Does the presence or lack of intellectual property right protection affect international trade flows in emerging market economies? An exploratory study.

Ngassam, Christopher

ABSTRACT

This study provides new evidence regarding the effects of patent protection on international trade in developing countries also known as "emerging market economies". It employs a gravity model of bilateral trade flows and estimates the effects of increased protection on a cross-section of 69x68 countries. It improves on previous studies in two respects. First, we estimate the gravity model for two different kinds of aggregates: total non-fuel trade and high technology trade. Second, it addresses the problem of zero trade flows between countries by adopting a bivariate distributed probit regression model. Third, to measure the strength of Intellectual Property Rights (IPRs) regimes, we make use of a fine tuned index on national IPRs systems developed by Park and Ginarte (1996). Our results confirm previous findings suggesting a positive link between IPRs protection and trade flows for the non-fuel trade aggregate. However, IPRs are not found to be significant for high technology trade flows.

INTRODUCTION

Intellectual property rights (IPRs) affect international trade flows when knowledge intensive goods move across national boundaries. The importance of IPRs for trade has gained more significance as the share of knowledge-intensive or high technology products in world trade has doubled between 1990 and 2003 from 12% to 24% (UN Comtrade Data Base). At the international level, IPRs have traditionally been governed by several conventions--most prominently the Paris Convention for patents and trademarks and the Berne Convention for copyright -, which are administered by the World Intellectual Property Organization (WIPO). In the 1980s, mounting disputes over IPRs lead to the inclusion of trade-related IPRs on the agenda of the GATT/WTO Uruguay round and the resulting "Trade Related Intellectual Property Rights Agreement, including Trade in Counterfeit Goods" (TRIPs) of 1994 represents the most far-reaching multilateral agreement towards global harmonization of IPRs.

Several studies have attempted to estimate the extent to which IPRs are trade-related. Maskus and Penubarti (1995) use an augmented version of the Helpman-Krugman model of monopolistic competition to estimate the effects of patent protection on international trade flows. Their results indicate that higher levels of protection have a positive impact on bilateral manufacturing imports into both small and large developing economies. These results are confirmed by Primo Braga and Fink (1997) where we estimated a similar model and found the same positive link between patent protection and trade flows.

The purpose of this study is to provides additional evidence regarding the effects of patent protection on the international trade patterns of developing economies. It employs a gravity model of bilateral trade flows and estimates the effects of increased protection on a cross-section of 69x68 countries. The next section presents the methodology. Section III describes the empirical results obtained while Section IV concludes the paper.

METHODOLOGY

To empirically estimate the effects of increased patent protection on bilateral trade flows we use a conventional gravity model. Gravity model has been applied successfully to explain different types of international flows, such as migration, commuting, recreational traffic, and trade. Typically, they specify that a flow from country i to country j can be explained by supply conditions in country i, by demand conditions in country j, and by forces either assisting or resisting the flow's movement. Gravity models were developed based on intuitive reasoning rather than economic modeling. Due to their empirical success, there have been numerous attempts to shed some light on the economic underpinnings of the gravity equation. Linneman (1966) showed how standard gravity equation can be derived from a quasi-Walrasian general equilibrium model of export supply and import demand. Bergstrand (1989) used a general equilibrium world trade model assuming utility-and profit-maximizing agent behavior and showed that the gravity model "fits in the Heckscher-Ohlin model of inter-industry trade and the Helpman-Krugman-Markusen of intra-industry trade.

Our dependent variables are bilateral trade flows for 69x68 countries which were extracted from the United Nations Comtrade database. The data refer to 2003 total non-fuel and high technology trade. The rationale for using high technology trade flows besides total non-fuel trade is based on the a priori expectation that the effects of IPRs protection are stronger for knowledge-intensive trade.

Following earlier specifications of gravity models, our explanatory variables are GDP and population of both countries i and j, geographical distance between the two countries, a dummy variable which is one of the two countries share a common border and zero otherwise, and a dummy variable which is one of the two countries share the same language and zero otherwise. See, for example, Tinbergen (1962), Linneman (1966), Aitken (1973), Pelzman (1977), and Primo Braga, Safadi and Yeats (1994). The coefficients on GDP are expected to be positive and around unity (Anderson 1979); the coefficients on population are expected to be small and negative, representing economies of scales (Linneman 1966). Positive geographic and cultural distance are expected to have a negative influence on bilateral trade flows, that is the coefficient on geographical distance is expected to be negative, the coefficients on common border and language are expected to be positive.

Finally to capture the effect of intellectual property rights on bilateral trade flows we use the IPRs index developed by Park and Ginarte (1996). This index grades national IPRs regimes of 110 countries on a scale from zero to five. To compute a country's ranking, Park and Ginarte (1996) create five different categories--extent of coverage, membership in international patent agreements, provisions for loss of protection, enforcement mechanisms and duration of protection. For each category, they use several benchmark criteria (e.g. patentability of pharmaceuticals for extent of coverage) and compute the share of "fulfilled" criteria. A country's score is the unweighted sum of these shares over all categories. The United States receives the highest score, 4.52; several nations without patent laws (e.g., Angola, Burma, Ethiopia, Papua New Guinea) receive a score of 0.

A common problem regarding the estimation of bilateral trade flows are reported as zero because countries do not trade with each other. For example, in our data set on average about 26% of the total non-fuel trade flows and 53% high technology trade flows are zero. A standard log-linear model with a log-normally distributed error term can not , by definition, explain these zero trade flows. Simple exclusion of zero trade flows would lead to potential sample selection bias. There are several ways how to address this problem. We follow Bikker and de Vos (1992), who propose a bivariate normally distributed probit regression. The model consists of an equation for the probability of zero observations and an equation for the magnitude of a positive action:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (1)

[[gamma].sub.ij] = [x.sub.ij] [beta] + [u.sub.ij] (2)

[I.sub.ij] is the observed phenomenon which is 0 if the bilateral trade flow between country i and j zero and [y.sub.ij]--the log of bilateral trade--if the trade flow is positive; [z.sub.ij] is the log of the variables explaining the probability of a positive observation (the gravity variables without the preferential trading dummies and the Park and Ginarte index), and [gamma] the corresponding vector of coefficients for these variables. [v.sub.ij] is a normally distributed error term with mean zero; the variance of [v.sub.ij] is normalized to one as all parameters [gamma] are determined apart from a constant. [x.sub.ij] is the logarithm of the explanatory variables for positive trade flows (the gravity variables and the Park and Ginarte index), [beta] the corresponding vector of coefficients to be estimated, and [u.sub.ij] a normally distributed error term with mean zero and variance [[sigma].sup.2]. The error terms [v.sub.ij] and [u.sub.ij] are correlated with each other and drawn from a bivariate normal distribution with a correlation coefficient equal to [rho]. Equations (1) and (2) are estimated by maximum likelihood technique.

Besides addressing the problem of sample selectivity, the bivariate probit regression model is attractive because it also estimates the effects of explanatory variables (such as IPRs) on the probability that two countries trade with each other. To evaluate the robustness of the results, we estimate these two model specifications for both exports--bilateral trade flows from country i to country j as reported by country i--and imports--bilateral trade flows from country j to country i as reported by country i. Since we are primarily interested in the role of IPRs in attracting trade flows and not in creating trade flows, we only use the Park and Ginarte index of the destination country of the trade flow as explanatory variable (that is country j in the case of exports and the country i in the case of imports).

EMPIRICAL RESULTS

Our estimation results are presented in Tables 1 through 3. The overall performance of the model is quite good. Most gravity variables have the expected signs and are statistically significant. Exceptions are for total non-fuel trade (Tables 1 and 2) the coefficient on the border dummy are, however, not significant. For the high technology aggregate (Table 3), the exceptions are similar: the coefficients on the border dummy is not statistically significant. Likelihood ratio tests indicate that for all alternative specifications estimated the explanatory variables are jointly significantly different from zero.

The estimated correlation coefficients between the probit and gravity equations [rho] are always close to zero and not statistically based on a likelihood ratio test for both total non-fuel and high technology trade. For both total non-fuel imports and exports, the Park and Ginarte index has only a small effect on the probability of positive trade flows between countries, although the effect is positive and statistically significant at the 5% level for total non-fuel exports. Turning to the gravity equation, IPRs have significantly positive impact on bilateral trade flows for both total non-fuel imports and exports. Comparisons of models (I) and (II) in Tables 1 and 2 suggests that inclusion of IPRs leads to relatively small changes in the coefficients of most gravity variables. The biggest changes occur in the coefficients on GDP and population of the destination country of the trade flow. These changes can be explained by the strong correlation strength of IPRs protection and the level of economic development as measured by per capita GDP. To what extent we pick up development related effects related to bilateral trade with the Park and Ginarte index remains open to discussion.

For high technology trade in Table 3 the evolving pattern is different. For both exports and imports, the Park and Ginarte index has a significantly negative impact on the probability that countries trade with each other. The impact of IPRs on positive trade flows, in turn is slightly negative but not statistically significant. This result is somewhat surprising. If IPRs influence trade flows, we would expect this influence to be most visible for trade in knowledge-intensive goods. Several explanations can be brought forward. First, strong market power effects in the case of high technology goods may offset positive market expansion effects caused by stronger IPRs regimes. Second, stronger IPRs regimes may cause high technology firms to serve foreign markets by FDI, in-part substituting for trade flows. Third, it may be that the Park and Ginarte index does not correctly capture the IPRs effect or that development related effects interplay with stronger IPRs protection.

SUMMARY AND CONCLUSION

With an increasing share of knowledge-intensive products in international trade and the inclusion of trade-related IPRs on the agenda of the GATT/WTO, IPRs have become an important trade issue.

Economic analysis suggests that the effects of IPRs protection on bilateral trade flows are theoretically ambiguous. Because of the complex static and dynamic considerations related to a policy of tighter protection, it is difficult to generate normative recommendations. When estimating the effects of IPRs protection in a gravity model of bilateral trade flows, our empirical results suggest that, on average, higher levels of protection have significantly positive impact on non-fuel trade. However, this result is not confirmed when confining the estimation to high technology goods where we found IPRs to have no statistically significant impact.

More empirical research is needed to gain more insight regarding the IPRs-trade link, especially at industry and firm level. The challenge of such research will be to find 'natural experiments' to overcome the colineraty and endogeneity problems of the cross-country type of analyses like the present study. One alternative, for instance, would be to consider a country which at some point in the past significantly changed its system of IPRs and to test for structural change. A further important field of research is to examine the impact of tighter IPRs on FDI and their interplay with trade flows.

REFERENCES

Aitken N.D. (1973). The Effect of the EEC and EFTA on European Trade: A Temporal Cross-Section Analysis, American Economic Review, 63, 881-892.

Anderson, J.E (1979). A Theoretical Foundation for the Gravity Equation, American Economic Review, 69, 106-116.

Bergstrand, J. H. (1989). The Generalized Gravity Equation, Monopolistic Competition, and the Factor-Proportions Theory in International Trade, Review of Economics and Statistics, 71, 143-153.

Bikker, J.A. & A.F. De Vos. (1992). An International Trade Flow Model With Zero Observations: An Extension of the Tobit Model, Cahiers Economiques de Bruxelles, Nr. 135, Amsterdam: Nederlandsche Bank.

Linneman, H. (1966). An Econometric Study of International Trade Flows, Amsterdam: North-Holland Publishing Company.

Maskus, K.E. & M. Penubarti. (1995). How Trade-Related are Intellectual Property Rights?, Journal of International Economics, 39, 227-248.

Park, W.G. & J.C. Ginarte. (1996). Determinants of Intellectual Property Rights: a Cross-National Study, Manuscript, (The American University).

Pelzman, J. (1977). Trade Creation and Trade Diversion in the Council of Mutual Economic Assistance: 1954-70, American Economic Review, .67, 713-722.

Primo Braga, C.A. (1996). Trade-Related Intellectual Property Issues: The Uruguay Round Agreement and Its Economic Implications, in W. Martin & L.A. Winters (eds), The Uruguay Round and the Developing Economies, World Bank Discussion Paper No. 307, pp. 381-411, (Washington, D.C.: The World Bank).

Primo Braga, C.A. & C. Fink. (1997). The Economic Justification for the Grant of Intellectual Property Rights: Patterns of Convergence and Conflict, in F.M. Abbott & D.J. Gerber (eds), Public Policy and Global Technological Integration, (The Netherlands: Kluwer Academic Publishers).

Primo Braga, C.A., R. Safadi & A. Yeats, (1994). Regional Integration in the Americas: 'Deja Vu All Over Again?, World Economy, 17, 577-601.

Tinbergen, J. (1962). Shaping the World Economy: Suggestions for an International Policy, New York: The Twentieth Century Fund.

Christopher Ngassam, Virginia State University

Table 1: Maximum Likelihood Estimates for Total Non-Fuel Imports (a)

 Model (I)
 Equation Probit Gravity

Intercept -7.000 (-27.40) -10.228 (-29.02)
GDP (i) 0.541 (31.47) 1.109 (51.73)
GDP (j) 0.567 (32.36) 1.341 (61.89)
Population (i) -.0194 (-9.80) -0.233 (-8.53)
Population (j) -0.058 (-3.03) -0.333 (-12.76)
Distance -0.435 (-12.17) -1.109 (-23.87)
Border -0.376 (-2.32) 0.179 (0.91)
Language 0.592 (8.67) 0.861 (9.50)
IPRs (b)
[rho] 2.100
Obs. 7304 5492
[rho] -0.034
-2ln[lambda] 0.853
([rho] = 0) (c)
-2ln [lambda]({[gamma], 8874.433
[beta]}= 0) (c)

 Model (II)
 Equation Probit Gravity

Intercept -6.960 (26.28) -10.956 (-30.58)
GDP (i) 0.545 (29.90) .949 (34.98)
GDP (j) 0.566 (32.33) 1.339 (62.12)
Population (i) -0.198 (-9.17) -0.082 (-2.64)
Population (j) -0.058 (-3.03) -0.336 (-12.97)
Distance -0.437 (-12.15) -1.060 (-23.20)
Border -0.378 (-2.33) 0.239 (1.27)
Language 0.591 (8.66) 0.867 (9.62)
IPRs (b) -0.014 (-0.53) 0.369 (9.59)
[rho] 2.083
Obs. 7304 5492
[rho] -0.043
-2ln[lambda] 1.346
([rho] = 0) (c)
-2ln[lambda]({[gamma], 8965.677
[beta]}= 0) (c)

(a) t-statistics in parentheses

(b) Park and Ginarte index of the destination country of
the trade flow, that is country (j) in the case of exports and
country (i) in the case of imports.

Table 2: Maximum Likelihood Estimates for Total Non-Fuel Exports (a)

Model (I)
Equation Probit Gravity

Intercept -6.631 (-27.77) -10.791 (-29.31)
GDP (i) 0.556 (33.86) 1.374 (60.26)
GDP (j) 0.458 (29.84) 1.017 (46.85)
Population (i) -0.052 (-2.84) -0.320 (-12.18)
Population (j) -0.153 (-8.15) -0.137 (-4.90)
Distance -0.473 (-13.55) -1.114 (-23.69)
Border -0.393 (-2.54) 0.301 (1.52)
Language 0.588 (8.96) 0.826 (8.95)
IPRs (b)
r 2.113
obs. 7309 5294
r 0.005
-2lnl (r = 0) (c) 0.016
-2lnl ({g,b}= 0) (c) 8520.968

Model (II)
Equation Probit Gravity

Intercept -6.766 (-27.10) -11.170 (-29.55)
GDP (i) 0.556 (33.85) 1.374 (60.38)
GDP (j) 0.443 (25.93) 0.945 (35.11)
Population (i) -0.052 (-2.83) -0.3320 (-12.20)
Population (j) -0.137 (-6.57) -0.070 (-2.17)
Distance -0.467 (-13.34) -1.100 (-23.41)
Border -0.381 (-2.47) 0.328 (1.65)
Language 0.588 (8.97) 0.826 (8.98)
IPRs (b) 0.047 (1.92) 0.176 (4.46)
r 2.109
obs. 7309 5294
r 0.002
-2lnl (r = 0) (c) 0.003
-2lnl ({g,b}= 0) (c) 8544.524

(a) t-statistics in parentheses

(b) Park and Ginarte index of the destination country
of the trade flow: country (j) for exports and country (i) for imports.

Table 3: Maximum Likelihood Estimates for High Technology Imports (a)

Model (I)

Equation Probit Gravity

Intercept -5.494 (-27.17) -14.487 (-26.21)
GDP (i) 0.568 (40.12) 0.911 (22.68)
GDP (j) 0.495 (36.36) 1.898 (52.12)
Population (i) -.0324 (-18.71) -0.086 (-2.06)
Population (j) -0.170 (-10.31) -0.733 (-20.70)
Distance -0.421 (-13.56) -1.115 (-19.11)
Border 0.011 (0.08) 0.157 (0.64)
Language 0.480 (8.54) 1.154 (9.53)
IPRs (b)
r 2.229
Obs. 7304 3548
r 0.066
-2lnl (r = 0) (c) 1.354
-2lnl ({g,b}= 0) (c) 7606.860

Model (II)

Equation Probit Gravity

Intercept -4.794 (-22.87) -14.313 (-26.95)
GDP (i) 0.717 (39.04) 0.960 (16.69)
GDP (j) 0.512 (36.45) 1.897 (52.38)
Population (i) -0.474 (-22.59) -0.132 (-2.38)
Population (j) -0.175 (-10.43) -0.731 (-20.70)
Distance -0.466 (-14.62) -1.124 (-19.00)
Border -0.110 (-0.78) 0.141 (0.61)
Language 0.488 (8.43) 1.146 (9.49)
IPRs (b) -0.340 (-14.09) -0.093 (-1.50)
r 2.228
Obs. 7304 3548
r 0.064
-2lnl (r = 0) (c) 1.309
-2lnl ({g,b}= 0) (c) 7812.274

(a) t-statistics in parentheses

(b) Park and Ginarte index of the destination country of the
trade flow: country (j) for exports and country (i) for imports.