|
|
|
|
Journal of International Relations and Development
Economic Integration in Trade and Foreign Direct Investment: Dynamic Considerations of Potential and Adjustment
by Peter Egger
*
Introduction
Projecting so-called natural trade flows or potential trade relations was a major task of empirical trade economics in the last decade. The main aim was to indicate the potential of economic integration with respect to trade that was not exhausted due to impediments of a mostly political but also economic type. This kind of research was usually based on the estimation of a gravity equation (Linnemann 1966). The gravity equation represents a reduced form which explains bilateral trade through supply and demand GDP (Gross Domestic Product) and population numbers between two partner-countries as well as factors of trade preferences (common language and borders, etc.) and resistance (distance between economic centres, etc.). Many studies tried to project the integration potential of Central and Eastern European countries (CEECs) after the fall of the Iron Curtain (Wang and Winters 1991; Hamilton and Winters 1992; Rosati 1992; Baldwin 1994; etc.).
In the first step, bilateral trade relations between a sample of reference market economies were estimated. In the second, it was assumed that trade relations between e.g. the CEECs and the reference market economies – mostly countries belonging to the European Union (EU) and the Organisation for Economic Co-operation and Development (OECD) – can be explained in the long-run by applying the estimated relationship to the cultural characteristics, geographical location and stage of economic development of the respective countries.
The abovementioned pioneers of such studies discovered substantial deviations in actual bilateral trade between the West (EU or OECD) and the CEECs in terms of their potential values. On the other hand, relatively strong relationships were found for intra-CEEC trade (i.e. trade between the CEECs). Thus, the high share of intra-CEEC trade and the closed nature of the respective economies facing their Western neighbours were seen to stem from pressures within the Council for Mutual Economic Aid (COMECON). Altogether, this was taken to indicate the large unexhausted East-West potential of integration at the time these economies started to open up in the early 1990s, which were accordingly expected to generate lively, dynamic processes involving their relations with Western Europe or the OECD.
Initially, it was pointed out that the adjustment of bilateral trade with the CEECs to its full potential would be a medium- or even long-term phenomenon. Therefore, it was somewhat surprising that after a time span of less than a decade the entire gap between potential and real trade relationships between the EU and the CEECs was closed, and no integration effects were left unexploited (Gros and Gonciarz 1996; Breuss and Egger 1999).
The purpose of this article is to investigate the adjustment processes in bilateral exports and stocks of outward foreign direct investment (FDI) which are associated with different kinds of shocks in explaining the factors underlying both. We present a case study focussing on bilateral relationships between the EU (as the sending countries) and three CEECs (Czech Republic, Hungary and Poland as the recipient countries). Previous studies (Wang and Winters 1991; Hamilton and Winters 1992; Baldwin 1994; etc.) were of a static type and only concerned the projection of long-term potential between the EU (OECD) and the CEECs, without consideration of the associated speed of adjustment. The present article demonstrates that a dynamic analysis provides information about the adjustment processes after shocks in the determinants that cannot be derived from simple static applications. A dynamic treatment of bilateral exports (and also of FDI) could perhaps help to better understand which time spans we should think of with regard to adjustment in the long-run after a shock or actual through to "natural" export and FDI relationships.
Bilateral Trade and Foreign Direct Investment
As mentioned above, the working model for analysing potential economic relationships was the gravity model. It was chosen because of its simple structure, its intuitive appeal and its success in exploring the volume and direction of bilateral trade. It has recently been applied, not only to bilateral trade, but also to FDI (e.g. Brenton et al. 1999). Although it has certainly demonstrated a remarkably good fit, its theoretical foundation is not entirely convincing (Hamilton and Winters 1992; Leamer and Levinsohn 1995).
More importantly, the work of the New Trade Theory focussing on the determinants of and relationship between multinational (MNE) and national (NE, i.e. exporters) enterprises (Markusen and Venables 1996; 1998) has largely been ignored in applications of the previous gravity approach. We shall therefore relate the empirical suggestions to this literature. This choice of theoretical background involves a set of explanatory variables that mainly captures overall bilateral country size, relative country size (similarity of size in terms of GDP), relative factor endowments, trade resistance factors (transport costs) and various economies of scale (Markusen and Maskus 1999a; 1999b; see Egger 2000 for a more detailed discussion of the implications).
So far, only static (cross-section as well as panel) analyses have been undertaken and used to analyse and project bilateral economic relationships. 1 On the one hand, this did not allow us to model the inertial reactions of both exports and FDI to changes in their determinants (theoretically motivated by the impact of adjustment costs of investment). On the other, the impact of linkages between trade and FDI (Caves 1996) were similarly neglected. 2 However, if adjustment costs do play any role then the relationship between trade and FDI should also exhibit inertia after a shock.
We propose to undertake an experiment, applying the parameter estimates presented in Egger (2000) also for relations between the EU and the CEECs. Hence, we take the parameters estimated for intra-EU relations as given and analyse the model projections for shocks in several exogenous determinants for the bilateral economic relations between the EU and three CEECs with respect to both the long-run and the dynamic paths of adjustment. 3
Table 1 summarises the parameter results from Egger (2000) for a specification in which endowment differences are twofold. Both the difference in the relationship of physical capital to unskilled labour (low enrolled people) and that of human capital (high enrolled people) to unskilled labour were arrived at in separate ways. For more details about the estimated equations and the construction of variables, readers are referred to the Appendix and to Egger (2000).
It should be noted that only the long-run effects of changes in the exogenous determinants on both exports and stocks of outward FDI are based on the abovementioned theoretical background. However, for a smooth process of adjustment we would also expect the (short-run) effects of the lagged endogenous variables to exhibit a positive sign and a coefficient smaller than one. The third column for both the export and FDI results indicates the sign of the effect of a positive shock in several exogenous determinants on bilateral exports and stocks of outward FDI as expected theoretically. In sum, we mainly have identical signs of the long-run effects of a shock in the exogenous determinants on both exports and FDI. Hence, exports and outward FDI are complementary with respect to changes in these variables. The only exceptions are relative corporate tax rates and transport costs, where the former would not be expected in theory, while the latter is consistent.
Of course, a larger bilateral economic space (combined amounts of GDP) positively influences both exports and FDI. On the other hand, there are positive effects on both, emerging from a higher degree of similarity in terms of the relative sizes of two countries. The more different countries become in terms of difference in the physical capital relative to the unskilled factor endowment, the higher both exports and FDI, while the opposite applies for the high-skilled relative to low-skilled endowment. See Egger (2000) for more detail on the theory, estimation results and test statistics.
Long-run relations between the European Union and Central and Eastern European Countries: Speeds of Adjustment
Assuming that these parameter estimates are also valid for relations between the eu and the ceecs, we turn to the following application. Given the corresponding variables for an average EU country and the three CEECs, we will undertake the experiments, assuming a shock in the explanatory variables pertaining to the CEECs. Specifically, we will change those variables making CEECs more (yet not fully) similar to the average EU member-country. We will derive the long-run effects on bilateral exports and FDI from the average EU member-state (as the source country) to the CEECs (as the host countries) and the associated speed of adjustment which can be simulated, relying on the relevant parameters for lagged exports and stocks of outward FDI in both the equations of bilateral exports and FDI.
Table 2 presents descriptive statistics for the relevant variables in our context. 4 First, we find that the real exports (stocks of outward FDI) to GDP ratio is much higher for intra-EU relations than for trade (FDI) of a typical EU member-state with one of the selected CEECs. Second, real GDP per capita is of course much higher for the average EU country than for any of the three CEECs. Third, the physical capital endowment relative to unskilled labour is higher for a typical EU-member than for a country from Central or Eastern Europe. Fourth, the ratio of high-skilled to unskilled labour is slightly lower for all CEECs than for the EU. Lastly, transport costs 5 are significantly lower for intra-EU exports than for trade with the CEECs, the only exception being EU trade with Poland.
In our experiment, we focus on six different types of shocks. We will assume a ceteris paribus shift in the explanatory variables of the CEECs and compare the outcome for exports and stocks of outward FDI for the relations of a typical EU member-state as the sending country with each of the CEECs with base-year values (i.e. for 1996). However, we are not only interested in the size of the long-run effect of a shock but also in the time span of adjustment. This depends on both the estimated parameters for adjustment costs (lagged endogenous effects) and linkages (cross-effects), and on the parameter size of the respective exogenous determinant. Therefore, we will refer to the cumulative impact on the dependent variables as a percentage of the long-run effect with respect to different periods of time.
We will generate ceteris paribus shocks in the following way. First, based on the gap in the openness of both exports and FDI between typical intra-EU relations and relations between the EU countries and the selected CEECs, we will envisage the effect of narrowing that gap by one percent (always assuming that the GDP figures remain constant). This means simply changing EU exports (FDI) to the respective country from Central or Eastern Europe exogenously and looking for the long-run effect generated by the endogenous impacts over time. Next, we observe a significant difference in GDP per capita between the average EU and CEECs. Hence, we will search for the impact of reducing the difference between them by one percent (always assuming that population numbers remain fixed). This shock is critical as we are altering two rather than just one variable. We should notice that both the bilateral combined sum of GDPs as well as the index of similarity in country size is affected by such a change. We can proceed in a similar way for the physical capital to unskilled and the high-skilled to low-skilled labour ratios, never assuming a change in the endowment with low-skilled workers. The last shock to analyse will be one that narrows the gap in transport costs by one percent.
Table 3 provides information on the long-run effects of shocks in different variables (exports and FDI, GDP, stocks of capital and human capital, and transport costs, respectively) which are associated with the narrowing of gaps as suggested above. We will shock the respective determinants in an arbitrary base year (1996 in our case) and analyse the long-run effects and adjustment processes with respect to this year. Of course, reducing the difference between a typical EU-member and each of the CEECs in the abovementioned variables means inducing different shocks of varying sizes for the bilateral economic relationships between the EU and the CEECs. The inclusion of not only lagged endogenous effects (lagged exports in the export specification and lagged FDI in the FDI specification) but also of cross-effects 6 (lagged exports in the FDI specification and lagged FDI in the export specification) in the estimated specification generates two different effects for both exports and FDI. We should note that any kind of shock we envisage can be discussed as a simultaneous shock in exports and FDI (of different size) in, say, period t. Therefore, Table 3 provides data on these effects and the overall change of exports is seen to consist not only of a multiple of the shock in the exports in period t (the own effect: XX), but also of a multiple of the shock in FDI in period t (the cross-effect: FX). The results show us that the own effects are much greater than the cross-effects and, hence, that none of the cross-effects outweighs the respective own effect (which could be the case in terms of parameter signs as the short-run effects of FDI on exports being negative). Of course, in terms of the long-run impact of a shock on exports and FDI the overall effects (X and F) are the sums of the two: X=XX+FX and F=FF+XF, respectively. Therefore, the own effect of FDI is greater than the overall change of FDI after a shock, because it is (slightly) reduced by the negative cross-impact of exports.
As noted above, the advantage of using a dynamic framework goes beyond the difference between short-run and long-run analyses. It also permits an examination of the speeds of adjustment between the short-run effect of a change and the long-run effect. Figure 1 shows the transformed adjustment paths for exports, stocks of outward FDI and their components (see the footnote to Figure 1). We should note that these paths are solely determined by the 2x2 parameter matrix for lagged exports and FDI in both the export and FDI specifications (i.e. a1, a2, b1, b2 in the Appendix). Hence, different sizes of shocks or shocks in different exogenous determinants only affect the cumulative impact of the shock on bilateral exports and FDI, but not the associated speed of adjustment. As the parameter estimates are derived from a panel estimation where the pooling assumption was applied (homogeneous parameters for all countries and periods of time), the adjustment paths are equal for all bilateral relationships and types of shocks.
Lines X and F in Figure 1 mark the respective adjustment lines for overall (bilateral) real exports and real stocks of outward FDI. The slopes of the curves demonstrate the associated speed of the adjustment process. The steeper a curve, the faster the adjustment will be. Hence, FDI adjusts at a slower pace than exports. From the different components we should recognise first that the own effects (XX and FF) are exhausted much faster than the cross-impacts (XF and FX). Secondly, both the cross-effect of exports on FDI and that of FDI on exports exhibit an equivalent speed of adjustment although their impact in size is different. 7 The latter is independent of the fact that, in our case, the two cross-effects have the opposite sign. Looking at the intersection of the several lines in Figure 1, we see that the underlying dynamic system yields relatively fast speeds of adjustment so that the overall effect of a shock is adjusted by 95 percent within four years for exports and in about six years for FDI. With respect to the discussion of potential and adjustment, we may conclude from our analysis that speeds of adjustment after shocks (like the fall of the Iron Curtain for CEECs in terms of shrinking trade costs, etc.) are faster than perhaps expected and suggested by previous work in that field. This, of course, does not say anything about the speed of the catching-up process in terms of per capita GDP and other variables, but it can help us understand the adjustment of bilateral relations to their "natural" levels given their exogenous determinants and the information about the parameters of the dynamic system.
Concluding Remarks
Empirical research on economic integration in the last decade has mainly been concerned with estimating (static) gravity models of trade and afterwards applying the estimated parameters and given exogenous determinants so as to project "natural" (i.e. potential) trade flows between a sample of reference (mostly eu or oecd) countries and other countries to be integrated (mostly central and eastern european economies). It was the merit of that literature to pioneer ways of figuring out the scope of the effects that we should expect. On the other hand, its disadvantage (at that time surely also because of a lack of longer time series) was that it could not answer the question of how long adjustment to the long-run should last. Additionally, some authors have shown that after a few years the – initially quite high – potential in actual trade flow relations was exhausted, which perhaps could be related to a nonstationarity problem in the data of the static approaches used.
This paper has tried to apply the results of a dynamic panel data analysis to the question of economic integration between the EU and three CEECs (Czech Republic, Hungary, and Poland). As usual, the parameter values for a panel of EU countries were assumed to also be valid for relations between the EU and the CEECs. Based on more recent theoretical work, this was done not only for trade (exports) but also for bilateral stocks of outward FDI. We studied in particular the long-run effects of partial catching-up in several determinants such as openness to exports and FDI from the EU as well as GDP per capita and other variables. Due to the dynamic structure of the underlying model, it was found that the long-run effects on exports and stocks of outward FDI from the EU to the CEECs are composites of own and cross-effects, which so far had not been derived from dynamic panel data analysis in this context.
In addition, the analysis also looked into the dynamic nature in terms of speeds of adjustment for overall exports and FDI. This goes beyond previous work that was exclusively based on static approaches. However, it has been suggested that we should expect exports to adjust more rapidly to their long-run level than (stocks of outward) FDI after a shock. It was demonstrated that exports would be expected to reach 95 percent of their long-run level after fewer than four years and stocks of outward FDI after about six years. This could help us understand better why, within a relatively short period of time (between 1991 and 1996), potential trade flows (always given their exogenous determinants) were shown to be fully adjusted, which initially had by no means been expected to occur at such a rapid pace (Gros and Gonciarz 1996).
References
Baldwin, Richard (1994) Towards an Integrated Europe. London: CEPR.
Brenton, Paul F. and Francesca DI MAURO (1999) The Potential Magnitude and Impact of FDI Flows to CEEC. Journal of Economic Integration 14(1), 59-74.
Brenton, Paul F., Francesca DI MAURO and Matthias LÜCKE (1999) Economic Integration and FDI: An Empirical Analysis of Foreign Investment in the EU and in Central and Eastern Europe. Empirica 26(2), 95-121.
Breuss, Fritz and Peter EGGER (1999) How Reliable Are Estimations of East-West Trade Potentials Based on Cross-Section Gravity Analyses? Empirica 26(2), 81-94.
Caves, Richard E. (1982/1996) Multinational Enterprise and Economic Analysis, 2nd edition. Cambridge: Cambridge University Press.
Egger, Peter (2000) "European Exports and Outward Foreign Direct Investment: A Dynamic Panel Data Approach". Vienna: WIFO Working Paper, No. 129.
GRAHAM, Edward M. (1996) "On the Relationship Among FDI and International Trade in the Manufacturing Sector: Empirical Results for the United States and Japan". Geneva: WTO-Staff Working Paper, No. RD-96-008.
Gros, Daniel and Andrzej GONCIARZ (1996) A Note on the Trade Potential of Central and Eastern Europe. European Journal of Political Economy 12(4), 709-21.
Hamilton, Carl B. and Alan L. WINTERS (1992) Opening up International Trade with Eastern Europe. Economic Policy 14, 77-116.
Leamer, Edward E. and James LEVINSOHN (1995) International Trade Theory: The Evidence. In Gene M. GROSSMAN and Kenneth ROGOFF (eds) Handbook of International Economics Vol. III, 1339-96. Amsterdam: Elsevier, North Holland.
Linnemann, Hans (1966) An Econometric Study of International Trade Flows. Amsterdam: Elsevier, North Holland.
Markusen, James R. and Anthony VENABLES (1996) "The Theory of Endowment, Intra-industry, and Multinational Trade". Cambridge, MA: NBER Working Paper, No. 5529.
Markusen, James R. and Anthony VENABLES (1998) Multinational Firms and the New Trade Theory. Journal of International Economics 46(2), 183-203.
Markusen, James R. and Keith E. MASKUS (1999a) "Multinational Firms: Reconciling Theory and Evidence". Cambridge, MA: NBER Working Paper, No. 7163.
Markusen, James R. and Keith E. MASKUS (1999b) "Discriminating Among Alternative Theories of the Multinational Enterprise". Cambridge, MA: NBER Working Paper, No. 7164.
Rosati, Dariusz K. (1992) "Problems of the Post-CMEA Trade and Payments". London: CEPR Discussion Paper, No. 650.
Wang, Zhen K. and Alan L. WINTERS (1991) "The Trading Potential of Eastern Europe". London: CEPR Discussion Paper, No. 610.
Appendix: Estimated Equations
The following two dynamic equations for exports and FDI have been estimated
DXijt = a0 + a1 DXij(t-1) + a2 DFij(t-1) + a3 DGDTijt + a4 DSIMIijt + a5
DRKLijt + a6 DRHSLSijt + a7 DRLTAXijt + a8 DTCFijt + d1t + u1ijt
DFijt = b0 + b1 DXij(t-1) + b2 DFij(t-1) + b3 DGDTijt + b4 DSIMIijt + b5
DRKLijt + b6 DRHSLSijt + b7 DRLTAXijt + b8 DTCFijt + d2t + u2ijt
where D indicates first differences in logs (hence, growth rates). X are exports, F are stocks of FDI, GDT is the bilateral combined sum of GDPs, SIMI is an index which measures relative country size, RKL measures the distance between two countries’ capital-low-skilled-labour ratios, RHSLS is the distance between two countries’ endowments with high-skilled in relation to low-skilled people. RLTAX measures the relationship between an exporter’s and an importer’s corporate tax rate, TCF is a transport cost factor reflecting the difference between c.i.f. and f.o.b. values from trade statistics, and dt is a time trend. Only real data (base year 1995) were used in the regressions which were run on bilateral export and FDI relations between EU member-countries using data over the period 1988-1996. The reader is referred to Egger (2000) for more details of the construction of variables. As usual, the estimated coefficients were then applied for the relationships between the EU and the CEECs.
Endnotes
Note *: Peter Egger is Reserch Fellow at the Austrian Institute of Economic Research (WIFO - Österreichisches Institut für Wirtschaftsforschung), Vienna. Back.
The author wishes to thank Christian Bellak, Fritz Breuss, Hartmut Egger, Wilhelm Kohler, Michael Pfaffermayr, and the anonymous referees for their helpful comments.
Note 1: To our knowledge, Egger (2000) is an exception, analysing intra-EU exports and FDI in a dynamic panel framework. The results presented in this article regarding long-term influences and adjustment paths are based on this work. Back.
Note 2: This holds true, albeit recently some authors have studied the relationship between exports and FDI through the analysis of residuals in the tradition of Graham (1996). Examples in this tradition are Brenton and Di Mauro (1999) and Brenton et al. (1999). Back.
Note 3: More precisely, we should define the "long-run" as the cumulative impact of a shock (either in exports, FDI or an exogenous determinant) on the dependent variables (bilateral exports and stocks of outward FDI). The cumulative impact is simply the integral of the resulting changes between periods (t=1), i.e. the period where the shock occurs, and (T=µ), i.e. infinity. Back.
Note 4: The following data sources were used: OECD Statistics of Foreign Trade (nominal bilateral exports of EU countries), the Vienna Institute for Comparative Economic Studies Data Base (nominal bilateral imports of CEECs), OECD International Direct Investment Statistics Yearbook (stocks of outward FDI). Reported bookvalues were deflated by use of the corresponding investment deflator and exchange rate index (see Egger 2000 for a justification), OECD Economic Outlook (export price indices), the IMF International Financial Statistics (exchange rate index), OECD National Accounts, Volume 1 (GDP, GDP deflator, gross fixed capital formation, and investment deflator), OECD Education Statistics 1985-1992, Education at a Glance, and UNESCO Statistical Yearbook (school enrolment, defining secondary or higher enrolled people as high-skilled and less enrolled as low-skilled). All variables are expressed in constant prices and US dollars taking 1995 as the base year. See Egger (2000) for further details of the construction of variables. Back.
Note 5: Measured as the difference between cost insurance freight (c.i.f.) and free on board (f.o.b.) values from bilateral trade statistics. To give an example: the difference between German exports to Hungary (reported in German export statistics) and Hungarian imports from Germany (reported in Hungarian import statistics) should be interpreted as transport costs (costs of exports) for German exports to Hungary. Of course, if there are huge differences in the reporting quality between Germany and Hungary, they would bias this measure. Back.
Note 6: The interaction of FDI and exports along their adjustment paths. Back.
Note 7: Note that lines XF and FX in Figure 1 fully coincide. Back.
May 2000