
Research Article


To Visualize Relationship Between Economic Growth, Government Spending and Money Supply: Evidence from Turkey 

Zahoor Hussain Javed
and
Ahmet Sahinoz



ABSTRACT

This study investigate relationships between economic growth, government spending and money supply over the period 1992:1 to 2003:3. Present findings investigate the directional and undirectional causality between economic growth and government spending in a bivariate system. On other hand, EngleGranger and PhillipsOuliaris tests shows no cointegration between economic growth and government spending. However, by the introduction of a third variable, money supply; in multivariate system the investigation indicates a directional causality from economic growth, government spending and money supply in the long run, with existence of undirectional causality from expenditure to money supply in the same period. Therefore, with exception to Johansen test, the EngleGranger test and PhillipsOuliaris show weak evidence for cointegration between economic growth, government spending and money supply. Nevertheless, present results show that economic growth is more volatile than the government spending and money supply, over the period under consideration, which is particularly important for policymaker in that it is the macroeconomic policy of government that affects growth and influences the money supply and government spending.





INTRODUCTION The purpose of the present study was to the examine relationship among economic growth (GNP) and government spending by introducing money supply as a third economic variable which may have significant effect on the causality and longrun relationship of economic growth and government spending in the Turkish economy by using causality tests of Granger^{[1]}, EngelGranger^{[2]}, PhillipsOuliaris^{[3]} and Johansen’s^{[4]} cointegration method. At first glance, we note the relationship between economic growth and government spending, then evaluate the relationship between economic growth, government spending and money supply. It is possible that expansion of government spending is proceeded by an increase in economic growth, with the initial effect felt largely within the national sector and only subsequently on the aggregate economy. If markets are interdependent, disturbances in one market will be transmitted to other markets. However, there have been few empirical studies of relationships between the government spending and the economic growth. The issue of concern herewith is whether the expenditure, money supply and the economic growth are segmented or interdependent and whether relationship between economic growth, government spending and money supply does exist.
Recent empirical investigation of the relationship between economic growth
and government spending has been conducted using the bivariate causality tests
of Granger^{[1]} and Sims^{[5]}. Keran^{[6]} suggested
that monetary influences had stronger, more predictable and faster impact on
economic growth than expenditure influences. Keran^{[7]} again argued
in empirical evidence that monetary influences had a stronger, more predictable
and faster impact on economic activity than expenditure influences in United
State. Andersen and Jorden^{[8]} suggested that changes in government
spending financed by monetary expansion are reflected in changes in the monetary
base and in the money stock and money that influence monetary actions on economic
activity is more certain than fiscal actions (expenditures). Henrekson^{[9]}
and Murthy^{[10]} have argued that the economic growthgovernment spending
link is a long term behaviourial relationship that should be tested on the basis
of their comovement over time. In this context, a number of empirical studies
examine the long run equilibrium condition between economic development and
public spending using the cointegration techniques developed by Engle and Granger^{[2]},
Johansen^{[4]}. Recently, Levine and Renelt^{[11]} using extreme
bound analysis, demonstrate that the relationship between economic growth and
some of its determinants was found to be fragile when using crosssectional
country data. Dacy^{[12]} find evidence for the existence of a long
run link between national income and government spending while Ashworth^{[13]}
and Hayo^{[14]} find little or no support for relationship between income
and expenditure. Ashworth^{[13]} has suggested that differences in the
time series properties of the underlying data and the choice of the estimation
procedure used may help explain diversity in cointegration results. Oxley^{[15]}
find evidence that long run relationship exist between economic growth and government
spending in Britain. On the whole, impact studies using time series support
our hypothesis, while causality investigations generate mixed results which
range from no causal relationship to bidirectional causality. Ahsan et al.^{[1619]}
have pointed out that the use of an additional fiscal or monetary variable can
change the causal relationship between public expenditure and national income.
More recently, Kennedy^{[20]} has suggested that cointegration tests
should fail to find cointegration among the variables under consideration if
a relevant variable is omitted from the analysis. This analysis of the long
run equilibrium condition between economic growth, government spending and money
supply involves three steps. This study attempts to fill this gap in the literature,
with special attention being paid to the how the long run relationship between
economic growth, government spending and money supply existence. Not withstanding,
our empirical analysis attempts to shed light on the channels through which
how economic growth, government spendings and money supply accelerate in Turkish
economy. Comparing present results with the earlier studies we then try to examine
in more detail the potential channels through which, how, economic growth, government
spendings and money supply work for Turkish economy with tools of Granger’s^{[1]}
causality, EngleGranger^{[2]}, PhillipsOuliaris^{[3]} and
Jhonsen’s^{[4]} cointegration tests.
MATERIALS AND METHODS We employ Granger’s^{[1]} causality, EngleGranger^{[2]}, PhillipsOuliaris^{[3]} and Jhonsen’s^{[4]} cointegration tests to find the interrelationship between economic growth, government spending and money supply. Firstly, we employ Granger’s^{[1]} causality test to analyse the interrelationship between economic growth and government spending. We use two sets of test to investigate relationship between the said variables. First, following Hashemzadeh et al.^{[21]}, we test the bivariate causality relationship between every pair of variables using the standerd twovariable approach, as described below:
Where, GNP is economic growth measured as percentage change in the same quarter
of the pervious year and EXP is government spending measured as percentage change
in the same quarter of the pervious year. Bivariate causality implies that EXP
is Grangercausing GNP growth provided that v_{t} is assumed as uncorrelated
disturbances in Eq. 1. Similarly, GNP is Grangercausing EXP
if η_{t }is assumed as uncorrelated disturbances in Eq.
2. If both of the above events occur, then feedback effects exist. The hypothesis
that government spending causes economic growth, if supported by the data, should
imply that the
null hypothesis of should be rejected by the calculated Fvalue when X is excluded
in the restricted form of Eq. 1. If there is bidirectional
causality then
and For
the bivariate analysis the Fvalue is calculated as:
Where, ESS_{R }and ESS_{U} are the sum of squared residuals for the constrained and unconstrained , Where ESS_{R }and ESS_{U} are sum of the squared residuals for the constrained and unconstrained causality regressions, respectively. n is the total number of observations and m is the number of lags per variable.
The second test examines the joint effect of two variables on the third variable.
The joint trivariate causality model is described as under:
Table 1: 
Unit root and stationarily tests 

Critical values for the ADF statistics from Fuller Table 8.5.2.
p:373.1976. ^{a} denotes 1% (3.58), ^{b} denotes 5% (2.93),
^{c} denotes 10% (2.60) 
In the trivariate causality system, we introduced a third variable M_{t
}money supply measured as percentage change in the same quarter of the
pervious year. The hypothesis being tested with Eq. 46
are, Whether EXP and M jointly cause GNP after controlling GNP’s own lags;
Whether GNP and M jointly cause EXP after controlling EXP’s own lags; Whether
EXP and GNP jointly cause M after controlling M’s own lags. Though questions
about optimal lags are raised in the literature. Jones^{[22]} demonstrates
that ad hoc methods for determining the lags to use in Granger’s causality
test performed better than some of statistical methods used to search for optimal
lags. Earlier, Thornton and Batten^{[23]} also found the final prediction
method to be a better technique for determining the optimal lags. Thus, the
issue of the best statistical method to use in determining the optimal lag is
unresolved. We, therefor, estimated Eq. 1, 2
and 46 assuming four lags for each variable.
The Fvalue for the trivariate causality test is calculated as:
Where the variables are as defined in Eq. 3. Unit root tests: The casual tests of Granger^{[1]} and Sims^{[5]} are essentially tests of the predictive ability of timeseries models. However, Granger causality tests require the use of stationary timeseries data^{[24,25]}. BhamaniOskooee et al.^{[26]} also use the bivariate causality tests of Granger to investigate relationship between export revenue and economic growth. Under current practice, developing such data requires testing of the observed data series for unit roots called DickeyFuller (DF) and augmented DickeyFuller (ADF)^{[27]} tests. Consequently, before testing the cointegrating regressions as given above, we need to examine the stationarily of respective time series. For this purpose, we test each series by well known Augmented DickeyFuller unit root test. Where, ΔY_{t }= Y_{t }Y_{t  1}, ε_{t} is the error term and δ is chosen to ensure serially uncorrelated residuals. The variables containing in Table 1 are examined for stationary. Data: Quarterly data on GNP growth, government spending (EXP) and money supply (M) of Turkish economy for the 1992:1 to 2003:3 period were taken from the various issues on Economic and Social Indicators of State Planning Organization and International Financial Statistics. RESULTS AND DISCUSSION
Table 2 reports the results of the bivariate analysis which
suggests that the undirectional positive causality is from GNP to EXP since
the estimated F value is statistically significant. Table 3
is based on three samples. Each sample is analysed for the total sample period
1992:12003:3 and two subperiods, 1992: 1997:4 and 1998: 2003:3. Under the consideration
period 1992:12003:3 trivariate analysis which finds undirectional and positive
causality between money supply and economic growth after excluding government
expenditure.
Table 2: 
Total sample: Bivariate analysis of causal relationship among
GNP growth and expenditure (EXP) for the 1992:12003:3 

Critical Fvalues 1%= 3.32, 5%= 2.37, 10%= 1.94, df (2, 138) 
Table 3: 
Total sample: Trivariate analysis of causal relationship
among GNP growth, spending (EXP) and money supply(M) for the 1992:12003:3 

Critical Fvalues 1%= 3.32, 5%= 2.37, 10%= 1.94, df (2, 138).
* Significant at the 1% level, indicating that there is a significant causal
relationship.
** significant at the 5% level, indicating that there is a significant causal
relationship.*** significant at the 10% level, indicating that there is
a significant causal relationship. EXP(M) → GNP ^{a }is interpreted
as EXP and M jointly cause GNP after excluding M. M(EXP) → GNP ^{b
}is interpreted as M and EXP jointly cause GNP after excluding EXP.
Figures on top are the p values, while those in parentheses below are the
Fvalues represented 
Table 4: 
Test for cointegration in the Model (GNP, EXP) 

^{a}Regression of EXP on GNP with a lag length of
0 chosen for unit root test. ^{b}Regression of GNP on EXP with a
lag length of 0 chosen for unit root test.
Johsen statistics computed from a VAR model in level specified with 3 lags
and a constant term 
Table 5: 
Test for cointegration in the model (GNP, EXP, M) 

The results of Johansen’s ^{[4]} trac statistics
are reported in Table 4 and 5. The first
row of numbers in parentheses gives the asymptotic significance level (p
values) estimated in MacKinnon et al. ^{[28]}. The second
row in parentheses gives the significance level estimated with the bootstrapped
simulations. ^{a}Regression of GNP on EXP and M with a lag length
of 2 chosen as for unit root test. ^{b}Regression of EXP on GNP
and M with a lag length of 2 chosen as for unit root test. ^{c}Regression
of M on GNP and EXP with a lag length of 2 chosen as for unit root test.
Johsen statistics computed from a VAR model in level specified with 3 lags
and a constant term 
In the same period empirical evidence suggests that bidirectional and positive
relationship between economic growth and money supply after excluding government
spendings. On the other hand during this period analysis shows undirectional
and positive relationship between government spendings and money supply after
excluding economic growth. Nevertheless, we look into subperiod 1992:1 to 1997:4,
which provide evidence of undirectional and positive causality between government
expenditures and money supply afer excluding economic growth. Not withstanding,
over the subperiod 1998:1 to 2003:3 empirical analysis addresses evidence of
bidirectional and positive causality from economic growth and government expenditure
after excluding money supply and undirectional and positive causality from money
supply and government expenditure after excluding economic growth. Table
4 suggest the absence of cointegration between economic growth and government
spendings. The EngleGranger and PhillipsOuliaris tests accept the null hypothesis
of no cointegration. Nevertheless, the Johansen test results are inconsistent.
The λ–Max test statistic accepts the null hypothesis of cointegration
and trace statistic suggests that each of the variable is stationary. Lastly,
tests for the existence of cointegrating relationship among economic growth,
spendings and money supply are shown in Table 5. The EngleGranger
test shows weak evidence for cointegration, with test statistic results of the
regressions showing only one regression is significant at the 5% level, while
PhillipsOuliaris test also show weak cointegration which shows only one regression
is significant at 10% level. Lastly, Jhonsen’s^{[4]} cointegration
test results emphasis that economic growth, government spending and money supply
accelerate in the same direction and shows robust relationship between the parameters.
CONCLUSIONS
The main object of the research reported on here to provide some assessment
relationship of economic growth, government spending and money supply. Empirical
studies of economic growth and government spending analysis which reports that
the undirectional positive causality is from economic growth to expenditure.
The main shortcoming of the bivariate causality analysis is omission of other
relevant variables. The empirical evidence emphasis that there is joint feedback
effect between economic growth and government spendings. However, The EngleGranger
and PhillipsOuliaris tests accept the null hypothesis of no cointegration,
while the Johansen test results are inconsistent with these results At the introduction
of a third variable money supply; within trivariate causality analysis of economic
growth and government spending for Turkish economy. The multivariate causality
empirical evidence shows that bidirectional and positive causality between economic
growth and money supply after excludings government spendings, while undirectional
and positive causality between government spendings and money supply after excludings
economic growth in the same period. However, in the 1998:12003:3, the structural
adjustment programmes which removed some of the economic distortions and increased
economic growth, government spending and money supply in same direction with
slightly having different magnitude. EngleGranger and PhillipsOuliaris empirical
shows weak cointegration between these variables, while Johonsen’s cointegration
test investigate, robust cointegration relationship between economic growth,
government spending and money supply. To round out the discussion, Turkish economy
can gain more economic growth and can remove economic distortions with tools
of constant growth rate of money supply, reduction of unproductive expenditures
and implementation of structural adjustment programmes.

REFERENCES 
1: Granger, C.W.J., 1969. Investigating causal relations by econometric models and crossspectral methods. Econometrica, 37: 424438. Direct Link 
2: Engle, R.F. and C.W.J. Granger, 1987. Cointegration and error correction: Representation, estimation and testing. Econometrica, 55: 251276. CrossRef  Direct Link 
3: Phillips, P. and S. Ouliaris, 1990. Asymptotic properties of residual based tests for cointegration. Econometrica, 1: 165193.
4: Johansen, S., 1991. Estimation and hypothesis testing of cointegration vectors in gaussian vector autoregressive models. Econometrica, 59: 15511580. CrossRef  Direct Link 
5: Sims, C.A., 1972. Money, income and causality. Am. Econ. Rev., 62: 540552. Direct Link 
6: Keran, W.M., 1969. Monetary and fiscal influences on Economic Activity: The Historical Evidence. 1st Edn., Fedral Reserve Bank, St. Lious, pp: 523.
7: Keran, W.M., 1970. Monetary and Fiscal Influences on Economic Activity: The Foreign Experince. 1st Edn., Fedral Reserve Bank, St. Lious, pp: 1628.
8: Andersen, C.L. and L.J. Jorden, 1968. Monetary and Fiscal Action: A Test of their Relative Importance in Economic Stabilization. 1st Edn., Fedral Reserve Bank, St. Lious, pp: 1123.
9: Henrekson, M., 1973. Wagner`s law: A spurious relationship? Public Finance, 48: 406415.
10: Murthy, N.R.V., 1993. Further evidence of Wagner`s law for Mexico: An application of cointegration analysis. Public Finance, 48: 9296.
11: Levine, R. and D. Renelt, 1992. A sensitivity analysis of crosscountry growth regressions. Am. Econ. Rev., 82: 942963. Direct Link 
12: Dacy, D.C., 1975. Foreign and government consumption, savings and growth in less developed countries. Econ. J., 65: 548561. Direct Link 
13: Ashworth, J., 1994. Spurious in Mexico: A comment on Wagner`s law. Public Finance, 49: 282286.
14: Hayo, B., 1994. No further evidence of Wagner`s law for Mexico. Public Finance, 49: 287294.
15: Oxley, L., 1994. Cointegration, causality and Wagner`s law: A test for Britain 18701913. Scottish J. Political Econ., 41: 286298. Direct Link 
16: Ahsan, S.M., A.C.C. Kwan and B.S. Sahni, 1989. Causality between government consumption expenditure and national income: OECD countries. Public Finance, 44: 204224.
17: Ahsan, S.M., A.C.C. Kwan and B.S. Sahni, 1992. Public expenditure and national income causality: Further evidence on the role of omitted variables. Southern Econ. J., 58: 623634. Direct Link 
18: Ahsan, S.M., A.C.C. Kwan and B.S. Sahni, 1990. Public expenditure and national income growth: US 195385. Public Finance and Steady Economic Growth, pp: 176178.
19: Ahsan, S.M., A.C.C. Kwan and B.S. Sahni, 1996. Cointegration and Wagner`s hypothesis: Time series evidence for Canada. Applied Econ., 28: 10551058. Direct Link 
20: Kennedy, P., 1998. A Guide to Econometrics. 4th Edn., MIT Press, Cambridge, MA., pp: 225239.
21: Hashemzadeh, N. and P. Taylor, 1988. Stock prices, money supply and interest rates: The question of causality. Applied Econ., 20: 16031611.
22: Jones, D.J., 1989. A comparison of laglenth selection techniques in test of Granger causality between money growth and inflation: Evidence for the US, 195986. Applied Econ., 21: 809822. CrossRef  Direct Link 
23: Thornton, D.L. and B.S. Dallas, 1985. Laglenth selection and test of Granger causlity between money and income. J. Money Credit Bank., 17: 164178.
24: Granger, C.W.J. and P. Newbold, 1974. Spurious regressions in econometrics. J. Econ., 2: 111120.
25: Ong, S.E., 1994. Structural and vector autoregressive approaches to modelling real estate and property stock prices in Singapore. J. Property Finance, 5: 418. CrossRef  Direct Link 
26: BahmaniOskooe, M., H. Mohtadi and G. Shabsigh, 1991. Exports, growth and causality in LDC`s: A reexamination. J. Dev. Econ., 36: 405415. Direct Link 
27: Dickey, D.A. and W.A. Fuller, 1979. Distribution of the estimators for autoregressive time series with a unit root. J. Am. Stat. Assoc., 74: 427431. CrossRef  Direct Link 
28: MacKinnon, J.G., A.A. Hang and L. Michelis, 1999. Numerical distribution of likelihood ratio test for co integration. J. Applied Econ., 14: 563577. Direct Link 



