Macroeconomic Determinants of Tariff Policy in Pakistan Dr. Mohammed Nishat Professor and Chairman, Department of Finance and Economics Institute of Business Administration-IBA University Road, Karachi Phones: 111-422-422 Ext. 222, Fax: 9243421 Email: mnishat@iba.edu.pk and Anjum Aqeel Assistant Professor Applied Economics Research Centre University of Karachi Phones: 021-9243175, Ext: 232 Email: anjumaq98@yahoo.com Abstract The purpose of this paper is to determine whether an empirical causal relationship exists between the level of tariffs and the key macroeconomic variables in Pakistan economy. The macroeconomic variables considered are GDP, unemployment, trade balance, wholesale price index, unit value of imports and indicator for import substitution policy. To analyze the relationship between tariffs and Pakistan economy during 1952 to 2003 period seven variable VAR model is used. Later final VAR model employs the Granger causality testing to explore patterns of causality between the variables. The bi-directional Granger causality test results validate the political economy logic thinking that macroeconomic variables influence tariffs in Pakistan. The results also highlight the feedback on various combinations of macroeconomic variables which again supports the political economy feedback from macroeconomic variables to the level of tariffs. This suggests that the tariff is endogenous in the political economic system of Pakistan. Also, the results indicate the same pattern of empirical relationship between tariffs and the other macroeconomic variables when structural reform and institutional development were observed during the study period.
Macroeconomic Determinants of Tariff Policy in Pakistan Mohammed Nishat Anjum Aqeel 1. Introduction The tariff remains the Pakistan s main trade policy instrument; its relative importance has increased as a result of the recent elimination of non-tariff barriers on several items. At the same time, it has been a major source of tax revenue. Despite severe economic and political difficulties Pakistan has by and large resisted protectionist pressure and opted for market based reforms including the adoption of a more liberal attitude to trade. As a result during last decades Pakistan s tariff has been considerably reduced. 1 The tariff protection is still relatively higher in Pakistan particularly for a few sensitive items, and it varied widely. Consequently, the tariff remains a potential restraint on domestic competition and thus obstacle to the efficient allocation of resources, with adverse consequences for the economy s productivity and local firms export competitiveness. Many studies have found evidence that the macroeconomic conditions determine the tariffs. 2 The usual argument is that protection is the means to improve the country s terms of trade and also improves the country s trade balance since it results in switch in pattern of demand toward home produced goods (Corden, 1987; and Bhagwati, 1988). However, this depends in part on the exchange rate regime prevailing in the country. 3 The unemployment and poor economic conditions also give rise to pressures for protection. 4 However, much would depend upon the source of the increase in unemployment or decline in economic activity. If unemployment was 1 The maximum tariff has now been reduced to 30% (with few exceptions that relate to automobiles and alcoholic beverages). It has already anticipated and decided to further reduce the maximum tariff to 25% (Trade Policy Reviews, Government of Pakistan). 2 Magee et. al 1989; Bohra and Kaempfer, 1991a,b; Das and Das, 1994; Thornton and Molyneux, 1997; Hall et. al., 1998; Brock and Magee, 1978; Magee, 1987; and Magee and Young, 1987) 3 For example with floating rate regime and no accompanying changes in monetary and fiscal policy the exchange rate is likely to appreciate when protection is increased. The ensuing appreciation would increase imports and reduce exports and the trade balance may not change at all. Moreover, the trade balance could actually worsen if the increase in protection stimulated protected industries more than if discouraged investment in industries adversely affected by appreciation, so that overall private investment rose. 4 In most developing countries we observe excess capacity and underutilization of labor, the protection switches the pattern of demand toward home-produced goods that raise aggregate output, and hence incomes and employment would rise. 1
the result of real wages were too high, tariff would increase employment in protected industries but raise costs in industries using protected goods as impetus, and hence increase unemployment there. If there were no initial excess capacity and available labor supply, protection would result in excess demand and inflationary pressures because of the switch in demand toward home produced goods. If domestic prices rose as a result, there would be a real appreciation of the exchange rate that would offset the effects of the protection on trade balance. Theoretically, protection is considered by some as means to improving a country s terms of trade. Restriction of the supply of exports to the world market may raise the prices of these exports and the reduction in demand for imports may reduce their prices. The major objections to protection for this purpose are that large countries engaging in this activity are likely to provoke retaliation, and that small countries can only influence their term of trade in the very short run. Recent explorations, both theoretical and empirical into the political economy of trade policy, have focused on endogeneity of the level and forms of protection (Baldwin, 1985). Tariffs on other forms of protection are the result of interaction in the political arena among various interest groups (Coughlin, 1985; McArther and Marks, 1988; Brock and Magee, 1978; Findlay and Wellisz, 1982; Ray, 1981; Lavergne, 1983; and Conybeare, 1983). Magee and et. al. (1989) suggests a rather intricate model of political pressure that establishes the level of protection. Interest groups for the factors of production and their lobbyists interact with political partied in the tariff formation process in order for all to further their own ends. Magee and Young (1987) test this model by using the rate of unemployment; for instance, will tend to lead affected industries their efforts for protection. The rate of growth of real GNP might also interact with the political pressures of protectionism. The rate of growth of real GNP is also likely to be interrelated to the unemployment rate. Similarly, changes in the trade balance may signal changes in the political effectiveness on the part of protectionist and anti-protectionist pressure groups since fear of retaliation is diminished. Periods in which imports exceed the exports are likely to allow increased political effectiveness on the part of protectionist groups. The trade balance should also interact with inflation with respect to political postures. 5 In literature empirical tests of tariffs that assume a priori causality 5 For instance, if there is trade deficit in inflationary times, this elicits and expansion of protectionist pressure. However, if inflation occurs when the trade balance is in surplus or improving, the argument that inflation causes a flood of import and lead to protection is weakened. Instead, anti-protection forces may arise in such circumstances to pressure for free trade by arguing that a lowering of tariffs will tend to lower inflation. 2
between macroeconomic events and the level of protection may be mis-specified. The level of protection may lead to certain economic consequences that have implications for aggregate macro variables. However, if the state of the macro-economy may lead to a realignment of the political forces that causes the endogenous level of protection to be established. Like many developing countries the pressure for trade protection in Pakistan comes from persistent trade deficit, high unemployment, declining or stagnant real incomes and inconsistency in various macroeconomic policies during last many decades. For improving the efficient allocation of resources through tariff policy in Pakistan we need to understand its relationship with other important macroeconomic variable in the economy. There are no previous studies available on this topic in Pakistan. Most studies in Pakistan focused on inter industry effective protection rates (Kemal, 1987; Kemal, Siddiqui and Siddiqui, and Kemal, 2000), as tax instrument (Floystad, 1985) and as issue of gains from trade (Khan and Lin, 1982). This paper attempts to fill the gap by examining the determinants of trade protection over time. More specifically this paper examines the relationship between the tariffs and GDP, trade balance, unemployment, wholesale price index, imports substitution measures, and unit value of imports. We also distinguish between the reform (1989 to 2003) and non-reform period (1952 to 1988) and during the periods of other structural changes regarding exchange rate observed after 1972 and after 1982. The rest of the paper is organized that section 2 describes the data and provides the definition of variables. Section 3 discusses the econometric methodology followed by empirical results in section 4. The concluding remarks are provided in section 5. 2. Data and Definition of Variables The annual data from 1952 to 2003 is used to determine the empirical relationship between tariffs and macroeconomics and policy variables (GDP, unemployment, trade balance, wholesale price index, unit value of imports and indicator for import substitution policy). The data is extracted from various issues of Economic Survey of Pakistan and statistical year book published by Federal Bureau of Statistics, Pakistan. The tariff is measured as total custom duties over total value of import. The import substitution measure is derived by the method suggested by Fane (1971) which takes into consideration of long run total import substitution and the same is taken 3
as sum of the amounts of import substitution calculated for the sub-periods rather than that calculated for the whole period [see Bacon (1976) for further details]. 3. The Econometric Methodology To analyze the relationship between tariffs and the Pakistan economy, a seven variables VAR model is used. The multivariate time series VAR model approach is employed rather than the single equation or a structural econometric model. The important characteristics of VAR modeling is that it does not require any stringent a priori assumption regarding exogeneity and endogeneity. The VAR form is simply a reduced form representation of some structural econometric model (Zellner and Palm, 1974). The other reason is that the causality tests on these VAR models are more powerful than a single-equation approach (Nelson and Schwert, 1982). The estimation technique used in this paper is suggested by Hsiao (1981) and Caines (1981). While the single equation model is easy and simple to estimate, as the equations are not derived explicitly from a larger model and therefore important feedback mechanism may be omitted. If the right hand side variables in the equations are exogenous, the equation may be part of a system of equations where the variables are interdependent. The extension of single equation approaches to models of interdependent variable, where feed back mechanism exists, went some way with the work of Sims (1972). Researcher in 1970s began developing two variable causality models. As an alternative to traditional econometrics system of equation in which variables are arbitrarily labeled as endogenous or exogenous, VAR models have emerged as powerful multivariate model since the early 1980s (Sims, 1980). In a vector autoregressive model each of a set of variables is regressed on passed values of itself and passed value of every other variable in the system. Cross variable linkages are incorporated because lags of all variables in each equation are included and also because of the existence of correlation among the disturbances of various equations. The present study employs F-test for joint significance of the lag coefficients and multivariate test to determine the direction of causality among the variables. An F-test is constructed under the null hypothesis that the coefficients on the lags of an independent variable in the equation for given dependent variable are jointly equal to zero. The multivariate generalization of Granger causality test has one unrestricted system containing lags of all the variables in the system and a restricted system which exclude lags of the variable(s) of interest. This cross equation restriction is tested by the following likelihood ratio test (See Enderes, 1995). 4
( T k){log log r u } Where T is the number of effective observations and k is the total number of parameters in the unrestricted system. r is the variance-covariance matrix of residuals of the restricted system and u is the variance-covariance matrix of residuals of the unrestricted system. The number of restrictions is equal to number of parameters reduced from the unrestricted system. The test statistics follows a chi-square distribution with degree of freedom equal to the number of restrictions. 4. Discussion of Results Determination of optimal lag length is important for the proper specification of a VAR model. We used the likelihood ratio test AIC and BIC criteria to determine the optimal lag length. To check the lag length we begin with 5 lags as the longest feasible lag length given the degrees of freedom consideration. The first step in estimating VAR model is to transform the data to induced stationarity by taking the variables in growth for as suggested by (Sherman, 2002). As presented in tables 1 and 2 the optimal lag on the basis of AIC and BIC and supplemented by likelihood ratio test is determined as 5. The bivariate F-test is presented in table 3. The causality test results for 12 hypotheses and feedback hypotheses H 1 to H 12 are highlighted in table 3. The H 1 to H 6 test for a causal relation from macroeconomic variables to the level of tariff, whereas H 7 to H 12 test the feedback effects of tariff on macroeconomic variables. The results indicate a bicausality between growth in GDP and tariff rates; unemployment and tariff; whole price and tariff. However, the causal relationship between trade balance and tariff is unidirectional as growth in trade balance does cause growth in tariff but not vice versa. The import substitution policy leads to growth in tariff rates. The unit value index and tariff do not have any causal relationship. Moreover, the wholesale price and tariffs have bidirectional causal relationship during the study period in Pakistan. This indicates higher prices cause lower tariffs, vice versa, higher tariff may curtail the demand for imported goods and hence the prices (The raw material consists of 60% of total imports). 5
The F statistics from these regressions suggest changes in all variables except unit value of imports Granger cause changes on tariff rates. 6 The positive sign of GDP indicates the pressure on tariff level. However, the positive association from unemployment, import substitution to tariffs is consistent with variables having triggered protectionist pressures. The negative relationship between trade balance and tariffs in case of Pakistan may be due to foreign tariffs retaliation as indicated by Bohara and Kaempfer (1990). The results presented in table 4 support the causality between tariffs and GDP and unemployment. It is further to note that GDP, unemployment and trade balance, simultaneously influence tariffs in Pakistan. The GDP, unemployment, trade balance and import substitution measures simultaneously cause tariffs. In last, except wholesale price index all variables simultaneously explain the tariffs in Pakistan. The results again support the political economy feedback from macroeconomic variables to the level of tariffs (GDP, unemployment, trade balance, import-substitution measure and unit value of imports). This suggests that the tariff is endogenous in the political economic system of Pakistan. 5. Summary and Concluding Remarks The purpose of this paper is to determine whether an empirical causal relationship exists between the level of tariffs and the key macroeconomic variables in Pakistan economy. The macroeconomic variables considered are GDP, unemployment, trade balance, wholesale price index, unit value of imports and indicator for import substitution policy. To analyse the relationship between tariffs and Pakistan economy during 1952 to 2003 period seven variable VAR model is used. Later final VAR model employs the Granger causality testing to explore patterns of causality between the variables. The bi-directional Granger causality test results validate the political economy logic thinking that macroeconomic variables influence tariffs in Pakistan. The results also highlight the feedback on various combinations of macroeconomic variables which again supports the political economy feedback from macroeconomic variables to the level of tariffs. This suggests that the tariff is endogenous in the political economic system of Pakistan. Also, the results indicate the same pattern of empirical relationship between tariffs and the other macroeconomic variables when structural reform and institutional development were observed during the study period. 6 The signs of the relationships are determined through residual correlation matrix. 6
References Baldwin, R. E. (1985), The Political Economy of U. S. Trade Policy, Cambridge, MA: MIT Press. Bacon H. K. (1976), Tariff Protection and Import Substitution in Post- War Greece, World Development, Vol. 4, No. 6, pp. 529 542. Bhagwati, J. (1988), Protectionism. Cambridge, MA; MIT Press. Bohara, A. K. and W. H. Kaempfer (1990), Retaliation in Trade Policy Between Canada and the United States, Working Paper, University of Colorado. Bohara, A. K. and W. H. Kaempfer (1991), A Test of Tariff Endogeneity in the United States, American Economic Review, Vol. 81, pp. 952 960. Brock, W. A. and S. P. Magee (1978), The Economics of Special Interest Politics: The Case of the Tariff, American Economic Review, Papers and Proceedings, 68, pp. 246-50. Caines, R. E., E. Keng and C. S. Sethi (1981), Causality Analysis and Multivariate Autoregressive Modelling with an Application to Supermarket Sales Analysis, Journal of Economic Dynamics and Control, 3, pp. 267-98. Conybeare, J. A. C. (1983), Tariff Protection in Developed and Developing Countries: A Cross- Sectional and Longitudinal Analysis, International Organisation, 37, pp. 441-63. Corden, W. M. (1987), Protection and Liberalisation: A Review of Analytical Issues, International Monetary Fund Occasional Paper 54, Washington D. C. Coughlin, C. C. (1985), Domestic Content Legislation: House Voting and Economic Theory of Regulation, Economic Inquiry, 23, pp. 437-48. Das, S. and S. P. Das (1994), Quantitative Assessment of Tariff Endogeneity: Interwar Vs. Postwar, Economic Letters, 44, pp. 139 146. Enders, W., (1995), Applied Econometric Time Series, John Wiley & Sons, Inc. Fane, G. (1971), Import Substitution and Export Expansion: Their Measurement and An Example of Their Application, Pakistan Development Review. Findlay, Ronald and Wellisz, Stanislaw, Endogenous Tariffs, The Political Economy of Trade Restrictions and Welfare, in J. N. Bhagwati, ed., Import Competition and Response, Chicago: University of Chicago Press, 1982, pp. 223-34. Floystad, G. (1985), On Tariffs and Optimal Policy in Developing Countries, The Pakistan Development Review, Vol. 24, No. 3&4. Hall, H. K., C. Kao and D. Nelson (1998), Woman and Tariffs: Testing the Gender Gap Hypothesis in a Downs-Mayer Political Economy Model, Economic Inquiry, 36, pp. 320-332. 7
Hsiao, Cheng, Autoregressive Modelling and Money Income Causality Detection, Journal of Monetary Economics, January 1981, 7, 85-106. Kemal, A. R. (1987), Effective Protection Rates A Guide to Tariff Making, The Pakistan Development Review, Vol. XXVI, No. 4. Kemal, A. R. Siddiqui, R. and R. Siddiqui and A. Kemal (2000), General Equilibrium Assessment of Trade Reforms on Poverty in Pakistan, Pakistan Institute of Development Economics, Islamabad. Khan, M. Ali and Po-Sheng (1982), Sub-Optimal Tariff Policy and Gains from Trade for LDCs with Urban Unemployment, The Pakistan Development Review, Vol. 21, No. 2. Lavergne, Real P., (1983), The Political Economy of u.s. Tariffs: An Empirical Analysis,Toronto: Academic Press. Magee, Stephen P., and Young, Leslie (1987), Endogenous Protection in the United States, 1900-1984, in Robert M. Stern, ed., U.S. Trade Policies in a Changing World Economy, Cambridge, MA: MIT Press, pp. 77-130. Magee, Stephen P., Brock, William A. and Young, Leslie (1989), Black Hole Tariffs and Endogenous Policy Theory: Political Economy in General Equilibrium, Cambridge: Cambridge University Press. McArthur, John and Marks, Stephen V. (1988), Constituent Interest vs. Legislator Ideology: The Role of Political Opportunity Cost, Economic Inquiry, 26, 461-70. Nelson, Charles R. and Schwert, G. William (1982), Tests for Predictive Relationships between Time Series Variables: A Monte Carlo Investigation, Journal of the American Statistical Association, 77, pp. 11-18. Ray, Edward J., The Determinants of Tariff and Non-tariff Restrictions in the United States, Journal of Political Economy, February 1981, 89, 105-21. Sims, C. A. (1972), Money, Income, and Causality, American Economic Review, Vol. 62, pp 540-552. Sims, C.A. (1980), Macroeconomics and Reality Econometrica, 48 (10), pp.1-48. Thornton, John and M. Philip (1995), Macroeconomic Determinants of Tariff Policy in a Developing Economy: Costa Rica 1963-92, Journal of Post Keynesian Economics, Vol. 17, No. 4. Thornton, J. and P. Molyneux (1997), Tariff Endogeneity: Evidence from 19 th Century Europe, Economic Letters, 56. Zellner, Arnold and Palm, Franz, Time Series Analysis and Simultaneous Equation Econometric Models, Journal of Econometrics, May 1974, 2, 17-54. 8
Table 1 AIC and SIC for Lag Length Lags AIC SIC 5 44.7028 55.4494 4 49.88 58.61 3 50.95 57.71 2 50.95 55.77 1 56.75 59.67 Table 2 Results of Likelihood Ratio Test for Lag Length 2 Remarks H : 4 lags against 5 lags 25.910 rejected H : 3 lags against 4 lags 11.340 not rejected H : 2 lags against 3 lags 7.294 not rejected H : one lags against 2 lags 27.190 not rejected 9
Table 3 Pair wise Granger Causality Tests Null Hypothesis F-Statistic Probability GDP does not Granger Cause Tariffs 1.488 0.217 Unemployment does not Granger Cause Tariffs 0.519 0.761 Trade Balance does not Granger Cause Tariffs 0.620 0.685 Unit Value of Imports does not Granger Cause Tariffs 2.663 0.037 Whole Sale Prices does not Granger Cause Tariffs 0.656 0.659 Import Substitution does not Granger Cause Tariffs 0.679 0.642 Tariffs does not Granger Cause Import substitution 2.271*** 0.068 Tariffs does not Granger Cause GDP 0.472 0.795 Tariffs does not Granger Cause Unemployment 0.429 0.825 Tariffs does not Granger Cause Trade Balance 3.864* 0.006 Tariffs does not Granger Cause Unit Value of Imports 2.724** 0.034 Tariffs does not Granger Cause Whole Sale Prices 1.585 0.189 * significant at 0.01 level ** significant at 0.05 level *** significant at 0.10 level 10
Table 4 Chi-Square Statistics for Various Hypothesis Test 2 Remarks GDP and unemployment do not cause tariffs 73.638* rejected GDP, unemployment and trade balance do not cause tariffs 154.702* rejected GDP, unemployment, trade balance and import substitutions do not cause tariffs 94.172* rejected GDP, unemployment, trade balance, import substitutions and unit value of imports do not cause tariffs GDP, unemployment, trade balance, import substitutions, unit value of imports and wholesale price index do not cause tariffs 70.228* rejected 27.052 not rejected Tariffs do not cause GDP and unemployment 55.961* rejected Tariffs do not cause GDP and unemployment and trade balance 38.296** rejected Tariffs do not cause GDP and unemployment, trade balance and import substitution Tariffs do not cause GDP and unemployment, trade balance, import substitution and unit value of imports Tariffs do not cause GDP and unemployment, trade balance, import substitution, unit value of imports and wholesale price index 23.758 not rejected 6.179 not rejected 7.748 not rejected * significant at 0.01 level ** significant at 0.05 level *** significant at 0.10 level 11