the effects of trade liberalization on imports in selected developing countries

The Effects of Trade Liberalization on Imports

in Selected Developing Countries

AMELIA U. SANTOS-PAULINO *

University of Kent at Canterbury, UK

Summary. — This paper analyses the impact of the reduction of tariff and nontariff barriers on theimports of 22 selected developing countries, utilizing dynamic panel data techniques. Domesticincome and relative prices are found to be significant determinants of import growth. In addition,the results indicate that import duties reduce import growth, but the effect varies according to theregion and the type of trade policy regime existing in the country. The results also show that theelimination of trade policy distortions has a strong, positive impact on import growth. Finally, it isfound that income and price elasticities are higher as a result of trade policy reform. � 2002Elsevier Science Ltd. All rights reserved.

Key words — trade liberalization, import growth, dynamic panel data, time series/cross-section,

developing countries

1. INTRODUCTION

In the formulation of trade and/or exchangerate policies, one of the major concerns ofpolicy makers is the responsiveness of tradeflows to relative price changes and incomevariations. The effect of trade and exchangerate policies is highly dependent upon the sizeof import and export price and income, elas-ticities.

As far as the analysis of import price andincome elasticities is concerned, the empiricalinvestigation of import demand functionshas been one of the most researched areasin international economics. 1 Economists havededicated a considerable amount of effort to theestimation of import demand functions, both atthe aggregate and disaggregated levels. Esti-mated elasticities are of significant practicalimportance to policy makers, where the elas-ticities derived are a crucial link betweeneconomies, and to determine the degree towhich the external balance constraint affects acountry’s growth performance.

Few studies however, have analyzed the im-pact of trade liberalization on import behavioracross developing countries (exceptions are, forexample, Bertola & Faini (1991) and Faini,Pritchett, & Clavijo (1992)). Knowledge of themajor variables that affect import performance,and the prediction of import flows, can help

policy makers to design and assess the overallsustainability of structural reforms. They areemployed, for example, as inputs into theconfiguration and implementation of structuraladjustment programs, for determining theappropriate speed of the trade liberalizationprocess, and for avoiding the possibility ofunexpected foreign exchange constraints en-dangering the reform effort.

The prediction of import response followingtrade liberalization measures is not an easytask, especially when extensive nontariff barri-ers on imports are present. Quotas, for in-stance, affect the responsiveness of imports toreal exchange rates, tariffs and activity levels(domestic output). But the combined effects ofimport barriers, both quantitative and non-quantitative, are hard to gauge because of theconstraints on data availability.

This paper specifies an import growthfunction, and presents new and relatively

World Development Vol. 30, No. 6, pp. 959–974, 2002� 2002 Elsevier Science Ltd. All rights reserved

Printed in Great Britain0305-750X/02/$ - see front matter

PII: S0305-750X(02)00014-1www.elsevier.com/locate/worlddev

* I thank Tony Thirlwall for valuable discussions. I am

also gratefull to Alessandra Guariglia, Andy Dickerson,

Miguel Le�oon-Ledesma, Valpy Fitzgerald and seminar

participants at the university of Kent and the Latin

American Economic Research Network (LAERN),

Oxford University, for helpful suggestions and com-

ments. The remaining errors are mine. Final revision

accepted: 6 February 2002.

959

comprehensive evidence regarding importgrowth in 22 selected developing countries,focusing on the impact of import controls, i.e.,tariff and nontariff barriers. The research ap-plies dynamic panel data models based on fixedeffects and generalized methods of moments(GMM). In addition, heterogeneous panels forthe complete sample, as well as for the differentregions of the world, are estimated using atime-series/cross-section (TSCS) technique. Thecountries are classified according the degree oftrade policy distortion based on the HeritageFoundation Index of Economic Freedom(O’Driscoll et al., 1999). The trade policy dis-tortion score is based on a country’s averagetariff rate, nontariff barriers, and corruption inthe Customs Services.

The plan of the paper is as follows. Section 2analyzes the theoretical specification and em-pirical evidence on import demand functions.Section 3 presents descriptive evidence of tradepolicy in the 22 countries considered. Section 4,specifies the import growth function to bestudied, and reports the empirical results. Fi-nally, conclusions are summarized in Section 5.

2. IMPORT DEMAND FUNCTIONS:THEORY AND EMPIRICAL EVIDENCE

Traditional import demand functions makeimports a function of domestic income anddomestic prices relative to the price of importsubstitutes. If the price and income elasticitiesof demand are assumed constant, the importfunction can be written as:

M ¼ PfEPd

� �w

Y p ð1Þ

where Y represents domestic income; Pf is for-eign prices; E is the nominal exchange rate; Pd isdomestic prices; w is the price elasticity of de-mand for imports; and p is the income elasticityof demand for imports. The price elasticity ofdemand for imports is expected to be negative,while the income elasticity is positive.

Taking logs of Eq. (1) and differentiatingwith respect to time, the growth of imports canbe expressed as:

m ¼ wðPf þ e� PdÞ þ pðyÞ ð2ÞThe partial adjustment form of the tradi-

tional import demand equation, in which im-port growth is assumed to adjust only partiallyto the difference between equilibrium import

growth in period t and the actual growth ofimports in the previous period, is representedas:

mt ¼ b0 þ b1pmþ b2y þ b3mt�1 þ lt ð3Þwhere b1 ¼ w and b2 ¼ p (i.e., the short-runprice and income elasticities); pm is the growthin relative prices; y is the growth in domestic(real) income and lt is the error term. The long-run price and income elasticities are given byb1=ð1 � b3Þ and b2=ð1 � b3Þ respectively.

As stressed earlier, there are a large numberof empirical studies, which estimate importdemand functions for both developed and de-veloping countries. These studies have usedmainly ordinary least-squares (OLS) and in-strumental variable techniques, assuming sta-tionary data. 2 Examples of import demandstudies for advanced economies are Kohli(1991), Urbain (1992), Deyak, Sawyer, andSprinkle (1993), Clarida (1994), Mah (1994),Marquez (1994), Carone (1996), and Masih andMasih (2000). Deyak, Sawyer, and Sprinkle(1989) and Pattichis (1999) estimate disaggre-gated import demand functions. Examples ofimport demand studies for developing countriesare Marquez and McNeilly (1988), Mah(1992, 1993, 1997), Bahmani-Oskooee and Rhee(1997), Bahmani-Oskooee and Niroomand(1998), Senhadji (1998), and Reinhart (1995).The conclusion of these studies is that, in gen-eral, income and relative prices are significantdeterminants of import performance, but theprice elasticities tend to be low, in most casesfar below unity. Income elasticities, however,tend to be above unity. Reinhart (1995) pro-vides a set of interesting results for developingcountries, where the elasticities differ consider-ably across regions. The price elasticities for theregions are: Latin America: �0.36; Asia: �0.40;Africa: �1.36; All countries: �0.53. The incomeelasticities are: Latin America: �0.96; Asia:�1.39; Africa: �1.14; All countries: �1.22.

(a) Import demand and trade liberalization

The influence of trade liberalization on im-port performance, and also the behavior ofimport demand elasticities during the processof reform, have been analyzed in different ways.Melo and Vogt (1984) propose two interestinghypotheses in this regard, for which they foundsupport by analyzing the case of Venezuela.First, they suggest that as the degree of importliberalization increases, the income elasticityof demand increases. That is, the relaxation of

WORLD DEVELOPMENT960

controls will tend to increase the incomeelasticity automatically. Second, as economicdevelopment proceeds, the price elasticity ofimport demand also rises as the ability to sub-stitute domestic production for imports (importsubstitution) become easier.

Nevertheless, the subsequent empirical evi-dence regarding the hypotheses has not beenconclusive. For instance, Boylan and Cuddy(1987) examined the two hypotheses for the caseof Ireland and did not find empirical support forthem. Mah (1999) argues, however, that Boylanand Cuddy’s findings are misleading because ofmethodological shortcomings. Mah (1999) ex-amines the Melo and Vogt (1984) hypothesesduring the process of economic development inThailand using, according to the author, a moreappropriate empirical technique. The resultssupport the hypothesis related to the incomeelasticity, showing that the income elasticityincreased as a result of trade liberalization. Theprice elasticity was not, however, found to beresponsive to trade liberalization.

Bertola and Faini (1991) provide one of theearliest studies of the impact of trade liberal-ization on import demand for a developingeconomy, accounting for the response of im-ports to the elimination of tariff and nontariffbarriers. Through the development of a theo-retical model and empirical application toMorocco, the authors show that quantitativerestrictions (QRs) had a significant impact notonly on the level of imports, but also on theirsensitivity to income and price variations. Forinstance, the authors demonstrated that, hadQRs for consumption goods been lifted in 1985(the date used for the prediction test), theirincome elasticity would have increased from0.93 to 1.20.

Faini et al. (1992) study the impact oftrade policy on import demand in developingcountries. The authors focus on the impact ofimport controls, assuming two categories ofimports, that is, those subject to quantitativerestrictions, and those that can freely enter thecountry. They show that estimated incomeelasticities in developing countries are generallyhigher than unity, and that relative prices aresignificant with an elasticity less than unity.Another finding is that when the lack of foreignexchange or, more generally, a restrictive traderegime effectively constrains import flows, themeasured impact of price and income elastici-ties becomes less evident. The Faini et al. re-sults suggest that the real effects of income andprice changes (e.g., a devaluation) on import

behavior are more evident when the impact ofimport controls and/or liberalization policies isalso included in the analysis. Thus, import de-mand studies, which do not evaluate the effectof import policy changes, should be interpretedwith caution, as far as the estimates of the in-come and price elasticities are concerned.

3. DESCRIPTIVE EVIDENCE ON TRADEPOLICY

One of the most important purposes of es-tablishing the General Agreement on Tariffsand Trade (GATT) in 1947 was to achieve asubstantial reduction of tariffs and nontariffbarriers to trade, and thus to secure freer accessof countries to international markets. Sincethen, trade policy reforms in developing coun-tries have been widely documented, mainlybecause trade reforms have become an integralpart of structural adjustment programs re-quired by international organizations, as a re-sult of international indebtedness (see IMF(1998); Dean, Desai, & Riedel (1994); UNC-TAD (1999, 2000)). Furthermore, the change inintellectual thinking, and the empirical evidenceprovided by multicountry studies (using differ-ent indicators of trade distortions), which an-alyze the virtues of a more outward-orientedeconomy, and the failures of protectionist pol-icies in some developing countries, were crucialfactors behind trade policy reform. The obvi-ous policy implication from this literature isthat developing countries should abandonprotectionist and restrictive trade strategies andopen their foreign trade sectors. 3

In relation to tariffs, there have been impor-tant developments in terms of their application,and substantial reforms and reductions of tar-iffs have been achieved in the successive roundsof trade negotiations. It is difficult to contendhowever, that the same achievement has beenreached regarding nontariff barriers, which areconsidered as a significant impediment to trade,and which now attract most of the effort ontrade negotiations and reforms.

Nontariff barriers affecting imports cantake various forms, among which can be men-tioned: quantitative import restraints (the pro-hibition or restriction of imports maintainedthrough import licensing requirements); stan-dard and administrative requirements, appliedfor ensuring quality of goods seeking accessinto domestic markets (which countries useas a protectionist measure); anti-dumping and

EFFECTS OF TRADE LIBERALIZATION 961

Countervailing Measures, used (and permittedby the WTO under special circumstances) toprotect domestic industry from material injuryaccording to GATT article VI, arising fromdumped or subsidized imports; governmentprocurement; services barriers; lack of adequateprotection of intellectual property rights; etc.The existence of such barriers is a reason ofconcern, but the WTO accepts the applicationof some of them as a form of protection ofdomestic markets, especially for health and/orsanitary reasons. Nevertheless, there is a moralhazard that countries might abuse the applica-tion of nontariff barriers, and such norms canbecome a disguised way of protectionism.

Table 7 in the appendix presents a summaryof the main trade policy reform measures un-dertaken in the countries analyzed in this study.It also provides the timing of the reforms,

where the ‘‘most recent’’ year of reform is usedas an indicator of trade liberalization in theempirical analysis. Even though the countriesreviewed undertook necessary reforms to re-duce trade distortions (especially imports), thecourse of reform was diverse, and in many casesreforms are still outstanding. The simplificationof import procedures, the reduction or elimi-nation of quotas, and the rationalization oftariff structures are the most widespread re-forms. As far as trade policy indicators areconcerned, import tariffs provide a convenientindicator of the negative impact that tradetaxes can have on import growth in quantita-tive terms.

Table 1 presents a summary of the behaviorof import taxes, as a percentage of total im-ports, before and after the identified date of themost significant trade reforms. In most of the

Table 1. Import duties and import growth before and after trade liberalizationa

Country Year of liberaliza-tion

Before liberalization (from 1976) After liberalization (up to 1998)

Import duty Import growth Import duty Import growth

06 d < 5%Indonesia 1986 4.82 7.55 5.11 8.59

56 d < 10%Costa Rica 1990 9.72 3.95 7.97 9.05Korea 1990 8.68 12.53 5.55 9.35Malaysia 1988 8.41 10.06 5.09 15.48Mexico 1986 8.27 7.94 4.69 13.44Paraguay 1989 8.59 11.02 4.88 22.88

106 d < 15%Chile 1976 13.93� 2.49 12.42 9.88Philippines 1986 13.48 2.84 13.97 13.12Sri Lanka 1990 13.38 7.93 13.41 8.77Thailand 1986 12.81 6.12 9.66 11.66Venezuela 1991 10.05 6.40 10.11 12.71Zambia 1990 10.07 �4.23 16.67 2.34

156 d < 20%Colombia 1991 15.04 4.99 9.17 14.52Dom. Rep. 1992 18.98 5.38 15.03 8.05Ecuador 1991 15.81 1.78 8.89 6.69Morocco 1984 19.11 3.31 16.65 6.49Uruguay 1985 16.27 �1.53 10.65 11.39

20%6 dCameroon 1991 21.65 6.77 20.98 3.40India 1991 38.59 6.79 27.77 10.93Malawi 1991 21.47 2.28 21.36 1.83Pakistan 1991 26.82 3.63 21.33 4.84Tunisia 1989 23.83 6.01 21.29 4.63

Sources: Dean et al. (1994), UNCTAD (1999), World Bank, World Development Indicators (WDI) (1999), WTO

Trade Policy Reviews (various issues).a d denotes import duties. The values are period averages, and are the author’s calculations. The data for Chile’simport duties before liberalization corresponds to the year 1975.


cases, the reforms, specifically import liberal-ization episodes, can be linked to reductions inimport taxes. Countries such as Colombia,Dominican Republic, Ecuador, India, Korea,Malaysia, Mexico, and Paraguay present no-table reductions in import duties, coincidingwith the periods of liberalization. Other econ-omies such as Chile, Thailand, Morocco, Uru-guay, Cameroon, Malawi and Tunisia alsomanaged to reduce import duty receipts as aproportion of total imports. In most of thecases, import growth increased following theliberalization policies. It would not be com-pletely appropriate however, to try to establisha relationship between trade liberalization andimport growth, based on import tariff reduc-tions alone, because some countries adopteddifferent approaches to the reform process, andthere are nontariff barriers to trade that mighthave also affected import growth. Moreover, insome countries import tariffs increased due tothe fact that such economies decided to elimi-nate nontariff barriers, and to convert theserestrictions into tariffs.

Regarding nontariff barriers, it is very diffi-cult to express in a single number or indicatortheir weight as a trade policy measure, mainlybecause they are very country specific, and be-cause the information is not always readilypublished, especially in the case of developingcountries. Some authors have used differentcriteria to try to measure nontariff barriers, andone of them is the use of dummy variables topresent structural breaks in policy reforms (seeGreenaway (1993)).

In connection to this, since 1995, the Heri-tage Foundation has developed an Index of

Economic Freedom, which provides an annualexamination of the factors that contribute mostdirectly to economic freedom and prosperity.The index includes the broadest array of insti-tutional factors, and one of the key elements inmeasuring economic freedom is trade policy.The five broad categories of countries, based onthe trade policy grading scale, are very low, low,moderate, high and very high.

Table 2 shows the trade policy scores, whichare based on a country’s average tariff rate thehigher the rate, the worse (or higher) thescore. 4 The other factors of trade policy arenontariff barriers and corruption in the Cus-toms Services. This index takes values of 1–5and tries to measure the extent to which gov-ernment policy acts as a disincentive to trade.As can be observed from Table 2, on average,most of the countries considered in this studyfall in the ‘‘moderate’’ category, which coincidewith the level of tariffs expressed in Table 1. Inaddition, countries that were acknowledged ashaving ‘‘very high’’ trade policy restrictionsappear to apply very high tariff levels. But,given the mixed evidence regarding trade policypractices, it is suitable to rely on the empiricalscrutiny, which is undertaken in the followingsection to analyze the relationship betweentrade liberalization and import performance.

4. THE MODEL AND RESULTS

(a) The model

Traditional dynamic import demand func-tions relating import flows to relative price and

Table 2. Classification of countries according to the Heritage Foundation trade policy grading scale: 1995–2000a

Level of protectionism Criteria Countries

Very low ATR6 4% and/or very lownon-tariff barriers.

Low 4 < ATR6 9% and/or lownon-tariff barriers.

Chile, Paraguay Uruguay

Moderate 9 < ATR6 14% and/or moderatenon-tariff barriers

Columbia, Costa Rica,Ecuador, Korea,Malaysia, Mexico

Philippines, Thailand, SriLanka, Venezuela, Zam-

biaHigh 14 < ATR6 19% and/or high

non-tariff barriers.Dominican Republic,

IndonesiaMorocco

Very high 19%6ATR and/or very highnon-tariff barriers that virtually

close the market to imports

Cameroon, India,Malawi

Pakistan, Tunisia

Source: Heritage Foundation Index of Economic Freedom, various issues (see Johnson & Sheehy (1995); Johnson

(1998a), Johnson (1998b); Johnson & Holmes (1998), O’Driscoll et al. (1999)).a ATR denotes average tariff rate. The validity of the Heritage’s classification of the countries was confirmed bycomparing with the IMF (1998) trade policy rating (for those countries for which the scores were available).


domestic incomes are estimated. Moreover, thefunctions include the effect of import duties onimport growth, and the effect of trade liberal-ization through the use of dummy variables.We also test if the income elasticity of importdemand changes with trade liberalization, andalso if the price elasticity changes as the abilityto substitute domestic production for importsbecomes easier. The assumption is that tradeliberalization (i.e., the reduction and/or elimi-nation of trade policy distortions) has a sig-nificant impact not only on the autonomousgrowth of imports, but on their sensitivity toincome and price variations as well.

Using the dynamic specification of the im-port growth function presented in Eq. (3), theaugmented import growth function which alsoallows for the effects of import duties and tradeliberalization on import growth, can be ex-pressed as:

mit ¼ ai þ b1pmit þ b2yit þ b3mit�1 þ b4dit

þ b5libit þ eit ð4Þ

where ai are country-specific effects (when paneldata are used), dit is import duties, and libit is ashift dummy variable for the years followingsignificant liberalization. The rest of the vari-ables are as defined earlier, and we expectb1 < 0, b2 > 0, 0 < b3 < 1, b4 < 0 and b5 > 0.

Trade liberalization can also affect the priceand income elasticities themselves, as suggestedby the Melo and Vogt (1984) hypotheses dis-cussed earlier. Such interaction effects can beestimated by including two slope dummy vari-ables, y � lib and pm� lib, to capture the jointeffects of the elimination of import distortionmeasures on income and price elasticities, re-spectively. Thus we also estimate:

mit ¼ ai þ b1pmit þ b2yit þ b3mit�1

þ b4dit þ b5libit þ b6ðpm� libÞitþ b7ðy � libÞit þ eit ð5Þ

(b) Empirical estimations

To estimate the import growth functionspresented in Eqs. (4) and (5), which allow forthe effects of import policy distortions and tradeliberalization on import performance, twoapproaches are used. The first is the estimationof dynamic panel models by fixed-effects(FE), and by GMM. The second approach is

the use of TSCS analysis for all the countries,as well as for the different geographic regionsstudied of Africa, East Asia, South Asia, andLatin America. Dynamic panel data analysis isused because of the assumption of the laggedadjustment of import growth to the growth ofthe previous period, as outlined in Section 2.

(i) Fixed effects and generalized methods ofmoments

In this section, the two forms of panel datamodel mentioned above are estimated. First,the FE estimator, based on the inclusion ofdummy variables to account for factors that arespecific to each country but constant over time(Greene, 1997); and, second, dynamic paneldata models based on GMM (Arellano, 1993;Arellano & Bond, 1998). 5

The results from both estimators are reportedin Table 3. The fixed effects results in column (i)show that all the arguments of the importgrowth function have the expected sign. Therelative price indicator is not however, statisti-cally significant. The coefficient on the laggeddependent variable is so small that there isvirtually no difference between short and longrun income and price elasticities. The importtariff coefficient (�0.20) and the trade lib-eralization dummy (3.20) are statistically sig-nificant, showing that both the reduction ofimports duties, as well as the reform of thetrade policy regime, had a marked impact onimport performance.

In order to assess the direct impact of tariffson import growth, import duty elasticities werecalculated from the estimated regression coef-ficients (see Table 6). The estimated elasticity is�0.42, and the impact of trade liberalization iscalculated to have raised the growth of importson average by 3.20 percentage points or 73%.

Column (ii) provides the estimates that testthe Melo and Vogt (1984) hypotheses. Theshort run income and price elasticities are bothstatistically significant as are the import dutyand trade liberalization coefficients. The importduty coefficient is �0.20, and the shift dummycoefficient of 6.19 shows that there is a signifi-cant import response to trade liberalization.With regard to the direct impact of tariffs onimport growth, the estimated import dutyelasticity is �0.43, and the calculated relativeimpact of liberalization on import growth is1.41, which implies that trade reforms increasedimport growth by over 100% (see Table 6).


Regarding the slope dummy variables, theMelo and Vogt (1984) hypotheses are con-firmed, since both the coefficients are signifi-cantly different from zero, and they show theexpected signs. Moreover, the two interactionvariables are jointly significant, with an F-sta-tistic of F ð2; 475Þ ¼ 9:61.

Turning now to the GMM estimates, theresults presented in columns (iii) and (iv) en-dorse the findings of the fixed effects estima-tions. Income and price elasticities are strongly

statistically significant and the magnitudes aresimilar to the fixed effect estimates. More in-teresting are the results concerning import du-ties and trade liberalization measures. As canbe seen from column (iii), import tariffs nega-tively affect imports: the estimated coefficient is�0.35, which implies an import duty elasticityof �0.67 (see Table 6). The independent impactof trade liberalization as shown by the tradeliberalization coefficient (1.99) is statisti-cally significant. That is, the implementation of

Table 3. Import performance in selected developing economies: 1976–97a

Explanatory variables Dependent variable: import growth mt

Fixed effects GMM

(i) (ii) (iii) (iv)

pm �0.10 �0.15 �0.18 �0.16(1.28)b (2.89)� (2.43)�� (2.66)�

y 1.65 1.91 1.31 2.60(10.03)� (8.47)� (5.48)� (5.41)�

mt�1 0.01 0.10 0.04 0.08(0.48) (2.53)�� (0.43) (0.20)

d �0.20 �0.20 �0.35 �0.43(2.26)�� (2.34)�� (1.86)�� (3.60)�

lib 3.20 6.19 1.99 9.10(2.17)�� (3.03)� (4.22)� (2.12)��

y � lib 0.59 0.93(1.81)�� (2.57)�

pm� lib �0.23 �0.40(2.12)�� (2.85)��

yLR 1.67 2.12 1.36 2.82pmLR �0.10 �0.17 �0.19 �0.17

Diagnostic statistics

R2 0.48 0.59Omit y � lib, pm� lib 9.61� 15.92�

Heteroskedasticity test 1.93 24.5��

Wald test [0.000] [0.000]Sargan test [0.452] [0.482]First-order serial correlation [0.000] [0.008]Second-order serial correlation [0.418] [0.436]Number of observations 504 504 386 386

a yLR and pmLR are the long run income and price elasticities respectively. Omit y � lib, pm� lib is the F-statistic forthe omission of these two variables from the regression. The heteroskedasticity test is based on a regression of theresiduals on the squared fitted values. The Wald test is for the joint significance of the regressors. The Sargan test is ofover-identifying restrictions. The tests for first and second order serial correlation are asymptotically distributed asstandard normal variables (see Arellano & Bond (1991)). The p-values report the probability of rejecting the nullhypothesis of serial correlation, where the first differencing will induce (MA1) serial correlation if the time-varyingcomponent of the error term in levels is a serially uncorrelated disturbance. The GMM estimations were performedusing the program DPD98 for Gauss (Arellano & Bond, 1998).b Figures in parentheses ( ) are absolute t-ratios.� p value 1%.�� p value 5%.�� p value 10%.


import liberalization policies raised importgrowth by 38%, according to the calculationspresented in Table 6.

Column (iv) shows the GMM results, whichconsider the interaction effects involving tradeliberalization and income and price elasticities,proposed by the Melo-Vogt hypotheses. Look-ing at the import duty coefficient, it is statisti-cally significant and the magnitude of thecoefficient is higher than in the previous case(�0.43). Also, the estimated import duty elas-ticity (�0.82) verifies the strong negative impactthat tariffs inflict on import growth. The tradeliberalization coefficient is 9.10, and the relativeimpact of liberalization on import growth iscalculated as 1.76, indicating that the lesseningof trade policy barriers contributes to an in-crease in import growth by more than 100%. 6

The Melo-Vogt hypotheses are again con-firmed.

(ii) Regional and trade policy disaggregationIn this section, a panel data model is imple-

mented which is appropriate to analyze dataobserved for a relatively large number of peri-ods and for a relatively small number of crosssectional units. The TSCS model allows for theerror term of each cross-section unit to be freelycorrelated across equations. The relevance ofthis type of model is that the error term doesnot need to be the same for each country; 7

thus, it is suitable to analyze region or groupspecific estimations.

The evaluation of trade policy reforms in thedifferent regions, and the classification ofcountries according to the degree of restrictionof the trade policy regime, suggest that theimpact of trade liberalization on import growthmight differ across those regions or groups ofcountries. Moreover, the price and incomeelasticities can vary across such groups, as sug-gested by studies, which deal with multicountrystudies of import demand functions (see Rein-hart (1995); Bahmani-Oskooee & Rhee (1997);and Senhadji (1998)).

(iii) The impact of liberalization according toregion

Table 4 presents the estimation of Eq. (5) forthe different geographical regions that comprisethis study. The countries in the sample wereclassified into four zones: Africa, East Asia,South Asia, and Latin America. The results forall the countries in column (i) validate thefindings of the dynamic panel data estimates.The growth of real GDP proves to have a sig-

nificant, positive impact on import growth, asshown by the short and long run income elas-ticity values (1.65 and 1.68, respectively). Theimport price elasticity has the expected sign,although is not statistically significant. The re-sults also confirm the significant influence thatimport tariffs and trade liberalization have onimport growth. Moreover, there is evidencethat trade liberalization raises the sensitivity ofimports to real income growth, and to relativeprice changes.

As expected, the region-specific results pre-sent more mixed outcomes. Regarding the in-come and price elasticities, the East Asia regionpresents the highest long run income elasticity(1.98), although Latin America and Africa alsoshow relatively high income elasticities (1.79and 1.33, respectively). Regarding the long runprice elasticity, Latin America possesses thehighest elasticity (�0.17), while Africa has thelowest elasticity, (�0.01), (which is not statis-tically different from zero). 8

The trade policy related indicators also pro-vide diverse results across the regions analyzed.The estimated import duty coefficients are sig-nificantly different from zero in all cases, exceptAfrica, and the magnitude of the coefficients isrelatively high in the different regions. To bemore specific, regarding the expected negativeimpact of tariffs on import growth, the calcu-lated import duty elasticities are very similar inEast Asia, South Asia and Latin America,where the elasticities are �0.65, �0.69, and�0.49, respectively (see Table 6). Although theimport duty coefficient is not significantlydifferent from zero in the African case, thecalculated elasticity (�1.04) shows a highernegative impact of tariffs on imports than in theother regions. It can be attributed to the factthat most African countries in our samplepossess a very high import duty ratio, bothbefore and after liberalization. 9

The results also show that trade policy re-forms emerge as a more important determinantof import growth in all of the regions. Ac-cording to the estimations, trade liberalizationhas a stronger impact on import growth inAfrica, that is 8.44 percentage points, whichimplies that liberalization increased importgrowth by more than 1,000% in this region(starting from a very low base). The impactof liberalization in East Asia, South Asiaand Latin America is also comparatively high�4.12, 1.41, and 1.10 percentage points re-spectively. This means that trade liberalizationincreased import growth by 61% in East Asia,


21% in South Asia, and 26% in Latin America(see Table 6). The results show that the elimi-nation, and/or reduction, of restrictive mea-sures affecting international trade has played amajor role in affecting import growth.

(iv) The impact of liberalization according todegree of protection

Another set of TSCS estimations are pre-sented in Table 5, which portrays the analysis ofimport growth in the sample of countries di-vided according to classification of the tradepolicy regimes based on the criteria discussed inSection 3 (see Table 2). 10 As in the region-specific case, the results vary according to thecharacteristics of the particular categories,providing insight into the differential impactthat a countrys trade policy (i.e., tariff andnontariff barriers) has on import growth, and onthe arguments of the import growth function.

The estimated import duties coefficient issignificantly different from zero in all the casesexcept in the ‘‘low-moderate’’ category, and thecoefficient is larger, i.e., more negative, in the‘‘high-very high’’ classification, as expected.Moreover, the coefficient is relatively impor-tant for ‘‘all countries.’’ Regarding the calcu-lated import duty elasticities (shown in Table6), import tariffs have a stronger negative effecton imports in countries with ‘‘high-very high’’restrictive trade regimes, and the elasticity is�0.74. In the ‘‘low-moderate’’ category, importtariffs have a smaller impact on import growth,�0.04.

The estimates for the trade regime liberal-ization, which involve the reduction and/or elimination of major import barriers, appearto be a more crucial determinant of importgrowth. Moreover, the findings confirm thepositive impact that trade reform has on import

Table 4. Two-step generalized least squares and maximum likelihood estimationa

Explanatory variables Dependent variable: import growth mt

All countries (i) Africa (ii) East Asia (iii) South Asia (iv) Latin America(v)

pm �0.11 �0.01 �0.16 �0.09 �0.19(0.60)b (1.40) (1.84)� (1.15) (4.21)��

1.65 1.44 1.92 0.80 2.01y (11.05)�� (4.27)�� (8.05)�� (1.87)� (8.37)��

mt�1 0.02 �0.08 0.03 0.03 �0.12(2.49)�� (0.91) (0.48) (0.23) (2.59)��

d �0.20 �0.15 �0.72 �0.20 �0.35(2.39)�� (0.69) (2.68)�� (2.03)�� (1.78)��

lib 6.73 8.44 4.12 1.41 1.10(3.37)�� (3.13)�� (1.99)�� (2.97)�� (1.93)�

y � lib 0.22 1.53 0.41 3.76 0.42(1.96)� (3.26)�� (1.79)�� (3.82)�� (1.70) ��

pm� lib �0.24 �0.34 �0.29 �0.05 �0.21(3.52)�� (2.34)�� (2.36)�� (0.18) (2.03)��

yLR 1.68 1.33 1.98 0.82 1.79pmLR �0.11 �0.01 �0.16 �0.09 �0.17

Diagnostic statistic

LRS 232.92 13.98 8.90 4.31 45.48[38.93] [18.31] [18.31] [7.81] [51.00]

Number of observations 462 105 105 63 189

a yLR and pmLR are the long run income and price elasticities respectively. Omit y � lib, pm� lib is the F-statistic forthe omission of these two variables from the regression. Likelihood ratio statistic (LRS) is a test for serial correlation.The numbers in brackets [ ] are the critical values.b Figures in parentheses ( ) are absolute t-ratios� Significant at the 5% level.�� Significant at the 1% level.�� Significant at the 5% level.


growth. According to the results, trade liber-alization has the strongest positive impact onimport growth in those countries with ‘‘high-very high’’ restrictive trade regimes. The influ-ence of trade liberalization on import growth inthis case is 4.93 percentage points, or in otherwords, liberalization increased imports by145%. The relative impact of liberalization incountries with ‘‘low-moderate’’ trade policydistortions is smaller in comparison to theother classifications, which is to be expectedgiven the initial conditions regarding tradepolicy and the use of instruments that directly

affect import flows. In this case, trade liberal-ization has a relatively small impact of only19% (or 1.02 points). It is clear from thisanalysis that initial trade policy conditions arealso of paramount importance for importgrowth in the liberalization process.

5. CONCLUSIONS AND POLICYIMPLICATIONS

This paper has examined, in a comprehensiveand systematic fashion, import growth func-tions for a selection of developing countries.This is a detailed analysis of the major tradepolicies implemented in the 22 countries studiedduring 1976–98, and dynamic panel data, andTSCS techniques have been applied to estimatethe impact of trade reforms on import growthon average, and across regions.

Over the period considered, estimates of theincome and price elasticities of demand forimports fall within the boundaries of the elas-ticities found in the previous empirical litera-ture. The price and income elasticities have theexpected sign and, in general, are statisticallysignificant.

Table 5. Two-step generalized least squares and maxi-mum likelihood estimationa

Explanatoryvariables

Dependent variable: import growth mt

All countries(i)

Low–moder-ate (ii)

High–veryhigh (iii)

pm �0.01 �0.11 �0.01(0.42)b (2.66)� (1.28)

y 1.66 2.00 1.03(11.03)� (11.45)� (4.82)�

mt�1 0.01 0.10 �0.04(0.32) (2.66)� (0.62)

d �0.25 �0.03 �0.16(2.75)� (1.68) (2.65)�

lib 5.94 1.02 4.93(2.76)� (3.49)� (1.85)��

y � lib 0.36 0.41 0.60(3.26)� (9.02)� (2.18)��

pm� lib �0.24 �0.19 �0.33(3.33)� (2.34)�� (1.34)

yLR 1.68 2.22 0.99pmLR �0.01 �0.12 �0.01

Diagnosticstatistic

LRS 228.40 232.97 32.11[36.19] [26.22] [32.67]

Number ofobservations

420 273 147

a yLR and pmLR are the long run income and price elas-ticities respectively. LRS is a test for serial correlation.The numbers in brackets [ ] are the critical values. In thisset of estimations, Indonesia and Zambia are not in-cluded because they switched regimes during the period.Hence, the sample size is slightly smaller than in Tables 3and 4.b Figures in parentheses are absolute t-ratios.� Significant at the 1% level.�� Significant at the 10% level.�� Significant at the 5% level.

Table 6. Import duty elasticities and relative impact oftrade liberalizationa

Estimationmethod

Import dutiesðldÞ

Liberalization(% lib)

Fixed effects:Eq. (4) �0.42 0.73Eq. (5) �0.43 1.41

GMM:Eq. (4) �0.67 0.38Eq. (5) �0.82 1.76

Time series/cross-section (Eq. (5)):All countries �0.42 1.53Africa �1.04 11.11East Asia �0.65 0.61South Asia �0.69 0.21Latin America �0.49 0.26

All countries �0.50 1.27Low–moderate �0.04 0.19High–very high �0.74 1.45

a The import duty elasticity is calculated as ld ¼ðd=mÞðom=odÞ ¼ ðd=mÞb1, where d and m are the meansof import duties and import growth respectively. Theproportionate impact of trade liberalization is calculatedas %lib ¼ b5=mlib¼0, where b5 is the coefficient of tradeliberalization and mlib¼0 is the mean of import growthbefore liberalization.


The main purpose of this paper, however, hasbeen to examine the extent to which trade-related measures affect import growth in devel-oping countries. One of the indicators used toassess the impact of trade policy distortions onimport growth is the level of import duties. Itwas found that import duties reduce importgrowth, but the effect varies according to theregion and the type of trade policy regimeprevailing in the country. The calculated im-port duty elasticities vary considerably amongregions. Africa seems to be the most affected bythe level of import tariffs, but the negative effectis also appreciable in the other regions. Inaddition, the impact of import duties differssignificantly according to the degree of protec-tionism and/or distortions of trade policy pre-vailing in the countries. Import duties affectmostly countries classified as having high andvery high levels of protectionism.

As described earlier, the countries analyzedhave all undertaken profound trade reforms, inwhich the reduction and simplification ofimport tariffs, as well as the elimination ofnonquantitative restrictions, were fundamentalelements of the liberalization process. The re-sults here provide empirical evidence support-ing the premise that the elimination of tradepolicy distortions has a strong, positive impacton import growth. Specifically, trade liberal-ization has more than doubled import growthon average across all countries, but, as the re-gionally disaggregated estimations show, theimport duty elasticities and the relative impactof trade liberalization vary considerably acrossregions and types of trade policy regimes.

We also tested for the Melo and Vogt (1984)hypotheses. The Melo-Vogt hypothesis con-cerning the increase of income elasticities fol-lowing import liberalization is supported by thedifferent exercises undertaken in this study. Thehigher income elasticities after liberalizationreflect an increase in the degree of openness tointernational trade of the countries analyzed.The hypothesis of an increase in price elastici-

ties as a result of import reform is also con-firmed in most cases.

The response of import growth to trade lib-eralization measures does not come as a sur-prise, in the light of trade policy practices indeveloping countries before undertaking thereforms. The policies include the use of devel-opment policies focused on restrictive tradepolicies such as import substitution strategy,which concentrate on enhancing the ability tosubstitute domestic production for imports.

Finally, the excessive import growth follow-ing trade liberalization episodes has seriouspolicy implications, especially for the balanceof trade and balance of payments, because inmost cases imports increase by more than ex-ports, causing trade imbalances. 11 The balanceof payments crises suffered by a large numberof developing countries have revealed the ex-tent to which growth rates have been con-strained by their balance of payments positions(see Khan & Zahler, 1985). In many cases,trade policy reforms have not been accompa-nied by an export promotion strategy, whichwould compensate for the higher imports gen-erated by the relaxation of import barriers. Thebalance of payments crises suffered by a largenumber of developing countries have revealedthe extent to which growth rates have come todepend on steadily rising export earnings andcapital inflows, and how disruptive an interrup-tion to these sources of foreign exchange can be.If import growth is faster than export growth inthe process of trade liberalization, an importantissue is the sustainability of the trade accountdeficit, and that depends on the sequence ofreforms, the efficacy of real exchange rates as abalance of payments adjustment mechanisms,and the inflows of foreign capital. Liberaliza-tion needs to take place in such a way as tomaintain a sustainable balance of paymentsposition; otherwise the resource gain from lib-eralization can easily be offset by real resourceslosses arising from the need for balance ofpayments adjustments (i.e., devaluation).

NOTES

1. See Khan (1974). Some of the earlier studies that

estimate import demand functions are surveyed by

Goldstein and Khan (1985). Other early studies on

import and export demand elasticities for different

countries are: Kreinin (1967, 1973); Houthakker and

Magee (1969); Khan (1974, 1975); Goldstein and Khan

(1976, 1978); Murray and Ginman (1976); Wilson and

Takacs (1979); Warner and Kreinin (1983); Haynes and


Stone (1983); Bahmani-Oskooee (1986); and Marquez

(1990). The literature on import demand functions is

discussed in more detail in Section 2.

2. The unit root tests and cointegration technique

provide a more appropriate method of estimating long

run elasticities in a time series framework. Bahmani-

Oskooee and Niroomand (1998) provide a discussion of

the issue.

3. See Edwards (1992, 1993); Krueger (1998); Ben-

David, Nordstr€oom, and Winters (1999); Rodr�ııguez and

Rodrik (2000); Thirlwall (2000).

4. If average tariff rates are not available, the average

rate is determined by calculating the revenue generated

from tariffs and duties as a percentage of total imports.

Moreover, the information on the overall tariff structure,

its various rates, and the items to which these rates apply

to estimate an effective tariff rate are analyzed.

5. The GMM estimator is preferred by some authors

to the fixed effects estimator to estimate dynamic panel

data models since the dynamic fixed-effects model

produces estimates that are inconsistent if the number

of ‘‘individuals’’ tends to infinity while the number of

time periods T is fixed (see Nerlove, 1967; Nickell, 1981;

and Harris & M�aaty�aas, 1986). More specifically, the bias

is of order (1/T). In the present case, the number of years

is large (T ¼ 23) and thus the bias should be minimal.

The GMM estimator is based on first differencing and

controls for the endogeneity of the lagged dependent

variable (and can also control for the potential endoge-

neity of other explanatory variables––see Arellano

(1993); Arellano & Bond (1998)).

6. It follows that these results could have implications

for the balance of trade and payments. There is evidence

that trade liberalization worsens the trade balance

significantly (see UNCTAD (1999)).

7. The estimator is a two-step generalized least

squares estimator with maximum likelihood esti-

mates (MLE) interaction. The model allows for group-

wise heteroscedasticity, cross-group correlation, and

within-group autocorrelation (see Greene (1997), Chap-

ter 16).

8. For Africa and Latin America, the negative coeffi-

cient on the lagged dependent variable indicates a

nonequilibrium (i.e., divergent) path for import growth.

9. As can be seen in Table 1, Cameroon, Malawi and

Tunisia fall into the categories of countries with high

import duties as a share of total imports, with duty

ratios above 20% before and after liberalization.

10. The Heritage classification starts from 1995. Thus,

in order to assess the impact of tariff and nontariff

barriers for the whole number of observations analyzed

in the present study, a new classification covering the

complete period (1976–98) was undertaken. This classi-

fication is based on the Heritage Foundation’s criteria

(see Table 8 in the appendix).

11. For example, Santos-Paulino (2002) finds that

trade liberalization is a significant determinant of export

performance, but its effect varies across continents while

export duties have a small detrimental effect on export

growth. In addition, Santos-Paulino and Thirlwall

(2001) show that liberalization has stimulated export

growth, but has raised import growth by more, leading

to a worsening of the balance of trade and the balance of

payments.

REFERENCES

Arellano, M. (1993). On testing of correlation effectswith panel data. Journal of Econometrics, 59(1), 87–97.

Arellano, M., & Bond, S. (1991). Some tests of speci-fication for panel data: Monte Carlo evidence and anapplication to employment equations. Review ofEconomic Studies, 58, 277–297.

Arellano, M., & Bond, S. (1998). Dynamic panel dataestimation using DPD98 for Gauss: a guide for users,Mimeo.

Bahmani-Oskooee, M. (1986). Determinants of interna-tional trade flows: the case of developing countries.Journal of Development Economics, 20, 107–123.

Bahmani-Oskooee, M., & Niroomand, F. (1998). Longrun price elasticities and the Marshall–Lerner con-dition revisited. Economics Letters, 61, 101–109.

Bahmani-Oskooee, M., & Rhee, H. (1997). Structuralchange in import demand behavior. Journal of PolicyModeling, 19, 345–350.

Bahmani-Oskooee, M., & Mirzai, A. (2000). Real andnominal effective exchange rates for developingcountries: 1973:1–1997:3. Applied Economics, 32,411–428.

Ben-David, D., Nordstr€oom, H., & Winters, A. (1999).Trade, income disparity and poverty. Geneva, Swit-zerland: World Trade Organization.


Bertola, G., & Faini, R. (1991). Import demand andnon-tariff barriers the impact of trade liberalization.Journal of Development Economics, 34, 269–286.

Boylan, T. A., & Cuddy, M. P. (1987). Elasticities ofimport demand and economic development. Journalof Development Economics, 26, 301–309.

Carone, G. (1996). Modeling the US demand forimports through cointegration and error correction.Journal of Policy Modeling, 18, 1–48.

Clarida, R. (1994). Cointegration, aggregate consump-tion, and the demand for imports: a structuraleconometric investigation. American Economic Re-view, 94, 298–308.

Dean, J. M., Desai, S., & Riedel, J. (1994). Trade policyreform in developing countries since 1985: a reviewof the evidence. World Bank Discussion Papers, 267.

Deyak, T. A., Sawyer, W. C., & Sprinkle, R. L. (1989).An empirical examination of the structural stabilityof disaggregated US import demand. Review ofEconomics and Statistics, 71(2), 337–341.

Deyak, T. A., Sawyer, W. C., & Sprinkle, R. L. (1993).The adjustment of Canadian import demand tochanges in income, prices, and exchange rates.Canadian Journal of Economics, 26, 890–900.

Edwards, S. (1992). Trade orientation, distortions andgrowth in developing countries. Journal of Develop-ment Economics, 39, 31–57.

Edwards, S. (1993). Openness trade liberalization andgrowth in developing countries. Journal of EconomicLiterature, 31, 1358–1393.

Faini, R., Pritchett, L., & Clavijo, F. (1992). Importdemand in developing countries. In M. G. Dagenais,& P.-A. Muet (Eds.), International Trade Modeling,International Studies in Economic Modeling No. 11(pp. 279–297).

Goldstein, M., & Khan, M. (1976). Large versus smallprice changes and the demand for imports. IMF StaffPapers, 22, 200–325.

Goldstein, M., & Khan, M. (1978). The supply anddemand for exports: a simultaneous approach. Re-view of Economics and Statistics, 60, 275–286.

Goldstein, M., & Khan, M. (1985). Income and priceeffects in foreign trade. In R. Jones, & P. Kenen(Eds.), Handbook of International Economics (pp.1041–1099). Amsterdam: North-Holland.

Greenaway, D. (1993). Liberalising foreign tradethrough rose tinted glasses. Economic Journal, 103,208–223.

Greene, W. (1997). Econometric Analysis (third ed.).London: Prentice Hall.

Harris, M. N., & M�aaty�aas, L. (1986). A comparativeanalysis of different estimators for dynamic paneldata models. Working Paper, Department of Econo-metrics, Monash University, May.

Haynes, S. E., & Stone, J. A. (1983). Secular and cyclicalresponses of US trade to income: an evaluation oftraditional models. Review of Economics and Statis-tics, 65(1), 87–95.

Houthakker, H. S., & Magee, S. P. (1969). Income andprice elasticities in world trade. Review of Economicsand Statistics, 51(2), 111–125.

International Monetary Fund (1998). Trade liberal-ization in IMF-supported programs. World Eco-

nomic and Financial Surveys. Washington DC:IMF.

International Monetary Fund (1999). Dominican Re-public: selected issues. IMF Staff Country Report(No. 99/117). Washington DC: IMF.

Johnson, B., & Sheehy, T. (1995). 1996 Index ofEconomic Freedom. Washington, DC: The HeritageFoundation.

Johnson, B., Holmes, K. R., & Kirkpatrick, M. (Eds.).(1998a). 1997 Index of Economic Freedom. Washing-ton, DC: The Heritage Foundation.

Johnson, B., Kirkpatrick, M., & Holmes, K. R. (Eds.).(1998b). 1998 Index of Economic Freedom. Washing-ton, DC: The Heritage Foundation.

Johnson, B., & Holmes, K. R. (Eds.). (1998). 1999 Indexof Economic Freedom. Washington, DC: The Heri-tage Foundation.

Khan, M. S. (1974). Import and export demand indeveloping countries. IMF Staff Papers, 21, 678–693.

Khan, M. S. (1975). The structure and behavior ofimports of Venezuela. Review of Economics andStatistics, 57, 221–224.

Khan, M. S., & Zahler, R. (1985). Trade and financialliberalisation given external shocks and inconsistentdomestic policies. IMF Staff Papers (March), 22–25.

Kohli, U. (1991). Technology duality and foreign trade.Ann Arbor: University of Michigan Press.

Kreinin, M. E. (1967). Price elasticities in internationaltrade. Review of Economics and Statistics, 49(4), 510–516.

Kreinin, M. E. (1973). Disaggregated import demandfunctions further results. Southern Economic Journal,40, 19–25.

Krueger, A. O. (1998). Why trade liberalization is goodfor growth. Economic Journal, 108(450), 1513–1522.

Mah, J. S. (1992). On the behavior of import demand ofKorea. Asian Economic Journal, 6, 115–119.

Mah, J. S. (1993). Structural change in import demandbehavior: the Korean experience. Journal of PolicyModeling, 15, 223–227.

Mah, J. S. (1994). Japanese import demand behavior:the cointegration approach. Journal of Policy Mod-eling, 16, 291–298.

Mah, J. S. (1997). Cyclical factors in import demand.Journal of Policy Modeling, 19, 323–331.

Mah, J. S. (1999). Import demand liberalization andeconomic development. Journal of Policy Modeling,21(4), 497–503.

Marquez, J. (1990). Bilateral trade elasticities. TheReview of Economics and Statistics, 72, 70–77.

Marquez, J. (1994). The econometrics of elasticities orthe elasticity of econometrics: an empirical analysisof the behavior of US imports. The Review ofEconomics and Statistics, 76, 471–488.

Marquez, J., & McNeilly, C. (1988). Income and priceelasticities for exports of developing countries. TheReview of Economics and Statistics, 72(2), 306–314.

Masih, R., & Masih, A. M. M. (2000). A reassessment oflong-run elasticities of Japanese import demand.Journal of Policy Modeling, 22(5), 625–639.

Melo, O., & Vogt, M. G. (1984). Determinants of thedemand for imports of Venezuela. Journal of Devel-opment Economics, 14, 351–358.


Michaely, M., Papageorgiou, D., & Choksi, A. (1991).Liberalizing Foreign Trade: vol. 7. Lessons of expe-rience in the developing world. Basil Blackwell:Oxford and Cambridge MA.

Murray, T., & Ginman, P. J. (1976). An empiricalexamination of the traditional aggregate importdemand model. Review of Economics and Statistics,58(1), 75–80.

Musonda, F., & Adam, C. (1999). Trade liberalization inZambia 1970–95. In A. B. Oyejide, A. B. Ndulu, & J.W. Gunning (Eds.), Regional Integration and TradeLiberalization in Sub-Saharan Africa (vol. 2: CountryCase-Studies) (pp. 450–547). London: Macmillan.

Nerlove, M. (1967). Experimental evidence on theestimation of dynamic economic relations from atime series of cross-sections. Economic Studies Quar-terly, 18, 42–74.

Nickell, S. (1981). Biases in dynamic models with fixedeffects. Econometrica, 49(6), 1417–1426.

O’Driscoll, Jr.,, G. P., Holmes, K. R., & Kirkpatrick, M.(Eds.). (1999). 2000 Index of Economic Freedom.Washington, DC: The Heritage Foundation.

Pattichis, C. A. (1999). Price and income elasticities ofdisaggregated import demand: results from UECMsand an application.AppliedEconomics, 31, 1061–1071.

Reinhart, C. (1995). Devaluation, relative prices, andinternational trade. IMF Staff Papers, 42(2), 290–312.

Rodrik, D., 1997. Trade policy and economic perfor-mance in Sub-Saharan Africa. A study commissionedby the Swedish Ministry of Foreign Affairs, Novem-ber.

Rodr�ııguez, F., & Rodrik, D. (2000). Trade policy andeconomic growth: a sceptic’s guide to the cross-national evidence. In B. Bernanke, & K. S. Rogoff(Eds.), Macroeconomics Annual 2000. The MITPress for NBER: Cambridge, MA.

Santos-Paulino, A. U. (2002). Trade liberalisation andexport performance in selected developing countries.Journal of Development Studies.

Santos-Paulino, A. U., & Thirlwall, A. P. (2001). Theimpact of trade liberalisation on export Growth,import growth, the balance of trade and the balanceof payments of developing countries. University ofKent at Canterbury, Mimeo.

Senhadji, A. (1998). Time series estimation of structuralimport demand equations: a cross-country analysis.IMF Staff Papers, 45(2), 236–268.

Thirlwall, A. P. (2000). Trade agreements trade liberal-ization and economic growth: a selective survey.African Development Review (December), 129–160.

UNCTAD (1999). Trade and development report,United Nations Conference on Trade and Develop-ment. Geneva: United Nations Conference on Tradeand Development (UNCTAD).

UNCTAD (2000). The least developed countries report2000. Geneva: United Nations Conference on Tradeand Development (UNCTAD).

Urbain, J. P. (1992). Error correction models foraggregate imports the case of two small openeconomies. In M. G. Dagenais, & P.-A. Muet(Eds.), International Trade Modeling, InternationalStudies in Economic Modeling No. 11 (pp. 237–279).New York: Chapman and Hall.

Warner, D., & Kreinin, M. E. (1983). Determinants ofinternational trade flows. Review of Economics andStatistics, 65(1), 96–104.

Wilson, J. F., & Takacs, W. E. (1979). Differentialresponses to price and exchange rate influences inthe foreign trade of selected industrial countries.Review of Economics and Statistics, 61(2), 267–279.

Winglee, P., Boonekamp, C., Cho, Y. J., Fritz-Krockow,B., & Uimonen, P. (1992). Issues and developmentsin international trade policy. In World Economic andFinancial Surveys, Ch. V (August). InternationalMonetary Fund: Washington DC.

World Bank (1999). World Development Indicators, CD.World Trade Organization (1995a). Cameroon trade

Policy Review. Geneva, Switzerland: WTO.World Trade Organization (1995b). Costa Rica trade

Policy Review. Geneva, Switzerland: WTO.World Trade Organization (1995c). Pakistan trade

Policy Review. Geneva, Switzerland: WTO.World Trade Organization (1995d). Sri Lanka trade

Policy Review. Geneva, Switzerland: WTO.World Trade Organization (1996a). Colombia trade

Policy Review. Geneva, Switzerland: WTO.World Trade Organization (1996b). Dominican Republic

trade Policy Review. Geneva, Switzerland: WTO.World Trade Organization (1996c). Zambia trade Policy

Review. Geneva, Switzerland: WTO.World Trade Organization (1997a). Chile trade Policy

Review. Geneva, Switzerland: WTO.World Trade Organization (1997b). Malaysia trade

Policy Review. Geneva, Switzerland: WTO.World Trade Organization (1997c). Mexico trade Policy

Review. Geneva, Switzerland: WTO.World Trade Organization (1997d). Paraguay trade

Policy Review. Geneva, Switzerland: WTO.World Trade Organization (1998a). India trade Policy

Review. Geneva, Switzerland: WTO.World Trade Organization (1998b). Indonesia trade

Policy Review. Geneva, Switzerland: WTO.World Trade Organization (1998c). Uruguay trade

Policy Review. Geneva, Switzerland: WTO.World Trade Organization (1999a). The Philippines

trade Policy Review. Geneva, Switzerland: WTO.World Trade Organization (1999b). Thailand trade

Policy Review. Geneva, Switzerland: WTO.

APPENDIX A. DATA DEFINITIONS ANDSOURCES

Import Growth (m): Imports of Goods andServices; annual percentage growth (constant1995 US$). Source: World Bank, World De-velopment Indicators (WDI) (1999).

Income Growth (y): GDP; annual percentagegrowth (constant 1995 US$). Source: WorldBank, World Development Indicators (WDI)(1999).

Import Duties (d): Import duties (percent ofimports). Import duties comprise all levies


Table 7. Imports and exchange liberalization in developing countries (1972–98)a

Region/country

Reform Trade reform measures

First Most recent

South AsiaIndia 1989 1991 Some tariff increases and net additions to the OGL. Significant

liberalization of tariffs and QRs in the 1991 reform program.Unification of the exchange rate regime.

Pakistan 1989 1991 Replaced nontariff barriers with tariff, reduction of maximum tariffrates and reduction of exemptions from tariff.Some restrictions in capital account transactions were removed in1991, and new instruments are not subject to exchange controls.

Sri Lanka 1987 1990 Reduction of the range and number of goods requiring licensing.Exchange rate reform started in 1984. By 1994 most exchangecontrols were removed.

East AsiaIndonesia 1985 1990 Reduction in the coverage of nontariff import barriers. Tariffs were

reduced to around 10% by 1993.Korea 1984 1990 Removed nonagricultural QRs.

1998 Reduction of unweighted average tariff.Malaysia 1986 1989 Tariffs reductions were made during 1988–92, in items including

food, household goods, clothing, and electronic goods.Philippines 1986 1989 Gradual replacement of QRs with tariffs. Reduction of tariff bands.

Reform of customs procedures.Thailand 1982 1990 Elimination of nonagricultural QRs.

Tariffs reductions program, but later reversed.AfricaCameroon 1989 1991 Elimination of QRs on imports.Malawi 1988 1991 Reduction of import duties. Limitation of foreign exchange alloca-

tion to a small negative list. Transfer QRs to surtaxes.Morocco 1983 1989 Reduction in QRs on noncompetitive goods. Sharp reduction of

maximum tariff; new tariff surcharge.Tunisia 1987 1990 Gradual replacement of QRs with surcharges. Tariff reduction;

increases in surcharges.Zambia 1990 1990 Gradual increase of OGL and exchange rate unification. Reduction

of maximum tariff rate and range.Latin AmericaChile 1985 1988 By 1985 Chile had virtually no QRs (and prohibited by the

constitution). Reduction of uniform nontariff rate.Colombia 1985 1991 Significant reduction in both levels and dispersion of tariff rates and

expanded the number of tariff positions on the free import list.Elimination of import licensing, reductions in the levels of tariffs,reduction of the number of tariffs from 14 in 1990 to four in 1993,and liberalization of the exchange rate.Supplement of tariff reductions by a competitive exchange rate. In1991 all foreign exchange operations were to be transacted at themarket determined exchange rate, and foreign exchange controlswere relaxed, and foreign licenses were abolished.

Costa Rica 1985 1990 Reduction in average tariff rates and a decrease in the dispersion ofrates.In January 1992 the foreign exchange system was deregulated, floatedthe exchange rate, opened the capital account, and eliminated foreignexchange controls.

DominicanRepublic

1990 1992 Nontariff barriers were largely dismantled. Tariffs reform (bothnumber and rates). Reduction of import surcharge, and furtherabolition in 1995. Customs modernization. Simplification of theexchange rate system.

Ecuador 1985 1991 Segmented elimination of QRs. Tariff maximum reduced to 35%.Mexico 1985 1988 Progressive removal of import restrictions and their replacement with

tariffs. In 1986 import-licensing coverage was reduced, and in 1987 allminimum prices were eliminated. The QRs have been almosteliminated from intermediate capital goods.

Continued next page


collected on goods at the point of entry into thecountry. They include levies for revenue pur-poses or import protection, whether on a spe-cific or ad-valorem basis, providing they arerestricted to imported products. Data areshown for central government only. Source:World Bank, World Development Indicators(WDI) (1999).

Real Effective Exchange Rate (pm): The rel-ative price of imports, ðPfE=PdÞ, where Pf isforeign prices; E is the nominal exchange rate;and Pd is domestic prices, is calculated as theinverse of the REER. Data for the REER forColombia, Costa Rica, Ecuador, India, Indo-nesia, Malaysia, Mexico, Pakistan, Philippines,Singapore, Sri Lanka, Thailand, and Tunisiaare from Bahmani-Oskooee and Mirzai (2000).The REERs for the remaining countries areconstructed from IMF’s International FinancialStatistics (various issues) (Tables 7 and 8).

Table 8. Classification of countries according to tradepolicy regimea

Classification/countries

Low–moderateChile ParaguayColombia PhilippinesCosta Rica Sri LankaEcuador ThailandKorea UruguayMalaysia VenezuelaMexico

High–very highCameroon MoroccoDominican Republic PakistanIndia TunisiaMalawi

a The classification presented in this table is based on theHeritage Foundation criteria in terms of tariffs andnontariffs barriers. The background information is takenfrom Tables 1, 2, and 7.

Table 7—continued

Region/country

Reform Trade reform measures

First Most recent

In 1991 the foreign exchange markets were unified, and a band withinthe peso was allowed to fluctuate was established.

Paraguay 1989 1995 Simplification of the tariff structure and reduction of rates. Nontariffbarriers applied to few agricultural products were replaced by tariff.Exchange controls were abolished, establishing a free-floatingexchange rate.

Uruguay 1983 1985 QRs and other barriers to trade were removed, trade regulations weresimplified, and a gradual process of reducing import duties wasestablished. Administrative controls have been reduced.

Venezuela 1989 1991 Virtual elimination of up front QRs (re-established by 1992).Reduction and rationalization of maximum tariff.Unification of the four markets exchange rates. Foreign exchangecontrols were abolished.

Sources: Dean et al. (1994); Greenaway (1993); IMF (1998, 1999); Michaely, Papageorgiou, and Choksi (1991);

Rodrik (1997); Musonda and Adam (1999); UNCTAD (1999); Winglee, Boonekamp, Cho, Fritz-Krockow, and

Uimonen (1992); WTO Trade Organisation (1995a–d, 1996a–c, 1997a–d, 1998a–c, 1999a,b).a OGL denotes open general license; QRs denotes quantitative restrictions.


the effects of trade liberalization on imports in selected developing countries

Documents