The Causal Relationship Between Government Revenue and Spending

260

JOURNAL ECONOMICSAND FINANCE 9 Volume28 9 Number 2 9 Summer 2004 OF

THE CAUSAL RELATION BETWEEN GOVERNMENT REVENUE AND SPENDING: EVIDENCE FROM EGYPT AND JORDAN
By Bassam AbuAI-Foul and Hamid Baghestani* Abstract In investigating the causal relation between government revenue and spending, our empirical results support the tax-and-spend hypothesis for Egypt and the fiscal synchronization hypothesis for Jordan. Breaking away from these historical trends is essential for both countries to eliminate the budget defa:it and therefore ensure the availability of domestic saving for private investment. To cope with unemployment and poverty, continuing privatization is recommended for both countries to improve productivity and efficiency in the domestic economy. Privatization should lead to higher domestic saving and investment and at the same time eliminate the budget deficit by enhancing revenue and curbing spending. (JEL H62, 1"163)

Introduction
The purpose of this paper is to investigate the causal relation between government revenue and spending for Egypt and Jordan. Given their fiscal circumstances, our investigation should help determine proper reforms for these countries to cope more effectively with their current economic challenges. A crucial challenge facing Egypt is unemployment. According to current official estimates, unemployment of about 8 percent is accompanied with an annual growth rate of 3 percent in labor force. To reduce unemployment to more manageable levels, it is estimated that Egypt needs to achieve a healthy and sustainable annual growth rate of at least 6 percent in real GDP. l In an attempt to reach this goal, Egypt has utilized a private-sector-led growth policy. Privatization and transition to a market economy are intended to improve productivity, efficiency, and competition in the domestic economy. However, the low levels of domestic saving and investment create an impediment for economic growth in Egypt. Improvements in the domestic saving rate stem from improving productivity, which, in turn, makes privatization an important factor in reducing unemployment and poverty. Promoting a more efficient pension system, restructuring the financial system, and further developing capital markets have provided additional ways to increase the domestic saving rate. Eliminating the budget deficit is a further step to ensure the availability of domestic saving for private investment. To this end, it is essential for the government to implement policies that reduce and eventually eliminate the budget deficit. It is therefore the aim of this paper to offer such policy actions, using the evidence on the causal relation between government revenue and spending in Egypt. Unemployment is also a crucial challenge facing Jordan. According to current official estimates, unemployment of about 15 percent is accompanied with an annual growth rate of 4-5 percent in the labor force. It is estimated that Jordan also needs a healthy and sustainable annual growth rate of at

* Bassam AbuAI-Foul,Departmentof Economics and Public Administration,American Universityof Sharjah, Sharjah, United Arab Emirates, [e-mail];Hamid Baghesteni, Department of Economics and Public Adminislration,American Universityof Sharjah, Sharjah, United Arab Emirates, [e-mail] authors gratefullyacknowledgethe commentsof an anonymousrefereeon an earlierdraftof this paper. l See the WorldBankGroup(2001).

JOURNAL O F ECONOMICS AND F I N A N C E * Volume 28 * Number 2 * Summer 2004

261

least 6 percent in real GDP to stabilize unemployment.2 To encourage economic progress, Jordan has focused on a private-sector export-oriented growth strategy. The government has aggressively pursued privatization of most public enterprises in transportation, electricity, water, and telecommunications. In addition to increasing efficiency, productivity, and competitiveness of privatized companies, the aim has been to encourage domestic saving and stimulate private investment. In terms of exportoriented growth, Jordan has established several free zones, including the Aqaba port along the Red Sea, Zarqa, the Sahab industrial estate, and lrbid. Private sector participation is encouraged through investment tax incentives. Licenses to operate within a free zone area are given to private companies if they have the potential of bringing new industries and technology to the country, utilizing local raw materials and components in the process of production, improving the Jordanian labor skills, and lowering the country's imports. 3 Despite these structural reforms, a healthy and sustainable growth in real GDP has not yet been realized. In addition to the lack of concrete export competitiveness, barriers to faster growth include the low levels of domestic saving and slow response of private investment.4 Besides privatization, other efforts to increase the saving rate include further development of the pension system, the financial system, and the capital markets. As in the case of Egypt, eliminating the budget deficit in Jordan is essential to ensure the availability of domestic saving for private investment. Providing evidence of the causal relation between government revenue and spending should thus help determine ways to reduce and eventually eliminate the budget deficit in Jordan. The rest of the paper is organized as follows. The next section theoretically discusses four hypotheses of government finance: (i) the tax-and-spend hypothesis, (ii) the spend-and-tax hypothesis, (iii) the hypothesis of causally independent tax and spending decisions, and (iv) the fiscal synchronization hypothesis. Citing some empirical evidence, mostly on developing countries, augments this theoretical discussion. The third section describes the data and the econometric methodology. Such tests as a unit root and cointegration are necessary to identify the appropriate bivarlate model for investigating the directions of causation between revenue and spending. However, our sample periods are not sufficiently long to get any power for these tests. To overcome this problem, the causality tests are performed using three bivariate models. These are (i) the vector autoregressive model in levels, (ii) the vector autoregressive model in first differences, and (iii) the error-correction model. The causality test results for Egypt and Jordan, presented in the fourth section, are not sensitive to the choice of the model. Accordingly, in the fifth section, we rely on these test results to discuss the policy implications and conclude the paper.

Theoretical and Empirical Background
Several alternative hypotheses of government finance characterize the causal relation between spending and revenue. The tax-and-spend hypothesis, championed by Friedman (1978), theorizes a causal relation running from revenue to spending. It views spending as adjusting, up or down, to whatever level can be supported by revenue. Control of taxation, according to Friedman (1978), is essential to limiting growth in government. In reducing the budget deficit, for instance, one should not rely on raising taxes, since higher revenue invites higher spending. Like Friedman, Buchanan and Wagner (1977, 1978) advocate the tax-and-spend hypothesis. But they warn that the tax-and-spend prediction may be distorted due to the fact that tax rate changes are accompanied by intense political debate and controversy over economic impact and income distributional issues. Deficit financing

2Seethe WorldBankGroup(2003). J See JordanlnvesanentBeard(2000). ' One should,of course, be mindfulof the fact that the political instabilityof the region is another impedimentto fitstereconomic8rowth,moreso in filecaseof Jordan than Egypt.

262

JOURNAL OF ECONOMICS AND FINANCE 9 Volume 28 9 Number 2 9 Summer 2004

rather than tax financing by politicians may then become the source of growth in spending. Empirical evidence in support of the tax-and-spend view is presented by Baffes and Shah (1994) for Brazil, by Dan'at (1998) for Turkey, by Darrat (2002) for Lebanon and Tunisia, by Cheng (1999) for Columbia, the Dominican Republic, Honduras, and Paraguay, and by Ewing and Payne (1998) for Colombia, Ecuador, and Guatemala. The spend-and-tax hypothesis relies on the reverse relation, with revenue responding to prior spending changes. In line with the Ricardian equivalence theorem, Barro (1974) maintains that the public fully anticipates and capitalizes the future tax liability implied by present government borrowing. Thus, in the absence of fiscal illusion, increases in government spending lead to increases in taxes. Peacock and Wiseman (1979) see natural, economic, or political crises as justifications for spending hikes that are subsequently approved by tax increases.According to this hypothesis, spending cuts are the desired solution to reducing the budget deficit, especially in the absence of crises. Empirical evidence by Mithani and Khoon (1999) and Ram (1988) supports the spend-and-tax hypothesis, respectively, for Malaysia and Honduras. The third hypothesis emphasizes the institutional separation of allocation and taxation functions of government and the independent determination of revenue and spending. With respect to the U.S., this hypothesis emphasizes the absence of coordination between spending and revenue decisions due to the lack of agreement between the executive and legislative branches of government participating in the budgetary process. [See Wildavsky (1988) and Hoover a~d Sheffrin (1992).] Consistent with this view, Baghestani and McNown (1994) conclude that neither the tax-and-spend nor the spend-and-tax hypothesis accounts for post-World War II budgetary expansion in the U.S. Instead, they show that both the expansion in revenue and spending is determined by iong-rnn economic growth. With respect to developing countries, Ram (1988) provides empirical evidence in support of the institutional separation hypothesis for India, Panama, Paraguay, and Sri Lanka. The fourth hypothesis indicates bidirectional causation between revenue and spending. [See Musgrave (1966) and Meitzer and Richard (1981).] This f ~ t l synchronization hypothesis postulates that the revenue and spending decisions are made simultaneously, by analyzing costs and benefits of alternative government programs. Therefore, this view precludes unidirectional causation from revenue to spending or from spending to revenue. Empirical evidence in support of the fiscal synchronization hypothesis is presented by Baffes and Shah (1994) for Argentina and Mexico, by Cheng (1999) for Chile, Panama, Brazil, and Peru, by Ewing and Payne (1998) for Chile and Paraguay, by Kimenyi (1990) for Kenya, and by Li (2001) for China. For a comprehensive survey of the empirical evidence on the tax-spend debate for both developed and developing countries, see Payne (2003).

Data and Methodology
This study utilizes the annual data on government spending, government revenue, and Gross Domestic Product (GDP). These data for Egypt (1977-1998) and Jordan (1975-2001) are obtained from the International Financial Statistics tapes. Following Bohn (1991), among others, government revenue and spending are expressed as a ratio of GDP. While controlling for GDP, this treaunent alleviates the question of whether the revenue and spending variables should be in nominal or real terms. Throughout this study, Rt denotes the logarithm of the government revenue/GDP ratio, and X~ denotes the logarithm of the government spending/GDP ratio. The transformation of the series to logarithms is intended to eliminate the problem of heteroskedasticity. Granger's (1969, 1980) notion of causality is based on a temporal ordering and incremental predictability criteria. That is, Xt is said to Granger-cause RI if the past history of Xt contains information useful for predicting R,. To formulate the appropriate model for investigating the causal relationship between R, and X,, however, one needs to determine the stochastic properties of

JOURNAL OF ECONOMICS AND F I N A N C E , Volume 28 * Number 2 9 Summer 2004

263

the individual time series. In case each variable is integrated of order zero, R, - I(0) and X, - I(0), the appropriate model is vector autoregressive in levels (VAR-L) as given by (1).

Rt =a,o + ~ : + b+,R,., + X+:I coX,-+ + E+,

(1)
Xt = dlo + ~,i:1 el, Xt., + z~,/:lflj Rt-j + wl,.

In case each variable is integrated of order one, R, ~ I(1) and Xt ~ I(1), and the variables are not cointegrated, the appropriate model is vector autoregressive in first differences (VAR-D) as given by (2).

ARt = a:o + Z,,:+ b2, AR,., + Z;:+ c~+ AX,.j + ~2t

(2)
AXt = d2o + ~:+ e~, AXI.~ + Z,j:tfzj ARt.j + w2,.

In case each variable is integrated of order one, R, ~ I(1) and X, ~ I(1), and the variables are cointegrated, the appropriate model is an error correction model (ECM). For the purpose of our study, however, we utilize the following unrestricted ECM as given by (3). [See Engle and Granger (1987), pp. 271-72.]
AR, = a~o + ~=+ b~,ARt.+ + ~=+ c3/ AX,.+ + 2H R,.+ + X+2X,.+ + ~3,

(3) aX, = d3o + s e~, aX,., + X+:+f~+ ~XR,.++ ~2+ R,-+ + ~2:X,.+ + w3,

The difference between the VAR-D in (2) and the ECM in (3) is the inclusion ofR,.I andX,.l. These series, of course, are from the long-run equilibrium relation in the ECM. The VAR-L in (1), as noted by Baghestani and McNown (1992, p. 130), is not inconsistent with the long-run equilibrium relation but requires restrictions across the two equations. 5 In order to determine the appropriate model, it is necessary to conduct unit root and cointegration tests. However, as also indicated by Hakkio and Rush (1991), a very long sample is required to get any power for such tests. The lack of a long sample for Egypt and Jordan, therefore, prevents us from conducting the tests of a unit root and cointegration. To overcome this problem, all three bivariate models in (1)-(3) are employed in this study to investigate the causal relation between Rt and X,. A common practice in causality testing is to specify a common lag length on RI and X, in the VARs and ECM. Lee (1997) warns against this practice due to inherent misspecifications. Therefore, in line with Cheng (1999), among others, we utilize Hsiao's (1981) version of the Granger causality test. Specifically, the optimal lag structure in the VARs and the ECM in (1)-(3) is determined according to the criterion of minimum final prediction error (FPE), as presented for the bivariate case by Hsiao (1981). The FPE search in this study allowed a maximum lag of four years. Extending the maximum lag to six years, however, did not alter the optimal lag structure for the three models. Before proceeding further, it is appropriate to describe Hsiao's two-step procedure of the Granger causality test employed here. In doing so, let's concentrate on the revenue equation of the VAR in levels in (1), which is formulated to test the null hypothesis that X, does not cause RI. The

5It is importantto notethat the formulationsof the three bivariatemodelsin (!)-(3) are basedon the assumptionthat the two series,RtandXhare botheitherl(0)or I(1).

264

JOURNAL OF ECONOMICS AND FINANCE 9 Volume 28 9 Number 2 9 Summer 2004

first step is to specify a univariate model o f the dependent variable R,, with varying order of lags from 1 to M, where M is the maximum order of lags. Estimating this univariate model at various lags allows us to calculate the corresponding FPEs, as defined by Akaike (1969) and given by (4), where T is the total number of observations, m is the order of lags varying from 1 to M, and SSE is the sum of squared errors. FPE(m) - (T + m + 1) (T-m-l)
SSE(m)

(4)

T

The smallest FPE, denoted FPE(m*), determines the optimal order o f lags, m*, for the univariate model o f Rt. The second step concentrates on a bivariate model of Rr, which includes the lagged variables on both R, and X~. Here, R,., is treated as the controlled variable and Xt.y as the manipulated variable. That is, we control for the lagged dependent variables, Rt-I..... R,.,. and then estimate the bivariate model with varying order of lags from 1 to N, where N is the maximum order of lags on X,. The corresponding FPEs are then calculated using (5), where n is the order of lags on X, varying from 1 to N. (T + m * +n + 1) SSE(m*, n) FPE(m*, n) = ( T - m * - n - 1 ) " T

(5)

The smallest FPE, denoted FPE(m* n*), determines the optimal order o f lags, n*, for the bivariate model of R,. The null hypothesis that X, does not cause R, is then rejected if FPE(m*, n*) < FPE(m*). Repeating this two-step procedure, we will test the null hypothesis that Rt does not cause Xr, using the spending equation in (1). In the same manner, the causal relation between R~ and Xr is tested using the VAR-D in (2). Based on the revenue equation in (2), when testing the null hypothesis that X~ does not cause R~, the controlled variable is ARt.~, and the manipulated variable is AXt.j. Based on the spending equation in (2), when testing the null hypothesis that Rt does not cause X,, the controlled variable is AXe.a,and the manipulated variable is AR~j. Again, the same procedure is employed to test the causal relation between R~ and Xt using the ECM in (3). Based on the revenue equation in (3), when testing the null hypothesis that X~ does not cause Rt, the controlled variables are ARf_~,R~-h and Xvl, and the manipulated variable is AXt.j. Based on the spending equation in (3), when testing the null hypothesis that Rt does not cause Xt, the controlled variables are AXt.~,Rt-i, and X~.I, and the manipulated variable is AR~.j.

Empirical ResuRs
Table 1 reports the causality test results for Egypt for 1977-1998. The estimates from the revenue equation o f the VAR-L, FPE(2,1) = 0.0131 > FPE(2) = 0.0122, support the null hypothesis that X~ does not cause RI. The same finding is reached with the estimates from the revenue equation of the VAR-D, as FPE(1,1) = 0.0141 > FPE(1) = 0.0128. The conclusion that spending does not cause revenue is also supported by the revenue equation estimates of the ECM, as FPE(I,I) -- 0.0147 > FPE(1) = 0.0123. 4

e Due to the additional controlledvariables in the ECM, the calculationsofth9 FPE statistics using (4) and (5) take into accountthe change in the degreesof freedom.

JOURNAL OF ECONOMICS AND FINANCE 9 Volume 28 9 Number 2 9 Summer 2004

265

Table 1: Causality Test Results for Egypt (1977-1998) Dependent Variable Controlled Variable Additional Controlled Variable Manipulated Variable Causality Inference

FPE

Vector autoregressive model in levels (VAR-L)

R R
X X

R 0=2) R 0--2)
X(i=l) X(i=l)

X(j=I)
R0=I)

0.0122 0.0131
0.0151 0.0131

X~-> R

R~X

Vector autoregressive model in first differences (VAR-D)

AR AR

AR (i=l) ~u~ (i=l)

AX(j=I)

0.0128 0.0141

X;e>R

AX AX

AX(i=I) AX(i=I)

AR (j=l)

0.0155 0.0134

R=::~X

Error-correction model (ECM)

AR AR

AR (i=l) AR (i=l)

R (-1) R (-I), X(-1)

AX(j=I)

0.0123 0.0147

X;~>R

AX S~"

AX(i=I) ~'(i=l)

X(-1) ~-1),R(-1)

~R (j=l)

0.0169 0.0153

R=~X

Notes: R denotes the logarithm of the governmentrevenue/GDPratio, and X denotes the logarithm of the government spending/GDPratio. Final predictionerror (FPE) statistics are calculated using the formulas in (4) and (5). Numbers in parenthesesfor i and j are the optimallag length for the respectivevariables.

The estimates from the spending equation o f the VAR-L reject the null hypothesis that Rt does not cause X,, since FPE(1,1) = 0.0131 < FPE(1) = 0.0151. This f'mding is consistent with the spending equation estimates of the VAR-D, as FPE(I,1) = 0.0134 < FPE(I) = 0.0155. The finding that revenue causes spending is also supported by the spending equation estimates o f the ECM, as FPE(1,1) = 0.0153 < FPE(1) = 0.0169. As seen, our test results for Egypt, indicating a unidirectional causation from revenue to spending, are not sensitive to the choice o f the bivariate model. A closer inspection of the parameter estimates o f the bivariate models (not reported here) reveals a positive causal effect o f taxes on spending. This, in line with Friedman (1978), implies that raising taxes in Egypt invites higher spending and therefore does not produce the desired outcome o f reducing the budget deficit. Table 2 reports the causality test results for Jordan for 1975-2001. The estimates from the revenue equation o f the VAR-L, FPE(2,1) = 0.0067 < FPE(2) = 0.0086, reject the null hypothesis

266

JOURNAL OF ECONOMICS AND FINANCE 9 Volume 28 9 Number 2 9 Summer 2004

that Xt does not cause RI. This finding is consistent with the estimates from the revenue equation of the VAR-D, as FPE(1,1) = 0.0074 < FPE(I) = 0.0085. The conclusion that spending causes revenue in Jordan is also supported by the revenue equation estimates of the ECM, as FPE(1,1) = 0.0070 < FPE(I) = 0.0086.

Table 2: Causality Test Results for Jordan (1975-2001)
Dependent Variable Controlled Variable Additional Controlled Variable Manipulated Variable Causality Inference

FPE

Vector autoregressive model in levels (VAR-L)

R
R X X

R (i=2) R (i=2)
X(i:l) X (i:l)

X(j=I)

0.0086 0.0067 0.0063 0.0047

X:=~ R

R (j:2)

R =:~ X

Vector autoregressi~e model in first differences (VAR-D)

AR AR AX AX

AR (i=l) AR (i=l) AX(i=l) AX(i=I)

AX(j=I)

0.0085 0.0074 0.0069 0.0055

X::~ R

AR (j=l)
Error-correction model (EMC)

R:=~X

AR AR

AR (i=I) AR (i=l)

R (-l)
R (-1), X(-1) X(-1) X(-l), R (-1) AX(j=I)

0.0086 0.0070

X=~R

AX AX

AX(i=l) AX(i=I)

AR (j=l)

0.0067 0.0051

R:=~X

Note: See the notes in tale ! for definitions.

The estimates from the spending equation o f the VARoL reject the null hypothesis that R, does not cause Xt, since FPE(I,2) = 0.0047 < FPE(I) = 0.0063. The finding is consistent with the spending equation estimates of the VAR-D, since FPE(1,1) = 0.0055 < FPE(I) = 0.0069. The conclusion that revenue causes spending in Jordan is also supported by the spending equation estimates of the ECM, as FPE(1,1) = 0.0051 < FPE(1) = 0.0067. As seen, our test results for Jordan, indicating bidirectional causation between revenue and spending, are not sensitive to the choice o f the bivariate model. A closer inspection of the parameter estimates o f the bivariate models (not reported here) reveals a positive causal effect o f

JOURNAL O F ECONOMICS AND FINANCE 9 Volume 28 * Number 2 9 S u m m e r 2004

267

taxes on spending as well as a p o s i t i v e causal effect of spending on taxes. These results imply that the revenue and spending decisions are intertwined in the same direction.7

Discussion and Conclusion
With young populations in both countries, a crucial challenge facing Egypt and Jordan is unemployment. To ease this problem, it is essential for them to achieve a healthy and sustainable growth in real GDP. However, an impediment for economic growth is the low level of domestic saving and investment. To increase the domestic saving rate, both Egypt and Jordan have been aggressively implementing policies toward privatization as well as promoting a more efficient pension system, restructuring the financial system, and further developing capital markets. To ensure the availability of domestic saving for private investment, it is also essential for the government to implement policies that reduce the budget deficit. Our empirical f'mdings for Egypt indicate a unidirectional causation from revenue to spending, with higher revenue leading to higher spending. This, consistent with Friedman (1978), suggests that raising taxes does not produce the desired outcome of reducing the budget deficit. Reducing the budget deficit requires a more equitable and transparent tax system. Recent tax reform proposals, which are intended to increase revenue, include a new sales tax, increased custom duties, and increased Suez Canal fees and stamp duties. To break away from the historical trend of tax-and-spend strategy, however, such improvements in tax revenue should accompany reforms in public spending. Consistent with a privatization policy, recent reductions in public employment and investment spending, elimination of government subsidies on most consumer goods, and a cutback in defense spending are seen as steps in the right direction. Our empirical results for Jordan indicate bidirectional causation between revenue and spending. This outcome, consistent with Musgrave (1966) and Meltzer and Richard (1981), suggests that fiscal policymakers in Jordan do not make spending (tax) decisions in isolation from tax (spending) decisions. The joint determination of revenue and spending is appealing as long as it effectively curbs the budget deficit. However, as indicated above, our results imply that the revenue and spending decisions are intertwined in the same direction. This means that higher revenue leads to higher spending and vice versa. Again, in order to break away from this trend, efforts to enhance sources of revenue should be accompanied with reductions in spending. Fiscal structural reforms include (i) the recent Income Tax Law, which is intended to extend the tax base and reduce exemptions, (ii) the transformation from an income tax-based system to an expenditure tax-based system, which is intended to increase the general sales tax from 10 percent to 13 percent, and (iii) instituting stamp and land registration fees.s In curbing spending, in addition to privatizing state-owned companies in such areas as transportation, telecommunication, and energy, Jordan has reduced public employment and investment spending and cut back on subsidies. Efforts toward privatization should help both Egypt and Jordan enhance revenue and curb spending, especially in the long run. Elimination of the budget deficit, while necessary to promote faster economic growth, is important for debt reduction. In fact, part of the increased revenue generated from privatization has been used to pay the interest on the outstanding public debt. Setting a ceiling on the outstanding public debt ratio to GDP, replacing high interest loans with lower interest government bonds, and debt rescheduling from the Paris Club and the World Bank are all steps in the right direction to more effective debt management for both countries.

7 The revenue and spending equations with the lowest FPE pass a series o f diagnostic tests including those for

autocorrclationand hetoroskedasticity. s See CentralBankof Jordan (2001).

268

JOURNAL ECONOMICSAND FINANCE 9 Volume 28 9 Number 2 9 Summer 2004 OF

R~erenc~
Akaike, Hirotugu. 1969. "Fitting Autoregressive Models for Prediction." Annals of the Institute of Statistical Mathematics 21: 243-247. Baffes, John, and Anwar Shah. 1994. "Causality and Comovement between Taxes and Expenditures: Historical Evidence from Argentina, Brazil, and Mexico." Journal of Development Economics 44:311-331. Baghestani, Hamid, and Robert McNown. 1992. "Forecasting the Federal Budget with Time-Series Models." Journal of Forecasting 11: 127-139. Baghestani, Harold, and Robert McNown. 1994. "Do Revenues or Expenditures Respond to Budgetary Disequilibria?" Southern Economic Journal 61: 311-322. Barro, Robert J. 1974. "Are Government Bonds Net Wealth?" Journal of Political Economy 82: 10951117. Bohn, Henning. 1991. "Budget Balance through Revenue or Spending Adjustments? Some Historical Evidence for the United States." Journal of Monetary Economics 27: 333-359. Buchanan, James, and Richard Wagner. 1977. Democracy in DeficiL New York: Academic Press. Buchanan, James, and Richard Wagner. 1978. "Dialogues Concerning Fiscal Religion." Journal of Monetary Economics 4: 627-636. Central Bank of Jordan. 2001. "Annual Report." Online at: http://www.cbj.gov.jo. Cheng, Benjamin S.1999. "Causality between Taxes and Expenditures: Evidence from Latin American Countries." Journal of Economics and Finance 23:184-192. Dan'at, All. 1998. "Tax and Spend, or Spend and Tax? An Inquiry into the Turkish Budgetary Process." Southern Economic Journal 64: 940-956. Darrat, All. 2002. "Budget Balance through Spending cuts or Tax Adjustments?" Contemporary Economic Policy 20:221-233. Engle, Robert F., and Clive W.J. Granger. 1987. "Cointegration and Error Correction: Representation Estimation and Testing." Econametrica 55:251-276. Ewing, Bradley T., and James E. Payne. 1998. "Government Revenue-Expenditure Nexus: Evidence from Latin America." Journal of Economic Development 23: 57-69. Friedman, Milton. 1978. "The Limitations of Tax Limitation." Policy Review: 7-14. Granger, Clive W.J. 1969. "Investigating Causal Relation by Econometric Models and Cross-Spectral Methods." Econometrica 37: 424-438. Granger, Clive W.J. 1980. "Testing for Causality." Journal of Economic Dynamic and Control 4: 16t-194. Hakkio, Craig S., and Mark Rush. 1991. "Cointegration: How Short Is the Long-RunT' Journal of International Money and Finance 10: 571-581. Hoover, Kevin D., and Steven M. Sheffi'in. 1992. "Causation, Spending, and Taxes: Sand in the Sandbox or Tax Collector for the Welfare State?" American Economic Review 82: 225-248. Hsiao, Cheng. 1981. "Autoragressive Modeling and Money Income and Causality Detection." Journal of Monetary Economics 7: 85-106. Jordan Investment Board. 2000. "Investing in Jordan: Free Zones." Online at: www.jordaninvestment.com. Kimenyi, Mwangi S. 1990. "The Causal Relationship between Revenues and Expenditures: A Developing Country Case Study." Journal of Public Finance and Public Choice 8:3-11. Lee, Jim. 1997. "Money, Income and Dynamic Lag Pattern." Southern Economic Journal 64: 97103. Li, Xiaoming. 2001. "Government Revenue, Government Expenditure, and Temporal Causality: Evidence from China." Applied Economics 33: 485-497.

JOURNAL OF ECONOMICS AND FINANCE 9 Volume 28 9 Number 2 9 Summer 2004

269

Meltzer, Allen H., and Scott F. Richard. 1981. "A Rational Theory of the Size of Government." dournal of Political Economy 89:914-927. Mithani, D.M., and Goh Soo Khoon. 1999. "'Causality between Government Expenditure and Revenue in Malaysia: A Seasonal Cointegration Test." ASEAN Economic Bulletin 16: 68-79. Musgrave, Richard. 1966. "Principles of Budget Determination." In Public Finance: Selected Readings, edited by H. Cameron and W. Henderson. Now York: Random House. Payne, James E. 2003. "A Survey of the International Empirical Evidence on the Tax-Spend Debate." Public Finance Review, 31 : 302-324. Peacock, Alan, and Jack Wiseman. 1979. "Approaches to the Analysis of Government Expenditures Growth." Public Finance Quarterly 7: 3-23. Ram, R. 1988. "A Multicountry Perspective on Causality between Government Revenue and Government Expenditure." Public Finance 43:261-269. The World Bank Group. 2001. "Egypt in Brief." Online at: http://www.worldbank.org (October). The World Bank Group. 2003. "Hashemite Kingdom of Jordan." Jordan Quarterly Publication First Quarter: 7-14. Wildavsky, Aaron. 1988. The New Politics of the Budgetary Process. Gienview, IL: Scott
Foresman.