Displacement Effect And Economic Growth In The Uk Finance Essay - Free Essay Example

In this chapter of the research, will discuss the assumption made by both the Peacock and Wiseman (1961) displacement hypothesis to explain the increases in the proportion of time government expenditure to economic growth in the United Kingdom. They found that government expenditure in the United Kingdom did not follow a smooth trend, but instead, it seems to jump up in separate times. Peacock and Wiseman (1961) proposed the displacement effect hypothesis. It had related to the Wagners law even though there are a few differences between them. Thus, they contend that under normal conditions of peace and economic stability, changes in public expenditure are quite limited. The effect of the public expenditure on the time pattern of the general government expenditure is that public sector size will tend to be constant over time, rather than increasing, unless same major crisis periods occur, which require an increase in government intervention. The equivalent expansion of the public sector will not be just temporary, since the new levels of government expenditure and taxation will be accepted by the electors, and therefore public sector size will remain stable at an higher level until the next shock. The data used in this study is the time series Quarterly data for two periods of (1980q1 to 1990q2), and (1990q3 to 2007q4), have utilized to analyze the relationship between government expenditures and economic growth by measuring the gross domestic product in the Saudi economy. The rest of the chapter is organizing as following: section one, presents some empirical results of relevant theoretical and empirical literature on the relationship between government expenditure and economic growth. Section tow, presents the version of Peacock and Wiseman and their formula to explain the Displacement Effect. Section three, investigates the data and empirical results and analysis by using the methods. In addition, Section four, presents the results of analysis by using the time series techniques , such as the Ordinary Least Square (OLS), Augmented Dickey-Fuller for stationary Unit Root Tests, co-integration test , Causality Granger test , and Error Correction Model (ECM) , that for real GDP and Non-Oil GDP . While section five, concludes the chapters and presents. 9.2. The Displacement Effect Hypothesis 9.2.1. Structural Break Theory As we mentioned before in chapter three, wars are capable of displacing this notion of tolerable tax rates. In addition, expenditure may fall again, but not to their previous levels. Therefore, public expenditure grows in a discontinuous and stepwise fashion, the steps coming at times of major social upheavals (Safa, 1998). According to, Tussing and Henning (1991:397) the upward displacement effect by Peacock and Wiseman is an example, but obviously not the only one of such a structural change. Nelson and Plosser (1982) examined the relationship between the unable to reject the null of a unit root against trend stationary alternatives their data set. They found that impact on the way economic series have viewed and treated subsequently, which have further discussed by Perrons (1989). Zivot and Andrews (1992) pointed out the specification argued in favour of the need to view break points as endogenous and to develop procedures, which considered this endogenously. Diamond (1977) presented the displacement effect as a theory of structural break, which means that the usual ceteris paribus assumption of unchanged tastes, preferences and institutions after the upheaval has denied. He has used the Chow test comparing two periods separated by a social upheaval, and he found that, if this shows significant structural change and there has been displacement. 9.2.2. A Ratchet Effect As mentioned previously, the main argument of the ratchet effect is that if there a crisis and GNP decline, then the public expenditure decline but less than GNP. According to Bird (1972), he has explained the displacement effect and called it the ratchet effect. Moreover, Bird (1972) has argued that crises are likely to have short-term implications for (E / GNP) rather than crises lead to a permanent upward displacement for (E / GNP). Henrekson (1992) argued that the (E / GNP) is fall in the short run in times of unexpectedly rapid GNP growth. Other study for Peacock and Wiseman (1979) they argued that at the extreme, the ratchet effect interpretation of the displacement effect leads to the denial of its very existence. 9.3. Empirical Testing of Displacement Effect: Previous Studies Gupta (1967) was the first attempt to subject the displacement effect to empirical testing. He found significant displacement after the world wars in all cases except for Sweden after World War II. However, this result seems to be due to an estimation error, he also found significant displacement caused by the Great Depression in the case of the U.S. and Canada. According to, Henrekson (1990:246) Peacock and Wiseman (1961), adopt a clearly inductive approach to explaining the growth of government expenditure. When Peacock and Wiseman observed that expenditures over time appeared to outline a series of plateaus separated by peaks, and that these peaks coincided with periods of war and preparation for war they were led to expound the displacement effect hypothesis. Legrenzi (2003) argued that the displacement effect for Italy within a multivariate revenue-expenditure model of government growth. His result for long-run analysis shows an effect of GDP on the governments growth. Otherwise, the short-run analysis shows some evidence for the displacement effect, in terms of a lower resistance against tax financing of government expenditures in the war. The similar test of Guptas version in many ways is for Bonin, Finch and Waters (1969); they have tested displacement effect in the U.K. after the two world wars. In addition, Peacock and Wiseman investigated that both citizens and government hold divergent views about the desirable size of public expenditures and the possible level of government taxation. This divergence can adjust by social disturbances that destroy established conceptions and produce a displacement effect. People will accept, in a period of crisis, tax levels and methods of raising revenue that in quieter times would have though intolerable, and this acceptance remains when the disturbance itself has disappeared. As a result, the revenue and expenditure statistics of the government show a displacement after periods of social disturbance. Expenditures may fall when the disturbance is over, but they are less likely to return to the old level. The state may begin doing some of the things it might formerly have wanted to, but for which it had hitherto felt politically unable to raise the necessary revenues (Peacock and Wiseman 1961: 26). Other study for Henry and Olekalns (2000), investigated the Peacock and Wisemans displacement effect to explain the increases in the ratio of government expenditure to GDP in the United Kingdom. They used a data set extending back to 1836; they found instances where displacement may say to have occurred. 9.4. The formulating of the versions of Displacement Effect We tested the Displacement Effect by reversing the Peacock-Wiseman version of Wagners, which are with real GDP (9.1): Table 9.1: The original Version of Peacock-Wiseman with real GDP No Function Version Year 1 L(GE) = ÃÆ'Ã… ½Ãƒâ€šÃ‚ ± + L(GDP) Peacock-Wiseman 1967 Moreover, we will use non-oil sector of Growth Domestic Product (GDP) table (9.2). Table 9.2: The Version of Peacock-Wiseman with real Non-Oil Sector of GDP No Function Version Year 1 L(GE) = ÃÆ'Ã… ½Ãƒâ€šÃ‚ ± + L(Non-Oil GDP) + e Peacock-Wiseman 1967 9.5. The Econometric Methodology and Analysis 9.5.1. Ordinary least square test (OLS) The ordinary Least Square test (OLS), has used to estimate the coefficients in the equations. The Durbin-Watson statistic indicates the absence of the serial correlation among the residuals; the closer the DW statistic and better result are to (2). Test reflects the regression equations ability to determine the dependent variables performance. In contrast, the coefficients of the logarithm model have an interpretation, as elasticises. The logarithm transformation is applicable only when all the observations in the data set are positive. In contrast, the parameters of the logarithm model have an interpretation as elasticises. The logarithm transformation is applicable only when all the observations in the data set are positive. According to, Gujarati (1995), the normal regression model by taking logs of both sides of the equation: Y = ÃÆ'Ã… ½Ãƒâ€šÃ‚ ± + X + e (9.1) To be: Log Y = ÃÆ'Ã… ½Ãƒâ€šÃ‚ ± + Log X + e (9.2) The slope is: Slope = (9.3) The elasticity is: Elasticity = = (9.4) For simplification, E can write as: = (9.5) The normal equation of Peacock and Wiseman version is: GE = f (GDP) f 0 f (9.6) Where: GE = Total Government Expenditure level in real terms. GDP= Gross Domestic Product in real terms. GE = ÃÆ'Ã… ½Ãƒâ€šÃ‚ ± + GDP + e (9.7) The equation by using logarithm model: L (GE) = ÃÆ'Ã… ½Ãƒâ€šÃ‚ ± + L (GDP) + e (9.8) E (Peacock Wiseman) = (9.9) 9.5.1.1. Structural Break Chow Test with Real GDP To find out whether there is a structural break between two periods we divide the observations, we need to calculate the chow test, which is like a F- test, the test statistic from the following formula (9.10): (9.10) The hypotheses tests are: Source RSSc RSS1 RSS2 df Model 9.33377 0.123032 4.65830 1 Residual 0.7410185 0.0067273 0.4136198 110 By using the formula above we can conclude, F-test (1, 110) = 83.914, and the critical value from the F-Table (5%) = 3.92. We have found that since the test F test (1, 110) = 83.914 is greater than the critical F- table = 3.92, we can reject the null hypothesis that there is no structural break and instead accept the alternative hypothesis that there is structural break, It means we have a structural break in the data. Thus, we need to divide the data into tow sup-samples. In the case of Saudi Arabia, we can analysis the Peacock-Wiseman version as: 9.5.1.1.1. Ordinary Least Square (OLS) with Real GDP Peacock-Wiseman (1980Q1 TO 1990Q2) The Peacock and Wiseman version would present as following: L(GE)= 6.43204+ 0.3737 L(GDP) (9.11) (14.44) (8.58) The numbers between parentheses are (t- statistics) for each estimated parameter and intercept. In the equation (9.11), we will get elasticity value directly as (E=0.3737) 0, that means an increase of (99.63%) unit in Government Expenditure (GE) generates a (99.63%) unit increase Gross Domestic Products (GDP). Moreover, the Government Expenditure (GE) explains (65%) change in Gross Domestic Products (GDP) (table (9.3). Table 9.3: Regression results for Peacock Wiseman Version for (OLS) test from (1980Q1) to (1990Q2) with Real GDP Versions D-Variable Constant In-Variable Coefficient R ² Peacock Wiseman L(GE) 6.43204 L (GDP) 0.3737203 0.6480 Peacock-Wiseman (1990Q2 TO 2007Q4) The Peacock and Wiseman version would present as following: L(GE)= 0.554041+ 0.94752 L(GDP) (9.12) (1.51) (27.67) The numbers between parentheses are (t- statistics) for each estimated parameter and intercept. In the equation (9.12), we will get elasticity value directly as (E = 0.94752) 0 , that means an increase of (99.05%) unit in Government Expenditure (GE) Gross Domestic Products (GDP) generates a (99.05%) unit increase Gross Domestic Products (Non-Oil GDP). Moreover, the Government Expenditure (GE) explains (91.8%) change in Gross Domestic Products (GDP) (table 9.4). Table 9.4: Regression results for Peacock Wiseman Version for (OLS) test from (1990Q3) to (2007Q4) with Real GDP Versions D-Variable Constant In-Variable Coefficient R ² Peacock Wiseman L(GE) 0.554041 L (GDP) 0.9475297 0.918 9.5.1.2. Structural Break Chow Test with Real Non-Oil-GDP To find out whether there is a structural break between two periods we divide the observations, we need to calculate the chow test, which is like a F- test, the test statistic from the following formula: The hypotheses tests are: Source RSSc RSS1 RSS2 df Model 8.62851 0.0943032 4.040055 1 Residual 1.44628 0.096802 1.0318692 110 By using the formula above we can conclude, F-test (1, 110) = 30.953, and the critical value from the F-Table (5%) = 3.92. We have found that since the test F test (1, 110) = 30.953 is greater than the critical F- table = 3.92, we can reject the null hypothesis that there is no structural break and instead accept the alternative hypothesis that there is structural break, It means we have a structural break in the data. Thus, we need to divide the data into tow sup-samples. 9.5.1.2.1. Ordinary Least Square (OLS) with Real NON-OIL-GDP Peacock-Wiseman (1980Q1 TO 1990Q2) The Peacock and Wiseman version would present as following: L(GE)=6.664974+0.3234 L(Non-Oil GDP) (9.13) (11.59) (6.24) The numbers between parentheses are (t- statistics) for each estimated parameter and intercept. In the equation (9.13), we will get elasticity value directly as (E=0.3234) 0, that means an increase of (99.68%) unit in Government Expenditure (GE) generates a (99.68%) unit increase Gross Domestic Products (GDP). Moreover, the Government Expenditure (GE) explains (65%) change in Gross Domestic Products (GDP) (table (9.5). Table 9.5: Regression results for Peacock Wiseman Version for (OLS) test from (1980Q1) to (1990Q2) with Real Non-Oil GDP Versions D-Variable Constant In-Variable Coefficient R ² Peacock Wiseman L(GE) 6.664974 L (Non-Oil GDP) 0.32340 0.6480 Peacock-Wiseman (1990Q3 TO 2007Q4) The Peacock and Wiseman version would present as following: L(GE)= -0.568392+ 0.9721244 L(Non-Oil GDP) (9.14) (-0.82) (16.32) The numbers between parentheses are (t- statistics) for each estimated parameter and intercept. In the equation (9.14), we will get elasticity value directly as (E = 0.9721244) 0 , that means an increase of (99.03%) unit in Government Expenditure (GE) Gross Domestic Products (Non-Oil GDP) generates a (99.03%) unit increase Gross Domestic Products (Non-Oil GDP). Moreover, the Government Expenditure (GE) explains (79.7%) change in Gross Domestic Products (Non-Oil GDP) (table 9.6). Table 9.6: Regression results for Peacock Wiseman Version for (OLS) test from (1990Q3) to (2007Q4) with Real Non-Oil GDP Versions D-Variable Constant In-Variable Coefficient R ² Peacock Wiseman L(GE) -0.568392 L (Non-Oil GDP) 0.9721244 0.797 9.5.2. Unit Roots Test Unit Root test aims to examine the properties of time series quarterly data for each of the Government Expenditures (LGE), Gross Domestic Product (LGDP), during the period from (1980q1-1990q2) to (1990q3-2007q4). To test the stationary time series model for the study variables, it requires the unit root test (Enders: 1995). Despite the multiplicity of the unit root tests, but we will use Augmented Dickey-Fuller for stationary Unit Root Tests, through the following equation:   (9.15) Where: = the first difference of the series. is the series under consideration (GDP, government expenditures, or government revenues), t = the time trend. k= the number of lag. is a t is a stationary random error (white noise residual). The hypotheses tests are: If we fail to reject the , then we have a unit root process. On the other hand , if the outcome indicates that the series are stationary after the first difference , in other words , the series integrated of order one I(1) , then we have to proceed to perform a co-integration test. Augmented Dickey-Fuller for stationary Unit Root Tests have used to test for unit roots. If the null hypothesis that the variable contains a unit root cannot be rejected, In this section we have to test the Unit Root Tests for Peacock and Wiseman version for real GDP and Non Oil GDP during two periods, firstly from (1980q1) to (1990q2) and from (1990q3) to (2007q4). Table (9.7) presents the calculated t-value from Augmented Dickey-Fuller for stationary Unit Root Tests on each variable. Table 9.7: Augmented Dickey-Fuller for stationary Unit Root Tests for Real GDP and Non Oil GDP from (1980Q1) to (1990Q2) Variables Augmented Dickey-Fuller for stationary Unit Root Test Statistics L(GDP) -2.725 L(GE) -3.514 L(Non-Oil GDP) -3.426 Critical Values 1% level -2.431 Critical Values 5% level -1.687 Critical Values 10% level -1.305 For the period during (1980Q1 to 1990Q2), according to the result in table (9.7), while all variables under examination are time-series variables, we needed first to test the properties of the series. In order to avoid the problem of spurious regression, each series has tested for stationary. To do so, we apply Augmented Dickey-Fuller for stationary Unit Root Tests, considering 5% level of significance, for the unit root test whether to accept or reject the null hypothesis. However, we found the results of each variable used in Peacock Wiseman version in Saudi Arabia indicate that the series are non-stationary in level but stationary after the first difference. The number of observation is 41 for Saudi Arabia and the following table (9.7) summarize the results of the unit root test for Saudi Arabia. Based on these test it can concluded that all variables tested (LGDP, LGE, LNON OIL GDP) are contained a unit root insignificant level of 5% for Augmented Dickey-Fuller for stationary Unit Root Tests. These results are consistent with the standard theory, which assumes that most macroeconomic variables are not static level, but become stationary in first difference (Enders: 1995).The next step would be to test for co-integration by testing the residual from the co-integrating regression. Table 9.8: Augmented Dickey-Fuller for stationary Unit Root Tests for Real GDP and Non Oil GDP from (1990Q3) to (2007Q4) Variables Augmented Dickey-Fuller for stationary Unit Root Test Statistics L(GDP) -4.199 L(GE) -7.332 L(Non-Oil GDP) -6.301 Critical Values 1% level -4.110 Critical Values 5% level -3.482 Critical Values 10% level -3.169 On the other hand, for the period during (1990Q3 to 2007Q4), according to the result in table (9.8), while all variables under examination are time-series variables, we needed first to test the properties of the series. In order to avoid the problem of spurious regression, each series has tested for stationary. To do so, we apply Augmented Dickey-Fuller for stationary Unit Root Tests, considering 5% level of significance, for the unit root test whether to accept or reject the null hypothesis. However, we found the results of each variable used in Peacock Wiseman version in Saudi Arabia indicate that the series are non-stationary in level but stationary after the first difference. The number of observation is 69 for Saudi Arabia and the following table (9.8) summarize the results of the unit root test for Saudi Arabia. Based on these test it can concluded that all variables tested (LGDP, LGE, LNON OIL GDP) are contained a unit root insignificant level of 5% for Augmented Dickey-Fuller for stationary Unit Root Tests. These results are consistent with the standard theory, which assumes that most macroeconomic variables are not static level, but become stationary in first difference (Enders: 1995).The next step would be to test for co-integration by testing the residual from the co-integrating regression. 9.5.3. Co-integration Test In this section we have to test the Co-integration Test for Peacock and Wiseman version for real GDP and Non Oil GDP during two periods, firstly from (1980q1) to (1990q2) and from (1990q3) to (2007q4). As mentioned previously, the concept of integration common that if the level variables of the form are non-stationary any package of first class, if possible, to generate a linear combination of these variables is characterized by a static zero-class integrated I (0). It is in this case, the  integrated real-time variables of the same rank co-integrated, and thus it can use the level variables in the regression, nor is the decline in this case a false spurious, (Rau, 1994). The null hypothesis is that the variables under investigation are not co-integrated. The rejection of the null hypothesis requires that the trace value of the co-integration test to be greater than at least one of the different critical values. Therefore, failing to reject the null hypothesis of no co-integration leads us to conclude that no relationship in the long-term equilibrium between government spending and national income. Co-integrating test in this study are conducted using the method developed by Johansen (1988), and Johansen and Juselius (1990). Many studies used the Engle Granger two-step, but there are those who used a Johansen and Juselius )1990) , for so many advantages, such as first, that tests for all of the variables and, secondly, all variables are treated as internal variables, so that the choice of the variable is not arbitrary. This procedure is the most reliable test for co-integration. To determine whether stochastic trends in series have related to each other or not, we will test for co-integration in Peacock Wiseman version. In addition, after determining the order of integration by Augmented Dickey-Fuller for stationary Unit Root Tests, we test whether the series are co-integrated or not, and if they are, we have to identify the co-integrating long-run equilibrium relationship (Brooks, 2008). In this section, we have to test the Co-integration Test with (Real GDP) and Co-integrati on Test with (Real Non-Oil GDP). 9.5.3.1. Co-integration Test with (Real GDP) In the case of Real GDP for the period during (1980q1 to 1990q2), table (9.9) shows that co-integration relationship were found and the test support the existence of one co-integration equation in the relationship between LGE and LGDP. By looking at the Trace Statistic value in table (9.9), we conclude that we must reject the null hypothesis of no co-integration in of Peacock Wiseman version with, because the Trace Statistic values are greater than the critical values at 5% levels. The existence of co-integration vector has pointed out by trace test since t-test value exceeds critical value in 5% level of significant. This means the co-integration tests are statistically significant at five percent level for determining the long-run relationship between LGE and LGDP. Otherwise, there is long run equilibrium relationship between Real GDP and Government Expenditures has found in Peacock Wiseman version that the trace tests indicates at 5%. At the Trace Statistic value in table (9.9), we can reject the null hypothesis of co-integration in Peacock Wiseman version with respect to real GDP, because the Trace Statistic values are greater than the critical values at 5% levels. Table 9.9: Johansen Co-integration Test results with (Real GDP) from (1980Q1) to (1990Q2) Versions Hypothesized No. of CE(s) Eigen value Trace Statistic (Long Run) Critical Value 5% Prob Peacock Wiseman None 0.51356   33.2534   15.41   0.0000 At most 1 0.08645   3.79   3.76   0.0000 On the other hand , In the case of Real GDP for the period during (1990q3 to 2007q4), table (9.10) shows that co-integration relationship were found and the test support the existence of one co-integration equation in the relationship between LGE and LGDP. By looking at the Trace Statistic value in table (9.9), we conclude that we must reject the null hypothesis of no co-integration in of Peacock Wiseman version with, because the Trace Statistic values are greater than the critical values at 5% levels. The existence of co-integration vector has pointed out by trace test since t-test value exceeds critical value in 5% level of significant. This means the co-integration tests are statistically significant at five percent level for determining the long-run relationship between LGE and LGDP. Otherwise, there is long run equilibrium relationship between Real GDP and Government Expenditures has found in Peacock Wiseman version that the trace tests indicates at 5%. At the Trace Statistic value in table (9.10), we can reject the null hypothesis of co-integration in Peacock Wiseman version with respect to real GDP, because the Trace Statistic values are greater than the critical values at 5% levels. Table 9.10: Johansen Co-integration Test results with (Real GDP) from (1990Q3) to (2007Q4) Versions Hypothesized No. of CE(s) Eigen value Trace Statistic (Long Run) Critical Value 5% Prob Peacock Wiseman None 0.75275   105.5668   15.41   0.0000 At most 1 0.12419 9.1496   3.76   0.0000 9.5.3.2. Co-integration Test with (Real Non-Oil GDP) In the case of Real Non-Oil GDP for the period during (1980q1 to 1990q2), table (9.11) shows that there is long run equilibrium relationship between Real GDP and Government Expenditures has found in Peacock Wiseman version with respect to real non-oil gross GDP at 5% levels . We can reject the null hypothesis of co-integration in Peacock Wiseman version with respect to real non-oil gross GDP table (9.11), because the Trace Statistic values are greater than the critical values at 5% levels. Table 9.11: Johansen Co-integration Test results with (Real Non-Oil GDP) from (1980Q1) to (1990Q2) Versions Hypothesized No. of CE(s) Eigen value Trace Statistic Critical Value 5% Prob Peacock Wiseman None   0.70444   79.2146   15.41   0.0000 At most 1   0.50992   29.2407   3.76   0.0000 On the other hand , In the case of Real Non-Oil GDP for the period during (1990q3 to 2007q4), table (9.12) shows that there is long run equilibrium relationship between Real GDP and Government Expenditures has found in Peacock Wiseman version with respect to real non-oil gross GDP at 5% levels . We can reject the null hypothesis of co-integration in Peacock Wiseman version with respect to real non-oil gross GDP table (9.12), because the Trace Statistic values are greater than the critical values at 5% levels. Table 9.12: Johansen Co-integration Test results with (Real Non-Oil GDP) from (1990Q3) to (2007Q4) Versions Hypothesized No. of CE(s) Eigen value Trace Statistic Critical Value 5% Prob Peacock Wiseman None   0.73329   158.7948   15.41   0.0000 At most 1   0.62460 67.6036   3.76   0.0000 9.5.4. Causality Test: After making sure of the time series model to study the variables that they are not stationary in the level and stationary in the difference, and then check it all-integrated joint, it is clear that there is a long-term equilibrium relationship. According to, Engle and Granger (1987), the variables that integrate common equilibrium reflects a long-term, it should be a representation of Error Correction Model (ECM), which has the potential to test and assess the relationship in the short and long term between the variables of the form, as it avoids  problems arising from the spurious correlation.   To apply the Error Correction Model (ECM) for Peacock Wiseman version in Saudi Arabia, we must employ Granger-causality as follows: In the context of error correction model (ECM) of the variables that are co-integrated. Standard Granger-Causal for the variables that do not co-integrated. 9.5.4.1. Granger Causality Test The Engle and Granger approach have two phases, the first: Assessing the relationship model equilibrium in the long term, called the decline of joint integration.  The second: an assessment error correction model to reflect the relationship in the short term or short-term volatility on the direction of the relationship in the long run, this model is estimated by the introduction of short-term residuum estimated long-term decline in the independent variable Decelerated for a single. In this section we have to test the Granger Causality for Peacock and Wiseman version for real GDP and Non Oil GDP during two periods, firstly from (1980q1) to (1990q2) and from (1990q3) to (2007q4). 9.5.4.1.1. Granger Causality Test from (1980q1) to (1990q2) with Real GDP Table (9.13) shows the probability values from Granger Causality Test for Peacock and Wiseman Version during periods from (1980q1) to (1990q2) with Real GDP. The reported F-statistics are standard test for the joint hypothesis that LGE does not Granger Cause LGDP. In the case of Saudi Arabia, the probability for accepting the Null-Hypothesis was only 0.1% while 99.9% rejecting this hypothesis, which means LGE, cause LGDP by around 99.9% all the time in Peacock and Wisemans Version. In table (9.13) the feedback of causality from LGDP to LGE has presented where the probability for accepting the Null-Hypothesis was, only 2.8% while 97.2% rejecting the hypothesis, which means LGDP cause LGE by about 97.2% all of them in the case of Saudi Arabia. Table 9.13: Granger Causality test for Peacock and Wiseman Version from (1980q1) to (1990q2) with Real GDP Null Hypothesis F-Statistic Prob. LGE does not Granger Cause LGDP 40.212 0.0010 LGDP does not Granger Cause LGE 7.1809 0.0280 The probability values from Granger Causality Test, table (9.14) present the causality test result from (1990q3) to (2007q4) with Real GDP. The reported F-statistics are standard test for the joint hypothesis that LGE does not Granger Cause LGDP. In the case of Saudi Arabia, the probability for accepting the Null-Hypothesis was only (1%) while 99% rejecting this hypothesis, which means LGE, cause LGDP by around 99% all the time in Peacock and Wisemans Version. In table (9.14) the feedback of causality from LGDP to LGE presented where the probability for accepting the Null-Hypothesis was, only 0.1% while 99.9% rejecting the hypothesis, which means LGDP cause LGE by about 99.9% all of them. Table 9.14: Granger Causality test for Peacock and Wiseman Version from (1990q3) to (2007q4) with Real GDP Null Hypothesis F-Statistic Prob. LGE does not Granger Cause LGDP 115.16 0.010 LGDP does not Granger Cause LGE 48.24 0.001 9.5.4.1.2. Granger Causality Test with Real Non-Oil GDP from (1990q3) to (2007q4) The probability values from Granger Causality Test, table (9.15) present the causality test result from (1980q1) to (1990q2) with (Real Non-Oil GDP). The reported F-statistics are standard test for the joint hypothesis that LGE does not Granger Cause LNON_OIL_GDP. In the case of Saudi Arabia, the probability for accepting the Null-Hypothesis was only 41% while 59% rejecting this hypothesis, which means LGE, cause LNON_OIL_GDP by around 59% all the time in Peacock and Wisemans Version. In table (9.15) the feedback of causality from LNON_OIL_GDP to LGE presented where the probability for accepting the Null-Hypothesis was, only 0.1% while 99.9% rejecting the hypothesis, which means LNON_OIL_GDP cause LGE by about 99.9% all of them. Table 9.15: Granger Causality test for Peacock and Wiseman Version from (1980q1) to (1990q2) with (Real Non-Oil GDP) Null Hypothesis F-Statistic Prob. LGE does not Granger Cause LNON_OIL_GDP 1.7821 0.410 LNON_OIL_GDP does not Granger Cause LGE 32.534 0.001 The probability values from Granger Causality Test, table (9.16) present the causality test result from (1990q3) to (2007q4) with (Real Non-Oil GDP). The reported F-statistics are standard test for the joint hypothesis that LGE does not Granger Cause LNON_OIL_GDP. In the case of Saudi Arabia, the probability for accepting the Null-Hypothesis was only 0.9% while 99.1% rejecting this hypothesis, which means LGE, cause LNON_OIL_GDP by around 99.1% all the time in Peacock and Wisemans Version. In table (9.16) the feedback of causality from LNON_OIL_GDP to LGE presented where the probability for accepting the Null-Hypothesis was, only 0.2% while 99.8% rejecting the hypothesis, which means LNON_OIL_GDP cause LGE by about 99.8% all of them. Table 9.16: Granger Causality test for Peacock and Wiseman Version from (1990q3) to (2007q4) with (Real Non-Oil GDP) Null Hypothesis F-Statistic Prob. LGE does not Granger Cause LNON_OIL_GDP 9.5193 0.009 LNON_OIL_GDP does not Granger Cause LGE 40.708 0.002 9.5.4.2. Error Correction Model (ECM) The Error Correction Model (ECM) differs as discussed by Granger (1988) for the number of error correction terms. The concept of error correction is related to co-integration because the co-integration relationship describes the long run equilibrium. If a set of variables are co-integrated, then there exists an error correction model to describe the short run adjustment to equilibrium Engle and Granger (1987). The incidence of mutual co-integration between the variable indicates that the Granger must be Causal in one direction, at least, but the rules of engagement did not refer to the direction of causality between the variables. To verify the rules of engagement we are conducting tests of causation in the context of Error Correction Model (ECM). With regard to periods of lag length, and use the same lag length for Peacock Wiseman version, which we were when we tested for co-integration. In addition, the t-statistics on the coefficients of the lagged error correction term (ECTt-1 (indicate the significance of the long-run causality between the two variables. The statistical significance of the t-statistics is in our tests should be at most 5% level. These analyses regarded as usual analyses of the displacement hypothesis and the hypothesis that the part of national income constant to government expenditure increases with income (Gupta 1967, Diamond 1977, Nomura 1991, 1995). Moreover, Peacock and Wiseman agree with Wagners version of Wagners law. In this section we have to test The Error Correction Model (ECM) for Peacock and Wiseman version for real GDP and Non Oil GDP during two periods, firstly from (1980q1) to (1990q2) and from (1990q3) to (2007q4). 9.5.4.2.1. Error Correction Model (ECM) from (1980q1) to (1990q2) with (real GDP) In the table (9.17), the results from (1980q1) to (1990q2) indicate that there is long-run unidirectional causality that runs from GDP to GE (Peacock Wiseman Version). We draw this conclusion because the sign for GE is positive, and at the same time, the coefficient is statistically significant at the 5%level, Thus, Peacock Wiseman version has found to hold for GDP in the case of Saudi Arabia. Table 9.17: Causality with Error Correction Model (ECM) test from (1980q1) to (1990q2) with (Real GDP) Versions Variables ECTt-1 T-Stat Peacock Wiseman L(GE) 0.0094398 7.69 L(GDP) 0.1036134 1.56 In the table (9.18), the results from (1990q3) to (2007q4) indicate that there is long-run unidirectional causality that runs from GDP to GE (Peacock Wiseman Version). We draw this conclusion because the sign for GE, positive, and at the same time it coefficient is statistically significant at the 5%level, while the signs for GDP is either positive, and/or the coefficient is statistically insignificant at the 5% level. Thus, Peacock Wiseman version has found to hold for GDP in the case of Saudi Arabia. Table 9.18: Causality with Error Correction Model (ECM) test from (1990q3) to (2007q4) with (Real GDP) Versions Variables ECTt-1 T-Stat Peacock Wiseman L(GE) 0.0086968 9.99 LGDP 0.2657211 3.53 9.5.4.2.2. Error Correction Model (ECM) from (1990q3) to (2007q4) with (real Non-Oil GDP) In the table (9.19), the results from (1980q1) to (1990q2) indicate that there is long-run unidirectional causality that runs from Non-Oil-GDP to GE (Peacock Wiseman Version). We draw this conclusion because the sign for GE is positive, and at the same time, the coefficient is statistically significant at the 5%level, Thus, Peacock Wiseman version has found to hold for Non-Oil-GDP in the case of Saudi Arabia. Table 9.19: Causality with Error Correction Model (ECM) test from (1980q1) to (1990q2) with (Real Non-Oil GDP) Versions Variables ECTt-1 T-Stat Peacock Wiseman L(GE) 0.0101674 1.73 L(Non-Oil GDP) 2.316124 6.43 In the table (9.20), the results from (1990q3) to (2007q4) indicate that there is long-run unidirectional causality that runs from Non-Oil-GDP to GE (Peacock Wiseman Version). we draw this conclusion because the sign for GE, is incorrect, negative, and at the same time it coefficient is statistically significant at the 5%level, while the signs for Non-Oil-GDP is either positive, and/or the coefficient is statistically insignificant at the 5% level. Thus, Peacock Wiseman version has found to hold for Non-Oil-GDP in the case of Saudi Arabia. Table 9.20: Causality with Error Correction Model (ECM) test from (1990q3) to (2007q4) with (Real Non-Oil GDP) Versions Variables ECTt-1 T-Stat Peacock Wiseman L(GE) -0.0037897 -3.42 L(Non-Oil GDP) 0.8948453 6.33 9.6. Conclusion According to, (Gupta, 1967) and (Diamond 1977,) argued that the displacement effect led to the share of national income devoted to government expenditures increases with GDP. In this chapter, we examined the relationship between the expenditures and economic growth of Peacock Wiseman version for Saudi Arabia by using time series quarterly data for the periods during (1980Q1 to 1990Q2) and during (1990Q3 to 2007Q4) . It has applied three distinct time series techniques. We have examined the regressions for Peacock Wiseman version by using Ordinary Least Square (OLS) for Real GDP and Non Oil GDP. The displacement literature surveys have shown that the earlier empirical tests of displacement suffer from several methodological compare between the studies has impaired by different choices of periods and data series. The next step is the Unit Root tests by using the Augmented Dickey-Fuller for stationary Unit Root Tests for Real GDP and Non Oil GDP, also we have used Co-integrating test for Real GDP and Non Oil GDP. Finally, Causality tests by using Granger causality tests and Error Correction Model (ECM). First, the regressions for Peacock Wiseman version by using Ordinary Least Square (OLS) for Real GDP and Non Oil GDP, to presents the probability of the equations and to analysis the R-square and DW, for Peacock Wiseman version. Second, The Unit Root tests by using the Augmented Dickey-Fuller for stationary Unit Root Tests for Real GDP and Non Oil GDP. In the case of the levels of the series, the null-hypothesis of non-stationary cannot reject for any of the series. Third, these results suggest that there is a co-integrating relationship between the share of government spending in national output and per capita income. In this situation, if co-integration exists between government expenditure and GDP, then Peacock Wiseman version holds. The equilibrium relationship indicates that the major determinant of government expenditure in Saudi Arabia, in the long run, is national income. Fourth, Granger causality tests have used to confirm the causality direction between the Variables. In the long run we found statistically significant evidence in favour of per capita GDP Granger-causing the share of government Expenditures in GDP, which is consistent with Peacock Wiseman version. The result of causality test indicate that the existence of strong feedback causality for Peacock Wiseman version in the long run. On the other hand, by using Error Correction Model (ECM), the concept of error correction, this has related to co-integration because the co-integration relationship describes the long run equilibrium. If a set of variables are co-integrated, then there exists an error correction model to describe the short run adjustment to equilibrium. Overall studies with the exception of Pryor (1968), the time dimension has completely suppressed, despite the fact that the Peacock and Wiseman hypothesis purports to explain the development of government expenditure over time.