Finding the impact of foreign debt servicing on per capita income growth rate: a case study of Pakistan

Among the wide array of macroeconomic problems confronted by Pakistan, foreign debt servicing, has occupied a substantial place. Persistent fiscal deficit since the independence, turned out to be the ground for various governments to rely on the internal or external borrowings. The acquisition of more external debt instead of internal resource mobilization results in higher level of debt stock. Due to soaring level of debt, Pakistan has been allocating a major chunk of resources to debt repayment, which is tarnishing its economic growth. The present study has been conducted to discover the impact of foreign debt servicing on per capita income growth rate of Pakistan for the period 1981-2010 by applying relatively new technique, called Auto-Regressive Distributed Lag (ARDL) of Co-integration. The results confirm that the foreign debt servicing has adversely and significantly affect the per capita income growth rate of Pakistan in both short-run and long-run in the specified period. There is an ardent need for comprehensive policy on part of the government of Pakistan to salvage the economy of such a financial loss.


INTRODUCTION
Pakistan is currently facing many economic and social issues which are collectively affecting the economic growth. The public debt and its servicing are the most crucial problems that Pakistan has been facing in recent years. Domestic and foreign borrowings are considered as normal phenomenon because countries at the initial stages of development need capital stock (Malik and Siddiqui 2001). Apparently, foreign debt increases the economic growth but if it gets accumulated beyond a specific limit, it can have devastating effects upon economic growth as proved by (Hasan 1999).
The poor countries like Pakistan are getting more and more loans just to survive but it creates dependency on donors. More debt is being acquired just to repay previous debt and it is neither being used for the development purposes nor for the human capital formation. Pakistan is spending its export revenues on debt servicing instead of utilizing it on human capital, investment in real assets and scientific research and development. Debt servicing is severely affecting economic growth of Pakistan by eating up major share of resources. The high level of debt leaves no incentives for the government to carry out the macroeconomics reforms and good effective policies because returns from these reforms will only be used to pay back outstanding debt and its services.
The history of Pakistan has remained caught in the debt trap. Debt trap is a situation where a country takes *Corresponding author. E-mail: azrabatools@yahoo.com Authors agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License debt and to pay interest payments, it acquires more debt due to unavailability of resource. Foreign debt amount increased from US $37.24 billion as of June, 2006 to US $57.21 billion as of June, 2010, which shows an increase of 53.63%. In fiscal year 2009 to 2010, the elected government of Pakistan spent a large amount of US $ 3.112 billion on debt servicing out of which US $ 2.3 billion was paid as principal amount.
There are a host of factors, which have contributed towards this dependence on foreign debt including unbalanced and wrong economic policies, inefficient governments and misappropriation by the political and administrative elite of the country during many successive regimes in Pakistan. The current account deficits, fiscal account deficit, macroeconomic mismanagement, non-development expenditures of the successive governments are largely responsible for this high level of debt servicing.
According to former president of state bank of Pakistan Mr. Syed Salim Raza (2009), fiscal deficit is almost half of the total budget, which is the result of non-development expenditures and considered as the biggest problem of Pakistan after inflation 1 . The main heads of nondevelopment expenditures include general administration expenditure on various departments of federal government, Law and order, defence expenditure, subsidies and debt servicing. Among all these, defence expenditures of Pakistan has been increasing in every budget with a constant rate and consuming a large share of income.
The fiscal deficit of Pakistan is around 5.4% of gross domestic product (GDP) and it has to rely again on foreign debt to fulfill this deficit. Most of the other countries around the globe fulfill their fiscal and current account deficit by mobilizing the internal resources or foreign direct investment but Pakistan always relies on foreign debt, which results in high debt accumulations. Debt servicing takes large benefits from the domestic economy as a large amount of foreign exchange reserve has been transferred to the lender countries. It reduces the country's ability to grow itself rather it raises its dependence on the lender countries.
The effects of the high debt can easily be observed by the budget allocation of Pakistan every year. Keeping in mind the severe effects of the debt repayments of the external debt, the present study investigates the impact of debt serving on the economic growth of Pakistan. The present study is organized as follows: Following the strong background of Pakistan debt problem, review of the literature is discussed in second section, theoretical frame work and model specification is discussed in third section, methodology and estimation is discussed in forth section and estimation results and 1 http://www.nation.com.pk/pakistan-news-newspaper-daily-englishonline/Politics/05-Feb-2009/Nondevelopment-expenditure-biggest-problem-SBP findings are discussed in section five and section six contains conclusion followed by policy recommendations.

Review of literature
There are voluminous studies in the literature in which the effect of debt on economic growth has been highlighted. Some studies discussed the debt overhang conditions and some showed significant and negative effect of debt and debt servicing on economic growth. Sinha (1999) examines the relationship between investment, export stability and growth for nine Asian countries 2 . Generalized Johansen framework of co-integration was used to carry out results. Results suggest investment in these countries was positive and significant for the economic growth. Dijkstra and Hermes (2001) concluded, by applying orthogonal least-squares (OLS) technique that uncertainty measure of total debt and long-term debt servicing payments by 104 HIPCs has negative and statistically significant relation with economic growth for the period of 1970 to 1998.  analyzed how debt burden affects economic growth of 53 lower and middle income countries both directly and indirectly in the period of 1970 to 1999. The author also concludes that debt servicing negatively affects investment and economic growth. Karagol (2002) undertakes an empirical study to analyze the effect of debt servicing on economic growth of Turkey by applying Johansen co-integration technique. The author proves that debt servicing has negative relation with economic growth in short-run and long-run. Same is confirmed by Adesola (2009) for Nigeria and Malik et al (2010) for Pakistan. Gupta et al. (2005) checks the expenditure composition and fiscal consolidation effect on the economic growth. Panel data was used for 39 underdeveloped countries by applying generalized method of moments. The results explore that reduction of 1% point in fiscal deficit to gross domestic product ratio led to average increase of 0.5% point in per capita growth. Wijeweera et al. (2005) empirically studies the checked relationship between external debt and GNP for Sri Lanka by applying Engle and Granger (1987) co-integration and extracellular matrix (ECM). Results suggests that the sign of debt servicing coefficient was negative and capital stock and human capital were having positive relation in short-run and long-run. McGrath (2006) analyzes the impact of industrial development and financial deregulation on economic growth of Czech Republic, Hungary and Poland. Industrial development was found significant and positively affects economic growth in three countries. Industrial production unidirectional caused GDP for all three countries. Osinubi et al. (2006) empirically examines how the budget deficit led to accumulation of external debt which affects the economic growth of Nigeria for the period of 1970 to 2003. The author explains the need for Nigeria to finance its fiscal deficit. Complete evasion of external debt as a means of financing budget deficits will not help the economy as debt at sustainable level leads to development. Hameed et al. (2008) verifies the debt overhang situation in Pakistan. Sultan (2008) empirically confirms the positive impact of industry value added on economic growth of Bangladesh by applying simple OLS for the period of 1965 to 2004.
After reviewing the literature, it can easily be evaluated that most of the studies have been conducted to check the effect of debt on economic growth. A fewer studies have been conducted to grasp the effect of debt servicing on single country's economic growth. The present study is conducted to fill the gap by checking the effect of foreign debt servicing along with some allied variables affecting on economic growth (in terms of per capita income growth) of Pakistan by using relatively new technique.

Theoretical frame work
The Pakistan economy has been undergoing severe economic pressure throughout the history. Pakistan has relied on the foreign debt to finance its deficit. The debt taken in the early decades was used appropriately which resulted in high growth and development of the economy. The situation started worsening after 1970's when the debt was not being used properly. The debt has devastating effect on the economic growth of Pakistan as its servicing is far more than its capacity. There are two main theories which elaborate the debt situation of an economy that is, debt overhang hypothesis and debt laffer curve. If the debt burden of an economy becomes so large that a country does not remain in a position to take additional debt to finance its future projects and even though these projects would be profitable enough to reduce indebtedness of a country over time. The debt laffer curve theory states that the higher level of debt stock is associated with the lower probabilities of the debt repayment. Debt overhang hypothesis seems to be in the sight for the case of Pakistan. The current capacity of the economy to repay its debt indicates that economy is in the second part of the debt laffer curve where expected repayments are decreasing with increasing debt stock. The effect of debt servicing can be interesting study to validate the above stated hypothesis for the case of Pakistan.

Model specification
The model is designed to investigate the effect of debt servicing on per capita income growth rate. The debt Jadoon et al. 41 servicing variable has been taken as the main variable along with other potential variables which affects the per capita economic growth rate.
GrPCY=β 0 +β 1 DS t +β 2 DI t +β 3 GrSEC t +β 4 GrIND t +β 5 FD t + Є t Where, GrPCY= Growth rate of Per Capita Income DS= Foreign Debt Servicing DI= Gross Domestic Investment as percentage of GDP GrSEC= Growth rate of Secondary School Enrolment GrIND= Growth rate of Industry Value added FD= Fiscal Deficit as percentage of GDP Є t = Error Term In the model foreign debt servicing has been taken as main variable which includes debt servicing payments of foreign medium and long term loans. The expected relationship between debt servicing variable and the economic growth is negative. Karagol (2002) proved the same negative relation for Turkey and Hameed et al. (2008) and Malik et al. (2010) for Pakistan. Gross domestic investment is defined as increase in stock of capital in an economy and it does not include deduction for depreciation which is previously produced. Here the ratio of gross domestic investment to GDP has been taken and a significant and positive relation is expected between gross domestic investment as a percentage of GDP and economic growth. The positive relation between domestic investment and economic growth for China was confirmed by Tang et al. (2008). Growth rate of secondary school enrolment has been taken as proxy for the quality of human capital as used by  and Clements et al. (2003). The positive and significant relation is expected between secondary school enrolment and economic growth as proved by Skipton (2007) and Afzal et al. (2010). The industry value added is the GDP share by industry or contribution of the industry to overall GDP. A positive and significant relation is expected between industrial value added and economic growth as industry is sharing major portion in the total production. The positive and significant relation between industrial development and economic growth was found by McGrath (2006) and confirmed by Sultan (2008). Fiscal deficit does always have a negative effect on the economic growth. Fiscal deficit increases the amount of debt and its repayment, which slows down the economic growth process. Expected sign of the coefficient of fiscal deficit is negative as proved by Gupta et al. (2005) for 39 underdeveloped countries.

DATA AND METHODOLOGY
Data set used in this study is comprised of last three decades (1981 to 2010). Annual data has been used and data up to the year 2008 has been taken from the world development indicators (WDI) CD-ROM (2008) and the remaining two years data has been taken from the World Bank website. Data on few variables has also been taken from, various issues of economic survey of Pakistan. After the collection of data the next mandatory task is the test, to check the stationarity of the variables in order to apply an appropriate econometric technique. In the present study, two tests Augmented Dickey Fuller (ADF) and Kwiatkowski-Phillips-Schmidt-Shin (KPSS) (1992) have been applied to check the stationarity.
For the test of existence of long-run relationship (co-integration) among the variables, a number of techniques are available. A relatively new technique ARDL bound testing approach has been used in this study due to the shortcomings of others techniques, covered by this technique. This technique is based on the general to specific modeling and has been developed by Pesaran and Pesaran (1997), Pesaran and Smith (1998), Pesaran and Shin (1999) and Pesaran at el (2001). ARDL technique of co-integration has been used due to the problems with other techniques of cointegration like Engel-Granger (1987) and maximum likelihood based Johansen (1988) and Johansen and Juselius (1990).
ARDL has the advantage that it can be applied irrespective of the order of integration of the variables used in study. "ARDL can be used whether the variable is I (0), I (1) or fractionally co-integrated" (Pesaran and Pesaran 1997). An advantage of using ARDL approach of co-integration is that the relationship can be estimated by simple OLS once the order of ARDL is recognized. The ARDL approach is suitable for small sample size. Another advantage of using ARDL bound testing approach is that "it takes satisfactory number of lags to confine the data generating process within the general-to-specific framework" (Laurenceson and Chai, 2003). In comparison with other vector autoregressive (VAR) models, the ARDL model accommodates greater number of variables. Moreover, "a dynamic error correction model (ECM) can be obtained from ARDL through simple linear transformation" (Banerjee et al. 1993).
To apply ARDL bound testing approach first of all stationarity level of the variables has been checked through unit root tests. The variables must be I (0) or I (1) for the application of ARDL bound testing approach. After checking the order of integration of variables, existence of the long-run relationship between the variables has been checked by applying the F-test. The F-test has been carried out by the imposition of the restriction on the coefficients with null hypothesis that is, there exists no long run relationship among the variables and with alternative hypothesis that is, there exists long run relationship among the variables. If the F-statistics lies below the lower bound then null hypothesis is not rejected and if the value of F-statistics is greater than the upper bound then the null hypothesis is rejected. The results remain inconclusive if the value of F-statistics lies between the lower and upper bound then.
In the next stage ARDL equation is estimated where optimal lag length is chosen according to one of the Akaike information or Schwartz Bayesian, which are considered standard criterion for choosing the maximum lag length. After that long-run solution has been obtained for the selected lag length. Diagnostic tests have been applied to check the validity of the model. Lagrange multiplier test has been applied to check either problem of serial correlation exists or not. Ramsey's reset test has been used for the functional form, Skewness and Kurtosis test has been used for the normality and lagrangian multiplier test has been used to check the problem of heteroscedasticity in the data. The equation for the model is given as under.
Where 'i' to 'm' shows the selected lag length.
In the next stage the error correction model has been estimated by using the differences of the variables and the lagged long-run solution. The coefficient of the error correction term shows at which speed the variables return to the new equilibrium.
Finally the cumulative recursive sum (CUSUM) and cumulative recursive sum of squares (CUSUMSQ) tests are employed to check the stability of the short-run and long-run coefficients.

Estimation results and findings
Table 1, 2, 3 and 4 shows the results 3 of the augmented Dickey-Fuller and Kwiatkowski-Phillips-Schmidt-Shin (KPSS) unit root tests at level. The unit root results at level suggested that DI and GSE were stationary at level. Other variables were checked at first difference and the results are given.
The table suggests that ARDL bound testing approach can be applied here because all the variables were stationary at level or at first difference. None of the included variable in the model were stationary at level two or at second difference. After checking the stationarity level of the variables, the next step was to select the maximum lag length. Most widely used criterions to select the maximum lag length are Akaike Information and Schwarz Bayesian. The Schwarz Bayesian criterion had been used for this study for the selection of lag length.
The optimum lag length of 1 has minimized the values of AIC and SCH criterions. The SBC criterion had been chosen to select the lag order for ARDL bound testing approach over the AIC as it has low prediction error (Ma and Jalil 2008). The partial F-test had been applied in the study to check the long-run relationship among the variables. The null hypothesis for the test was that no long-run relationship exists among variables with alternative hypothesis that long-run relationship exists. The following table shows the long-run relationship results of bound test for the economic growth model. Table 4.
The results show that F-value was greater than the UB at 5 and 10 %t level of significance. This test provided the 3 Eviews 7, econometric software has been used to obtain results.  Note. * Shows significance of the variable at 1% and ** shows at 5% ** and *** shows at 10% Note. * shows significance of the variable at 1% and ** shows at 5% ** and *** shows at 10% base to apply ARDL technique of co-integration as it showed the existence of long-run relationship among the variables. As shown in table 5. All the variables of the models had expected signs. The debt servicing coefficient showed that one million dollar increase in debt servicing results in 0.12% decrease in the economic growth. The logic behind this inverse relationship is that although Pakistani government takes debt to finance its projects or to retreat its deficit in balance of payment but as we know the whole amount of debt is not utilized for such purposes rather a very big chunk is paid back in shape of debt repayment. So such money being out of the flow (Leakage) of Pakistani economy, cast negative effect on economic growth. Question arises here is that, why debt is taken on some part of the developing countries like Pakistan? The answer lies in the fact that internal sources of revenue generation are not enough. So the developing countries should wake up from this ignorance and must avoid the horrors of debt and its repayment and try indigenous resource generation. These results were consistent with the results of study by Kargaol (2002) that also showed the negative effect of debt servicing on economic growth.
The fiscal deficit as percentage of GDP negatively affected the economic growth; one percent increase in fiscal deficit as percentage of GDP leads to 40 % decrease in the economic growth. A country does financial deficit due to short of revenue. So the purpose of fiscal deficit continue to use such capital for economic growth. But unfortunately, the capital so collected cannot avert the decree of the fate of developing countries unless they avoid mismanagement or corruption. It means no use of making fiscal deficit. Rather it causes harms to the growth rate of country like Pakistan. Industrial growth has played significant role in the economic growth of Pakistan. In the present study  industrial value added growth had been found highly significant. One percent increase in the growth rate of industry value added leads to 24% increase in the economic growth of Pakistan. Such result is also shown by McGrath (2006) and Sultan (2008). The industrial growth as the theories suggests is one of the strong indicators of economic growth. As we know when any industry promotes, it creates economic activities in the economy that is, other allied industries also get boost up. Hence employment is generated and demand for goods services, further increases, total national income increases and causes GDP per capita to increase. Growth rate of secondary school enrolment in this study had has been taken as a proxy of human capital. The study has revealed that secondary school enrollment also influences the economics growth of Pakistan positively in the long run, as it had has been found highly significant. The background reason is that human capital can be tapped only through education, skill and training. An educated person becomes resourceful and finds ways to be productive. Hence as secondary school enrolment increases, it also causes an escalation in economic growth of Pakistan The domestic investment coefficient had has been found positive but insignificant for the economic growth. Domestic investment means a step toward prosperity because investment is also accompanied by employment and twin increase in employment and output accelerates the pace of economic growth of a country. The value of R-square showed that 79 % of total variation in the economic growth had been explained by the independent variables. The value of R-bar-square shows the goodness fit of the model adjusted to the degree of freedom and it had a value of 0.72 in this model.

Regressors Coefficients Standard Errors T-Ratios Probabilities
The Durbin's h-statistic had been used to check the problem of auto correlation. From the results, it can safely be said that problem of auto correlation does not exist in the data. As shown in table 6.The diagnostic tests had been applied for the robustness of results. The results, indicates that the data is not suffering from the problem of serial correlation. The results confirmed the normality of the data set and correct functional form of the model. The white test showed that the error terms had constant variance, which confirmed the absence of hetero scedasticity. Table 7, shows the long-run coefficients of the variables by applying ARDL. All the coefficients had expected signs. One million dollar increase in debt servicing leads to 0.2 % decrease in economic growth in the long-run. As expected, the school enrolment had long-run positive effect on the economic growth. The positive and significant relationship between growth rate of secondary enrolment and economic growth was also confirmed by Skipton (2007) and Afzal et al (2010).
Fiscal deficit as percentage of GDP had long-run significant and negative effect on the economic growth. In the long-run, one percent increase in the fiscal deficit as percentage of GDP leads to 52 % decrease in growth rate. Gupta et al. (2005) also confirms the same in his panel study. Growth rate of domestic investment had been found insignificant in the long-run. The industrial value added had been found to have a positive and highly significant relation with the economic growth, which was also confirmed by Sultan (2008). One percent increase in the growth rate of industrial value added caused causes 32 percent increase in the economic growth.
Table 8, reveal that error correction term had been found highly significant. The negative sign with ECM shows the convergence of the dependent variable general practitioner for children and the young (GPCY) towards long-run equilibrium path in response to the changes in the independent variables in the model. The ECM (-1) had been found significant at 1% level with coefficient equal to -.76321. The coefficient of ECM (-1) confirmed the co-integration among variables and deviation from the equilibrium level of economic growth during current period will be corrected by 76.32 % in the next period.
The CUSUM and CUSUMSQ tests confirmed the stability of the model if the test representing line stays within the critical bounds at 5% level of significance. The graphs of the CUSUM and CUSUMSQ tests had been presented in Figure 1 and 2. The results of both tests confirmed the correctness of parameters of short-run and long-run variables for growth rate and also verified the structural stability of the model.

CONCLUSION AND RECOMMENDED POLICIES
In the literature, most of the studies on Pakistan have tried to explain the effects of the debt on economic growth. A few numbers of studies have been conducted to check the effects of debt servicing on the economic growth of Pakistan. High levels of debt repayments are negatively affecting the growth process in Pakistan. High levels of debt accumulation over the last few decades and its repayments have become a burning issue in recent times. The paper aimed to check the impact of debt servicing on the economics growth of Pakistan by taking data of last three decades. The ARDL bound testing has been applied and results showed that debt servicing has affected economic growth in the long-run more than the short-run. The reduction in economic growth had been found to be 0.12 and 0.15 % in the short-run and long-run respectively due to one million dollar increase in the debt servicing. The results highlight the need of less reliance on the foreign resources and the importance of domestic resource mobilization. The industrial value added growth had been found positive and significant in both short-run and long-run in the economic growth model, which needs special attention in policy making.
The growth rate of secondary school enrolment had been found highly significant for growth model in the longrun. The fiscal deficit impact had been found negative and significant on economic growth of Pakistan. The fiscal deficit in Pakistan is due to non-development expenditure, the defense expenditures and interest payments, which are more than the development expenditures. The debt dependence policy always put extra pressure on the domestic resources. By acquiring more and more foreign debt with strict conditionalities 4 , the economy is falling into a debt trap 5 .Keeping in view the results of the present study following policy recommendations can be helpful to enhance the growth rate and to tackle the rising problem of the debt.
The government of Pakistan has been spending major share of its revenues for the non-development expenditures which has brought fiscal deficit to the level of 50% of the total budget for the year 2010 to 2011. The government has to finance its projects by printing new money which not only creates inflation but also crowds out investment due to increase in the interest rate. The proper resource utilization is need of the time in education and health sector and institution building and reconstruction.
Pakistan is blessed with affluent human capital which can be used for the self-dependent macro-economic performance. The quality education, training, research and development (R & D) facilities can enhance efficiency of the labor force and help to increase productivity, which will ultimately lead to the higher growth with lower costs of production. The industry value added of Pakistan is 25% of its GDP; this sector has great potential to contribute more in the GDP of the economy. Proper attention and policies are needed to give it a boost.