The Growth of Government Spending in Lesotho

Transcription

1 The Growth of Government Spending in Lesotho Retselisitsoe I. Thamae Department of Economics, National University of Lesotho, P.O. Roma 180, Maseru, Lesotho ( Abstract: The paper analyses the dynamics of government spending growth in Lesotho using the multivariate cointegration techniques for the period The results indicate that government spending is positively related to income and population while negatively related to tax share in the long-run. The latter supports the idea of fiscal illusion caused by budget deficits, which lessen the perceived cost of public spending to taxpayers. The role of internal and external shocks on government spending is also investigated in this study but such factors are found to be less important in determining the growth of government expenditure in Lesotho. I. INTRODUCTION An excessive growth of government spending is often alleged to be economically destructive and this is because different ways of financing government expenditure such as taxation, borrowing and creating money are considered to have adverse effects on the economy. As suggested by Alm and Embaye (2010), these negative effects include slow economic growth, large budget deficits, mounting inflation and interest rates, trade deficits as well as falling exchange rates. In addition, Mitchell (2005) indicates that the growth of government spending does not only retards economic performance through its requirement of costly financing decisions, but also by crowding out private sector activities, funding harmful interventions, and hindering innovation and resource allocation. Therefore, in order to address these issues, the understanding of the reasons behind the growth of public spending is necessary. According to Borcherding and Lee (2004), the analysis of government spending growth is generally classified into two categories, namely, a-institutional and institutional approaches. The a-institutional approach argues that the growth of the level of government expenditure depends on the changing social and market conditions. Under this approach, the median voter is assumed to play a significant role in determining the level of spending by the government and

2 hence the demand for public services is considered to be driven by factors such as the voter s preferences, income, tax-price and relative price of private goods and services. Population also falls under the a-institutional approach as one of the demographic factors affecting the level of public spending and thus captures the publicness of services delivered by the government. The institutional approach, on the other hand, emphasizes the role of rent-seeking activities, structural changes and major economic or political shocks on the growth of government spending. The approach adds more factors missing under the a-institutional paradigm and offers a wider range of frameworks in explaining the growth of public expenditure. For example, the institutional theory is related to the concepts of Wagner s (1893) Law and the displacement effect of Peacock and Wiseman (1961). The Wagner s Law predicts that government expenditure increases at the faster rate than the growth of the income level while the notion of displacement effect argues that government spending may shift permanently to a new level as a result of major disturbances such as wars and economic crises. Openness is also proposed as an additional factor that has a positive effect on the scope of government, with the relationship being robust when the risk associated with terms of trade is highest (Rodrik, 1998). In Lesotho, the level of government spending as measured by real per capita government spending and government spending as a percentage of gross domestic product (GDP) has experienced an upward trend since From that period till 2010, the real per capita government spending averaged M1,684 at 2004 constant prices. In 2009, it reached the maximum of M3,665, which is about 5 times as much as the real per capita spending of M711 during 1980, and slightly went down to M3,326 in On the other hand, the real per capita GDP has approximately doubled from M2,531 in 1980 to M5,396 in All these show a more than proportionate increase in real per capita government spending relative to that of real per capita GDP. Thus, the aim of this paper is to investigate the factors behind the dynamics of government spending growth in Lesotho using the multivariate cointegration techniques. The study contributes to the empirical literature of the determinants of public spending growth in developing countries. The rest of the paper is structured as follows. Section II provides Lesotho s brief economic background. Section III reviews the literature on public spending growth and section IV outlines the model. The results and analyses are reported in section V while section VI offers a conclusion. 2

3 II. ECONOMIC CONTEXT As a small open economy, Lesotho has experienced an annual average growth rate of approximately 3.8 percent over the past ten years and currently has a per capita income of about US$1000. The growth is attributed mainly to the contribution of the secondary sector, of which manufacturing is the largest, owing to the African Growth and Opportunity Act (AGOA) of the United States (US). The economic progress within the country generally depends mainly on the developments in the neighboring Republic of South Africa (RSA), which surrounds it entirely, through factors such as the remittances of workers employed in the RSA mines as well as the royalties from the transfer of water resources to RSA under a relatively large government infrastructure project the Lesotho Highlands Water Project (LHWP). Furthermore, RSA is Lesotho s major trading partner, with about 80 percent of Lesotho s imports coming from RSA and RSA buying approximately 25 percent of Lesotho s exports (IMF, 2012). Lesotho also shares similar trading, exchange rate and monetary policies with the RSA that continue to benefit the country since they are both members of the Common Monetary Area (CMA), the Southern African Customs Union (SACU) and the Southern African Development Community (SADC). For instance, the net transfers from the SACU revenue pool account for about more than half of the total government revenue in Lesotho and hence support a considerable amount of government expenditure (Masenyetse and Sephoko, 2012). Nevertheless, the high dependency of Lesotho on SACU revenues makes the country to be more vulnerable to external shocks. In recent years, for example, Lesotho had to deal with several macroeconomic challenges as a result of a sharp decline in SACU revenues, which resulted in rising budget deficits as well as external debt burden. The country, therefore, had to enhance its domestic revenue collections and implement some austerity measures by cutting its spending, particularly on recurrent expenditures. III. LITERATURE REVIEW The theories of public spending growth usually begin with some variant of the median voter hypothesis, which assumes that the median voter plays a significant role in determining the level of spending by the government (Alm and Embaye 2010). Consequently, the demand for public services is considered to be driven by factors such as the median voter s preferences, income, tax-price and relative price of private goods and services (Borcherding and Lee 2004, Bowen 1943). One of the earliest studies offering a formal representation and empirical estimation of the median voter model is that of Borcherding and Deacon (1972), which analyses the demand for 3

4 public services provided by the non-federal governments in the US. Niskaken (1978) also extends the empirical application of the same model by examining the relationship between government budget deficits, public spending and inflation within the US economy. While these studies follow the a-institutional approach in determining the factors behind the growth of government expenditure, some empirical work, like that of Wagner (1893), is more consistent with the institutional approach. Richter and Paparas (2012), for example, provide support of the validity of Wagner s Law in Greece, which shows a stable relationship between public spending and national income, with the causality running from national income to government expenditure. The conditions in Greece were also found to be conducive for Wagner s Law since the period covered includes early stages of growth, industrialisation and modernisation of the Greek economy. Conversely, Durevall and Henrekson (2011) show that Wagner s Law generally does not hold in the long-run in Sweden and the United Kingdom (UK). Although the existence of Wagner s Law occurred between 1860 and mid 1970s due to the formation of the modern public sector, the study indicates that Wagner s Law did not hold during the initial industrialisation era. Furthermore, the results reveal that in recent periods, income affects public spending only when the population age structure is controlled. However, the study suggests that the ratchet or displacement effect, which occurs as a result of wars, social upheavals and recession is not a general cause of public expenditure growth in Sweden and the UK. Alternatively, Menyah and Wolde-Rufael (2012) present the evidence that the growth of government spending in RSA seems to occur as a result economic growth or social progress brought by industrialisation in accordance with Wagnerian proposition. Nonetheless, the study points out that RSA should also aim to achieve higher economic growth in order to meet the growing demand for social and infrastructure expenditure. In addition, the analysis of RSA s public spending growth by Alm and Embaye (2010) supports the idea that government expenditure is not only related to national income and the true cost public service provision as captured by the wage rate but it is also associated with fiscal illusion caused by the budget deficits. On the other hand, the external conditions such as wars and oil shocks are found to play an important role in explaining the dynamics of public expenditure and this renders support for notion of ratchet or displacement effect. Therefore, the study concludes that both institutional and a-institutional factors explain quite well the relationship between public spending, national income, tax share and wage rate in RSA. In the case of five South East Asian countries Indonesia, Malaysia, Philippines, Singapore and Thailand, the empirical results from the study by Dogan and Tang (2006) on 4

5 government expenditure and national income provides no evidence for the Wagner s Law. The findings also indicate that public spending do not play a major role in terms of promoting economic growth in those countries, with the exception of Philippines, and such a result was unexpected given the general belief that the government in those countries is considered to have contributed a lot towards development through its spending. Even though several studies find weak support for validity of Wagner s Law in developing countries (see Gandhi, 1971 and Ram, 1987), the results from the analysis by Akitoby et al. (2006) reveal that public expenditure and national output in majority of developing economies are cointegrated, implying the existence of long-term relationship between government spending and output in line with Wagner s Law. IV. ANALYTICAL FRAMEWORK 4.1 Model Specification The paper adopts the standard median voter framework that is extended to allow for other factors such as internal and external shocks. Following Niskanen (1978) and Alm and Embaye (2010), the median voter s demand function consists of the quantity of public good demanded by the median voter, the perceived per unit price of government services paid by the median voter, the median voter s income and other exogenous conditions affecting the demand for government services, and it is specified as follows: where is the scale parameter;, and are parameters of the median voter s demand function with, and. Since the variables and are unobservable, it is assumed that the cost of is priced by the government at a unit marginal cost equal to. Then, given the median voter s share of the unit cost of government services, the perceived per unit price of public services paid by the median voter is given by the product while gives the government spending per capita. This leads to the following median voter s demand function: 5

6 The variable, which represents the median voter s tax share, is assumed to be a function of the fraction of government expenditure financed by tax revenues and the total number of taxpayers, as follows: where is the total tax revenues, is the total government spending and is the total number of voter-taxpayers. In the case where the median voter has perfect foresight, government spending would not be affected by deficit financing. However, it is assumed that the median voter may be subject to fiscal illusion due to debt financing and thus, the higher deficit would result in higher demand for government spending. It is also assumed that the marginal cost is a function of the private sector wage rate and the total number of voter-taxpayers, as follows: where is the scale parameter, and measures the rate of increase in the price of government services relative to that of services in other sectors while captures the degree of publicness of services offered by the government. The parameter is positive if the productivity growth in the public sector is lower than in the private sector and it is unity if there is no growth in public sector productivity. On the other hand, is equal to zero if government services are pure public goods; it is unity if the government provides purely private goods; and it is greater than unity if there is crowding out of public services. Substituting equations and into equation leads to the following: and given that and log linearising, where,,,, and. 6

7 The variables in equation are all measurable and the sign of the coefficient on income is expected to be positive while that of on tax share variable is expected to be negative. The coefficient on population is expected to have a positive sign when demand for public services is price inelastic (that is, ) but any sign when demand is price elastic. On the other hand, the expected sings for coefficients and on wage rate and exogenous conditions, respectively, are ambiguous. 4.2 Estimation Strategy The Johansen s (1988, 1995) multivariate cointegration procedure is applied to estimate equation (6) since it performs better in terms of determining the long-run relationship among series with the same order of integration. The paper, however, begins with the investigation of the presence of unit root among the variables using the augmented Dickey Fuller (ADF) test (Dickey and Fuller 1979, 1981). If the variables are found to be integrated of the same order, the maximum likelihood estimation of the unrestricted vector autoregression (VAR) model is then undertaken to determine if the variables are cointegrated. The lag length is chosen on the basis of the Akaike information criterion (AIC) and the Schwarz-Bayesian information criterion (SBIC) while the cointegrating rank is determined by applying two likelihood ratio tests, namely, trace statistic and maximum eigenvalue statistic. When variables are cointegrated, the error correction model (ECM) is estimated to analyse the developments between short-run and long-run relationships among them. Finally, the ECM within the multivariate formulation is used to investigate Granger causality between the variables (Engle and Granger, 1987). V. DATA AND ESTIMATION RESULTS 5.1 Data and Descriptive Statistics The paper employs annual time-series data for the period 1980 to Table 1 presents the descriptive statistics for the main variables used in the estimation and some of their trends are depicted in Figures A1 to A4 in the appendices. The dependent variable is captured by real per capita government spending (G) while real GDP per capita (Y) is a proxy for the median voter s income (Niskanen 1978). The ratio of government revenue to government spending (R/E) represents the tax share and the total number of voter-taxpayers is given by population (N). The data on all these variables is obtained from the International Monetary Fund (IMF) World Economic Outlook Database (October, 2012) and the nominal variables are expressed in 2004 constant prices using the consumer price index. The other data on openness, which is defined as 7

8 an interaction between international trade and terms of trade, as one of the exogenous factors is sourced from the World Bank Development Indicators (2012). The wage rate (W), however, is not included in the estimation because of the unavailability of data. Table 1: Descriptive Statistics Variable Mean Standard Deviation Minimum Maximum Real per capita government spending, G 1, ,665 Real GDP per capita, Y 3, ,531 5,396 Revenue/spending, R/E Population, N International trade (as fraction of GDP) Terms of Trade Unit Root Test Results The ADF unit root test is applied to determine the order of cointegration of the variables. The lag length for each series is selected based on AIC. The null hypothesis states that the variable is non-stationary and failure to reject the null indicates that there is a unit root. Table 2 reports the results for unit root test both in levels and first differences and on the basis of the p-values, all variables are found to be integrated of order one or at 1% significance level. Therefore, this implies that there might be an existence of long-run relationship between real per capita government spending, real GDP per capita, tax share and population. Table 2: Augmented Dickey-Fuller (ADF) Unit Root Tests Variable non-stationary in levels non-stationary in first differences Test statistic p-value Test statistic p-value G Y R/E N Openness Notes: All variables are in logarithmic form. 5.3 Cointegration Test Results 8

9 Since the variables are found to be integrated of order one, the Johansen multivariate cointegration technique is employed to ascertain the presence of long-run relationship among them. The optimal lag specification of the VAR model based on AIC and SBIC is 1 and the results for the test of cointegration are presented in Table 3, with the Johansen trace test results given in part A while part B provides the estimated parameters of the normalised cointegrating equation. The trace test confirms the existence of one cointegrating vector at 5% level of significance and all the normalised cointegrating coefficients are statistically significant at that level. Hence, there is a stable relationship between real per capita government spending, real GDP per capita, tax share and population. The long-run estimates for real GDP per capita, tax share and population from the cointegration test results also prove to be consistent with the theory. For example, real per capita government spending is found to be positively related to both real GDP per capita and population while there is a negative relationship between tax share and real per capita government spending. The latter seems to support the proposed idea of fiscal illusion caused by budget deficits or tax cuts, which lessen the perceived cost of government spending to taxpayers. Hence, higher fiscal deficits (surpluses) in Lesotho are associated with higher (lower) levels of real per capita government spending. These cointegration test results are similar to the ones reported by Alm and Embaye (2010) since they show that public expenditure is not only related to national income and population but it is also associated with fiscal illusion as a result of the budget deficits. Part A: Johansen trace test for cointegration Table 3: Cointegration Test Results Maximum rank Eigenvalue Trace statistic 0.05 critical value Part B: Normailsed cointegrating coefficients G Y R/E N Constant (-2.757) Notes: t-statistics in parentheses (2.432) (-1.896) (3.335) 9

10 Alternatively, the long-run positive relationship between government spending and national income serves as an implication for the existence of Wagner s Law in Lesotho and this is in line with the empirical findings for majority of developing countries (Akitoby et al. 2006). Although the Granger causality tests has to be undertaken to provide further support for the validity of Wagner s Law, the implication offered seems to be plausible given the fact that Lesotho, as one of the developing economies, is undergoing through stages of growth, industrilisation and modernisation the conditions that are conducive for Wagner s Law (Richter and Paparas 2012). Furthermore, the public sector in Lesotho has to supply services that are normally rising over time such as education, healthcare and pensions, and these social expenditures coupled with the spending in infrastructure are credible reasons why government expenditure seems to increase at the faster rate than the growth of the national income (Wagner 1893). The findings from this cointegration analysis have some important implications for Lesotho. First, given the presence of fiscal illusion, the electorate has to be clearly presented with the fiscal budget to discern the true cost of government spending, rather than being expected to be indulgent towards any debt financing or tax-cut strategy (Young 2009). Second, Lesotho has to maintain its budget deficit and external debt levels within some tolerable limits in order to avoid the adverse effects brought by excessive growth of public spending (Alm and Embaye 2010, Mitchell 2005). Lastly, the rising population (see Figure A4 in the appendices) and the industrialisation in Lesotho requires the country to have a right mixture of economic policies that will enable it to attain higher economic growth in order to meet the growing demand for social and infrastructure expenditure (Menyah and Wolde-Rufael 2012). 5.4 Error Correction Model (ECM) Estimation Results Table 4 presents the estimation results from the ECM based on the assumption of one cointegration relationship. The residuals in the model are shown to be approximately white noise by the Lagrange multiplier (LM) and the Jarque-Bera diagnostic tests at 1% level of significance. The estimates of the tax share and population adjustment parameters, however, were highly insignificant and consequently, the model was re-estimated with the restrictions of weak exogeneity on tax share and population variables. The p-value for the chi-square test justifies that tax share and population are weakly exogenous at 5% level of significance. Therefore, the equations of both variables in the estimation of the ECM have no error correction terms since their behaviour is explained only by the short-run dynamics. 10

11 Table 4: Estimation Results of the Error Correction Mmodel (ECM) Error correction [0.823] LM-statistic (8 lags) [0.068] (-0.388) (-0.215) (1.076) (0.698) (0.472) (6.834) (-0.793) (-3.276) (0.354) (-2.153) Jarque-Bera statistic (with 8 degrees of freedom) [0.021] Cointegrating vector, Notes: t-statistics in parentheses and p-values in brackets (0.362) (-0.225) (-0.075) (0.666) (0.530) (1.848) (0.158) (1.710) In order to capture the effects of internal and external shocks on the growth of government spending, the paper also included openness and a dummy variable (1998 dummy) for the political unrest of 1998 in the estimation of the baseline ECM shown in Table 4 with the first-order lag of the endogenous variables. Nevertheless, both variables were found to be insignificant as suggested by the specification test results reported in Table A1 (in the appendices) and thus have little effect on real per capita government spending. Therefore, in accordance with the conclusions of the study by Durevall and Henrekson (2011), the empirical results from the estimations of this paper suggest that the ratchet or displacement effect, which occurs as a result of internal and external shocks, has not been a general cause of the growth of government expenditure in Lesotho during the period under consideration. Furthermore, the results have revealed the baseline ECM to be the preferred specification since it fits the data well as shown by the lowest value of the SBIC. However, the adjustment coefficient on real per capita government spending from this baseline model in Table 4 has the appropriate sign but its magnitude is small and insignificant. Similarly, Table 2A (in the appendices) gives no strong evidence for the Granger causality running from real GDP per capita, tax share or population to real per capita government spending, indicating lack of support for the validity of Wagner s Law in Lesotho under the Granger causality approach. Nonetheless, these shortcomings might be attributed to the small sample size employed in the estimations of 11

12 this paper as well as the low power problem as a result of the over-fitting needed under the multivariate Granger causality tests (Dolado and Lutkepohl 1996). VI. CONCLUSION The paper analyses the dynamics of government spending growth in Lesotho using the multivariate cointegration techniques on time-series data covering the period of 1980 to The results from the cointegration test analysis indicate that there is a stable relationship between real per capita government spending, real GDP per capita, tax share and population in Lesotho. The test results are also consistent with the theory since real per capita government spending is found to be positively related to both real GDP per capita and population while there is a negative relationship between tax share and real per capita government spending. The latter seems to support the proposed idea of fiscal illusion caused by budget deficits, which lessen the perceived cost of government spending to taxpayers in Lesotho. The long-run positive relationship between government spending and national income, on the other hand, serves as an implication for the existence of Wagner s Law in Lesotho. Although the Granger causality tests provide no further support for the validity of Wagner s Law, the implication offered seems to be plausible since Lesotho is one of the developing economies and it is undergoing through stages of growth, industrilisation and modernisation the conditions that are conducive for Wagner s Law. Furthermore, the public sector in Lesotho has to supply services that are rising over time such as education, healthcare and pensions, and these social expenditures coupled with the spending in infrastructure are credible reasons why government expenditure seems to increase at the faster rate than the growth of income in Lesotho. Alternatively, the empirical results suggest that the ratchet or displacement effect is not a general cause of the growth of government expenditure in Lesotho. These findings have some important policy implications for Lesotho. First, given the presence of fiscal illusion, the electorate has to be clearly presented with the fiscal budget to discern the true cost of government spending, rather than being expected to be indulgent towards any debt financing or tax-cut strategy. Second, Lesotho has to maintain its budget deficit and external debt levels within some tolerable limits in order to avoid the adverse effects brought by excessive growth of public spending. Lastly, the rising population and the industrialisation in Lesotho requires the country to have a right mixture of economic policies that will enable it to attain higher economic growth in order to meet the growing demand for social and infrastructure expenditure. 12

13 REFERENCES Akitoby, B., B. Clements, S. Gupter and G. Inchauste (2006). Public Spending, Voracity, and Wagner s Law in Developing Countries, European Journal of Political Economy. 22: Alm, J. and A. Embaye (2010). Explaining the Growth of Government Spending in South Africa, South African Journal of Economics. 78: Borcherding, T.E. and D. Lee (1977). The Growth of Relative Size of Government. C. K. Rowley and F. Schneider (ed.), Encyclopedia of Public Choice. Dordrecht, The Netherlands: Kluwer Academic Publishers: Borcherding, T.E. and R.T. Deacon (1972). The Demand for the Services of Non-federal Governments, The American Economic Review. 62: Bowen, H.R. (1943). The Interpretation of Voting in the Allocation of Economic Resources, The Quarterly Journal of Economics. 58: Dickey, D.A. and W.A. Fuller (1979). Distribution of the Estimators for Autoregressive Time Series with a Unit Root, Journal of the American Statistical Association. 74: Dickey, D.A. and W.A. Fuller (1981). Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root, Econometrica. 49: Dogan, E. and T.C. Tang (2006). Government Expenditure and National Income: Causality Tests for Five South Asian Countries, International Business and Economics Research Journal. 5: Dolado, J. and H. Lutkepohl (1996). Making Wald Tests Work for Cointegrated VAR Systems, Econometric Reviews. 15: Durevall, D. and M. Henrekson (2011). The Futile Quest for a Grand Explanation of Long-run Government Expenditure, Journal of Public Economics. 95: Engle, R.F. and C.W.J. Granger (1987). Cointegration and Error Correction: Representation, Estimation, and Testing, Econometrica. 55: Gandhi, V.P. (1971). Wagner s Law of Public Expenditure: Do Recent Cross Section Studies Confirm It? Public Finance. 26: IMF (2012). Kingdom of Lesotho: Poverty Reduction Strategy Paper National Strategic Development Plan, Country Report No. 12/102. Washington: International Monetary Fund. Johansen, S. (1988). Statistical Analysis of Cointegration Vectors, Journal of Economic Dynamics and Control. 12: Johansen, S. (1995). Likelihood-Based Inference in Cointegrated Vector Autoregressive Models. Oxford: Oxford University Press. Masenyetse, R. F. and I. M. Sephooko (2012). Government Revenue-Expenditure Nexus in Lesotho: the Decline in SACU Revenue, American Journal of Economics. 2: Menyah, K. and Y. Wolde-Rufael (2012). Wagner s Law Revisited: A Note from South Africa, South African Journal of Economics. 80:

14 Mitchell, D.J. (2005). The Impact of Government Spending on Economic Growth, Backgrounder No Washington: The Heritage Foundation. Niskanen, W.A. (1978). Deficits, Government Spending, and Inflation: What is the Evidence?, Journal of Monetary Economics. 4: Peacock, A.T. and J. Wiseman (1961). The Growth of Public Spending in the United Kingdom. Princeton, NJ: Princeton University Press. Ram, R. (1987). Wagner s Hypothesis in Time-series and Cross-section Perspectives: Evidence from Real Data for 115 Countries, Review of Economics and Statistics. 69: Richter, C. and D. Paparas (2012). The Validity of Wagner s Law in Greece During the Last 2 Centuries, INFER Working Paper 2012(2). Bonn: International Network for Economic Research. Rodrik, D. (1998). Why do More Open Economies have Bigger Governments? The Journal of Political Economy. 106: Wagner, A. (1893). Grundlegung der Politschen Okonomie. Liepzig, Germany: C.F. Winter. Young, A. T. (2009). Tax-Spend or Fiscal Illusion? Cato Journal. 29: ,000 3,500 APPENDIX Real per Capita Government Revenue Real per Capita Government Spending 3,000 2,500 2,000 1,500 1, Figure A1: Real Government Rrevenue and Spending Per Ccapita (Maloti),

15 Government Revenue (% of GDP) Government Spending (% of GDP) Figure A2: Government Revenue and Spending as a Percentage of GDP, ,500 5,000 4,500 4,000 3,500 3,000 2,500 Figure A3: Real Gross Domestic Product (GDP) Per Capita (Maloti),

16 Figure A4: Population (Logarithmic Scale) Table A1: Specification Tests Specification AIC SBIC Is the variable significant? Baseline Openness No 1998 Dummy No Table A2: Granger Causality Test Results Null hypothesis Wald p-value Y does not Granger-cause G (R/E) does not Granger-cause G N does not Granger-cause G G does not Granger-cause Y (R/E) does not Granger-cause Y N does not Granger-cause Y