Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005)

The objectives of this paper are to critically evaluate causality, vulnerability to innovation of consumption, income and economic growth. In methodological terms the paper uses annual data for the period 1960-2005, taken from Economic Survey of Pakistan (various issues) and International Financial Statistics (2005). Vector Autoregressive (VAR) model with impulse response function (IRF), error variance decomposition and Granger Causality test is used for the analysis. The study showed that any innovation of one standard deviation took seven years for economic growth and more than ten years for consumption and income. The variation in consumption is mostly explained in their own. The variation in income is mostly explained by consumption. The variation in economic growth is slightly explained by consumption and income. Bilateral causality is not found, and mostly independent type relationships are detected. Based on the finding of the study, it is recommended to harmonize fiscal policies with monetary policy. The gap between policy formation and its implementation specifically in monetary policy required to reduce. Through fiscal policies the government can easily enhance income, consumption level, productive capacity of the economy, employment opportunities and reduce poverty level. But at the same time effective managing of monetary and fiscal policies are needed to accommodate the enhanced consumption from indigenous production rather than concentrating on import.  Dr. Abdul Qayyum Khan, Assistant Professor of Economics, Dept. of Economics, Hazara University Mansehra. Email: [email protected]  Dr. Muhammad Azam, Assistant Professor of Economics, Dept. Management Sciences, Abdul Wali Khan University Mardan. Email: [email protected] Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005) Abdul Qayyum , M. Azam Journal of Managerial Sciences Volume III, Number 1 24 Introduction The government adopted a deliberate policy (in 1960s), of concentrating national income in the hands of the upper income groups on the basis of economic assumption that the rich save a larger proportion of their income and hence a higher national savings rate could be achieved with an unequal distribution of income. In practice the assumption that it would elevate domestic savings over the time failed to become visible, while the policy of dispensing incomes in favour of the economic elite succeeded. In the rural sector, 15 percent of resources generated annually were moved to the urban industrialists and 63 to 85 per cent of these moved resources went into increased urban consumption. The actual savings rate remained below 12 per cent and never raise to targeted domestic saving rate of 25 per cent (Griffin, 1965). The majority of Pakistan’s population was suffering an absolute decline in their living standards, while an elite and highly monopolistic class was accruing wealth, during the process of rapid economic growth of the 1960s. In 1969-70 per capita consumption of food grain of the poorest 60 per cent of Pakistan’s urban population declined to 96.1 from an index of 100 in 1963-64. Over the same period in the case of the poorest 60 per cent of rural population the decline was even greater. In 1969-70 per capita consumption of food grain declined to only 91 from an index of 100 in 1963-64, in case of the poorest 60 per cent of rural population (Hamid, 1974). Rural sector poverty was so grave in 197172, that 82 per cent of rural households could not afford to provide even 2,100 calories per day per family member (Naseem, 1977). Investment undertaken were hardly finance from internally generated funds, thereby requiring heavy borrowing from foreign governments. The ability to finance increased government expenditures Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005) Abdul Qayyum , M. Azam Journal of Managerial Sciences Volume III, Number 1 25 from tax revenue were constrained by two factors: (i) slowing down of GDP growth, and (ii) inability of government to improve direct taxation coverage, as a result, the deficit increased rapidly. The government reduced subsidies on consumption goods and increase indirect taxation in order to control the rising budget deficit. However, in the face of increasing current expenditures these measures failed to reduce the budget deficit. Monetary expansions were approached to finance budget deficit, ensuing in accelerated inflation. The booms in the construction and consumption linked with Middle East remittances coupled with the easing of budgetary pressures, helped in stimulating economic growth. GDP average growth reached to 6.6 per cent during the Zia period 1978-88, as it was about 5 per cent during the Z.A. Bhutto period 1973-77. The gross fixed capital formation as a percentage of GDP was 15.5 per cent in the Bhutto period and reached to 16.8 per cent in the Zia period (Economic Survey of Pakistan, 1980). Blanchard and Perotti (2002) and Fatás and Mihov (2001) identified exogenous shocks to government consumption by assuming that the government consumption is prearranged comparative to the other variables included in their VAR model. Their most applicable conclusions for our interests can be summarized as follows. First, a positive shock to government consumption brings about an unrelenting rise in that consumption variable. Second, the fiscal extension raises a positive response in output, with the implied multiplier being greater than one in Fatás and Mihov (2001), but close to one in Blanchard and Perotti (2002). Third, in both these studies the fiscal expansion leads to large (and significant) increases in consumption. Fourth, the response of investment to the consumption shock is found to be insignificant in Fatás Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005) Abdul Qayyum , M. Azam Journal of Managerial Sciences Volume III, Number 1 26 and Mihov (2001), but negative (and significant) in Blanchard and Perotti (2002). Perotti (2002) extended the methodology of Blanchard and Perotti (2002) to data for the Australia, Canada, U.K. and Germany. Their qualitatively findings similar to the ones obtained for the U.S. concerning the response of consumption (positive) and investment (negative) to an exogenous raise in government spending. In similar work, Mountford and Uhlig (2002) used the skeptic classification procedure originally proposed in Uhlig (1997) (anchored in sign and near-zero limitations on impulse responses) to categorize and estimate the effects of a “balanced budget” and a “deficit spending” shock. As in Blanchard and Perotti (2002), Mountford and Uhlig (2002) found that government consumption shocks crowd out both residential and non-residential investment, but did not decrease private consumption. Overall, the evidence discussed above as tending to support the predictions of the Keynesian model, over those of the Neoclassical model. In order to evaluate the strength of the above findings and the behavior of alternative variables of interest, in this paper attempt has been made to provide some complementary evidence using the Impulse Response Function (IRF) and variance decomposition strategies for Pakistan. Objectives The main objectives of the study are (i) To appreciate the interrelationships among the consumption, income and economic growth (ii) To detect which of the three variables are more vulnerable to innovation (iii) To verify if we can detect causal links among some of the three variables. Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005) Abdul Qayyum , M. Azam Journal of Managerial Sciences Volume III, Number 1 27 Materials and Methods Time series data for the sample period 1960-2005, which are taken from Economic survey of Pakistan various issues, and International Financial Statistics is used. To determine the stationarity of data, an Augmented Dickey-Fuller (ADF) test is used. The Akaike information criterion is used to select the optimum ADF lag. Stationarity of the variables are checked once with an intercept is included only, and again when both an intercept and a linear deterministic trend is included. Variables which are non-stationary at level make stationary after taking first difference. Vector Autoregression (VAR) model, which treats all the variables in the system as endogenous is used to analyze the dynamic impact of the random errors on the variable’s system. In order to encapsulate the causality among the three main variables of the study (consumption, income and economic growth) Granger causality test is used. A statistical package Eview is used for deriving the results. More specifically, the following multivariate VAR model of order P is used for estimation: n n yt = K + αi xt +  βi yt-1 + Ut (1) i=1 i=1 Where xt and yt is a (n × 1) vector of endogenous variables being considered (consumption, income and economic growth) αi and βi is (n × n ) matrix of coefficient, K is the vector of constant, P is the number of lags and Ut is a (n × 1) vector of uncorrelated white noise disturbances. Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005) Abdul Qayyum , M. Azam Journal of Managerial Sciences Volume III, Number 1 28 Results and Discussion Non Stationarity of the Time Series Table I presents the results of the unit root test. All the three variables are non-stationary when intercept is included only, and after inclusion of trend the variables remain non-stationary. Table I ADF Test for Stationarity Variables Include intercept only Include intercept and trend Result Test statistics Critical Value Test statistics Critical Value PC 0.4985[1] (6.5707)[0] -3.5814 -3.5889 -2.7179[0] (-5.9008) [2] -4.1781 -4.1896 I(1)* I(1)** PI 0.2681[1] -6.8389[2] -3.5814 -3.5850 -2.1363[2] -6.8160[2] -4.1728 -4.1896 I(1)* I(1)** EG -2.3555[2] -5.1552[0] -3.5814 -3.5850 -2.6314[2] -4.9631[2] -4.1728 -4.1896 I(1)* I(1)** Figures in square brackets besides each statistics represent optimum lags selected using the minimum AIC value. Figures in Parentheses are first difference of variables, * shows result when intercept is included only, ** Show results when intercept and trend is included. Co-integration of the Variables– The Johansen Test Johansen Likelihood Ratio (LR) test is used to find out the co integration in the regressions used for analysis. The result of Likelihood Ratio (LR) test is depicted in table II. The Likelihood Ratio (LR) test results point out that the assumption of no co integration has been rejected for all equations by Likelihood Ratio (LR) statistics. The test denotes the existence of two co integrating equations as the calculated values of Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005) Abdul Qayyum , M. Azam Journal of Managerial Sciences Volume III, Number 1 29 Likelihood Ratio (LR) statistics are greater than the critical values at 5 percent as well as 1 percent. The test results show that the variables are co integrating and they have long-term relationships. Table II Johansen co integration test result with intercept (no trend) in CE and no intercept in VAR. (Variables included in the co integrating vector: PC, PI and EG). Test assumption: No deterministic trend in the data. Lag interval is 1 to 1 Eigenvalue Likelihood Ratio 5 Percent Critical Value 1 Percent Critical Value Hypothesized No. of CE(s) 0.5093 60.64 34.91 41.07 None ** 0.3829 29.32 19.96 24.60 At most 1 ** 0.1678 8.08 9.24 12.97 At most 2 *(**) denotes rejection of the hypothesis at 5%(1%) significance levelL.R. test indicates 2 co integrating equation(s) at 5% significance level Impulse Response FunctionsGraphical Analysis Response of Variables to Impulse of 1 SDCombined Graphics The graphical illustration gives the evaluation of the three variables in IRF terms to variations, unitary innovation or shocks (of one standard deviation). Figure 1 translates the rapidity of absorption of the innovation by the three variables. It is found that the absorption takes 7 years for economic growth, and more than 10 years for income and consumption. Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005) Abdul Qayyum , M. Azam Journal of Managerial Sciences Volume III, Number 1 30 Fig. 1 Economical response to impulses of 1 standard deviation (D.V) innovations 0. 02 0. 00 0. 02 0. 04 0. 06 1 2 3 4 5 6 7 8 9 10 DPC DPI DEG Resp o n se o f DP C to On e S .D. In n o v a tio n s 0. 01 0. 00 0. 01 0. 02 0. 03 0. 04 1 2 3 4 5 6 7 8 9 10 DPC DPI DEG Resp o n se o f DP I to On e S . D. In n o v a tio n s 0. 4 0. 2 0. 0 0. 2 0. 4 0. 6 1 2 3 4 5 6 7 8 9 10 DPC DPI DEG Resp o n se o f DEG to On e S . D. In n o v a tio n s Economical Response to Impulses of Standard Deviation +/-2 Standard ErrorsMultiple Response Graphics Figure 2 and table III indicate the three variables response to innovations introduced in the VAR model structure. The result indicates the response or the absorption rhythm of each one of the three variables to innovation or impulses of size 1 s.d. +/2 s.e. The first, second and third graph in row 1 of Figure 2 and first, second and third columns of “Response of DPC” of table III give the response of the consumption to innovations or impulses introduced by itself, by income and by economic growth respectively. The first, second and third graph in row 2 of Figure 2 and first, second and third columns of “Response of DPI” of table III give the response of the income to innovations or impulses introduced by Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005) Abdul Qayyum , M. Azam Journal of Managerial Sciences Volume III, Number 1 31 consumption, by itself and by economic growth respectively. In similar fashion the first, second and third graph in row 3 of Figure 2 and first, second and third columns of “Response of DEG” of table III give the response of the economic growth to innovations or impulses introduced by consumption, by income and by itself. Fig. 2 Economical responses to impulses of 1 standard deviation (D.V) +/2 standard errors 0. 02 0. 00 0. 02 0. 04 0. 06 0. 08 1 2 3 4 5 6 7 8 9 10 Resp o n se o f DP C to DP C 0. 02 0. 00 0. 02 0. 04 0. 06 0. 08 1 2 3 4 5 6 7 8 9 10 Resp o n se o f DP C to DP I 0. 02 0. 00 0. 02 0. 04 0. 06 0. 08 1 2 3 4 5 6 7 8 9 10 Resp o n se o f DP C to DEG 0. 02 0. 01 0. 00 0. 01 0. 02 0. 03 0. 04 0. 05 1 2 3 4 5 6 7 8 9 10 Resp o n se o f DP I to DP C 0. 02 0. 01 0. 00 0. 01 0. 02 0. 03 0. 04 0. 05 1 2 3 4 5 6 7 8 9 10 Resp o n se o f DP I to DP I 0. 02 0. 01 0. 00 0. 01 0. 02 0. 03 0. 04 0. 05 1 2 3 4 5 6 7 8 9 10 Resp o n se o f DP I to DEG 0. 6 0. 4 0. 2 0. 0 0. 2 0. 4 0. 6 1 2 3 4 5 6 7 8 9 10 Resp o n se o f DE G to DP C 0. 6 0. 4 0. 2 0. 0 0. 2 0. 4 0. 6 1 2 3 4 5 6 7 8 9 10 Resp o n se o f DEG to DP I 0. 6 0. 4 0. 2 0. 0 0. 2 0. 4 0. 6 1 2 3 4 5 6 7 8 9 10 Resp o n se o f DEG to DEG Respo nse to One S .D. Inno v a tio ns ± 2 S .E. Table III: Values of the impulse response function (IRF) Response of DPC: Period DPC DPI DEG 1 0.057655 0.000000 0.000000 3 0.027606 0.011748 0.003688 6 0.010498 0.005926 0.001767 9 0.005556 0.002492 0.000481 10 0.004145 0.001931 0.000361 Response of DPI: Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005) Abdul Qayyum , M. Azam Journal of Managerial Sciences Volume III, Number 1 32 Period DPC DPI DEG 1 0.028886 0.032056 0.000000 3 0.010847 0.000772 -0.002585 6 0.005604 0.001686 -0.000385 9 0.002838 0.001278 0.000193 10 0.002319 0.000837 -1.03E-05 Response of DFB: Period DPC DPI DEG 1 0.206821 0.149692 0.448084 3 0.116525 -0.102349 -0.099959 6 -0.014591 -0.025323 -0.023138 9 -0.001825 0.000648 0.000460 10 -6.69E-05 -0.004556 -0.004183 Ordering: DPC DPI DEG Variance Decomposition In Figure 3 and Table IV the values of variance decomposition of the three variables are given. This table values show how the variance of each one of the series is decomposed during a period of ten years. The first group of columns in Table IV is referred to consumption. Those values of standard errors that consumption explains by itself lies between 87% to 100%, with values descending slowly. Income is the second variable to explain most the variation in consumption with values ranging from 0 to 12.38%. Economic growth explains 0 to .62% variations in consumption. The second group of columns refers to the income variance decomposition. Income by itself explains variation between 55.19% to 43.15%. Consumption and economic growth explain 44.81% to 55.95% and 0 to .89% of variation in income. The third group Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005) Abdul Qayyum , M. Azam Journal of Managerial Sciences Volume III, Number 1 33 of columns shows the economic growth variance decomposition. Economic growth by itself explains variation between 75% to 58%. Consumption and income explain 16% to 32% and 8% to 9% of variation in economic growth. Table IV: Values of the variance decomposition Variance Decomposition of DPC: Period S.E. DPC DPI DEG 1 0.057655 100.0000 0.000000 0.000000 3 0.071378 87.20190 12.19187 0.606229 6 0.075829 87.04456 12.31822 0.637221 9 0.076967 87.01211 12.36380 0.624088 10 0.077103 86.99314 12.38278 0.624076 Variance Decomposition of DPI: Period S.E. DPC DPI DEG 1 0.043151 44.81223 55.18777 0.000000 3 0.047037 52.58045 46.61254 0.807008 6 0.049924 55.23186 43.86518 0.902967 9 0.050492 55.87346 43.22928 0.897259 10 0.050553 55.95096 43.15391 0.895130 Variance Decomposition of DEG: Period S.E. DPC DPI DEG 1 0.515715 16.08308 8.425212 75.49170 3 0.616954 32.61412 8.644204 58.74167 6 0.622943 32.06165 9.352199 58.58615 9 0.623334 32.05500 9.393808 58.55120 10 0.623364 32.05184 9.398225 58.54993 Ordering: DPC DPI DEG Inter-Dependency and Causality in Consumption, Income and Economic Growth in Pakistan (1960-2005) Abdul Qayyum , M. Azam Journal of Managerial Sciences Volume III, Number 1 34 Fig. 3 Variance decomposition