The effects of fiscal policy on economic growth: empirical evidences based on time series data from Pakistan.

Ali, Shahid ; Ahmad, Naved

This study investigates the effectiveness of fiscal policy and its impact on macroeconomic activities in Pakistan during the period 1972-2008. It examines the role of fiscal policy under democratic and military regimes. Recognising the shortcomings of traditional procedures, this study adopts modern econometric techniques to identify the effects of fiscal policy on macroeconomic activities. The sensitivity analysis procedure has been carried out to select the robust variables, which are not sensitive to different econometric techniques. Dynamic simulations have also been performed to observe the reaction of unexpected structural and policy shocks. Using Auto Regressive Distribute Lag (ARDL) model, we find that the overall fiscal deficit exerts a negative effect on economic growth in the long run. Hence, this study concludes that expansionary fiscal contraction occurs in Pakistan. In order to estimate short-run dynamics, we use the Error Correction Mechanism (ECM). In the short run, the overall fiscal deficit has a significant impact on economic growth. The study recommends that the budget deficit should be in the narrow band of 3 to 4 percent of GDP. Beyond this limit, the unsustainable budget deficit could have undesirable macroeconomic costs and the government's macroeconomic objectives such as low inflation and high economic growth might be in jeopardy. The government should also curtail nonproductive expenditures. Particular attention should be given to the Public Sector Development Plan (PSDP), as it is the complementary of private investment and has a longterm impact on macroeconomic activities.

JEL classification: E62, O4 and C32

Keywords: Fiscal Policy, Economic Growth, ARDL

1. INTRODUCTION

Fiscal policy refers to government's efforts to influence the direction of the economy through changes in taxes or expenditures. Optimal fiscal policy in Pakistan and in other developing countries plays a pivotal role in growth process and, hence, serves as a vital instrument for economic growth. The efficacy of fiscal policy in improving economic conditions in the long run is, however, a controversial issue and needs further investigation.

In conventional model, a federal tax cut without a corresponding reduction in federal expenditures will encourage consumption expenditures and interest earnings due to increase in personal disposable income. Contrarily, according to Ricardian Equivalence Theorem (RET), the same change in fiscal policy will not result in any of the above mentioned macroeconomic impacts. In other words, a reduction in deficit-financed federal tax cut will not affect macroeconomic outcomes [Saxton (1999)].

The empirical literature on the effects of fiscal policy on Pakistan's economic growth is still at its infancy, we surmise. Shabbir and Mahmood (1992), Iqbal (1995. 1994, 1998), Khilji and Mahmood (1997) have concluded that fiscal deficit is one of the significant variables that affects economic growth in Pakistan. Haq (2003), on the other hand, has argued that fiscal deficits do not have any effect on key macroeconomic indicators such as investment, inflation and GDP growth. The impact of fiscal policy on economic growth can also be demonstrated and explored through transmission mechanism; it affects economic growth via demand and supply sides. According to Khalid, et al. (2008) fiscal policy is considered to have dynamic transmission mechanism, as it carries longer policy lags for different macroeconomic variables and hence, it has different impacts on key macroeconomic variables.

Recognising the importance of sound fiscal policy, the present study explores the link between fiscal policy and economic growth for Pakistan's economy for the period 1972-2008. The study also examines the effectiveness of fiscal policy in different political regimes. Using dynamic model and various econometric techniques, this study tests the significance of various empirical models. The study also imparts some policy recommendations for the development of sound fiscal policy in Pakistan. This study is the first empirical analysis on the "effectiveness of fiscal policy and its impact on economic growth" in Pakistan.

The rest of this study is organised as follow: Section 2 presents the summary of review of literature on the effects of fiscal policy on economic growth in different parts of the world. Section 3 presents the model specification and methodology. Section 4 represents the empirical findings and the last section provides concluding remarks and policy implications.

2. LITERATURE REVIEW

"The macroeconomic relationship between fiscal policy and economic growth has long fascinated economists. Unfortunately, analyses of that relationship have frustrated empiricists for almost as long. One root of that frustration is the array of possible policy indicators" [Fu, et al. (2003)].

A large number of studies have been carried out to examine the impact of fiscal policy variables on economic growth, investment, consumption, inflation, exchange rate, external deficit and other macroeconomic activities [Landau (1986); Huoppner (2003); Perotti (2005), Amanja and Morrissey (2005); Falk, et al. (2006); Rezk (2006); Castro, et al. (2006); Fatas and Mihov (1998); Sinha (1998); William and Orszag (2003); Claus, et al. (2006) and Kukk (2006)]. Government spending, tax revenues and budget deficits as fiscal policy variables have been used by these authors and found different responses of macroeconomic activities to fiscal innovations. According to Huoppner (2003), Claus, et al. (2006), Esau (2006), Heppke-Falk, et al. (2006) and Castro, et al. (2006), shocks to government spending positively affect GDP growth rate, whereas shocks to taxes inversely affect GDP growth rate. Furthermore, GDP growth rate responds negatively to budget deficit in the long run [Balassa (1988); Iqbal and Zahid (1998); Jafri, et al. (2006)]. Many researchers [Barro and Sala-i-Martin (1995); Sala-i-Martin (1997); Mendoza, et al. (1997); Tanzi and Zee (1997); Kneller and Gemmell (1999); Odedokun (2001); and Bose, et al. (2003); Amanja and Morrissey (2005); Romero de Avila and Strauch (2007)] have used fiscal policy variables in the growth equations and have found their significant contribution. The rising budget deficit has been considered as one of the main constraints to economic growth [Iqbal and Zahid (1998); Fischer (1993); Easterly and Rebelo (1992); Levine and Zervos (1993); Barro (1991); Mwebaze (2002) and Balassa (1988)]. From the relevant literature it is clear that fiscal policy affects economic growth. However, the sign and magnitude of the effects of different tools of fiscal policy are ambiguous.

Only few studies have examined the effects of fiscal policy on specific macroeconomic variables in Pakistan [Ahmad and Qayyum (2008); Haque and Montiel (1991); Khalid, et al. (2008)]. Few studies have included budget deficit in growth equations and have found that budget deficit is one the significant variables affecting economic growth [Shabbir and Mahmood (1992); Iqbal (1994, 1995, 1998); Khilji and Mahmood (1997)]. As far as theoretical work regarding the relationship between fiscal policy and economic growth is concerned, the most notable work has been done by Trevala (2005) and Blinder and Solow (1972). Tervala (2005) argued that fiscal growth raises the output of non traded goods and crowds out private consumption of non traded goods. However, Blinder and Solow (1972) argued that in the simplified IS-LM framework the long run sign of the pure fiscal multiplier is undermined a priori, fiscal policy only acts perversely in unstable system.

3. MODEL SPECIFICATION, DATA AND METHODOLOGY

3.1. Model Specification and Data

In order to examine the effects of fiscal policy on economic growth, we estimate the following equation.

[Y.sub.t] [[lambda].sub.0] + [[lambda].sub.1] F[P.sub.t] + [[lambda].sub.2] [X.sub.t] +[[lambda].sub.3] [(FP*DUM).sub.t] ... (1)

Where Y = Growth rate of GDP per capita, vector X represents the set of control variables i.e., private investment (PINV), inflation (INF), current account deficit (CAD) and FP represents Fiscal Policy variables. In the above equation changes in FP variables has a dynamic impact on Y. Further, to capture the effects of fiscal policy in democratic and military regimes, we include the interaction term of fiscal policy with political dummy.

We use overall fiscal deficit as a proxy of fiscal policy.

The data for this study consist of annual observations for the period 1972-2008. The most important data source is Economic Survey of Pakistan (Government of Pakistan). A multivariate framework is employed in this study. (1)

3.2. Methodology

This study concentrates on the ADF and PP and Ng-Perron unit root tests. To test the long run relationship, this study uses the robust econometric technique, Autoregressive Distributed Lag model (ARDL), popularised by Pesaran and Shin (1998), and Pesaran, et al. (2001).

The error correction version of ARDL model is given below for the above given Equation (1).

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] ... (2)

Where Y represents real GDP growth rate, FP represent fiscal policy variables such as fiscal deficit as a percent of GDP (FD), current expenditures as a percent of total expenditures (CE) and development expenditures as a percent of total expenditures (DE). X represents control variables. [[beta].sub.0] is drift component and [mu] is white noise.

In order to find out the short run coefficients, we use the following equation:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] ... (3)

[eta] is the error correction term in the model indicates the pace of adjustment reverse to long run equilibrium following a short run shock.

Private investment is measured by the sum of business fixed investment, residential investment and inventory investment. Moreover, current account balance is measured by the sum of net exports of goods and services, net income from abroad (Net Factor Payment) and net unilateral transfers.

Samudram and Vaithilingam (2009) in case of Malaysia and Mohammadi, et al. (2008) in case of Turkey used Autoregressive Distributed Lag model (ARDL) to examine the impact of public expenditure on economic growth.

To cope up with the endogeneity of explanatory variables, and to avoid inconsistent results, this study uses two-stage least Square (2SLS) instrumental variable techniques.

4. EMPIRICAL RESULTS OF GROWTH EQUATION

4.1. Testing of the Unit Root Hypothesis

ADF test, PP test and Ng-Perron unit root test were applied in order to test the unit root hypothesis to all variables. A summary of these test results is reported in Table 1.

Results show that each of the variables is integrated of different order. The results of the unit root tests enable us to apply any cointegration technique. The results of ADF and PP unit root tests show that all variables are integrated of order one except PINV and CAD. The results of Ng-Perron unit toot test show that all variables are integrated of order one except CAD. The results of Ng-Perron unit root test are given in Table 2.

To choose a robust model for estimation of growth equation, we estimate different growth equations and select three of them for comparison. These equations have been estimated via ARDL co-integration technique.

4.2. Lag Selection of ARDL

After finding integrating order of all variable, the ARDL co-integration system is implemented for Pakistan utilising annual data over the period 1972-2008. In the first stage, the order of lag length is usually obtained from unrestricted vector autoregressive (VAR) via Schwartz Bayesian Criteria (SBC) and Akaike Information Criteria (AIC). The order of lag length is "2" which is selected through the minimum value of SBC as shown in Table 3.

Therefore, lag order 2 is selected on lowest value of SBC in Table 3 for the growth equation. In the next step, we determine individual lag order for the estimation of ARDL, which is (2, 2, 2, 2, and 0). Finally, the F-test Statistics is estimated on the basis of Wald-test. The results are given in the following Table 4.

The results of bound testing approach show that calculated F statistics is statistically significant for growth equation and higher than upper bound critical value at 5 percent level of significance implying that there is a co-integration among the variables in the models. The stability of long run relationship among the variables in the model is also clear from the cumulative sum (CUSUM) stability test. (2) Having found a long run relationship, we apply the ARDL method to estimate the long run and short run coefficients. (3) Long run results are shown in Table 5.

We use fiscal deficit as a percent of GDP (FD), Private investment as a percent of GDP (PINV), inflation rate (INF), current account deficit as a percent of GDP (CAD) as explanatory variables in growth equation. An interaction term of fiscal deficit with dummy of democracy is also included in the growth equation.

ARDL technique provides best results in the presence of endogeneity. (5) The explanatory variables and their lags are used as instruments. It is clear from Table 5 that all variables have expected signs and parameters are significant. The long run results suggest that all variables are important factors affecting economic growth. The coefficient of fiscal deficit is negative and significant at 1 percent level of significance indicating that expansionary fiscal contraction occurs in Pakistan. In the long run rising fiscal deficit reduces national savings and slows down economic growth. These results support the findings of other studies, which evidenced that fiscal deficit negatively affects economic growth [Balassa (1988); Barro (1991); Easterly and Rebelo (1992); Levine and Zervos (1993); Fischer (1993); Barro and Sala-i-Martin (1995); Mendoza, et al. (1997); Tanzi and Zee (1997); Kneller and Gemmell (1999); Odedokun (2001); Mwebaze (2002); Bose, et al. (2003); Ali (2005); Amanja and Morrissey (2005); Jafari, et al. (2006); Kukk (2006); Romero de Avila and Strauch (2007)]. The results of this study also support the findings of the studies in Pakistan [Shabbir and Mahmood (1992); Iqbal (1994, 1995); Khilji and Mahmood (1997); Iqbal and Zahid (1998)]. The main reason of expansionary fiscal contraction in Pakistan is that government activities are mostly politically motivated and unproductive and therefore restrains growth. Moreover, the huge fiscal deficit is due to non development expenditures. Only interest payment of public debt and defence expenditures exceed the development expenditures. Due to these reasons fiscal deficit negatively affects economic growth in the long run. The coefficient of Private investment is significant and its positive sign indicates that high level of investment increases the productivity and, hence, accelerates economic growth. The results show that inflation negatively affects economic growth. This is due to the fact that inflation decreases domestic demand and increases the cost of production. These factors decelerate economic growth. Another important inference drawn from the above result is that the sign of interaction term is negative and significant indicating that fiscal deficit is negatively affecting economic growth in military regime. The sign of current account balance is negative and significant at 1 percent level of significance; it indicates that an increase in current account deficit decreases the foreign exchange reserves with host country and hence, reduces economic growth. The coefficient of fiscal deficit is positive when the square term of fiscal deficit is introduced in the model. The square term with negative coefficient is the indication of fiscal deficit Laffer curve in case of Pakistan. It means that fiscal deficit is not a problem up to some threshold level.

The estimated lagged error correction term EC (-1) is negative and highly significant. The negative and significant error correction term also indicates that there is a long run relationship among the variables Y, FD, PINV, INF and CAD. The feedback coefficient is -0.43. It suggests that about 43 percent disequilibrium is corrected in the current year. The result also suggests that in the short run fiscal deficit has significant impact on economic growth. In the short run, increase in fiscal deficit leads to a decrease in the real gross domestic product. However, in the short run changes in CAD and [DELTA]FD*DUM have insignificant impact on economic growth.

4.3. Sensitivity Analysis

Even though we have given the model specification, yet for the purpose of estimation, we conduct sensitivity analysis and use only robust variables, which are not sensitive to different econometric techniques. For this purpose, we run a lot of regressions and choose the most robust variables for our analysis. The robustness of the variables is also apparent from the short run diagnostic test. From the results of the short run diagnostic tests it is clear that there is no serial correlation and hetroscedasticity in the model. To detect the problem of autocorrelation and hetroscedasticty, we use serial correlation Lagrangian Multiplier (LM) and autoregressive conditional heteroskedasticity tests respectively. In order to test the normality of error term, we use Jarque-Bera test. From the calculated value of Ramsey RESET test it is clear that the functional forms of the models are correctly specified. Moreover, the data is normally distributed. In order to analyse the stability of long run and short run coefficients, the CUSUM and CUSUMsq stability test are applied. The results of CUSUM and CUSUMsquare show that all variables are cointegrated. Moreover, the results show that neither the CUSUM nor the CUSUMsq test statistics exceed the critical values, which ensure that all models are stable and correctly specified. Furthermore, the robustness of the variables is also apparent from the constancy of parameters by using both ARDL and 2SLS econometric techniques. The models are not sensitive to changes in econometric techniques.

5. CONCLUSIONS AND POLICY IMPLICATION

In this study we examine the dynamic effects of fiscal policy on macroeconomic activities over the period 1972-2008. ADF test, PP test and Ng Perron unit root test are applied to test the unit root hypothesis to all variables. The results of ADF and PP unit root tests show that all variables are integrated of order one except CAD and PINV. The results of Ng-Perron unit root test show that all variables are integrated of order one except CAD. The results of the unit root tests enable us to apply ARDL co integration techniques.

Using modern econometric approaches, the results show that there is a long run relationship between overall fiscal deficit and economic growth. It is clear from growth equation that all variables are important factors affecting economic growth. The negative and significant coefficient of fiscal deficit indicates that expansionary fiscal contraction occurs in Pakistan. The main reason of expansionary fiscal contraction in Pakistan is that government activities are mostly politically motivated and unproductive and therefore restrains growth. The huge fiscal deficit is due to non development expenditures.

Using the non linear equation, we find that fiscal deficit positively affects economic growth up to some threshold level. Beyond that threshold level, fiscal deficit negatively affects economic growth and has some serious macroeconomic consequences.

For short run dynamics. Error Correction Mechanism (ECM) has been used. The results of ECM suggest that in the short run overall fiscal deficit exert significant impact on economic growth. This reveals the fact that in the short run rising fiscal deficit creates excess demand, which encourages firms to use more of their existing capacity and people to spend more, and hence economic situation in the short run improves, but in the long run rising fiscal deficit has some serious implication for economic growth. The feed back coefficient is negative and significant suggesting that about 0.43 percent disequilibrium in the previous period is corrected in current year.

The study recommends that the government should keep its budget deficit in the narrow band of 3 to 4 percent of GDP. Beyond this limit the unsustainable budget deficit could have undesirable macroeconomic costs and the government's macroeconomic objectives such as low inflation and high economic growth might be in jeopardy. If the government is able to reduce its budget deficit, eventually she would get rid of the vicious circle of debt overhanging problem, because the debt-GDP ratio would increase only if the fiscal deficit as a percentage of GDP exceeds the real GDP growth rate. However, the reduction in fiscal deficit must be due to reduction in the public expenditure rather than an increase in resource mobilisation. The government should curtail non productive expenditures; high attention should also be given to the Public Sector Development Plan (PSDP), as it has a long term impact on economic growth.

Appendices

APPENDIX 1

DEFINITION OF THE VARIABLES

The definitions of all variables (explanatory variables and instrumental variables) used in this study are given below.

Overall Budget Deficit/Surplus = (Current Account Expenditures + Development Expenditures) - (Repayment of Foreign Debt) - (Net Revenue Receipts) - (the contribution by autonomous bodies) - (The amount earned by disinvestment of shares).

Economic Growth = Growth rate in Real Gross Domestic Product (GDP)

Gross Private Domestic Investment = (Business Fixed Investment + Residential Investment) + (Inventory Investment)

Current Account Balance = Net Exports of Goods and Services + Net Income from abroad (NFP) + Net Unilateral Transfers

Inflation = Consumer Price Index (Inflation rate)

Public Debt = Total public debt as a percent of GDP.

Exchange Rate = Real exchange rate

Interest Rate = 6 months T- bill rate for short run and 9 months T-bill rate for long run.

Money Supply = M1 + Saving Deposits including MMDAs (Money Market Deposit Accounts) + Small Denomination time Deposits + MMMFs (Money Market Mutual Funds).

APPENDIX 2

EMPIRICAL RESULTS USING 2SLS

Table 1
Estimated Coefficients Using 2SLS Techniques

Dependent Variable 2SLS
Real GDP Growth Rate (Y) Technique (6)
Regressors Coefficients

FD -1.11 **
PINV 0.21
INF -0.03 ***
CAD -0.69 ***
[FD.sup.2] --
FD *DUM -0.12 **
 [R.sup.2] = 0.97
 Adjusted [R.sup.2] = 0.96
 F-statistics = 1532.06
 Dh Stat = 1.86

Note: *, ** and*** represent Significant at 1 percent,
5 percent and 10 percent level of significance.

APPENDIX 3

RESULTS OF CUSUM AND [CUSUM.sub.SQ] (7)

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

REFERENCES

Ahmad, I. and A. Qayyum (2008) Effect of Government Spending and Macro-Economic Uncertainty on Private Investment in Services Sector: Evidence from Pakistan. European Journal of Economics, Finance and Administrative Sciences ISSN 14502275 Issue 11.

Ali, A. M. (2005) Fiscal Policy and Economic Growth: The Effect of Fiscal Volatility. Journal of Business and Economics Research 3:5.

Amanja, D. M. and O. Morrissey (2005) Fiscal Policy and Economic Growth in Kenya. (CREDIT Research Paper No. 05/06).

Balassa, Bela (1988) Public Finance and Economic Growth. Policy, Planning and Research Department. (Working Papers; Vol. 1, No. WPS 31).

Barro, R. J. (1991) Economic Growth in a Cross-Section of Countries. Quarterly Journal of Economics 106:2, 407-443.

Barro, R. J. and Xavier Sala-i-Martin (1995) Economic Growth. McGraw-Hill.

Blinder, A. S. and R. M. Solow (1972) Does Fiscal Policy Matter? Econometric Research Programme Research Memorandum No. 144.

Bose, N., M. E. Haque, and D. R. Osborn (2003) Public Expenditure and Economic Growth: A Disaggregated Analysis for Developing Countries. Centre for Growth and Business Cycle Research Discussion Paper Series, University of Manchester.

Castro, D. F., F. and Hernandez D. C., P. (2006) The Economic Effects of Exogenous Fiscal Shocks in Spain: A SVAR Approach. (ECB Working Paper No. 647).

Claus, I., A. Gill, B. Lee, and N. McLellan (2006) An Empirical Investigation of Fiscal Policy in New Zealand. (New Zealand Treasury Working Paper, 06/08).

Dickey, D. A. and W. A. Fuller (1981) Likelihood Ratio Statistics for Autoregressive Time Series With A Unit Root. Econometrica 49, 1057-72.

Easterly, W. R. K, R. Levine, and S. Rebelo (1992) How Do National Polices Affect Long-run Growth? A Research Agenda. (World Bank Discussion Paper No. 164).

Easterly, W. E. and K. S. Hebbel (1993) Fiscal Deficits and Macroeconomic Performance in Developing Countries. World Bank Research Observer 8:2, 211-37.

Fischer, S. (1993) The Role of Macroeconomic Factors in Growth. Journal of Monetary Economics 32. M.I.T. Cambridge, USA.

Fu, D., L. L. Taylor, and K. M. Yucel (2003) Fiscal Policy and Growth. Federal Reserve Bank of Dallas. (Research Department Working Paper 0301).

Haq, A. T. (2003) Fiscal Strategy for Growth and Employment in Pakistan: An alternative Consideration. Islamabad: International Labour Office. (Employment Paper 56).

Haque, Nadeem U. and P. Montiel (1991) The Macroeconomics of Public Sector Deficits: The Case of Pakistan. The World Bank. (Policy Research Working Paper Series 673).

Heppke-Falk, K. H., J. Tenhofen, and G. B. Wolff (2006) The Macroeconomic Effects of Exogenous Fiscal Policy Shocks in Germany: A Disaggregated SVAR Analysis. Deutsche Bundesbank. (Discussion Paper No. 41).

Hoeppner, F. (2003) Business Cycle Effects of Fiscal Policy: Empirical Evidence from Germany. dissertation.de.

Iqbal, Z. and G. M. Zahid (1998) Macroeconomic Determinants of Economic Growth in Pakistan. The Pakistan Development Review 37:2, 125-148.

Iqbal, Z. (1995) Constraints to the Economic Growth of Pakistan: A Three-Gap Approach. The Pakistan Development Review 34:4, 1119-1133.

Iqbal, Z. (1994) Macroeconomic Effects of Adjustment Lending in Pakistan. The Pakistan Development Review 33:4, 1011-1031.

Jafari, S. A., M. Alizadeh, and K. Azizi (2006) Long-Run Relationship Between Budget Deficit and Macroeconomic Performance of the Iranian Economy: A Theoretical and Empirical Analysis. Quarterly-Journal of the Economic Research 10:4, 25-46.

Kaakuga, Esau (2006) The Impact of Fiscal Policy on Economic Growth in Namibia. South African Journal of Economic Growth and Management Sciences, 102-112.

Khilji, N. M. and A. Mahmood (1997) Military Expenditures and Economic Growth in Pakistan. The Pakistan Development Review 36:4, 791-808.

Khalid, M. W. S. Malik, and A. Sattar (2007) The Fiscal Reaction Function and the Transmission Mechanism for Pakistan. The Pakistan Development Review 46:4, 435-447.

Kneller, R., M. B., and N. Gemmell (1999) Public Policy and the Government Budget Constraint: Evidence from the OECD. Journal of Public Economics 74. 171-190.

Kremers, J. J., N. R. Ericson, and J. J. Dolado (1992) The Power of Cointegration Tests. Oxford Bulletin of Economics and Statistics 54, 325-47.

Kukk, K. (2006) Fiscal Policy Effects on Economic Growth: Short Run vs Long Run, Department of Economics. Tallinn University of Technology. (TTUWPE No. 167). (Unpublished).

Landau, D. (1986) Government and Economic Growth in the LDC's: An Empirical Study for 1960-1980. Economic Development and Cultural Change 35:1, 35-76.

Levine, R. and D. Renelt (1992) A Sensitivity Analysis of Cross-Country Growth Regressions. American Economic Review 82:4, 942-963.

Levine, R. and S. Zervos (1993) Looking at Facts: What We Know about Policy and Growth from Cross-Country Analysis. World Bank Policy Research Papers. (WPS 1115).

Mwebaze, T. (2002) Determinants of Economic Growth: a Panel Data of 60 Countries. Electronic Publication, Uganda Martyrs University, NKOZI, UGANDA. (http://www.fiuc.org/iaup/esap/publications/umu/paneldata.php)

Mendoza, E. G., G. M. Milesi-Ferretti, and P. Asea (1997) On the Ineffectiveness of Tax Policy in Altering Long-run Growth: Harberger's Superneutrality Conjecture. Journal of Public Economics 66, 99-126.

Mohammadi, H., M. Cak, and D. Cak (2008) Wagner's Hypothesis: New Evidence from Turkey using the Bounds Testing Approach. Journal of Economic Studies 35, 341-351.

Ng, S. and P. Perron (2001) Lag Length Selection and the Construction of Unit Root Test with Good Size and Power. Econometrica 69, 1519-54.

Pakistan, Government of (Various Issues) Pakistan Economic Survey. Islamabad: Economic Advisors Wing, Ministry of Finance.

Perotti, R. (2005) Estimating the Effects of Fiscal Policy in OECD Countries. (CEPR Discussion Paper Series, No. 4842).

Pesaran, H. M. and Y. Shin (1998) An Autoregressive Distributed Lag Modeling Approach to Cointegration Analysis, In S. Storm (ed.) Econometrics and Economic Theory in the 20th Century: The Ragnar Frisch Centennial Symposium. Cambridge University Press.

Pesaran, M. H., Y. Shin, and R. Smith (2001) Bound Testing Approaches to the Analysis of Level Relationships. Journal of Applied Econometrics 16, 289-326.

Phillips, P. C. and P. Perron (1988) Testing for a Unit Root in a Time Series Regression. Biometrica 75, 335-346.

Rezk, K., M. C. Avramovich, and M. Basso (2006) Dynamic Effects of Fiscal Shocks upon Diverse Macroeconomic Variables: A Structural VAR Analysis For Argentina. XLI Annual Reunion of the Argentine Association of Political Economy, National University of Salta.

Romero, D.A., D. and R. Strauch (2007) Public Finances and Long Term Growth in Europe. Evidence from a Panel Data Analysis. European Journal of Political Economy.

Sala-i-Martin, X. (1997) I Just Ran Two Million Regressions. American Economic Review 87, 178-183.

Samudram, M., M. Nair, and S. Vaithilingam (2009) Keynes and Wagner on Government Expenditures and Economic Development: The Case of a Developing Economy. Empirical Economics 36, 697-712.

Saxton, Jim (1999) Tax Expenditures: A Review and Analysis. Joint Economic Committee Study (http://www.house.gov/jec/fiscal/tax/expend.htm)

Shabbir, T. and A. Mahmood (1992) The Effects of Foreign Private Investment on Economic Growth in Pakistan. The Pakistan Development Review 31:4, 831-841.

Sinha, D. (1998) Government Expenditure and Economic Growth in Malaysia. Journal of Economic Development 23:2.

Tanzi, V. and H. Zee Howell (1997) Fiscal Policy and Long-Run Growth. IMF Staff Papers 44:2, 179-209.

Tariq, A. Haq (2003) Fiscal Strategy for Growth and Employment in Pakistan: An Alternative Consideration. A Case Study Prepared within the Framework of the ILO's Global Employment Agenda. Employment Paper 56.

Tervala, J. (2005) Fiscal Policy in a Two-Sector Model of a Small Open Economy. University of Helsinki, Faculty of Social Sciences, Department of Economics, Discussion Papers, ISSN 1459-3696. (Discussion Paper No. 619).

William, G. Gale and Peter R. Orszag (2003) Economic Effects of Sustained Budget Deficits. National Tax Journal 56:3.

Comments

Fiscal policy is now considered as a staunch option for macroeconomic stabilisation programmes across many countries. Considering the importance of the subject the authors have investigated the effectiveness of fiscal policy measured as Budget deficit and government consumption/investment expenditures interms of its effect on macroeconomic activities in Pakistan by applying modern econometric techniques. Further they have done the same by identifying different governance regimes also.

My first comment is that most economist are using the ARDL methodology, so it may not be called new, and further in econometrics it's the suitability rather then complexity and time reference for choosing the econometric technique.

Secondly the paper has a very broad objective and that too without any theoretical/behavioural equations and that creates a problem of where to start and where to end. e.g. had it just been contained on the aggregate economy and identified some theoretical channels then it would have been easy for the flow and conclusion of the paper. That is why we see the crowding out, expansionary fiscal contraction, debt sustainability, inflation, twin deficit etc. and so many other issues taken care off in this paper.

Thirdly the paper must have been prepared in haste as it contains many typographic mistakes referring to it being taken from a larger document (may be a thesis).

Fourth it has so many issues contained in it that it confuses the reader and to some extent the author as well, as there are contradictory statements.

Further in the end it has policy implications which do not emerge from the paper, e.g. proposing a threshold level of FD to GDP is not estimated some where.

Specifically, in my opinion using just the criteria of SBC may not be enough as there are other methods such as akaike and modeified form of SBC and akaike which can be used to verify, above all the sheer justification based on parsimonious model is also required for a result to be true.

I hope the authors have adjusted the GDP after its rebasing in 2000, similarly the private consumption expenditures also need to be adjusted.

Now I come to the most crucial part, the results, since the paper does not theoretically identify the channels hence the justifications of the results are all spontaneous and ad-hoc, without referring to peer results or to a theory.

e.g. in the first model where growth in GDP is related to certain factor, it is reported that in the long run inflation effects the output growth negatively as it decreases aggregate demand and increases cost of production. But with reference to table 63 (inflation); with money supply increase inflation increases which stimulates the private investment and hence increases the supply to cover for the increase in aggregate demand and hence prices reduce. So there are conflicting strong result and need referencing to some peers or to a theory.

In Equation 6.1.1 the fiscal deficit is shown to impact the private investment negatively, as the access to credit for private borrower reduces which in case of Pakistan is proved to be not holding by many authors.

Further results such as pertaining to Table 21 we see that with appreciation of exchange rate exports decrease and the private investment increases; meaning as Dr Nadeem UI Haque would say domestic commerce has increased.

In the VAR, although referred to the SVAR but the author has used the Choleski decomposition which is arbitrary and highly dependent on the ordering of the variable, therefore the variance decomposition and impulse can be erroneous.

In the end I would still congratulate the author to bring upon a good topic for discussion, but also encourage them to improve it in the light of above comments.

Mahmood Khalid

Pakistan Institute of Development Economics, Islamabad.

(1) See Appendix 1 for the definitions of variables.

(2) The results of CUSUM are given in Appendix 3.

(3) For details see Pesran, et al. (2001).

(4) ARDL order is (2, 2, 2, 2, 0) selected based on SBC.

(5) To check the robustness of the model, we provide the results of 2SLS in Appendix 2. From the results of both techniques (ARDL and 2SLS) it is clear that the parameters of the model are not sensitive to change in econometric technique and hence, it shows the robustness of the model.

(6) INT, M2, ER, PD and all of the variables in the growth equation that are believed to be uncorrelated with the disturbances are used as instrumental variables.

(7) The straight lines represent critical bounds at 5 percent significance level.

Shahid Ali <shahid_aerc@yahoo.com> is Lecturer in Economics, University of Swat, Khyber Pakhtunkhwa. Naved Ahmad <navedahmad@hotmail.com> is Associate Professor and Chairman, Department of Economics and Finance, Institute of Business Administration (IBA), Karachi.

Authors' Note: We are thankful to Syed Akbar Zaidi, Haroon Jamal and Khalid Mahmood for their helpful comments and suggestions. We are, however, responsible for the views expressed in this paper and any remaining errors.

Table 1

Unit Root results

 ADF (Drift and Trend) P-P (Drift and Trend)

Variables Level 1st Diff Level 1st Diff

Y -0.56 -4.24 ** -0.09 -4.87 *
FD -1.84 -7.91 * -1.84 -7.91
PINV -6.06 * -3.92 *** -1.41 -10.30 *
PCON -1.52 -X1.88 * -1.61 -7.23 *
INF -0.72 -4.67 ** -1.41 -4.89 *
CAD -4.14 * -6.67 * -5.98 * 13.09 *

Notes: *(**) Shows significance at 1 percent (5 percent) level.

Table 2

Ng-Perron Unit Root Results

 Ng-Perron Test Statistics

 At Level

 MZa MZt MSB MPT

Y -0.62 -0.23 0.37 37.15
FD -36.0 -134.29 0.01 0.01
PINY -9.69 -2.12 0.21 9.71
PCON -1.79 -0.49 0.27 24.08
INF -1.86 -0.84 0.45 40.55
CAD -17.96 ** -2.99 0.16 5.07

 At 1st
 Difference

 MZa MZt MSB MPT

Y -17.61 * -2.96 0.16 5.19
FD -3.76 * -39.11 10.37 3.10
PINV -12.13 *** -2.46 0.20 7.51
PCON -15.03 *** -2.719 0.18 6.18
INF -55.82 * -4.80 0.08 3.77
CAD -13.39 ** -2.58 0.19 6.80

Notes: *(**) Shows significance
at 1 percent (5 percent) level.

Table 3
Lags Defined through VAR-SBC for Overall Model

 Lag Selected through VAR-SBC Lag

 Growth Equation

0 104.69
1 91.73
2 90.33 *
3 90.55

Notes: * Indicates minimum Schwarz SBC at the corresponding lag.

Table 4
Lag Length Selection and Bound Testing for Cointegration

 Modal 1 (Growth Equation)

Order of the lags AIC HQ SBC F-test
 Statistics

K = 1 116.65 118.57 117.28 2.31
K = 2 113.98 * 117.55 * 115.17 * 5.75 **

 Short run Diagnostic Tests
 Serial Correlation LM tests = 1.65 (0.32)
 ARCH Tests: 1.54 (0.24)
 White Hetroscedasticity Test: 0.76 (0.34)
 Ramsey RESET = 1.02 (0.87)
 Jarque-Bera Tests= 897.45 (0.00)

*(**) Significant at 10 percent (5 percent) level of significant
according to Pesaran, et al. (2001) and Narayan (2005).

Table 5
Estimated Long Run Coefficients Using the ARDL (4)

Dependent Variable ARDL Technique
Real GDP Growth Rate (Y) Order (2, 2, 2, 2, 0)
Regessors
 Coefficients

LFD -1.64 *
PINY 0.26 *
INF -0.05 **
CAD -0.83 *
[FD.sup.2] -
FD * DUM -0.51 ***
 R2 = 0.99
 Adjusted [R.sup.2] = 0.99
 F-statistics = 1298.2
 Dh Stat = 2.14

Dependent Variable ARDL Technique
Real GDP Growth Rate (Y) Order (2, 2, 2, 2, 0)
Re gessors
 Coefficients

LFD 1.04 *
PINY 0.19 *
INF -0.06 ***
CAD -0.91
[FD.sup.2] -0.04 *
FD * DUM -
 [R.sup.2] = 0.99
 Adjusted [R.sup.2] = 0.99
 F-statistics = 1576.2
 Dh Stat = 1.81

Note: *, ** and *** represent Significant at 1 percent,
5 percent and 10 percent level of significance.

Table 6
Estimated Short Run Coefficients Using the ECM

Dependent Variable ARDL Technique
Change in Real GDP Growth Order (2, 2, 2, 2, 0)
Rate ([DELTA]Y) Regressors Coefficients

[DELTA]FD -0.28 **
[DELTA]PINV 0.17 ***
[DELTA]INF 0.08
[DELTA]CAD 0.98
[DELTA]FD*DUM -0.56
[EC.sub.t-1] -0.43
 [R.sup.2] = 0.81
 [R.sup.2] adjusted = 0.79

Note: *, ** and *** represent Significant at 1 percent,
5 percent and 10 percent level of significance.