AN EMPIRICAL INVESTIGATION OF CASUAL RELATIONSHIP BETWEEN INTERNATIONAL TRADE AND ECONOMIC GROWTH DETERMINANTS IN PAKISTAN

Data

The data for this study is collected through world development indicators of the World Bank. All data series are in (constant 2005) US dollars unless indicated otherwise. Economic growth is measured by ‘gross domestic products’ (GDP), volume of trade is measured though exports and imports, gross investment, foreign assistance (measured by official development assistance), remittances, and trade as

a percentage of GDP are also used to check the robustness of the models. The details of the data are provided in summary statistics in table 1 below:

Table 1: Summary Statistics

Mean 9.28E+09 1.32E+10 1.03E+10 1.10E+09 1.84E+09 32.45688 5.96E+10 Median

6.97E+09 9.04E+09 6.66E+09 9.01E+08 5.30E+08 33.19500 3.34E+10 Maximum

3.07E+10 4.65E+10 3.45E+10 3.51E+09 1.34E+10 38.91000 2.47E+11 Minimum

6.79E+08 1.03E+09 6.84E+08 2.53E+08 -9.36E+08 19.93000 3.71E+09 Std. Dev.

8.87E+09 1.33E+10 1.04E+10 7.38E+08 3.17E+09 4.245770 6.58E+10 Skewness

1.081957 1.398448 1.184843 1.316038 2.338760 -1.047281 1.488891 Kurtosis

Jarque-Bera 9.401147 16.72567 12.68875 19.42794 105.3444 10.52058 23.57593 Probability

Sum 4.46E+11 6.35E+11 5.57E+11 5.81E+10 9.96E+10 1557.930 3.28E+12 Sum Sq. Dev. 3.70E+21 8.27E+21 5.72E+21 2.83E+19 5.32E+20 847.2486 2.34E+23

This study uses Granger causality in Error Correction Model within a Vector Autoregressive framework following the model used by Meraj (2013). The general specification of ‘auto regressive distribution lags’ (ARDL) is listed below:

y t =  0 +   i  y t − i +   i  x t − i +  EC t − 1 +  t ------------------------ (1)

ARDL specification

The following specification of the ARDL model is used for this study:

Lgdp t =  0 +   1 i  Lgdp t − i +   2 i  Lex t − i +   3 i  Lim t − i +   4 i  Linv t − i +   5 i  Lod t − i +

6 i  Ltrd t − i +   7 i  Lrmt t − i +  0 Lgdp t − 1 +  1 Lex t − 1 +  2 Lim t − 1 +  3 Linv t − 1 +  4 Lod t − 1 +

 5 Ltrd t − 1 +  6 Lrmt t − 1 +  t − − − − − − − ( 2 )

Error Correction Model in Vector Auto Regressive Framework

From the equation (2), the following error correction model is derived:

Lgdp t =  0 +   1 i  Lgdp t − i +   2 i  Lex t − i +   3 i  Lim t − i +   4 i  Linv t − i +   5 i  Lod t − i +

6 i  Ltrd t − i +   7 i  Lrmt t − i +  0 Lgdp t − 1 +  1 Lex t − 1 +  2 Lim t − 1 +  3 Linv t − 1 +  4 Lod t − 1 +

 5 Ltrd t − 1 +  6 Lrmt t − 1 +  0 EC t − 1 +  t − − − − − − − ( 3 )

Empirical Results and Findings

The figure 1 below shows the graphical trend of seven data series over time. Since all the variables depict a trend over time, the probability of causal association among variables is high.

3E+11 2.5E+11 2E+11 1.5E+11

US$ 1E+11

5E+10

6 0 6 2 6 4 6 6 8 6 7 0 2 7 4 7 6 7 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 -5E+10

Exports

Imports

GDP (Currnt US$)

Investment

ODA

Trade as a percentage of GDP

Remittances

Figure 1: Trend of macroeconomic indicators during 1960 –2014

During 1970 to 2013, the GDP is propagated by US$ 212.3 billion (17.2 times) to US$ 225.4 billion; variation ensued US$ 27.3 billion with a growing population of 123.1 million. In addition, US$ 185 billion raise the GDP per capita to US$ 1016. The lowest GDP was recorded in 1972 (US$ 8.3 billion) and the highest was in 2013 (US$ 225.4 billion). The GDP per capita rose by US$1016 (5.6 times) to US$1238. The average annual growth of GDP per capita was US$ 23.6 (10.6 percent).

At the end of 1950, the share of primary goods in exports was dropped to 75 percent of exports’ income owing to a modification in the policies of industrial development (Chaudhary and Ahmed, 2004). During the early 60s, the variables compacted to follow liberal trade policies but the battle with India in 1965 resulted the penalization of foreign exchange, to compensate that high duties were imposed on imports. During the 70s, the average growth rate of exports was recorded at 13.5 percent and imports was 16.6 percent and the average growth rate of trade deficit was 20.5 percent. In the 80s, the average growth rate of exports was 8.5 percent, imports was 4.5 percent, and the trade deficit was 0.9 percent. During the 90s, the average annual exports growth was 5.6 percent, imports was 3.2 percent and trade deficit was -0.6 percent. In the first decade of the 21 st century, the average growth in exports was 9.9 percent, imports was 13.7 percent, and trade deficit was around 60 percent. Exports amounted to US$15.9 billion in 2009- 10, that was higher than the preceding year’s figure of US$14.7 billion with the growth rate of 8 percent compared to growth rate of 3 percent in the preceding year. The imports progress in 2009-2010 was weaken by 2.8 percent as compared to the previous year. Reduced global prices, compacted internal demand, exchange rate devaluation and better production of cotton crops were the main factors for an overall drop in imports bills. Therefore, general representation of Paki stan’s trade displays that the nature of trade has transformed from primary goods to industrial products. Most of the exported goods are still primary agricultural products specifically raw cotton.

The United States started offering economic aid and martial assistance to Pakistan soon after its inception in 1947. Altogether, the US indebted approximately US$ 67 billion between 1951 and 2011 (see appendix A). The US, at numerous stages including a recently in the 1990, completely stopped its financial assistance and locked the doors of the USAID organizations. Such pattern condensed the US a The United States started offering economic aid and martial assistance to Pakistan soon after its inception in 1947. Altogether, the US indebted approximately US$ 67 billion between 1951 and 2011 (see appendix A). The US, at numerous stages including a recently in the 1990, completely stopped its financial assistance and locked the doors of the USAID organizations. Such pattern condensed the US a

7.5 billion in five years’ period starting from 2010 for advancing Pakistan’s supremacy, economic development, and capitalization of inhabitants.

Remittances provides a momentous and emerging foundation of foreign exchange. The earlier upsurge of remittances inflow was started in the 70s when thousands of Pakistani laborers flew to the Gulf States in the search of a bright future. In the 2013, around 5.7 million Pakistani migrants were projected to be located abroad, compared with 3.7 million in 2000 and 3.6 million in 1990 1 (United Nation, 2014). It displays that around 54 percent of this evolution in the stock of immigrant took place throughout the period of 2000-2013. The elements motivating this upsurge movement consist of economic slowdown, growing poverty, speedy growth of population and considerable wage differences (Ministry of Finance, 2013). Pakistan is fronting a feeble balance of payments conditions and in these circumstances the remittances have appeared as a great source of foreign exchange income. The streams touched around US$14 billion in 2013 compared to only US$1 billion in 2001. Likewise, this proliferation in remittances has outstripped the net Official Development Assistance and FDI, which accounted for merely US$ 2.02 billion and US$1.31 billion in 2012 and 2013 respectively (WDI, 2014). Similarly, related to FDI and ODA, remittances has a tendency to robust and upsurge during economic chaos (Ahmed and Martinez, 2013; Mughal and Makhlouf, 2011).

We use equation (3) to get the empirical results, the following tests are applied to check the validity of the data:

Unit Root Testing

To check the stationarity of the data, two renowned tests are used i.e. ADF (augmented Dickey Fuller’ and PP (Philips Perron); the results are exhibited in table 2 below:

1 This agrees to about 2.2 percent of the country population in 2013 living overseas compared to 2.9 percent in 2000 and 5.9 percent in 1990.

Table 2: Unit Root Testing

Philips-Perron Variables

Augmented Dickey-Fuller

At Level

At Δ

At Level

-6.241* Note: *, **, *** indicate significance at 1%, 5%, 10% level of significance.

Note: All variables were I (I) non-stationary and transformed to stationary I (0) after first difference method applied; the confirmation is provided by both ADF and PP tests.

Johansen-Juselius Cointegration Test

After getting the confirmation of stationarity of the data, the next step is to check the cointegration between the variables; two different approaches are used to check this i.e. Johansen-Juselius and ARDL. The Johansen-Juselius method uses two different tests to check the cointegration i.e. Trace Statistics and Maximum Eigenvalues. The results of the tests are presented in table 3 below:

Table 3: Trace Statistics Unrestricted Cointegration Rank Test (Trace)

Prob.** No. of CE(s)

Eigenvalue

Statistic

Critical Value

0.0000 At most 1*

0.0000 At most 2

0.1052 At most 3

0.8620 Trace test indicates 2 cointegrating eqn(s) at the 0.05 level * denotes rejection of the hypothesis at the 0.05 level **MacKinnon-Haug-Michelis (1999) p-values

The trace statistics confirms that there are at least two cointegrating vectors. We get further confirmation by Maximum Eigen Value Statistics whose results are presented in table-4 below:

Table 4: Max-Eigen Value Statistics Unrestricted Cointegration Rank Test (Maximum Eigenvalue)

Hypothesized No. of CE(s)

Critical Value None *

Statistic

27.58434 0.0000 At most 1 *

21.13162 0.0001 At most 2

14.26460 0.0720 At most 3

3.841466 0.8620 Max-eigenvalue test indicates 2 cointegrating eqn(s) at the 0.05 level * denotes rejection of the hypothesis at the 0.05 level **MacKinnon-Haug-Michelis (1999) p-values

The Trace statistics and Maximum-Eigen value statistics of Johansen cointegration test suggest that there are two cointegrating vectors in our VAR.

Cointegration in ARDL

ARDL is another method developed by Pesaran et al. (2001) which is being used to get the confirmation of cointegration. The substantial feature of using ARDL is that it uses both I (0) and I (I) variables as the issues of pre-test bias are well-handled in this method. By using the equation (3) mentioned above, the null hypothesis of no cointegration asserts that lag values of the model should be equal to zero. It can be written empirically as:

Wald Test of Zero Restrictions

The results of Wald test of zero restrictions explain that the null hypothesis of no cointegration could be rejected at 5 percent level of significance; the results are presented in table-5 below:

Table 5: Wald Test

Equation: ARDL

Test Statistic

Value

df Probability

Granger Causality Tests

The empirical findings suggest that the export led growth hypothesis is valid for the case of Pakistan. The results of Granger Causality is presented in table 6 below:

Table 6: Granger Causality Test

Dependent

Significant level of F Statistics

ΔLtrd ΔLrmt ΔLgdp

4.39 - Note: (*), (**) ,(***) indicate significance at 1%, 5% and 10% respectively

The table 7 below shows the Granger causality relationship among the variables:

Table 7: Granger Causality Relationships

GDP Relationship

Significant at

Result

1. GDP and Exports

Exports Granger causes GDP

GDP doesn’t Granger cause Exports Unilateral Granger causality

2. GDP and Imports

Imports Granger causes GDP

Bilateral Granger causality GDP Granger causes Imports

3. GDP and Investment

Investment Granger causes GDP

Bilateral Granger causality GDP Granger causes Investment

4. GDP and ODA

ODA Granger causes GDP

GDP doesn’t Granger cause ODA Unilateral Granger causality

5. GDP and Trade Openness

Trade Openness Granger causes GDP

GDP doesn’t Granger cause Trade Openness Unilateral Granger causality

6. GDP and Remittances

Remittances Granger causes GDP

GDP doesn’t Granger cause Remittances Unilateral Granger causality

EXPORTS

7. Imports and Exports

Imports Granger causes Exports

Bilateral Granger causality Exports Granger causes Imports

8. Investment and Exports

Investment doesn’t’ Granger cause Exports

Unilateral Granger causality Exports Granger causes Investment

9. ODA and Exports

ODA doesn’t’ Granger cause Exports

Unilateral Granger causality Exports Granger causes ODA

10. Trade Openness and Exports

Trade Openness Granger causes Exports

Exports doesn’t Granger cause Trade Openness Unilateral Granger causality

11. Remittances and Exports

R emittances doesn’t Granger cause Exports

Exports doesn’t Granger cause Remittances No causality

IMPORTS

12. Investment and Imports

Investment Granger causes Imports

Bilateral Granger causality Imports Granger causes Investment

13. ODA and Imports

ODA doesn’t Granger cause Imports

Imports doesn’t Granger cause ODA No causality

14. Trade Openness and Imports

Trade Openness Granger causes Imports

Bilateral Granger causality Imports Granger causes Trade Openness

15. Remittances and Imports

Remittances doesn’t Granger cause Imports

Imports doesn’t Granger cause Remittances No causality

INVESTMENT

16. ODA and Investment

ODA doesn’t Granger cause Investment

Investment doesn’t Granger cause ODA No causality

17. Remittances and Investment

Remittances doesn’t Granger cause Investment

Investment doesn’t Granger cause Remittances No causality

18. Trade Openness and Investment

Trade Openness doesn’t Granger cause Investment

Unilateral Granger causality Investment Granger causes Trade Openness

ODA

19. Trade Openness and ODA

Trade Openness Granger causes ODA

Unilateral Granger ODA doesn’t Granger cause Trade Openness

causality

20. Remittances and ODA

Remittances Granger causes ODA

Unilateral Granger ODA doesn’t Granger cause Remittances

causality

TRADE OPENNESS

21. Remittances and Trade Openness

Remittances doesn’t Granger cause Trade Openness

Trade Openness doesn’t Granger cause Remittances No causality

The empirical findings suggest that exports, ODA, trade openness and remittances have a unilateral Granger causality with GDP and imports, investment show a bilateral Granger causality with GDP. Imports show a bilateral Granger causality with exports while investment, ODA and trade openness specify unilateral Granger causality with exports. Remittances and exports show no causal relationships with each other. Imports display a bilateral Granger causality with Investment and trade openness, and no causality with ODA and remittances. Investment portray a unilateral Granger causality with trade openness and no causality with ODA and remittances. Trade openness and remittances are found to have unilateral Granger causality with ODA and remittances and trade openness have no causal relationship.

Conclusion

We have estimated the impacts of globalization and free trade on economic growth of Pakistan by using econometric techniques of ‘auto regressive distributed lags’ (ARDL) and Granger causality, to indicate t

he significance of globalization. The empirical results approve that Pakistan’s economic growth has significantly been influenced by international trade and investment. A prime focus on investment and globalization can catalyze the process of growth in due course of time. Also, these vital linkages may become a footstep for many other least developed countries (LDCs) who are still reluctant to adopt globalization oriented policies. Also, the analysis of international trade reveals subsequent policy approvals. To enhance economic growth, the government of Pakistan can adopt exports-oriented policies that will be useful for intra-regional trade. The reduction in trade and tariff barriers and relaxation of quota items will enhance the trade with neighbouring countries which could be used to get the status of ‘most favorite nation’ (MFN) from trading partner countries. Not to mention, uninterrupted energy supplies and infrastructure facilitations can boost the exports. In the wake of globalization and trade openness, the exports could efficiently be used as an engine of growth for Pakistan.