THE CAUSALITY BETWEEN GOVERNMENT REVENUES AND EXPENDITURES IN BOTSWANA

Transcription

1 International Journal of Economic Issues, Vol. 8, No. 1 (January-June, 2015) : 1-14 International Science Press THE CAUSALITY BETWEEN GOVERNMENT REVENUES AND EXPENDITURES IN BOTSWANA CHRISTOPHER MUPIMPILA *, LESEGO DINALE * AND BOITUMELO MOFFAT * 1 Department of Economics, University of Botswana, Private Bag 0022, Gaborone, Botswana mupimpi@mopipi.ub.bw (Corresponding author), moffatb@mopipi.ub.bw The direction of causality between government revenues and expenditures has implications on budget deficits. Empirical evidence on the causality suggests ways of reducing government expenditures and budget deficits. This study uses a vector error correction model (VECM), and Granger causality; and finds support for the tax-andspend hypothesis. In Botswana, there is a negative and unidirectional causality from tax revenues to government expenditures. Therefore, to reduce budget deficits, it is necessary to increase tax revenues by broadening the tax base, improving tax compliance and administration in the country. These measures are pertinent because Botswana now faces persistent budget deficits, unlike in the past budget surpluses due to mineral windfalls. Key words: Government revenues, expenditures, budget deficits, causality, Botswana. JEL classification: C12, H20, H50, H62 1. INTRODUCTION The causality between government revenues and expenditures has been researched extensively. This is because the research is expected to yield information on how to reduce government budget deficits (Keho, 2013; Elyasi and Rahimi, 2012; Parida, 2012; Ho and Huang, 2009). In turn, reducing budget deficits is a priority because the deficits are associated with a wide range of problems, such as inflation, national indebtedness, and balance of payments crises (Easterly et. al., 1994). This paper analyzes the causality between government revenues and expenditures in Botswana. Since 2009 to-date, Botswana faces persistent government budget deficits, a situation which is a reversal of past experience when a boom in mining led to nearly two decades of budget surpluses. The boom was driven by diamond production and exports, Botswana s main foreign exchange earner. However, the 2008 world recession resulted in a slump in diamond exports, and successive budget deficits since In the efforts to rebalance the budget in Botswana, it is necessary to know the causality between government revenues and expenditures. Moalusi (2004) and Wolde-Rufael (2008) examined the causality between government revenues and expenditures in Botswana. However, we employ quarterly

2 2 / CHRISTOPHER MUPIMPILA, LESEGO DINALE & BOITUMELO MOFFAT data for the period 1995 to 2010, unlike Moalusi who used annual data for 1976 to We also focus on a country-specific study, whereas Wolde-Rufael studied 13 African countries, including Botswana, for the period 1971 to Devoting our attention to a country-specific study enables us to generate additional insights than otherwise would be the case. The rest of the paper is organized as follows: in section 2, we provide the review of the literature on the causality of government revenues and expenditures. We describe the data and model specification in section 3. In section 4, we analyze the empirical results, while in section 5 we present a summary and conclusions. 2. REVIEW OF LITERATURE Four hypotheses have been advanced on the causality between government revenues and expenditures. First, there is the tax-and-spend hypothesis, which holds that the direction of budget determination is from government revenues to expenditures. The view is that government revenues determine government expenditures, and as a result, higher levels of government revenues lead to increased government spending. This hypothesis was advocated by Friedman (1978), and Buchanan and Wagner (1978). However, the authors had different perspectives on the causality. According to Friedman, the causality is positive: higher levels of government revenues lead to increase in government spending. Therefore, reducing tax revenues can reduce budget deficits. For Buchanan and Wagner, the causality is negative. This is because a reduction in taxes leads to fiscal illusion whereby there is a perception, by the general public, that government programmes cost less than was previously the case. Consequently, the public increase their demand for government programmes and spending, resulting in higher deficits. Since raising taxes avoids a fiscal illusion, it is therefore necessary to increase taxes in order to reduce budget deficits. Empirically, a test of the tax-and-spend hypothesis should show a unidirectional causality from government revenues to expenditures. Studies on Africa which support this hypothesis include Moalusi (2004) for Botswana, Eita and Mbazima (2008) for Namibia, Keho (2010, 2013) for the Ivory Coast, Amoah and Loloh (2008) for Ghana in the short-run, and Wolde-Rufael (2008) for Ethiopia, Ghana, Kenya, Nigeria, Mali and Zambia. Other studies elsewhere that support this hypothesis are: AbuAl-Foul and Baghestani (2004) for Egypt, Al-Khulaifi (2012) for Qatar, Magazzino (2010) for Italy during the period 1862 to 1913, Garcia (2012) for 15 Spanish regions, Owoye (1995) for Japan and Italy. The tax-and-spend hypothesis is also supported by evidence from the following countries: Garcia and Henin (1999) for the USA, Germany and France; Cheng (1999) for Columbia, the Dominican Republic, Honduras and Paraguay; Narayan (2005) for Indonesia, Singapore, Sri Lanka in the short-run, and Nepal in both the short-run and long-run. Secondly, there is the spend-and-tax hypothesis advanced by Peacock and Wiseman (1979, 1961). This hypothesis suggests that the direction of budget determination is from expenditures to revenues; that government expenditures determine government revenues. According to Peacock and Wiseman, a crisis, such as a war, can cause a permanent increase in government expenditures, which therefore requires an increase

3 THE CAUSALITY BETWEEN GOVERNMENT REVENUES AND EXPENDITURES IN BOTSWANA / 3 in revenues. The authors referred to this phenomenon as the displacement effect. Clearly, in the context of the spend-and-tax hypothesis, increasing tax revenues can reduce budget deficits. The spend-and-tax hypothesis implies a unidirectional causality from government expenditures to revenues. In Africa, the evidence comes from Carneiro et.al.(2004) for Guinea-Bissau, Wolde-Rufael (2008) for Burkina Faso, and Amoah and Loloh (2008) for Ghana in the long-run. In other regions, the hypothesis is confirmed by Parida (2012) for India, and Narayan (2005) for Indonesia and Sri Lanka in the long-run. Thirdly, we have the fiscal synchronization hypothesis of Musgrave (1966), and Meltzer and Richard (1981). This hypothesis asserts that budget determination is a simultaneous process in which the government synchronizes revenue and expenditure decisions. In other words, there is interdependence between government revenues and expenditures; the government decides on revenues and expenditures concurrently. Therefore, controlling budget deficits requires a simultaneous increase in revenues and decrease in expenditures. In empirical tests, the fiscal synchronization hypothesis is expected to yield a bidirectional causality between government revenues and expenditures. In Africa, this has been found to be the case by Aregbeyen and Insah (2013) in Ghana, Nyamongo et. al.(2007) in South Africa in the long-run, and Wolde-Rufael (2008) in Mauritius, Swaziland and Zimbabwe. Outside the continent, the evidence is widespread, such as Hye and Jalil (2010) in Romania, Magazzino (2010) in Italy during the post-world War 2 period of , AbuAl-Foul and Beghestani (2004) and A-Zeaud (2012) in Jordan, Elyasi and Rahimi (2012) in Iran, Li (2001) in China, Ho and Haung (2009) in 31 Chinese provinces in the long-run, and Mehrara et. al.(2011) in 40 Asian countries. The fiscal synchronization hypothesis is also supported by evidence from multi-country studies, such as Owoye (1995) in the USA, Germany, UK, France and Canada; Cheng (1999) in Chile, Panama, Brazil, and Peru. Finally, there is the institutional separation hypothesis advanced by Wildavsky (1984) and Baghestani and McNown (1994). This is also known as the neutrality hypothesis. The view in this case is that budget determination is a process where revenue decisions are separate and independent of expenditure decisions. Revenues are determined by taxable capacity and the tax handles available to government. Expenditure decisions are determined by the requirements of the general public. Thus, there is institutional separation of revenue and expenditure decisions. The institutional separation hypothesis suggests that, empirically, there is no causality between government revenues and expenditures. Nyamongo et. al. (2007) found such evidence for South Africa in the short-run, and Wolde-Rufael (2008) for Botswana, Burundi and Rwanda. Elsewhere, the evidence is provided by Garcia and Henin (1999) for Canada and Japan, Ho and Haung (2009) for 31 Chinese provinces in the short-run, and Narayan (2005) for India, Malaysia, Pakistan, Philippines and Thailand. Clearly, all four hypotheses are supported by empirical evidence. However, the evidence seems to be contradictory. For instance, according to Moalusi (2004), tax-

4 4 / CHRISTOPHER MUPIMPILA, LESEGO DINALE & BOITUMELO MOFFAT and-spend is evident in Botswana, while Wolde-Rufael (2008) suggests there is institutional separation in the country. What could be the cause of the apparent contradictions in the research findings? There have been attempts to synthesize the empirical studies on the revenueexpenditure nexus, notably by Elyasi and Rahimi (2012), Narayan (2005), and Payne (2003). The authors show that contradictions in support of the four hypotheses can be attributed to differences in data coverage, methodologies and the variables used in the studies. Besides, in the multivariate models, only two additional variables are often included, namely; the gross domestic product (GDP), and interest rate. This practice poses the danger of omitting other relevant variables. Therefore, in the present study, we include an extra variable, trade openness. 3. DATA AND MODEL SPECIFICATION In this study, we use quarterly data for the period 1995 to In order to control for the effects of inflation, all the variables are in real terms, with the consumer price index as the deflator (2002:Q4=100). The data is from the Banks of Botswana and Statistics Botswana. We extend by an extra variable, the multivariate model adopted by Moalusi (2004). The additional variable is trade openness, based on Mupimpila (2005), who found trade openness to be a significant determinant of government expenditure in Botswana. The estimated equations are: ln EXP ln TAX ln TRADE lngdppc ln INT (1) t 0 1 t 2 t 3 t 4 t t lntax ln EXP lntrade lngdppc ln INT (2) t 0 1 t 2 t 3 t 4 t t Where EXP t is government expenditure, consisting of both development and recurrent expenditures. The variable TAX t is tax revenue, both direct and indirect tax revenue. TRADE t denotes trade openness, defined as the ration of the sum of exports and imports to GDP. The variable GDPPC t represents gross domestic product per capita, and unlike GDP by itself, GDPPC t takes into account the population size. INT t is the interest rate, the Bank Rate, the rate at which the Bank of Botswana lends to commercial banks. Banks operating in Botswana typically set their prime lending rate in relation to the Bank Rate (Bank of Botswana, 2010). The variables t and t are the error terms for equations (1) and (2) respectively, while t denotes time. As expected, we test for unit roots and cointegration in order to determine how to perform the Granger causality, whether through a vector autoregressive (VAR) or a vector error correction model (VECM). If cointegration is present, Granger causality is tested by means of the following error correction model: ln EXP ln EXP ln TAX lntrade p p p p t 0 1i t 1 2i t i 3i t i i 1 i 1 i 1 p ln GDPPC ln INT 4i t i 5i t i t 1 1t i 1 i 1 (3)

5 THE CAUSALITY BETWEEN GOVERNMENT REVENUES AND EXPENDITURES IN BOTSWANA / 5 ln TAX lntax ln EXP lntrade q q q Q t 0 1i t 1 2i t i 3i t i i 1 i 1 I 1 q ln GDPPC ln INT 4i t i 5i t i t 1 2t i 1 i 1 (4) Where EXP t 1 and TAX t 1 are the lagged values of EXP t and TAX t, t 1 and t 1 are the lagged values of the error terms from equations (1) and (2) respectively. Thus, the lagged changes in EXP t and TAX t represent the short-run causality between government revenues and expenditures, while the lagged values of the error terms provide the long-run effects. 4. EMPIRICAL RESULTS The descriptive statistics of the variables are in Table A.1 in the Appendix, and they show that the variables are normally distributed. We employ the Augmented Dickey- Fuller (ADF) and the KPSS 1 unit root tests to determine the order of integration. The results are in Table A.2 and A.3 in the Appendix. The ADF results show that with trend and intercept, tax revenue, GDP and trade openness are stationary at levels while government expenditure and the interest rate are integrated of order 1, that is, they become stationary after the first difference. However, the ADF test without trend and intercept reveals that all the variables are non-stationary and they are I(1) variables. In addition, the KPSS test indicates that all the variables are integrated of order 1. Carneiro et. al. (2004) suggest that the power of the ADF test is limited, therefore, in case the ADF and KPSS reveal conflicting results, the KPSS is preferred. Thus, we adopt the view that the variables in this study are integrated of order 1. Since the variables are I(1), we then test for cointegration using the Johansen test with two statistics, the trace and maximum eigenvalue statistics. The trace statistic test indicates that there are no cointegrating equations (Table A.4). However, the maximum eigenvalue test shows that there is at most, one cointegrating equation (Table A.5). Given that eigenvalue results are generally considered more reliable in small samples, we conclude that there exists a long term relationship between the variables in the model. Therefore, we conduct Granger causality by means of a vector error correction model (VECM). We use the Akaike Information Criterion (AIC) to select an optimal lag length of 2 in estimating the VECM. Table 1 presents the VECM results depicting the short-run relationships among the variables. Evidently, past values of tax revenues are significant in explaining current levels of government expenditures in the short-run, and the relationship is negative. On the other hand, past values of government expenditures do not have any significant impact on current levels of tax revenues in the short-run. Table 1 also shows the error correction coefficients, which measure the speed of adjustment of the variables before they return to their long term equilibrium levels. For both government expenditures and revenues, the error correction coefficient is statistically significant and also bears the expected sign. Equation (5) shows the

6 6 / CHRISTOPHER MUPIMPILA, LESEGO DINALE & BOITUMELO MOFFAT Table 1 Vector Error Correction Model lnexp lngdppc lnint lntax lntrade ECT * * ** *** [ ] [ ] [ ] [ ] [ ] (lnexp(-1)) * [ ] [ ] [ ] [ ] [ ] (lnexp(-2)) *** [ ] [ ] [ ] [ ] [ ] (lngdppc(-1)) * *** [ ] [ ] [ ] [ ] [ ] (lngdppc(-2)) [ ] [ ] [ ] [ ] [ ] (lnint(-1)) * *** [ ] [ ] [ ] [ ] [ ] (lnint(-2)) [ ] [ ] [ ] [ ] [ ] (lntax(-1)) * *** *** [ ] [ ] [ ] [ ] [ ] (lntax(-2)) * *** [ ] [ ] [ ] [ ] [ ] (lntrade(-1)) *** [ ] [ ] [ ] [ ] [ ] (lntrade(-2)) [ ] [ ] [ ] [ ] [ ] Constant * *** *** [ ] [ ] [ ] [ ] [ ] R Note: *, **, *** denote significance at the 1, 5, and 10 percent levels. normalized cointegration coefficients depicting the impact of the other variables on government expenditures in Botswana in the long-run. The t-statistics are in the brackets. In equation (5), the relationship between tax revenues and government expenditures is negative and significant, just as in Table 1. This implies that a reduction in tax revenues leads to an increase in government expenditures, and is in line with that of Moalusi (2004), who also found a significant negative relationship between government revenues and expenditures in Botswana. Ordinarily, one would expect a reduction in tax revenues to reduce government spending and deficits (Friedman, 1978). What then explains the fact that a reduction in tax revenues leads to an increase in government spending? As indicated by Buchanan and Wagner (1978), the reduction in tax rates causes a false belief in the public that the cost of government programmes is low. This perception of low cost government programmes prompts the public to demand more government programmes, which in turn increases government spending; hence the negative relationship between tax revenues and government expenditures.

7 THE CAUSALITY BETWEEN GOVERNMENT REVENUES AND EXPENDITURES IN BOTSWANA / 7 Figures 1 and A.1 show the impulse responses of the variables to shocks in themselves and other variables in the model. The impulses support the results of the regressions. In Figure 1, it is clear that government expenditures respond negatively to shocks in tax revenues in all the periods. This supports the regression results, which show that there is a negative relationship between government revenues and expenditures in Botswana. In Figure A.1, the impulse responses indicate a negligible response of tax revenues to government expenditures. Thus, the impulse responses suggest a unidirectional causality from tax revenues to expenditures. In this study, we also employ the variance decomposition technique, which shows how each variable can predict future values of itself and of other variables, thereby determining the degree of causality (Hye and Jalil, 2010). We present these results in Table A.6 and Table A.7. The variance decomposition results are consistent with the impulse responses, which indicate that government expenditures are more responsive to changes in tax revenues than tax revenues are to changes in government expenditures. In all the periods, government expenditures explain less than 2% of the forecast error variance in tax revenues (Table A.7). All other variables explain the forecast error variance in tax revenues better than government expenditures. Thus, the picture that emerges from the regression results, impulse responses, and the variance decomposition is that, in Botswana, there is a unidirectional causality from government revenues to expenditures, and the relationship is negative. Figure 1: Multivariate Model Impulse Responses

8 8 / CHRISTOPHER MUPIMPILA, LESEGO DINALE & BOITUMELO MOFFAT More precisely, we use the VECM Granger causality to examine the relationship between government revenues and expenditures in Botswana. The results of the Granger causality block exogeneity Wald Test are in Table 2, and indicate that we reject the null hypothesis that tax revenue does not Granger-cause government expenditures in Botswana. The null hypothesis is rejected at the 5% significance level. This implies that there is a causal relationship from tax revenues to government expenditures. In other words, there is a unidirectional causality from revenues to expenditures in Botswana. This supports the tax-and-spend hypothesis in the country. Therefore the Granger causality results support the regression results, the impulse responses, and the variance decomposition; that government expenditures are more responsive to changes in tax revenues than tax revenues are to changes in government expenditures. Table 2 Multivariate Granger Causality Results Dependent variable: D(LNEXP) Excluded Chi-sq df Prob. D(LNGDPPC) D(LNINT) D(LNTAX) D(LNTRADE) All Dependent variable: D(LNTAX) Excluded Chi-sq df Prob. D(LNEXP) D(LNGDPPC) D(LNINT) D(LNTRADE) All Concerning other African countries, the Granger causality results of the present study are consistent with the findings of Moalusi (2004) for Botswana, Eita and Mbazima (2008) for Namibia, Keho (2010, 2013) for the Ivory Coast, Amoah and Loloh (2008) for Ghana, and Wolde-Rufael (2008) for Ethiopia, Ghana, Kenya, Nigeria, Mali and Zambia. As indicated in section 2 of this paper, the findings of these studies support the tax-and-spend hypothesis, that there is a unidirectional causality running from revenues to expenditures. This is besides the evidence from other countries, which also confirms the validity of the tax-and-spend hypothesis (section 2). Furthermore, the present study shows the other variables, apart from government revenues, which significantly influence government expenditures. These variables are: GDP per capita and interest rate. They both have a positive and significant impact on government expenditures in Botswana (equation 5). In the case of GDP per capita, the result supports Wagner s law of increasing state activities. The law holds that, as it develops, a country is likely to enact more laws, govern large towns, and build

9 THE CAUSALITY BETWEEN GOVERNMENT REVENUES AND EXPENDITURES IN BOTSWANA / 9 additional infrastructure; as a result, there would be increased public expenditures. However, the positive and significant impact of the interest rate on government expenditures is unexpected; because it implies that an increase in interest rates leads to an increase in government expenditures in Botswana. How can this possibly happen? Ussher (1998) suggests that rising interest rates can lead to reduced investment and thereby induce the government to undertake expansionary fiscal policies, especially through spending on public projects. Under these circumstances there would be a positive relationship between interest rates and government expenditures. It is worth noting that, unlike Mupimpila (2005), we find trade openness to have a negative and insignificant impact on government expenditures in Botswana. This can be attributed to differences in methodologies, variables, and sample periods. Mupimpila assessed the determinants of government expenditures in Botswana for the period 1980 to 2000, using annual data, while our sample period is 1995 to 2010; using quarterly data. He also included two extra variables, the level of monetization in the economy and the expenditures on social services. In contrast to the present study, Mupimpila (2005) did not test for causality between government revenues and expenditures. 5. SUMMARY AND CONCLUSIONS In this paper, the causality between government revenues and expenditures in Botswana is analyzed for the period 1995 to 2010, using quarterly data. The study employs a vector error correction model (VECM), and Granger causality; and finds support for the tax-and-spend hypothesis in both the short-run and long-run. In Botswana, there is a negative and unidirectional causality from tax revenues to government expenditures. According to Buchanan and Wagner (1978), this phenomenon is caused by a fiscal illusion perceived by the general public, such that a reduction in revenues raises government expenditures. Thus, reducing government expenditures and deficits requires that the general public should have a clear understanding of the actual cost of providing government programmes. This can be done through cost-sharing and cost-recovery schemes. The empirical evidence in support of the tax-and-spend hypothesis has other implications on budget deficits in Botswana. Since raising taxes avoids a fiscal illusion, it is therefore necessary to increase taxes in order to reduce budget deficits. Tax revenues should be increased by broadening the tax base, improving tax compliance and tax administration procedures. These measures are necessary in the present era when Botswana faces persistent budget deficits, unlike in the past when a boom in mining led to almost two decades of budget surpluses. References AbuAl-Foul, Bassam, and Hamid Baghestani (2004), The Causal Relation Between Government Revenue and Spending: Evidence from Egypt and Jordan, Journal of Economics and Finance, 28(2): Al-Khulaifi, Abdulla S. (2012), The Relationship Between Government Revenue and Expenditure in Qatar: A Cointegration and Causality Investigation, International Journal of Economics and Finance, 4(9):

10 10 / CHRISTOPHER MUPIMPILA, LESEGO DINALE & BOITUMELO MOFFAT Al-Zeaud, Hussein Ali (2012), Government Revenues and Expenditures: Causality Tests for Jordan, International Journal of Contemporary Research in Business, 4(7): Amoah, Benjamin, and Francis White Loloh (2008), Casual Linkage Between Revenue and Spending: Evidence from Ghana, Bank of Ghana Working Paper WP/BOG-2008, Bank of Ghana. Aregbeyen, Omo, and Baba Insah (2013), A Dynamic Analysis of the Link between Public Expenditure and Public Revenue in Nigeria and Ghana, Journal of Economics and Sustainable Development, 4(4): Baghestani, Hamid, and Robert McNown (1994), Do Revenues or Expenditures Respond to Budgetary Disequilibria? Southern Economic Journal, 6: Buchanan, James, and Richard Wagner (1978), Dialogue Concerning Fiscal Religion, Journal of Monetary Economics,4: Bank of Botswana (2010), Annual Report, Gaborone: Bank of Botswana. Carneiro, Francisco G., Joao R. Faria, and Boubacar Sid Barry (2004), Government Revenues and Expenditures in Guinea-Bissau: Causality and Cointegration, African Region Working Paper Series No. 65, The World Bank. Cheng, Benjamin S. (1999), Causality Between Taxes and Expenditures: Evidence from Latin American Countries, Journal of Economics and Finance, 23 (2): Easterly, William, Carlos Afredo Rodriquez, and Klaus Schmidt-Hebbel (1994), Public Sector Deficits and Macroeconomic Performance, New York: Oxford University Press for the World Bank. Eita, Joel Hinaunye, and Daisy Mbazima (2008), The Causal Relationship Between Government Revenue and Expenditure in Namibia, MPRA Paper No. 9154, Munich Personal RecPEc Archive. Elyasi, Yousef, and Mohammad Rahimi (2012), The Causality between Government Revenue and Expenditure, International Journal of Economic Sciences and Applied Research, 5(1): Friendman, Milton (1978), The Limitations of Tax Limitation, Policy Review, (June): Garcia, Manuel Jaen (2012), The Revenues-Expenditures Nexus: A Panel Data Analysis of Spain s Regions, International Journal of Academic Research in Economics and Management Sciences, 1(1): Garcia, Sophie, and Pierre-Yves Henin (1999), Balancing Budget Through Tax Increases or Expenditure Cuts: Is it Neutral? Economic Modelling, 16: Hye, Qazi Muhammad Adnan, and M. Anwar Jalil (2010), Revenue and Expenditure Nexus: A Case Study of Romania, Romanian Journal of Fiscal Policy, 1(1): Ho, Yuan-Hong, and Chiung-Ju Huang (2009), Tax-Spend, Spend-Tax, or Fiscal Synchronization: A Panel Analysis of the Chinese Provincial Real Data, Journal of Economics and Management, 5(2): Keho, Yaya (2010), Budget Balance Through Revenue or Spending Adjustment? An Econometric Analysis of the Ivorian Budgetary Process, , Journal of Economics and International Finance, 2(1): Keho, Yaya (2013), Threshold Cointegration and Asymmetric Adjustment Between Government Spending and Revenue in Cote D Ivoire, Asian Journal Of Empirical Research, 3(4): Li, Xiaoming (2001), Government Revenue, Government Expenditure, and Temporal Causality: Evidence from China, Applied Economics, 33: Magazzino, Cosimo (2010), Public Expenditure and Revenue in Italy, , MPRA Paper No.2730, Munich Personal RePEc Archive. Mehrara, Mohsen, Mosayeb Pahlavani, and Yousef Elyasi (2011), Government Revenue and Government Expenditure Nexus in Asian Countries: Panel Cointegration and Causality, International Journal of Business and Social Science, 2(7):

11 THE CAUSALITY BETWEEN GOVERNMENT REVENUES AND EXPENDITURES IN BOTSWANA / 11 Meltzer, Allen H., and Scott F. Richard (1981), A Rational Theory of the Size of Government, Journal of Political Economy, 89: Moalusi, Daniel K. (2004), Causal Link between Government Spending and Revenue: A Case Study of Botswana Fordham Economics Discussion Paper Series, dp , Department of Economics, Fordham University, New York. Mupimpila, Christopher (2005), Determinants of Government Expenditures in Botswana, The ICFAI Journal of Applied Economics, 4(4): Musgrave, Richard (1966), Principles of Budget Determination, in Public Finance: Selected Readings, edited by H. Cameron and W. Henderson, New York: Random House. Narayan, Paresh Kumar (2005), The Government Revenue and Government Expenditure Nexus: Empirical Evidence from Nine Asian Countries, Journal of Asian Economics, 15: Nyamongo, Morekwa Esman, Moses M Sichei, and Niek J Schoeman (2007), Government Revenue and Expenditure Nexus in South Africa, SAJEMS NS 10 (2): Owoye, Oluwole (1995), The Casual Relationship Between Taxes and Expenditures in the G7 Countries: Cointegration and Error-Corretction Models, Applied Economics Letters, 2: Parida, Yashobanta (2012), Casual Link between Central Government Revenue and Expenditure: Evidence from India, Economic Bulletin, 32(4): Payne, James E. (2003), A Survey of the International Empirical Evidence on the Tax-Spend Debate, Public Finance Review, 31(3): Peacock, Alan, and Jack Wiseman (1961), The Growth of Public Expenditure in the United Kingdom, Princeton, NJ: Princeton University Press for the National Bureau of Economic Research. Peacock, Alan, and Jack Wiseman (1979), Approaches to the Analysis of Government Expenditures Growth, Public Finance Quarterly, 7:3-23. Ussher, Leanne (1998), Do Budget Deficits Raise Interest Rates? A Survey of the Empirical Literature, Transformational Growth and Full Employment Project, Working Paper No. 3, Economics Department, New School Research, New York. Wildavsky, Aaron (1984), The Politics of the Budgetary Process, Boston: Little, Brown and Company. Wolde-Rufael, Yemane (2008), The Renue-Expenditure Nexus: The Experience of 13 African Countries, African Development Review, 20: APPENDIX Table A.1 Descriptive Statistics LNEXP LNGDPPC LNINT LNTAX LNTRADE Mean Median Maximum Minimum Std. Dev Skewness Kurtosis Jarque-Bera Prob

12 12 / CHRISTOPHER MUPIMPILA, LESEGO DINALE & BOITUMELO MOFFAT Variables Unit Roots ADF Table A.2 ADF Unit Root Results Levels 5% Crit. Prob. First 5% Crit. Prob. Order of Values Diff. Values Integration With trend and intercept lnexp I (1) lntax I (0) lngdppc I (0) lntrade I (0) lnint I (1) Without trend and intercept lnexp I (1) lntax I (1) lngdppc I (1) lntrade I (1) lnint I (1) Variables With intercept Table A. 3 KPSS Unit Root Results Unit Roots KPSS 5% Crit. Probability First 5% Crit. Probability Order of Levels Values Difference Values Integration lnexp I (1) lntax I (1) lngdppc I (1) lntrade I (1) lnint I (1) With trend and intercept lnexp I (1) lntax I (1) lngdppc I (1) lntrade I (1) lnint I (1) Table A.3 presents results for the KPSS unit root test which indicates that all variables are integrated of order 1. Similar to the ADF test the probability of less than 0.05 was used to reject the null hypothesis of stationarity. The results of the KPSS test are chosen on the basis that it is more superior to the ADF test.

13 THE CAUSALITY BETWEEN GOVERNMENT REVENUES AND EXPENDITURES IN BOTSWANA / 13 Table A.4 Unrestricted Cointegration Rank Test (Trace) Hypothesized Trace 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.**4 r = r r r r Trace statistic test indicates no cointegrating eqn(s) at the 0.05 level Table A.5 Unrestricted Cointegration Rank Test (Max. Eigenvalue) Hypothesized Max-Eigen 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** r = 0* r r r r Max-eigenvalue test indicates 1 cointegrating equation at the 0.05 level * denotes rejection of the hypothesis at the 0.05 level Table A.6 Variance Decomposition of Government Expenditure (Multivariate) Period Std. Error DLEXP DLGDPPC DLINT DLTAX DLTRADE Table A.7 Variance Decomposition of Tax Revenue (Multivariate) Period Std. Error DLEXP DLGDPPC DLINT DLTAX DLTRADE

14 14 / CHRISTOPHER MUPIMPILA, LESEGO DINALE & BOITUMELO MOFFAT Figure A.1: Multivariate Model Impulse Responses

15

16