Journal Issue

Portugal: Selected Issues

International Monetary Fund
Published Date:
March 2004
  • ShareShare
Show Summary Details

III. The Impact of EU Enlargement on Portuguese Trade15

A. Introduction and Stylized Facts

47. Enlargement of the European Union (EU) from 15 to 25 members on May 1, 2004, is expected to bring important economic benefits for old and new members, largely through an expansion of trade. This chapter explores the potential trade implications of EU accession for Portugal. With currently very little trade between Portugal and the accession countries, model-based estimates suggest a potentially large trade expansion—perhaps by some two-thirds or more above current levels. However, the experience of earlier EU accession countries and of monetary union in Europe also indicated that the trade benefits do not accrue automatically: further structural reforms (in labor and product markets) as well as steps to strengthen competitiveness may be critical to secure the fruits of trade integration.

48. The recent declines in Portugal’s cost and price competitiveness and structural weaknesses in some areas—relatively low productivity, low educational attainment, and high labor costs compared with the accession countries—raise some concern about increased competition from the accession countries (Table 1). Is Portugal well positioned to benefit from trade gains after the EU enlargement? Is the current trade structure of Portugal and the accession countries relevant to assess the trade effects of the expanded European Union? This chapter does not aim to give definite answers to these questions but to shed some light on the potential impact of EU enlargement on Portugal’s trade flows by presenting stylized facts and empirical evidence from recent gravity model estimates.

Table 1.Productivity, Wages, and Schooling
Productivity 1/Wage Level 2/Schooling
European Union100.0100.09.0
Czech Republic27.917.29.5
Note: Productivity, wages, and schooling data are respectively for 2002, 2000, and 2000.Sources: IMF, WEO; Euroslat; and Barco Lee (2002) dataset.

49. At present, most of Portugal’s international trade takes place within the European Union, while trade shares with the 10 accession countries (AC-10) are low. Specifically, more than three-quarters of Portuguese exports and imports are with other EU countries. At the same time, Portugal’s exports to the 10 accession countries are small, accounting for 1.5 percent of total exports in 2002, compared with 12.6 for the present EU member countries (EU-15; Table 2). Viewed from the perspective of the accession countries, their exports to Portugal were, on average, only ½ percent of total AC-10 exports in 2002 (while exports to the EU accounted for about two thirds of total AC-10 exports). Notwithstanding some dispersion among the accession countries (Table 3), none of them had export shares to Portugal above 1 percent in 2002. The empirical results below suggest that this leaves considerable room for trade expansion after EU accession.

Table 2.Export Shares, 2002
PortugalSpainEU-15AC-10Rest of the World
Source: IMF, Direction of Trade Statistics (2003).
Table 3.Portugal, Spain and Accession Countries: Export Shares
PortugalSpainEU-15ACRest of the World
Czeck Republic0.42.068.616.315.1
Slovak Republic0.21.660.625.511.9
Source: IMF, Direction of Trade Statistics (2003).

B. Gravity Model Estimates

50. A large body of literature has studied the trade creation effects of free trade arrangements, single markets, and currency unions.16 Most of these studies find sizeable effects on trade flows. However, there is controversy over the exact size of the gains. For example, Rose (2000) finds that membership in a currency union triples trade with other currency union members, using a panel of cross-country data covering bilateral trade between 186 countries at five-year intervals between 1970 and 1990. Persson (2001) and Tenreyro (2001) correct Roses’s estimate for selection bias and simultaneity bias, respectively, and find that currency union increases trade by some 60 percent. The positive effects on trade of the European Economic Community and the European Free Trade Agreement have been highlighted by Bayoumi and Eichengreen (1995) and Frankel and Wei (1993), among others. More recently, Micco and others (2003) and Faruqee (2004), using post-1999 data, have examined the trade effects of European monetary union. They find that the euro’s impact on trade ranges between 4 and 16 percent.

51. Most of these studies estimate so-called gravity regressions for international trade. Gravity models examine how trade between pairs of countries is affected by monetary union or a common market after controlling for size, geographical distance, and other variables (such as, colonial links, shared border and common language) that might affect trade flows between two countries. In its simple form, the gravity equation for trade—in analogy to Newton’s law of universal gravitation—states that trade flows between two countries are proportional to the product of the two countries’ economic sizes (GDPs) and inversely proportional to distance (broadly understood as including all factors that might create trade resistance)17.

52. Using the methodology and database in Dabán, García-Escribano, and Hoffmaister (2004), the current trade flows between Portugal and the AC-10 are examined within a gravity model.18 Tables 2 and 3 above indicate that Portuguese trade with the AC-10 is small compared with trade flows between the AC10 and the rest of the EU. In order to assess whether these trade patterns can be explained by economic size, distance, and other “cultural” factors the following gravity regression is estimated:

where Tradeijt is bilateral trade in goods between trading countries i and j at time t, γij represents the fixed effect in trade between countries i and j τt represents common time effects for a particular year, Y and P represent gross domestic product in level and population, Xij is a matrix of variables including the geographical distance between i and j, and dummy variables for shared borders, common language and access to sea, and ε is the error term. DEUAC is a dummy variable taking on the value “1” if one trading partner belongs to the EU-15 and the other to AC-10 (and is “0” otherwise). The coefficient δ is the key parameter to assess the degree of “undertrading” or “overtrading” between the EU-15 and AC-10, i.e., of trade not explained by other observable factors. DPRT is a country dummy variable for Portugal. Thus, λ gives the deviation of Portugal with respect to the average EU-15 trade effect. To allow a comparison with other trading partners, some regressions also include country-specific effects for France, Germany, Italy, Spain, and the United Kingdom.

53. The estimation results in Table 4 suggest that the degree of undertrade between the EU-15 and AC-10 is around 28 percent for total exports19 and somewhat less (18 percent) for imports. Thus, the fact that countries in the EU trade more among themselves than with the AC-10 (as indicated in Table 2 above) can be largely explained by gravity and cultural factors. Nevertheless, there is also a significant amount of the trade differential that cannot be explained by these observable factors. This level of undertrade might be interpreted as the estimate of the potential trade gains after EU enlargement.

Table 4.Trade Estimates for EU-15 and AC-10 Countries
Total ExportsTotal Imports
Adjusted R0.900.900.900.91
Source: IMF, Direction of Trade Statistics (2003).Note: The dependent variable is log exports (imports) from country /to county j DEUAC10 is a dummy variable taking the value of one when trade is between an EU-15 country and one AC-10 country; and zero otherwise. Robust standard errors in parentheses. * indicates significant at 5%; ** significant at 1%.

54. In order to examine whether the average effect is similar across the present EU-15 member states, country dummies are interacted with the dummy variable DEUAC. Table 5 shows that undertrade varies widely across countries. While some countries like the France, Germany, and the United Kingdom are overtrading with the AC-10, the level of Portugal’s undertrade—that is, trade shortfalls relative to the trade volume explained by distance and other cultural factors—is above 70 percent for exports20 (and 66 percent for imports). This is substantially larger than the undertrade of neighboring Spain.

Table 5.Undertrade Between the EU-15 and AC-10 Countries
United Kingdom-12-35
Note: Undertrade is computed from the regression coefficients using the following formula: -(exp(beta)-l)*100.For EU-15 estimates in columns (1) and (3) are used.

55. The results above imply that the current level of trade between the EU-15 and the AC-10—and especially between Portugal and the AC-10—is significantly smaller than what distance and other fundamentals would justify. This indicates room for potentially large trade expansions.

56. Available evidence suggests, however, that the gains from trade integration are not always distributed equally and depend importantly on structural features of the integrating economies. Faruqee (2004) and other studies on optimal currency unions21 suggest that the actual realization of these gains is influenced by the share of total trade that is accounted for by intra-industry trade—which is the component of external trade that has been found to be particularly sensitive to exchange rate movements—and by other characteristics of the product market that determine the ability to cope with asymmetric shocks. Similarly, the degree of price and wage flexibility as well as factor mobility appear to be key factors influencing the gains from trade integration, especially when the latter is accompanied by relinquishing an independent monetary policy.

57. Portugal’s past experience in the EU and in monetary union may provide some guidance about its ability to benefit from EU enlargement. To get a rough bearing in this regard, the analysis below draws on the results in Faruqee (2004), who examines the impact of membership in the EU or monetary union on trade integration within the euro area. He estimates the following gravity-equation model for a panel of 22 industrialized countries:

where the variables are as defined in equation (1) above, and y represent per capita GDP, FTA is a dummy variable for free trade agreements, EU and EMU are dummy variables equaling one if both partners belong to the EU or the euro area, respectively, and zero otherwise. To assess differences among the euro-area countries, the coefficient of interest is λit: it captures the specific trade effect for country i (for example, Portugal) following monetary union, an effect that is in addition to common effect for all other euro-area members (which is captured by the common coefficient λt).

58. The results in Faruqee (2004) suggest significant trade benefits from economic integration, here proxied by EMU, but also important differences across individual countries. Overall, he finds that euro-area membership boosted trade among member countries by an average 7–8 percent and that trade gains from EMU are rising over time.22 His results also suggest that monetary union has not had trade diverting effects, since both intra- and extra-area trade increased under the single currency. Concerning differences across countries, Portugal (together with Finland) stands out as the country that harnessed least the trade benefits of monetary union. On average, Portugal’s trade gains were some 11 percent below the euro-area average, and the estimated trade benefits were not statistically significant for Portugal (see Figure 1).

Figure 1.Dynamic Patterns of EMU Trade Effects, 1997-2002

(Logarithmic change in trade)

59. Some factors seem to boost the trade benefits from economic integration and monetary union, but considerable uncertainty remains as to the precise channels explaining the estimated cross-country differences. As noted in Faruqee (2004), there appears to be a significant correlation between trade gains from monetary union membership and the share of intra-industry trade (Table 6). Finland, Ireland, and Portugal are the countries with the smallest initial shares of intra-industry trade within the EU, and these are also the countries estimated to have benefited the least from monetary union. However, this relationship is not that close for those countries on the top end of the trade gains (see Spain and Netherlands, for instance). This suggests that other factors—such as firm entry and exit policies in the product market and labor market flexibility—might also be relevant for a country to take advantage of trade gains from monetary union. Moreover, some of the divergent trade gains could have reflected different developments in competitiveness. As indicated in Figure 2, Portugal is estimated to have recorded a relatively large appreciation of its real effective exchange rate during the period covered in the empirical estimates. These losses (not captured in the empirical analysis based on equation 2) are likely to have contributed to the relatively low estimated trade benefits from monetary union in Portugal.

Figure 2.Cumulated Change in Real Effective Exchange Rate (ULC), 1997-2002

Source: Eurostat.

Table 6.Intra-Industry Trade Shares, 1997

(In percent) 1/


C. Conclusions

14. EU accession by 10 countries in May 2004 could bring important trade benefits to both old and new EU member countries. The results reported in this chapter suggest potentially large effects for Portugal, which presently trades significantly less with the accession countries than suggested by standard gravity models of external trade. However, there is also some evidence—including from the experience of present euro-area members—that the economic gains from trade integration do not accrue automatically and evenly across countries. The chapter highlights several factors that may influence Portugal’s potential gains from further economic integration in Europe (including those related to its trade structure and competitiveness), but investigating the precise channels that determine a country’s gains from integration remains an area for further research.


Prepared by Marta Ruiz-Arranz.

The theoretical foundations of the gravity equation can be found in Anderson (1979), Deardorff (1984), Helpman and Krugman (1985), and Helpman (1987).

The dataset covers the EU-15 and AC-10 countries and spans the period 1978–2002.

The coefficient of the dummy variable DEUAC on the log of export flows regression is -0.33 (column 1). Thus, the impact on trade is computed as ((e–033–1)xl00 = –28.l).

(e–031–1.02–1)x100 = –73.55

See IMF, SM/04/02 (1/8/04), for a review of the literature.

In another study for the countries in the European Union, Micco and others (2003) find that the euro’s impact on trade ranged between 4 and 16 percent.

Other Resources Citing This Publication