Does Fiscal Policy Stimulate Growth? The case of Zambia a Small Open developing Country. Spyton Phiri Advisor: Prof. Takashi Fukushima MEF07103 Summary Is government involvement vital for economic growth? This question has surprisingly dominated both theoretical and empirical debate amongst economists for a long time. In the theoretical literature, some schools of thought believe that government involvement in economic activity is essential for growth, whereas those of an opposing thought retains that government operations are inherently bureaucratic and inefficient and therefore holds back rather than stimulate growth. In the empirical literature, equally mixed results abound. The objective of this paper henceforth is not to reach a decision on the remarkable debate but to add to the fiscal policy-growth knowledge of works by examining the case of Zambia, a small open developing country. Time series techniques were used to investigate the relationship between various measures of fiscal and non-fiscal policy on growth on annual data for the period 1964 – 2007. Classifying total government expenditure into productive and unproductive expenditures, it was found that unproductive expenditure was neutral to growth as predicted by economic theory. Further it was found that productive government expenditure and private investment have a strong beneficial effect on growth in the long run. Hence these results should prove useful to policy formulators in Zambia in devising expenditure policies to make certain that unproductive expenditures are cut short and productive expenditures are enhanced while at the same time boosting private investment by putting in an enabling environment. 1.00 Introduction Advocates of government interference in economic activity maintain that such interference can result into long term growth. They quote government’s role in ensuring efficiency in resource allocation, stabilization of the economy and general regulation of markets as some of the methods in which government could expedite economic growth. On the other hand, antagonists hold the opinion that government operations are inherently bureaucratic and inefficient and consequently hold back rather than stimulate growth. It therefore then appears like whether government’s fiscal policy stimulates or holds back growth still remains a big empirical question to be resolved. Nonetheless, the current empirical literature is mixed, with some researchers finding the relationship between fiscal policy and growth as being positive meaning that fiscal policy stimulates growth, others as negative meaning that fiscal policy does not stimulate growth and still others with results as, not precisely determined. The objective of this paper henceforth is not to reach a decision on the remarkable debate but to add to the literature by examining the effects of fiscal policy on growth in a small developing economy, Zambia. Barro (1990) and Kneller et al (1999) as cited by M’Amanja and Morrissey provide a theoretical ground for, as well as empirical evidence of the beneficial effect of productive government expenditure and the harmful effect of taxation. In this paper, total government expenditure has been classified into productive and unproductive expenditures.
Time series techniques with incorporation of two time lags on annual time series data covering the period 1964 – 2007 was used to carry out this analysis for the case of Zambia. Zambia has had a mixed economic performance since independence and therefore it would be interesting to know the role of fiscal and related variables over the period in question. When Zambia became independent in 1964, its rich endowment of copper seemed to guarantee a bright future. However, state dominated economic policies coupled with falling copper revenues, led to plummeting standards of living. By the early 1990’s, Zambia was among the most heavily indebted and poorest countries in Africa. Following elections in 1991, the new government embarked upon a far reaching economic reform program. Despite substantial achievements in liberalization and establishing a market based economy, sustained growth has remained elusive. Positive growth rates in 1996 and 1997 were followed by a 2% decline in GDP in 1998. The big question to Zambian policy makers and indeed many other observers of the economy is what has been going wrong? And what could be the answer for Zambia’s economy if any? It will therefore be attempted in this paper to explain some of the causal factors. 2.00 Theoretical issues and empirical evidence 2.10 Theoretical issues According to literature available on endogenous growth theory, fiscal policy carried out by the government can affect both the level and growth rate of per capita output. An illustration in much details of the mechanism through which fiscal policy impacts on growth can be found in, among others, Barro (1990), Aghion and Durlauf (2005) and Barro and Sala-i-Martin (1995). These authors employ a Cobb-Douglas type production function with government provided goods and services (g) as an input to show the positive effect of productive spending and the adverse effects associated with distortionary taxes. The production function, in per capita terms can be given as follows, y=Ak1-αgα (1) Where y is the per capita output, k is per capita private capital and A is a productivity factor. If its assumed that the government runs a balanced budget during each period by raising a proportional tax on output at rate (τ) and lump sum taxes (L), the government budget constraint then can be expressed as, Ng+C=L+τny (2) where n is the number of the producers in that economy and C is the government consumption which is assumed to be unproductive. Theoretically, it has been found that a proportional tax on output affects the private incentives to invest, but a lump sum tax does not. Therefore, subject to a specified utility function, Barro (1990) and Barro and Sala-i-Martin (1995) derive the long run growth rate (γ) in the model as below, γ=λ (1-τ) (1-α) A1/(1-α) (g/y)α/(1-α) –μ (3) where λ and μ stand for parameters in the assumed utility function. From equation (3), it can be seen that the growth rate is a decreasing function of the distortionary tax rate (τ) and an increasing function of the productive government expenditure (g). It can also be seen that the growth rate is not affected by both the non-distortionary taxes (L) and the unproductive government expenditure (C). This specification above assumes that the government balances its budget in each period. The empirical model to be adopted therefore follows Kneller et al (1999) and Bleaney et al (2000) as cited in M’Amanja and Morrisey (2005) in which a more practical view is taken by assuming a non-balancing government budget constraint
in some periods. Equation (3) can therefore be re-written when this fact has been taken into account to hence obtain the following below expression, ng+C+b=L+τny (4) where b is the budget deficit or surplus in a given period. Since “g” is taken to be productive, its predicted sign will therefore be positive, but τ is negative because it distorts incentives of the private agents. But both C and L are hypothesized to have zero effects on growth. In the same manner, the effect of b is expected to be zero so long as Ricardian equivalence holds, but may be nonzero otherwise (Bleaney et al, 2000) as cited in M’Amanja and Morrisey (2005). The growth equation to be adopted in this paper is specified in the spirit of Kneller et al (1999) as cited in M’Amanja and Morrisey (2005) by considering both fiscal variables (Xit) and non-fiscal (Zit) variables so that the growth equation becomes, yt = α + βiZit + γjXjt +εit (5) where yt is the growth rate of output, X is the vector of fiscal variables, Z is the vector of non-fiscal variables, and εit are the white noise error terms. Its interesting to note that, if the budget constraint is fully specified, then γjXjt = 0 because expenditures must balance the revenues. For this to be avoided there is need to omit at least one element of X (say Xm) to avoid perfect collinearity (Kneller et al, 1999 as cited in M’Amanja and Morrisey, (2005)). But obviously, the omitted element must be that which theory suggests has neutral effect on growth, for if any other is selected will result in the introduction of substantial bias in the parameter estimates. Hence, equation (5) can be re-written as follows, yt = α + βiZit + γjXjt +γmXmt +εit (6) From equation (6), Xmt can then be omitted to obtain the final growth equation given as below, yit = α + βiZit + (γj-γm)Xjt + εit (7) Equation (7) therefore, as specified in Kneller et al (1999) as cited in M’Amanja and Morrisey (2005), constitutes the main idea of the model to be estimated. When specified in this manner, the interpretation of the coefficients of the fiscal variables should be seen in terms of implied financing. This therefore means that, the null hypothesis to be tested is (γj-γm) =0 instead of the conventional null that γj = 0. Thus accordingly, the interpretation of the coefficient of the fiscal variables is the “Effect of a unit change in the relevant variable offset by a unit change in the element omitted from the regression” (Kneller et al (1999) as cited in M’Amanja and Morrisey (2005). If it happens that the null is rejected, then more parameter estimates can be obtained if the neutral elements are eliminated from the model. According to literature that is available currently, there is no growth model that is generally accepted with regard to what factors are to be included in the growth equation. Therefore, those fiscal variables that are found to have neutral effect on growth as stated above are to be dropped. In the formulation of the variants of the growth equation (7), a model is estimated in which all the fiscal variables (except budget deficit because it’s an identity and is assumed to have no long term growth effect but is likely to have adverse short run effects) are included. Next, unproductive government expenditure is dropped from the equation (7) while retaining all the other expenditure and revenue items. To follow in line, is the dropping of the tax revenue item, but retaining all the other variables including unproductive expenditure and test for zero coefficient of the other neutral element (i.e. unproductive expenditure). Theoretically, the neutral elements of fiscal policy should be insignificant in the model and therefore in the final specification that is to be carried out, are to be dropped. This is because the expectation, based on literature and theory, is that, these
neutral fiscal variable(s) would have no effect on the long run growth. If at all, it happens that the null is not rejected, then the final model should yield more precise parameter estimates, with lower standard errors of the remaining fiscal variables. 2.20 Empirical evidence. The literature tracing the effects of fiscal variables on growth is abundant. It will not be attempted here to summarise it, noting the summaries in Sala-i-Martin (2002), Temple (1999), Kenny and Williams (2001), and Easterly and Levine (2001) all as cited in Aghion & Durlauf, (2005). All these research papers have at least one common feature, in that, they all found that some indicator of national policy is strongly linked with economic growth. Furthermore, the last two decades have seen an increase in empirical research targeted at resolving the relationship between various measures of fiscal variables and economic growth. In all these attempts at the subject, crosssectional, panel, and time series data have all been used. Therefore the efforts to point out the fiscal variables-growth relationship are undermined by conceptual, statistical and estimation concerns. Not surprisingly, empirical findings have differed from one another. For example, Nijkamp and Poot (2002) as cited in M’Amanja and Morrisey (2005) conducted a meta-analysis of past empirical studies of fiscal policy and growth and found that in a sample of 41 studies, 29% indicated a negative relationship between fiscal policy and growth, 17% a positive one, and 54% an inconclusive relationship. The great contributory factor to these varied empirical results is the measure used to proxy for fiscal policy. The other problems contributing to the mixed results include the use of different model specifications and estimation techniques, sample sizes, quality of the data, and limited availability of the data on certain relevant variables . The choice whether to take government expenditure as an exogenous variable or an endogenous one is also a related problem. Either choice might generate different results. In spite of this, most researchers are agreeable to the fact that if government policy influences growth, then it could be a very important ingredient in explaining variations in the long run growth among countries. Using of functional categories of public expenditure in regressions by some certain researchers have also yielded mixed results. For example, Devarajan et al. (1993) as cited by M’Amanja and Morrisey (2005) found government expenditure on health, transport and communications to be growth promoting but no positive impact of education and military spending. Albala and Mamatzakis (2001), as cited by M’Amanja and Morrisey (2005), using time series data covering 1960 – 1995 to estimate a Cobb-Douglas production function that includes public infrastructure for Chile, found a positive and significant correlation between public infrastructure and economic growth. These findings concretize the argument that empirical outcomes are likely to differ from country to country, region to region and time to time even when the same model of estimation techniques is to be used. It is therefore highly believed that the solution to the fiscal policy – growth paradox rests in particular country studies. 3.00 Econometric model and data. 3.10 Econometric model The empirical analysis of fiscal policy and growth will be started by the formulation of the model. The model that will be estimated is as below; Y = f (TRGDP, AGDP, BDGDP, PIGDP, UGGDP, PEGDP) (8) where, Y is the real gross domestic product (RGDP); TRGDP is the total tax revenue as a percentage of the RGDP;
BD is the Budget Deficit and it’s the difference between government total revenue (including all the grants) and total government expenditure (including net repayment); But this variable was not used in the estimation because it’s an identity but its worth analysing. AGDP is aid as a percentage of RGDP which are the receipts from abroad in the form of grants; PIGDP is the private investment as a percentage of the RGDP in the economy; UGGDP is the unproductive government expenditure (Total expenditure less expenditure on health, education and other economic activities) as a percentage of RGDP; PEGDP is the productive government expenditure (Defined to include the expenditure on health, education and other economic services) as a percentage of RGDP. The framework given in equation (8) incorporated with lags of the second degree will therefore be used to estimate the final growth equation given in (7). The choice of this distributed lag model as opposed to a static one, is motivated by the fact that, there’s need to capture all the dynamic responses in the dependent variable brought about by changes in its own lags. Estimating the model in this way is expected to yield valid t-statistics even if some of the explanatory variables are endogenous. From the model given in equation (8) above, it can be seen that real gross domestic product depends on fiscal policy variables being TRGDP, UGGDP, PEGDP, the identity BD and also on the non fiscal policy variables among them AGDP and PIGDP. The relationship in equation (8) can therefore be rewritten in an equation form as; Y = α + β1TRGDP + β2AGDP + β3PIGDP + β4UGGDP + β5PEGDP + εit (9) Where, α, β1, β2, β3, β4 and β5 are the parameters to be estimated. The expected signs of the coefficients of the explanatory variables are such that β3 and β5 > 0, β1 and β4 < 0 and β2 can be either positive or negative depending on the usage by the government. β1 < 0: The relationship between taxes and the real Growth rate (Y) is expected to be in the negative direction. This is because, tax revenue especially distortionary tax revenue distorts the incentives of the private agents meaning that an increase in taxes reduces the disposable income and consequently the savings and therefore investments. This will result into a drop in the output (Y) cetaris paribus. β2 > 0 or β2 < 0: The relationship between Aid and the real Growth rate (Y) can either be positive or negative depending on how the government utilizes it and also on the presence of other supporting policies on how to use Aid. β3 > 0: The relationship between Private Investment and real Growth rate (Y) is expected to be positive and to be significant to output. β4 < 0: The relationship between Unproductive government expenditure and the real Growth rate (Y) is expected to be negative and to be insignificant on growth. β5 >0: The relationship between productive government expenditure and the real Growth rate (Y) is expected to be positive and to have significant impact on growth. The relationship however can also be negative depending on the actual composition of this category of expenditure. 3.20 Tests for Cointegration. Cointergration occurs when the variables in a model are nonstationary, but the trends of the variables are related in a way so that the error term observations are stationary. The test for Cointegration in a single equation situation usually involves the test for unit roots in the residues of the cointegrating relationship from the
long run equation. The null hypothesis in that case then, is that the residues are nonstationary (have unit roots) with the alternative being they are stationary. One widely used test for Cointegration is the Dickey-Fuller test for unit roots. This is done by using the error term observations from the regression as the variable being tested. If the null hypothesis of the Dickey-Fuller test is rejected, then the error term observations may be stationary and therefore unit roots may not be present. If the null hypothesis of the Dickey-Fuller test is otherwise, that is to fail to reject it, then the error term observations may be non-stationary and therefore unit roots may be present. Table 1: Test for unit roots for all the variables in the model. Date: 04/06/2008 Time: 18:22 hours Sample: 1964 – 2007 Included observations: 42. Software: Stata Version 8.2 Variable Dickey Fuller Test statistic Critical value Stationarity test RGDP -6.444 [-2.950]** Stationary TRGDP -2.552 [-2.950]** Non-stationary AGDP -1.761 [-2.950]** Non-stationary PIGDP -1.196 [-2.950]** Non-stationary PEGDP 3.386 [-2.950]** Stationary UGGDP -0.757 [-2.950]** Non-stationary Source: Stata version 8.2 results ** Significant at 5 % level 3.30 Causality analysis Granger causality occurs when a variable X changes and changes follow in another variable Y thereafter. It is said that X “Granger-causes” Y. Hence, a variable X is said to granger cause another variable Y if past values of X can predict present values of Y. Causality analysis was conducted based on the model as stated in equation (9). Table 2 below provides a summary of the results of Granger causality test for the model.
Table 2: Summary of Causality analysis tests. Date: 05/06/2008 Time: 21:44 hours Sample: 1964 – 2007 Included observations: 42 Software: Stata Version 8.2 Direction TRGDP→PIGDP and Private PIGDP→TRGDP PIGDP→AGDP
Chi-sq (X2) 7.2136 Investment. 11.304 2.8469
P-value [0.037]*
Conclusion Bi-directional causality between Tax revenue
[0.004]* [0.107]***
Weak uni-directional causality between Private
Investment and Aid. PIGDP→UGGDP 8.3872 [0.005]* Bi-directional causality between Private investment and Unproductive Government expenditure. UGGDP→PIGDP 11.875 [0.099]** PEGDP→AGDP 4.9274 [0.027]* Uni-directional causality between Productive Government expenditure and Aid.
PEGDP→PIGDP 4.2133 [0.104]*** Uni-directional causality between Productive Government expenditure and Private investment. PEGDP→UGGDP 11.284 [0.010]* Bi-directional causality between Productive Government expenditure and Unproductive Government expenditure. UGGDP→PEGDP 9.2843 [0.003]* UGGDP→AGDP 8.9272 [0.004]* Uni-directional causality between Unproductive Government expenditure and Aid. RGDP→PEGDP 6.075 [0.197] No causality in both directions between real GDP (output) and Productive government expenditure. RGDP→PIGDP 1.6367 [0.201] No causality in both directions between real GDP (output) and Private investment. Source: Stata version 8.2 results Notes: /1 *, **, *** indicate rejection of the null of non-causality at the 5%, 10% and 15% significance levels respectively. /2 In most cases, where there is No causality in either direction, results are not reported in the table above. However, causality results between Productive government expenditure and output (RGDP) and Private investment and output (RGDP) are included to highlight the surprising finding of non-causality amongst these variables. 3.40 Data and the variables. All the data series on fiscal and non – fiscal variables were obtained from various data sources which include Alan Heston, Robert Summers and Bettina Aten, Penn World Tables Version 6.2, Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania, September 2006., IMF International Government Finance Statistics (Various issues), Bank of Zambia, Central Statistical Office and the World development Indicators website (2006). In this study, expenditure has been further divided into productive and unproductive expenditure. This classification is after the work of Barro and Salai-Martin (1995) as cited by M’Amanja and Morrissey who defines productive expenditure as that which enters into the production function of the private agent and unproductive expenditure as that which enters into the private agent’s utility function. Theoretically, it’s not so clear as to which items of public expenditure fall under the Barro categories. Therefore as a consequence, some subjectivity can not be exclusively ruled out. For the purpose of this paper, expenditure on health, education and other economic activities was taken as productive and the remaining of recurrent expenditure was assumed as unproductive.
Figure 1: Expenditure Trends (as Shares of GDP) for Zambia, 1964 – 2007. Source: Author, based on Alan Heston, Robert Summers and Bettina Aten, “Penn World Tables version 6.2 center for international comparisons of production, Income and Prices at university of Pennsylvania, September 2006 and Bank of Zambia various issues, 2007. Figure 1 above reveals that the share of unproductive government expenditure (Total expenditure less expenditure on health, education and other economic activities) has been steadily growing while that of productive government expenditure rapidly grew and reached its peak at 56% in 1994 and then started dropping until year 2000 before it started rising again.
Figure 2: Tax revenue as a percentage of RGDP trends from 1964-2007. Source: Author, based on Alan Heston, Robert Summers and Bettina Aten, “Penn World Tables version 6.2 center for international comparisons of production, Income and Prices at university of Pennsylvania, September 2006 and Bank of Zambia various issues, 2007. Figure 2 above shows the total tax revenue as a percentage of GDP trend for Zambia from 1964 to 2007. It’s surprising to note that since 1964, tax revenue has ever been going down. This can be attributed to a lot of factors happening in the economy predominantly being the increase in the contribution of agriculture to Zambia’s GDP. This has continued to rise over the years. Figure 3: Investment trends in Zambia from 1964-2007. Source: Author, based on Alan Heston, Robert Summers and Bettina Aten, “Penn World Tables version 6.2 center for international comparisons of production, Income and Prices at university of Pennsylvania, September 2006 and Bank of Zambia various issues, 2007. Figure 3 above shows the investment trends in Zambia. The figure shows that private investment as a percentage of GDP has generally been increasing whereas government investment as a percentage of GDP has not been doing so fine. Between 1981 and 1995, it was at its lowest. Thereafter, it rose steadily. Private investment has ever been increasing because of the good support environment like a stable political system that has prevailed in Zambia since independence.
Figure 4: Real growth trends for Zambia from 1964-2007 Source: Author, based on Alan Heston, Robert Summers and Bettina Aten, “Penn World Tables version 6.2 center for international comparisons of production, Income and Prices at university of Pennsylvania, September 2006 and Bank of Zambia various issues, 2007. The figure above shows that the trend of real GDP growth in Zambia started on a high note from 1964 to 1965. Thereafter it fluctuated between -1 and 10%, until 1993 when it started rising steadily. The rise after 1993 can be attributed to the good policies that came with the new multi-party government. Figure 5: Flow of Aid trends for Zambia from 1964-2007. Source: Author, based on Alan Heston, Robert Summers and Bettina Aten, “Penn World Tables version 6.2 center for international comparisons of production, Income and Prices at university of Pennsylvania, September 2006 and Bank of Zambia various issues, 2007. The figure above shows the trend of Aid as a percentage of RGDP. From 1964 to 1998, Aid was at a constant flow of between 1.9 and 3.8%. From thereon, it tremendously increased and the reason for this can be attributed to policy changes of good governance that led to both the bilateral and multilateral donors to
increase their faith in the Zambian government and therefore give more official development assistance (ODA). If this ODA has been effective for the Zambian case, is another study worth pursuing. Figure 6: Budget deficit trends for Zambia from 1964-2007. Source: Author, based on Alan Heston, Robert Summers and Bettina Aten, “Penn World Tables version 6.2 center for international comparisons of production, Income and Prices at university of Pennsylvania, September 2006 and Bank of Zambia various issues, 2007. The figure above shows the trend of budget deficit with grants (BDGDP) and without grants (BDGDP1). Very notable from the graphs is the sharp fall in the budget deficit (BD) around 1991. This could be attributed to the transition in the politics of the country at the time, in that, Zambia changed from the second republic and major elections were held which affected the economic affairs of the country so drastically. For estimation purposes and because of the fact that, budget deficits are most likely to significantly affect growth in the short run than in the long run, the BD variable was excluded in the long run analysis. 4.00 Regression results. Table 3: Summary of the main regression results for the model with all the variables included. Dependent variable: RGDP Method: Ordinary Least squares Date: 02/06/2008 Time: 19:22 hours Sample: 1964 – 2007 Included observations: 42, with 2 time lags. Software: Stata Version 8.2 Variable Coefficient p-value (P>|t|) TRGDP -0.0848406 [0.733] AGDP -1.000432 [0.119] PIGDP 0.2455009 [0.083]*** PEGDP 1.219868 [0.027]** UGGDP -0.3131165 [0.166] CONSTANT -3.106157 [0.797] R – Squared 0.5024 Adjusted R – squared 0.2154 Root MSE 3.6711 Source: Stata results version 8.2 *** Significant at 10 % ** Significant at 5 % Table 4: Summary of the main regression results for the model with unproductive government expenditure omitted. Dependent variable: RGDP Method: Ordinary Least squares Date: 02/06/2008 Time: 19:36 hours Sample: 1964 – 2007 Included observations: 42, with 2 time lags. Software: Stata Version 8.2 Variable
Coefficient p-value (P>|t|)
TRGDP -0.0984374 [0.699] AGDP -0.6065438 [0.289] PIGDP 0.0870854 [0.474] PEGDP 0.783229 [0.030]** CONSTANT 0.3545423 [0.977] R – Squared 0.4014 Adjusted R – squared 0.1537 Root MSE 3.8127 Source: Stata version 8.2 ** Significant at 5 %
5.00 Discussion of the empirical results. Following Barro (1990) & Kneller et al (1999) as cited in M’Amanja and Morrissey in their paper, Barro & Kneller et al postulated that removing unproductive government expenditure from the model should have no significant effect on the magnitudes and/or signs of the other variables in the model. Using panel data for 22 OECD countries, Kneller et al (1999) as cited in M’Amanja and Morrissey precisely discovered this to be true. It was one of the aims of this paper to use exactly the same general idea on a single country Zambia, but using time series techniques on annual data. The results found by this author are in agreement with their findings and the signs for all the variables were as hypothesized. Comparing Tables 3 and 4, it can be seen that although the magnitudes of the coefficients are not exactly the same, the differences are not significant and can be ascribed to the collinearity between some of the variables and/or poor quality data which is most common in developing countries. This poor quality in most cases is as a result of errors in measurement. However, it’s worth pointing out that in Table 3, Private investment which was significant at 10% significance levels is no longer significant when Unproductive government expenditure is omitted as shown in Table 4. Finally, the omission of the neutral variable (Unproductive government expenditure) did not lead to a more precise parameter estimates of the remaining variables as was found by Kneller et al (1999) for OECD countries, as cited in M’Amanja and Morrissey. This is because there was a not so significant reduction in the coefficients of some of the variables like PEGDP which reduced from 1.219868 to 0.783229 and also the R’s reduced. Because of this, all the interpretations and discussions on results will be based on Table 3 results. In agreement to predictions in theory, this author found a positive and significant correlation between productive government expenditure and real GDP (Output). Its elasticity with respect to output was 1.219868. The implication is that Zambia’s economy is likely to perform better if more resources are applied on truly productive expenditures such as preventive health care, medicines, and doctors in hospitals, building roads and bridges e.t.c. Private investment has a positive and significant coefficient with an output elasticity of 0.2455009. This is consistent with what was hypothesized that private investment should be positively related to growth. The policy implication is that, as is the government’s current view, private investment remains the “engine of growth” in Zambia. This is because private investment varies more widely and is very sensitive to factors such as political uncertainty, corruption, poor macroeconomic environment and so on. Because of this, the Zambian government must ensure that these factors are well in place just like it has declared a zero
tolerance to corruption if private investment is to continue fostering economic growth in Zambia. Aid which is foreign based was hypothesized to have either a positive or negative relationship with output depending on the usage. What was found in this study was a negative and non-significant relationship with the output. The elasticity was -1.000432. This finding is a surprise. Normally Aid is suppose to be effective relative to economic growth. The implication is that the usage of foreign Aid by the government has not been “right” and also the other supporting policies are not in place or are abused. Therefore the usage of foreign aid should be revised by the Zambian policy formulators so as to make sure that it’s used in gainful expenditures. The total tax revenue has a negative and insignificant coefficient with an output elasticity of -0.0848406. The intention initially was to segment total tax revenue into distortionary and non-distortionary tax revenues. But unfortunately this was not achieved as there was no annual data available for the case of Zambia. This result with regards to the sign of the coefficient was as hypothesized. It was going to be very interesting to see the impact of distortionary tax revenue on output for the case of Zambia. 6.00 Conclusions and Recommendations. It was one of the objectives of this study to investigate the impression of fiscal policy and non-fiscal policy variables on growth in Zambia. It was sought to omit unproductive government expenditure from the growth model (equation 9) without loss of information and robustness of the model. Stata version 8.2, one of the software in econometrics was used to analyse some of the variables affecting growth in Zambia. The fact that fiscal policy stimulates economic growth is one of the key findings of this research. Productive government expenditure and private investment have a great role in determining the growth of real output in Zambia. Consistent with theoretical prognosis, unproductive government expenditure has a neutral effect on growth. Therefore unproductive government expenditure should be reduced so that productive government expenditure is supported and enhanced. This policy recommendation is worthy accomplishing. Volatility of private investment to both internal and external forces is unquestionable in theory and practice. As a result, it’s the duty of the government to institute policies that protect and advance private sector investment for growth and well being to be attained at higher levels. Political stability and zero tolerance to corruption are some of the stands that the government should continue embracing. Inspite of the fact that this study has some limitations like failure to segregate total tax revenue into distortionary and non-distortionary tax revenues and those originating from variable measurements, its findings do provide some important policy matters for Zambia’s growth plan of action in as far as fiscal and nonfiscal policies are concerned. It’s hoped that the findings of this study will invigorate some exciting debate on whether fiscal policy stimulates growth in Zambia and indeed many other nations especially in the sub-Saharan region. Finally, further research should be carried out; (i) to ascertain why Aid has not been effective towards growth in the case of Zambia and (ii) to study the relationship between distortionary/ non-distortionary tax revenues and growth for the case of Zambia by attempting to segregate the total tax revenue into the 2 categories. References. Aghion & Durlauf, (2005). “Handbook of Economic Growth”. © 2005 Elsevier B.V.
Alan Heston, Robert Summers and Bettina Aten, (September 2006) “Penn World Tables version 6.2 center for international comparisons of production, Income and Prices at university of Pennsylvania, Barro, R. J and Sala-I-Martin, X. (1995) “Economic Growth”, McGraw-Hill, Inc. Fu, D. et al. (2003). “Fiscal policy and Growth” IMF Research department working paper 0301, Dallas. Gerson, P. (1998). “The impact of fiscal policy variables on output Growth”. IMF working paper WP/98/1. Harriss, L. C. (1956). “Government spending and long – run economic growth”. The American Economic Review, vol. 46, No 2, papers and proceedings of the sixtyeighth annual meeting of the American Economic Association. Retrieved on March 03, 2008. M’Amanja, D. & Morrissey, O. (2005). “Fiscal policy and Economic Growth in Kenya”. Credit Research paper 05/06 https://remote.grips.ac.jp/stable/pdfplus/,DanaInfo=.awxyCnxzvzIy2s+1884356.pdf Retrieved February 01, 2008. Ministry of Finance and National Panning. (2007). “Government of the Republic of Zambia, Monthly Digest of Statistics”, CSO, Lusaka, Various Issues Mwanawina, I. and Mulungushi, J. (2002). “Explaining African Economic Growth performance”. The case of Zambia. The Economic Association of Zambia. http://www.gdnet.org/pdf/draft_country_studies/Zambia-Mwanawina-RIR.pdf Downloaded on April 22, 2008. Nurul, I. (1972). “Foreign Assistance and Economic Development”: The case of Pakistan. The economic journal, vol. 82, No 325. https://remote.grips.ac.jp/stable/pdfplus/,DanaInfo=.awxyCnxzvzIy2s+1884356.pdf Retrieved February 04, 2008. Phiri F. (MEF03093 – 2003). “Fiscal policy and Economic Growth in Zambia”. Grips policy paper 2003. The World Bank. (2007). “Fiscal policy and Economic Growth”. Edited by Gray, C., Lane, T &. Varoudakis, A. http://siteresources.worldbank.org/INTECA/Resources/2578961182288383968/FiscalPolicy&EconomicGrowthinECA_FullReport.pdf Downloaded on April 22, 2008. World Bank. (2007). “World development Indicators 2007”.Washington DC: World Bank. APPENDIX 1: RAW DATA. YEAR 1964 1965 1966 1967 1968
RGDP 12.93 29.38 -4.05 5.04 2.58
BDGDP 26.54 25.62 21.29 24.23 23.90
28.58 27.89 23.50 26.71 26.46
TRGDP -2.04 -2.27 -2.21 -2.48 -2.56
2.89 3.10 4.34 5.09 4.26
AGDP 1.36 1.43 1.26 1.27 1.31
PIGDP 2.04 2.27 2.21 2.48 2.56
PEGDP UGGDP
1969 3.17 26.40 29.17 -2.77 4.31 1.29 2.77 1970 3.24 32.65 35.99 -3.34 4.67 0.98 3.34 1971 0.06 21.01 25.00 -3.99 5.71 0.92 3.99 1972 9.84 16.22 20.53 -4.31 5.66 1.40 4.31 1973 -0.96 23.38 27.98 -4.60 8.24 1.82 4.60 1974 6.74 27.33 32.87 -5.54 9.51 2.06 5.54 1975 -2.43 20.08 26.80 -6.72 5.19 2.38 6.72 1976 4.31 14.63 22.39 -7.76 7.80 2.46 7.76 1977 -4.34 14.20 23.57 -9.37 12.07 2.66 9.37 1978 0.57 11.98 23.20 -11.22 15.96 3.10 11.22 1979 -3.04 8.31 20.83 -12.52 10.84 3.48 12.52 1980 3.85 9.73 23.46 -13.73 19.12 4.42 13.73 1981 6.63 6.26 21.55 -15.29 19.46 4.95 15.29 1982 -2.91 5.36 21.59 -16.23 16.73 4.92 16.23 1983 -1.14 6.51 22.93 -16.42 18.94 5.25 16.42 1984 -1.72 2.96 20.63 -17.67 17.35 6.35 17.67 1985 1.24 1.44 20.18 -18.74 26.37 6.82 18.74 1986 1.70 1.02 21.95 -20.93 25.40 7.56 20.93 1987 1.49 -4.14 20.13 -24.27 31.38 7.56 24.27 1988 9.27 -10.65 15.65 -26.30 35.25 6.90 26.30 1989 -3.66 -10.75 16.87 -27.62 36.22 7.35 27.62 1990 -0.48 -10.42 19.50 -29.92 36.44 7.52 29.92 1991 -0.04 -15.06 17.41 -32.47 35.46 8.14 32.47 1992 -1.73 -17.87 18.74 -36.61 34.23 9.74 36.61 1993 6.81 -29.22 15.47 -44.69 46.01 11.17 44.69 1994 -8.62 -37.17 18.83 -56.00 60.31 12.80 56.00 1995 -4.34 -36.79 17.21 -54.00 35.00 14.84 54.00 1996 6.45 -30.93 17.63 -48.56 45.71 16.34 48.56 1997 3.51 -24.94 17.61 -42.55 42.34 17.34 42.55 1998 -1.95 -27.59 17.53 -45.12 38.45 17.99 45.12 1999 2.20 -20.13 17.10 -37.23 51.50 19.46 37.23 2000 3.60 -10.67 19.20 -29.87 38.20 21.50 29.87 2001 4.90 -19.51 18.72 -38.23 41.87 22.00 38.23 2002 3.30 -21.82 18.00 -39.82 42.00 22.44 39.82 2003 4.30 -25.34 16.00 4.25 38.00 22.02 41.34 2004 5.40 -26.13 15.87 5.25 45.33 23.00 42.00 2005 5.20 -28.98 18.23 7.84 43.00 24.00 47.21 2006 6.20 -27.34 17.98 8.90 55.00 26.76 45.32 2007 5.70 -26.14 17.10 7.40 52.12 27.00 43.24 Source: Author, based on Alan Heston, Robert Summers and Bettina Aten, “Penn World Tables version 6.2 center for international comparisons of production, Income and Prices at university of Pennsylvania, September 2006 , Bank of Zambia various issues 2007, Central statistical office, Republic of Zambia 2007, World development indicators (WDI), 2006