Relationship Between Trade Liberalization, Growth, and Balance of Payments in Developing Countries: An Econometric Study

Abstract

The objectives of this article are to study the impact of liberalization on growth, the trade deficits, and the current accounts of developing countries. It is expected that trade liberalization would promote economic growth from the supply side by leading to a more efficient use of resources, by encouraging competition, and by increasing the flow of ideas and knowledge across national boundaries. Trade liberalization could lead to faster import growth than export growth, and, hence, the supply side benefits may be offset by the unsustainable balance of payments position. This study uses panel data for 42 countries (both time-series and cross-sectional) to estimate the effect of trade liberalization on growth and growth on trade balance while controlling for other factors such as the income terms of trade. The major finding of the study is that trade liberalization promotes growth in most countries, but the growth itself has a negative impact on trade balance. Countries following trade liberalization could have high real exchange rates, a worsening savings-investment balance and consequently a worsening current account and trade balance and that this would be consistent with the rational forward looking agents in which inter-temporal budget constraints are satisfied.

This study was prepared while the author was visiting the Hamburg Institute of International Economics, Hamburg, Germany. The author wishes to thank Carsten Hefeker and participants at the seminar for their useful comments, the Leverhulme Trust for the Emeritus award, T. R. Gourvish for improving the article, the referees and editor of the journal for their helpful comments, and Gurleen Popli for her statistical assistance.

Notes

¹ Edwards (1993) considers the openness-growth relationship using existing indicators. Overall, nine indicators were used

1. The Sachs-Warner index

2. The World Bank Integration index

3. the Edward Leamer Openness Index

4. the average black market premium

5. the average tariff rate as developed by Barro and Lee (1994),

6. the average coverage of non-tariff barriers (Barro and Lee)

7. the Heritage Foundation index of distortions

8. the ratio of total revenues on trade taxes to total trade and

9. the regression index of Holger Wolf on import distortions.

His conclusion on various measures was that “in spite of significant efforts and ingenuity, there has not been too much progress in this area.” There existed positive relationship between trade intensity and growth performance on the basis of cross-country plot of average annual growth rate against the average annual growth rate in trade.

²The relationship between trade balance and GDP growth rates differs between Latin American and East Asian regions because of the lack of establishment of new trading regions in Latin America and slow growth in those countries as compared to East Asia.

³Socio-political and environmental variables have played a crucial role in economic development, growth and international trade. Countries that maintain permanent peace, and domestic and political stability, and are endowed with abundant resources, including a skilled labor force, have a high potential for growth and development. Socio-political variables cannot be easily quantified. Political scientists have constructed indices measuring key socio-political conditions in each economy. Many such variables were employed in the literature to capture the presence or absence of political stability, economic freedom, corruption, democracy, economic efficiency, and other conditions affecting economic performance of countries.

⁴The date based indicator of liberalization was used in a three period analysis using a panel of three cross-sections. A panel was constructed with 3 periods, 1970–79, 1980–89, and 1990–98 in order to estimate the effects of the openness indicator on growth over a longer span of time. The estimated coefficient on liberalization variable lies between insignificant and 1.19. For the 1989–98 period, the liberalization coefficient is not significantly different from zero. In Table III, when estimates are constrained, it was found that liberalization has an effect on growth and the impact for 1 unit change is 1.62 per cent point change on growth. The conditional convergence hypothesis holds. Extreme political repression tends to reduce economic growth while an increase in density of population increases economic growth.

⁵This is based on the assumption that lenders will not permit an individual to die with unpaid debts and it would not be optimal for the individual to disappear leaving unused resources. Given the infinite time horizon, the countries are assumed to have the same behavior as individuals.

⁶The slow growing economies are likely to suffer a debt burden in the short-run. In 1991, the slow growing Argentina and Nigeria had external debt as 3.9 percent and 4.8 percent of GDP, respectively while for fast growing Thailand the proportion was only 0.2 percent (Obstfeld and Rogoff 1997).

⁷It is now a well-known proposition that with integrated global capital markets there can be no intercountry differences in returns to capital (risk-adjusted) and as capital flows to the country where the rate of return is higher the cross-country differences in marginal products of capital would disappear leading to a convergence in output per worker. This proposition has been empirically tested and a plausible conclusion to draw from the debate is that there is convergence in output per worker but it has been very slow (See Table II, Coefficient of LRGDPCH70 and LRGDPCH89).

⁸In the representative agent model, higher productivity growth will tend to weaken the current account as people borrow today against higher future income. In the overlapping generations model, productivity growth could raise the labor income of young workers but does not affect the wage incomes of older workers. As young savers will count more heavily in aggregate saving than old dissavers, saving will tend to rise and the current account to improve (Obstfeld 1995).

⁹We do not distinguish between permanent and temporary deterioration in terms of trade in the empirical study because many developing countries over a long period export primary commodities and terms of trade have either moved against primary products or have not changed drastically during the last 30 years.

¹⁰World rate of interest on loans provided by the World Bank and other multilateral bodies charge the interest rate prevailing in developed countries. In the long-run growth rates and real interest rates are not likely to diverge from each other.

¹¹Liberalization could introduce non-stationarity in output behavior which we unfortunately could not investigate with merely 30 observations for each country and in some cases it is even less. Moreover, liberalization could bring structural change and even when the unit root hypothesis was not rejected it could be due to liberalization leading to economy wide changes and repercussions. Hence we have not used unit root tests to discover non-stationarity in output behavior.

¹² Results of these can be obtained from the author. These models using panel data analysis differ according to whether they treat intercept parameters as random or fixed across the sample. The estimators in the random effects model are the generalized least squares estimators, and they combine the within and between country estimators using the corresponding residual variances as weights. For elementary panel data techniques see Johnston (1996). For special treatment of panel data models, see Baltagi (1995) and Wooldridge (2002).

¹³We expect the current account to deteriorate with an increase in debt, increase in service payments, and an increase in world interest rates. Expected signs are negative with respect to each one of them.

¹⁴Monte Carlo simulation has shown that for panels with t = 5 or 6, the bias of the coefficients of lagged dependent variable can be significant, although the bias for the coefficients on other right hand side variables tends to be minor.

¹⁵First differenced GMM estimator (equation GMM) and the extended GMM estimator (system GMM) were studied through Monte Carlo experiments by Blundell and Bond and they showed that the extended estimator has a much smaller finite sample bias and much greater precision when estimating autoregressive parameters using persistent series. Both level and first differenced instruments are used in the system GMM while the equation GMM uses only the instruments in level form.

¹⁶China is excluded from our sample of countries because it has not adopted the market system and it is hard to state that it was a liberalized economy in any of the time periods.

¹⁷We used other variations of the equations for current account and trade balance variables. Growth was never significantly related to trade balance or current account in the first period. Liberalization has reduced both trade account and current account balances in Africa in the period 1990–2000. The dominant negative effect on trade balance comes from its interactions with terms of trade. Liberalization may deteriorate terms of trade as imports expand at a faster rate and import prices might rise faster than export prices leading to deterioration in terms of trade and this in turn could lead to trade balance deterioration. For African economies, liberalization interacting with growth has a positive impact on trade balance.

¹⁸We only considered trade balance to GDP and two models, one with time dummies and another without time dummies are estimated. Growth is significantly negatively related to trade balance in both models. What we can infer here is that faster economic growth in Asian and Latin American countries would lead to a decline in trade balance as import growth far exceeds export growth. Surprisingly, growth in advanced countries does not remove pressure on trade balance or current account balance.

Bond, S. 2002. Dynamic Panel Data Models: A Guide to Micro Data Methods and Practice Institute of Fiscal Studies, London, Paper 09/02.

International Monetary Fund, 1998. World Economic Outlook, and World Bank. 1998.

Rose, A., and Ostry, J. 1989. Tariffs and the Macroeconomy: Evidence from the USA. International Finance Discussion Papers 365, Board of Governors of the Federal Reserve System.

Summers, R., Heston, A., and Aten, B. 2001. Penn World Table Version 6.0 Center for International Comparisons at the University of Pennsylvania (CICUP).