 4.

Causality Test Based on Error-Correction Model

Note: t- ratios are in parentheses.

Table 4.

Causality Test Based on Error-Correction Model

	Bahrain		Oman
	∆Et	∆Rt	∆Et	∆Rt
ECTt-1	−0.98 (-3.49)	0.07 (0.19)	−0.39 (-1.70)	0.60 (1.25)
∆Et-1	0.07 (0.32)	−0.18 (-0.69)	0.18 (0.83)	0.34 (0.78)
∆Rt-1	−0.25 (-0.79)	0.42 (1.12)	0.01 (0.04)	0.20 (0.54)
Constant	0.03 (2.06)	0.02 (1.42)	0.02 (2.00)	0.02 (0.84)
Dummy	—	—	—	—
	Qatar		Saudi Arabia
	∆Et	∆Rt	∆Et	∆Rt
ECTt-1	−0.13 (-1.36)	0.38 (2.44)	−0.86 (-4.13)	−0.81 (-2.04)
∆Et-1	−0.33 (-1.63)	−0.25 (-0.76)	−0.19 (-0.72)	−0.70 (-1.37)
∆Rt-1	0.37 (2.54)	0.27 (1.14)	0.13 (0.70)	0.32 (0.89)
Constant	0.01 (0.55)	0.04 (1.87)	0.01 (1.10)	0.02 (0.75)
Dummy	0.06 (1.39)	−0.016 (-2.24)	—	—
	U.A.E.		Kuwait
	∆Et	∆Rt	∆Et	∆Rt
ECTt-1	−0.40 (-2.86)	−0.24 (-1.13)	−0.42 (-6.22)	0.48 (3.39)
∆Et-1	−0.09 (-0.31)	0.38 (0.87)	−0.09 (-0.23)	0.24 (0.30)
∆Rt-1	0.22 (0.94)	−0.13 (-0.36)	−0.11 (-0.91)	0.20 (0.78)
Constant	0.035 (1.96)	0.03 (0.10)	0.001 (0.06)	0.05 (1.71)
Dummy	—	—	0.22 (4.96)	−0.29 (-3.12)

Note: t- ratios are in parentheses.

View Table

Table 4.

Causality Test Based on Error-Correction Model

	Bahrain		Oman
	∆Et	∆Rt	∆Et	∆Rt
ECTt-1	−0.98 (-3.49)	0.07 (0.19)	−0.39 (-1.70)	0.60 (1.25)
∆Et-1	0.07 (0.32)	−0.18 (-0.69)	0.18 (0.83)	0.34 (0.78)
∆Rt-1	−0.25 (-0.79)	0.42 (1.12)	0.01 (0.04)	0.20 (0.54)
Constant	0.03 (2.06)	0.02 (1.42)	0.02 (2.00)	0.02 (0.84)
Dummy	—	—	—	—
	Qatar		Saudi Arabia
	∆Et	∆Rt	∆Et	∆Rt
ECTt-1	−0.13 (-1.36)	0.38 (2.44)	−0.86 (-4.13)	−0.81 (-2.04)
∆Et-1	−0.33 (-1.63)	−0.25 (-0.76)	−0.19 (-0.72)	−0.70 (-1.37)
∆Rt-1	0.37 (2.54)	0.27 (1.14)	0.13 (0.70)	0.32 (0.89)
Constant	0.01 (0.55)	0.04 (1.87)	0.01 (1.10)	0.02 (0.75)
Dummy	0.06 (1.39)	−0.016 (-2.24)	—	—
	U.A.E.		Kuwait
	∆Et	∆Rt	∆Et	∆Rt
ECTt-1	−0.40 (-2.86)	−0.24 (-1.13)	−0.42 (-6.22)	0.48 (3.39)
∆Et-1	−0.09 (-0.31)	0.38 (0.87)	−0.09 (-0.23)	0.24 (0.30)
∆Rt-1	0.22 (0.94)	−0.13 (-0.36)	−0.11 (-0.91)	0.20 (0.78)
Constant	0.035 (1.96)	0.03 (0.10)	0.001 (0.06)	0.05 (1.71)
Dummy	—	—	0.22 (4.96)	−0.29 (-3.12)

Note: t- ratios are in parentheses.

Moreover, the estimated parameter on λ gives an indication of how fast, in the short run, expenditure growth adjusts itself toward the long-run equilibrium level. For Oman, the United Arab Emirates, and Kuwait, the adjustments were made at a speed of around 2.5 years (a coefficient of around 0.4 on annual data). For Bahrain and Saudi Arabia, the adjustments were close to one year, while Qatar’s adjustment was very slow—more than 7 years. These estimated speeds of adjustment are in general slower than those estimated for other countries in the literature. This, however, could be partly explained by the relatively sizable foreign assets held by the GCC governments, which serve as a cushion and allow the government to adjust at a relatively slow pace.¹⁰

Given that a bi-directional causality was initially observed for Kuwait, Qatar, and Saudi Arabia, variance decomposition was calculated to further determine which direction of causality is the strongest.^11,12 The results show that the causality from revenue to expenditure tends to be stronger (Table 5). In Saudi Arabia, over 80 percent of the variation of expenditures was accounted for by past revenues, while 20 percent was accounted for by past expenditures. This pattern seems to be largely consistent throughout the whole period examined, except for the first period, when the corresponding ratio was 64 percent and 36 percent. For Kuwait and Qatar, while innovations in expenditures account for most of the variations in future expenditures in the first or second periods, the reverse is consistently the case in periods beyond. That is, variation in revenues is of increasing importance in explaining future variation in expenditures. Therefore, the results from variance decomposition analysis together with the test for causality tend to support the conclusion that the revenue-spend hypothesis has prevailed in all GCC countries since the mid-1970s.

Table 5.

Proportion in Percent of Forecast Error Variance K-Periods Ahead Produced by Each Innovation

Table 5.

Proportion in Percent of Forecast Error Variance K-Periods Ahead Produced by Each Innovation

Error in:	No. of periods	Kuwait		Qatar		Saudi Arabia
		Innovation in:		Innovation in:		Innovation in:
		Et	Rt	Et	Rt	Et	Rt
Et	1	98.4	1.6	71.6	28.3	35.7	64.3
	2	65.2	34.8	35.9	64.1	19.7	80.3
	3	46.6	53.4	33.8	66.2	17.8	82.2
	4	39.9	60.1	34.0	66.0	18.1	81.9
	5	36.3	63.7	34.4	65.6	18.3	81.7
Rt	1	0.0	100.0	0.0	100.0	0.0	100.0
	2	8.2	91.8	0.3	99.7	9.6	90.4
	3	11.0	89.0	2.7	97.3	14.6	85.4
	4	12.7	87.3	5.6	94.4	16.3	83.7
	5	14.1	85.9	8.3	91.7	17.0	83.0

View Table

Table 5.

Proportion in Percent of Forecast Error Variance K-Periods Ahead Produced by Each Innovation

Error in:	No. of periods	Kuwait		Qatar		Saudi Arabia
		Innovation in:		Innovation in:		Innovation in:
		Et	Rt	Et	Rt	Et	Rt
Et	1	98.4	1.6	71.6	28.3	35.7	64.3
	2	65.2	34.8	35.9	64.1	19.7	80.3
	3	46.6	53.4	33.8	66.2	17.8	82.2
	4	39.9	60.1	34.0	66.0	18.1	81.9
	5	36.3	63.7	34.4	65.6	18.3	81.7
Rt	1	0.0	100.0	0.0	100.0	0.0	100.0
	2	8.2	91.8	0.3	99.7	9.6	90.4
	3	11.0	89.0	2.7	97.3	14.6	85.4
	4	12.7	87.3	5.6	94.4	16.3	83.7
	5	14.1	85.9	8.3	91.7	17.0	83.0

B. Impulse Response Analysis

Taking the causality from revenue to expenditure as given, we then examine how a shock to revenue is transmitted to expenditure by examining the impulse responses. Impulse responses give the dynamic response of each variable to innovations of this variable as well as of the other variables included in the VAR system.¹³

Figures 7 and 8 depict intertemporal and accumulated responses from a one-standard deviation shock on revenue to expenditure over time, respectively.¹⁴ As shown in Figure 7, increases in revenue lead to increases in expenditure in the first period in all the GCC countries. However, the patterns differ across countries thereafter. For Oman and Kuwait, expenditure growth rates remain positive throughout the period but decelerate gradually until reaching zero. For Saudi Arabia, the United Arab Emirates, and Qatar, expenditure growth turns negative in the second period and the rate of decline gradually slows down to zero. For Bahrain, expenditure growth does not turn negative until the fourth period, after when the rate of decline gradually slows down to zero. Figure 8 shows the corresponding cumulated responses. For Oman and Kuwait, expenditure level will rise gradually and monotonically before reaching the equilibrium; for Saudi Arabia, the United Arab Emirates, and Qatar, expenditure level will rise sharply and overshoot the equilibrium within two periods following the shock, after which it will go down gradually and reach the equilibrium. Bahrain displays a similar pattern, but it takes four years for expenditure level to rise before going down somewhat and reaching the equilibrium. It is not clear the reasons behind the smoother reaction of expenditure to revenue changes observed in Kuwait and Oman. However, it is important to note that these two countries generally articulate their fiscal priorities in a medium term framework, and have adopted formal oil savings/stabilization funds for more than two decades.¹⁵

View Full Size

GCC Countries: Intertemporal Response of Government Expenditure to Cholesky One Standard Deviation Innovation in Government Revenue

Citation: IMF Working Papers 2002, 201; 10.5089/9781451874365.001.A001

Source: Authors’ estimates.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

GCC Countries: Accumulated Responses of Government Expenditure to Cholesky One Standard Deviation Innovation in Government Revenue

Citation: IMF Working Papers 2002, 201; 10.5089/9781451874365.001.A001

Source: Authors’ estimates.

• Download Figure
• Download figure as PowerPoint slide

IV. Conclusion and Policy Implications

Not surprisingly, our results—based on a cointegration and error-correction model and a variance decomposition analysis—support the revenue-spend hypothesis for all GCC countries. Thus, government spending follows revenue, suggesting a pro-cyclical expenditure policy to variations in oil revenue—the largest budgetary revenue component. In this context, GCC countries could enhance the effectiveness of fiscal policy by making budget expenditure less driven by revenue availability. This would avoid the costs and instability that variations in public spending generate, compounding the boost-bust economic cycle associated with oil price shocks. The authorities could resort to a medium-term expenditure framework, so that expenditures can be planned and insulated from volatile short-term revenue availability. Fiscal rules could be also adopted to constrain expenditure, the non-oil fiscal imbalance, and/or borrowing by the government. As a complement, the GCC authorities could adopt formal oil stabilization funds—as some have already done—to mitigate the pressure on government to overspend in periods of rising revenue by channeling a significant portion of the increase in oil revenue away from the budget. Additionally, GCC countries could also strengthen their budget preparation and execution process.