Flávio Vilela Vieira a, Eduardo Amaral Haddad b & Carlos Roberto Azzoni b a Universidade Federal de Uberlândia, Uberlândia, Brazil

Transcription

1 This article was downloaded by: [Sistema Integrado de Bibliotecas USP] On: 10 September 2014, At: 03:55 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: Registered office: Mortimer House, Mortimer Street, London W1T 3JH, UK Latin American Business Review Publication details, including instructions for authors and subscription information: Export Performance of Brazilian States to Mercosul and Non-Mercosul Partners Flávio Vilela Vieira a, Eduardo Amaral Haddad b & Carlos Roberto Azzoni b a Universidade Federal de Uberlândia, Uberlândia, Brazil b Universidade de São Paulo, São Paulo, Brazil Published online: 09 Sep To cite this article: Flávio Vilela Vieira, Eduardo Amaral Haddad & Carlos Roberto Azzoni (2014) Export Performance of Brazilian States to Mercosul and Non-Mercosul Partners, Latin American Business Review, 15:3-4, To link to this article: PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content ) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at

2 Latin American Business Review, 15: , 2014 Copyright # Taylor & Francis Group, LLC ISSN: print= online DOI: / Export Performance of Brazilian States to Mercosul and Non-Mercosul Partners FLÁVIO VILELA VIEIRA Universidade Federal de Uberlândia, Uberlândia, Brazil EDUARDO AMARAL HADDAD and CARLOS ROBERTO AZZONI Universidade de São Paulo, São Paulo, Brazil ABSTRACT. We investigate the role of the trade-weighted real exchange rate and foreign income on the export performance of Brazilian states, differentiating between Mercosur and non- Mercosur partners. The results indicate that state exports are price and income inelastic. There are differences in the influence of the different trade factors between the two groups of partners. One crucial difference is the relevance of the real exchange rate effect for non-mercosur partners. This might be associated with the existence of specific rules for Mercosur that can overcome the usual effect of relative competitiveness associated with movements in the real exchange rate. RESUMEN. Investigamos el papel que juegan el tipo de cambio real ponderado en función del comercio, y los ingresos extranjeros sobre el desempeño de las exportaciones de los estados brasileños, trazando un diferencial entre los socios del Mercosur y los que no son sus miembros. Los resultados indican que, tanto el precio como el ingreso proveniente de las exportaciones de los estados son inelásticos. Existen diferencias en la influencia ejercida por los distintos factores comerciales que existen entre los dos grupos de socios. Una diferencia decisiva es la importancia del efecto que el tipo de cambio real tiene para los socios que no son miembros del Mercosur. Este hecho puede estar asociado a la existencia de normas específicas para el Mercosur que pueden dominar el acostumbrado efecto Received July 23, 2013; revised May 5, 2014; accepted May 7, Address correspondence to Carlos Roberto Azzoni, Universidade de São Paulo, Av. Prof. Luciano Gualberto 908, Cidade Universitária, São Paulo-SP, , Brazil. cazzoni@usp.br 253

3 254 F. V. Vieira et al. de la competitividad relativa, asociada a las oscilaciones en el tipo de cambio real. RESUMO. Investigamos o papel da taxa de câmbio, ponderada pelo volume de comércio, e da renda do país importador sobre a performance exportadora dos estados brasileiros, diferenciando os parceiros comerciais entre aqueles pertencentes ao Mercosul e os que não são membros. Os resultados indicam que as exportações são inelásticas a preço e renda em geral, mas há diferenças na influência dos distintos fatores de comércio entre os grupos de parceiros. Uma diferença crucial está no papel da taxa de câmbio, que é mais importante para o comércio com outros países do que com os do Mercosul. Isso pode estar associado às regras específicas do Mercosul, que podem anular o efeito de competitividade associado com a taxa de câmbio. KEYWORDS. Brazilian states export performance, foreign income, Mercosur and non-mercosur, panel data, trade-weighted real exchange rate INTRODUCTION The development of the Mercosul in the mid-1990s has been the focus of a lot of debate and analysis in terms of its trade and economic impact for Brazil, Argentina, Uruguay, and Paraguay; however, there is still no consensus on these issues. One of the early studies addressing the creation of Mercosul and how Brazil should deal with it on trade, political, and economic grounds is Batista (1994). One of the arguments is that Brazil should be seen as a global trader with interests that go beyond trade in goods, especially financial flows. Dupeyron (2009) provided an historical analysis of the creation of Mercosul, starting with the negotiations between Brazil and Argentina in the second half of 1980s on issues of regional integration. In 1991, Argentina, Brazil, Paraguay, and Uruguay signed the Mercosul founding treaty (Treaty of Asunción). The idea was to create a common market that would result in the free movement of goods, services, capital, and labor. How effective Mercosul has been in enhancing trade between the countries involved is still an open empirical question. Most papers deal with the issue on a country-to-country level, ignoring the regional differences within each country. On the Brazilian side, Sá Porto (2002a) found that although trade between Brazilian states and the Mercosul countries did increase between 1990 and 1998, the impact was not homogeneous across Brazilian states. Sá Porto and Canuto (2002, 2004) continued that study (see also Sá Porto, 2002b), showing that the impacts of Mercosul on Brazilian states trade

4 Export Performance of Brazilian States 255 flows are robust, regardless of the model used. Azzoni and Sá Porto (2007) indicated that Mercosul presented a positive effect for the export of Brazilian states. A more recent assessment can be found in Baumann (2011). Any study on the impact of the creation of Mercosul must take into account the traditional variables considered in the study of international trade basically, income and exchange rates. Most of the literature on the role of real exchange rate and foreign income for export performance uses national-level trade weights. In this article we construct specific trade weights for Brazilian states in their relationship with Mercosul and non-mercosul countries. We use a dynamic specification to estimate state export performance, which is a clear advance in the treatment of endogeneity and instrument proliferation issues. The panel data System Generalized Method of Moments (GMM) results for the Mercosul trade partners reveal that lagged exports and commodity prices are statistically significant; there is only scarce evidence for the relevance of tradeweighted gross domestic product (GDP) and lagged trade-weighted real exchange rate,with no statistical significance for current trade-weighted real exchange rate and state size. The results for the non-mercosul trade partners revealthatlaggedexports,currentandlaggedtrade-weightedrealexchangerate are statistically significant; commodity prices and state GDP are only significant in some specifications; and trade-weighted GDP or lagged State GDP are not significant. The estimated coefficients for the trade-weighted price and foreign income for the Mercosul and non-mercosul trade partners indicate inelastic demands. Exports to Mercosul partners show a higher influence of the tradeweighted real exchange rate than to non-mercosul partners; trade-weighted GDP is quantitatively more relevant for exports to non-mercosul partners. One crucial difference for the non-mercosul trade partners is the effect of the real exchange rate, which might be associated with the fact that export flows to Mercosul are subject to specific rules (common external tariff and trade agreements) as compared to exports to other countries. The effects of this regulation can overcome the usual effect of relative competitiveness associated to movements in the real exchange rate. The article is divided in three sections, besides this introduction. Section 1 presents a brief summary of the empirical literature on export performance. Section 2 presents the methodology and the empirical results. The final section presents our concluding remarks. BRIEF REVIEW OF THE LITERATURE ON EXPORT PERFORMANCE One of the pioneer empirical works on state export performance using state-level trade weights to construct real exchange rate and foreign income is Cronovich and Gazel (1998). They argued that there is a problem of

5 256 F. V. Vieira et al. misspecification in using national trade weights when evaluating the role of the exchange rate and foreign income to explain state export performance. Another limitation of the few studies at the subnational level analyzing export performance is that they frequently focus on the supply side of foreign trade (Carlino, Voith, and Cody, 1994). Our empirical work deals with these limitations by using state-specific trade weights and by considering the influence of both demand and supply factors. Cronovich and Gazel (1998) estimate a fixed effect panel data model and demonstrate that real exchange rate and foreign income do matter for exports at the state (subnational) level, thus influencing long-term economic growth. Gazel and Schwer (1998) and Erickson and Hayward (1991) are other examples of the relevance of the role played by demand factors at the subnational level. The results for the United States suggest that regional exports are positively correlated to size (GDP) and negatively correlated to the distance to the foreign trade partner. The literature on Brazilian export performance is scarce but has recently grown, shedding light on price (exchange rate) and income effects, mostly using national trade weights. Cavalcanti and Ribeiro (1998) found that relative prices (exchange rates) are relevant in explaining Brazilian exports. Neves, Cristina, and Lélis (2007) estimated price and income elasticities for the exports of seven states, concluding that state exports are price and income inelastic. After the change in the Brazilian exchange rate regime in 1999, the empirical literature on export performance shifted to investigate the slow response of exports to real exchange rate movements (hysteresis hypothesis). Markwald and Puga (2002) found evidence that the inflow and outflow of export companies are correlated with the exchange rate policy. The positive impact of exchange rate depreciation is higher, and the negative impact of appreciation is lower, supporting the hysteresis hypothesis. They suggested that periods of exchange rate depreciation are associated with an increase in the size (number of companies) of the export base. Scarpelli (2010) suggested that deviations from a long-term relationship between exchange rate and exports take a long period to be corrected. Other studies have focused on measuring border effects. Silva, Almeida, and Oliveira (2007) measured the border effect using a cross-sectional gravity model for the 27 Brazilian states and 46 countries. The evidence indicates that trade among Brazilian states is 33 times higher than trade with a foreign country, which can be explained by the low degree of substitutability for domestic and foreign goods and by international trade barriers. Daumal and Zignago (2010) developed an empirical investigation on the determinants of border effects of Brazilian states by estimating the degree of trade integration between states and the magnitude of the barriers faced by their exporters in the 1990s. The results suggest that the Brazilian market was highly fragmented: in 1991, an average Brazilian state traded 37 times more with itself than with other Brazilian states. Leusin and Azevedo (2009) also dealt with border effects for the Brazilian regions.

6 Export Performance of Brazilian States 257 METHODOLOGY AND EMPIRICAL RESULTS This section describes the procedures to construct the trade-weighted real exchange rate and foreign income, considering two sets of trade partners: Mercosul (Argentina, Uruguay, and Paraguay) and non-mercosul (all the remaining trade partners). Trade-Weighted Real Exchange Rate and Foreign Income We use as the dependent variable in our equations the log of exports of each state (see list of states in Table A1), from 1996 to The state-specific weights used to construct the trade-weighted variables (labeled LRERTW and LGDPTW) are the share of each foreign country in the state s exports. For year t, the share of country j in state i s exports is: Share t;i;j ¼ X t;i;j = X j X t;i;j Where X t,i,j is state i s (i ¼ 1 to 27) exports to country j in year t, with j changing over time and across states. The trade-weighted GDP (TWGDP) of each state s trading partners use GDP Purchasing Power Parity (PPP) from the World Development Indicators (2010), in 2005 constant US$. Let GDP t,j denote real GDP for country j in year t. The trade-weighted foreign GDP for state i in year t is given by: TWGDP t;i ¼ X j Share t;i;j GDP t;j The trade-weighted real exchange rate (TWRER) for each state is constructed with data from: nominal exchange rates (foreign currency per Real), denoted by E t,j, from the International Financial Statistics (IFS) 1 ; Consumer Price Index for each country j, denoted by P j ; Consumer Price Index for Brazil, denoted by P i. The real exchange rate between state i and country j, at time t, is given by: 2 RER t;i;j ¼ E t;j P t;i =P t;j The trade-weighted RER for the state i, in time t, is given by: ð1þ ð2þ ð3þ TWRER t;i ¼ X j Share t;i;j RER t;i;j ð4þ The general equation estimated is: LEXPORT it ¼ b 0 þ b 1 LEXPORT it 1 þ b 2 LTWRER it þ b 3 LTWGDP it þ b 4 LPCOM it þ b 5 LGDPSTATE it þ u it ð5þ

7 258 F. V. Vieira et al. where i indicates state and t indicates time, illustrating the time dimension of our panel; LEXPORT is the log of exports of goods and services; LTWRER is the log of trade-weighted real exchange rate at the state level, using local currency and domestic and foreign price indexes; LTWGDP is the log of trade-weighted GDP at the state level; LPCOM is the log of commodity prices; and LGDPSTATE is the log of the states GDP. We expect a positive sign for the coefficient of trade-weighted foreign income (LTWGDP), since, all other things being equal, states exporting to countries with larger incomes should export more than states exporting to countries with lower levels of income. We expect a negative coefficient for the trade-weighted real exchange rate (LTWRER), since an increase (decrease) in LTWRER is associated to a more appreciated (depreciated) real exchange rate. Given that other empirical works found that exports do not respond immediately to changes in the exchange rate, we allow exchange rates to affect exports with a one-year lag. A Dynamic Panel Data Analysis: System GMM The GMM method is adequate for our case, because it is designed to estimate dynamic panel data models and is able to deal with the following requirements and data characteristics: small number of years (time dimension 14 in our case); large number of states (cross-section dimension 27 in our analysis); and lagged dependent variable (LEXPORT) as an additional explanatory variable. Furthermore, explanatory variables do not need to be strictly exogenous, it is possible to include individual fixed effects, and there is no second order autocorrelation. One of the empirical challenges of this investigation is how to deal with the use of weak instruments. When the set of instruments is weak, the estimated models will be associated with an asymptotical increase in coefficient variance. Moreover, in small samples the use of weak instruments produces biased coefficients. To reduce the potential bias and inaccuracy associated with the use of difference GMM, Arellano and Bover (1995) and Blundell and Bond (1998) developed a system of regressions in differences and levels. The instruments for the regression in differences (in levels) are the lagged levels (differences) of the explanatory variables. They can be considered appropriate instruments under the assumption that, if there is correlation between the levels of the explanatory variables and the country-specific effects, such correlation does not exist when those variables are in differences. Roodman (2009a, 2009b) developed a Stata command to deal with the issue of the excessive use of instruments, especially for System GMM estimations (xtabond2). Instrument proliferation frequently arises in System GMM estimation when the time dimension increases. The number of instruments can be too large compared to the sample size, invalidating some asymptotic

8 Export Performance of Brazilian States 259 results and specification tests. The outcome of instrument proliferation is to overfit endogenous variables and to fail to expunge their endogenous components, resulting in biased coefficients. Another argument is that the Hansen and Difference-in-Hansen tests can be weak in the presence of overidentification. We used the collapse suboption (Stata 11) to deal with the excessive number of instruments. The final outcome is to divide the GMM-style moment conditions into groups, and sum the conditions in each group to form a smaller set of conditions. Estimated Models and Analysis The results of the estimation of Equation (5) for Mercosul trade partners are presented in Tables 1 and 2, and for non-mercosul partners, in Table 3. The time span is 1996 to 2009, except for GDPSTATE, which is 1996 to Therefore, we have 351 observations for GDPSTATE, and 378 for all the other variables. Table A2 in the Appendix presents the state export shares to Mercosul (Argentina, Paraguay, and Uruguay) and to the United States and China, for the two end years of our period of analysis. There is a clear increasing trend in the share with China and a decreasing trend in the United States share, for most states. 3 STATE EXPORT PERFORMANCE: MERCOSUL PARTNERS The results for the Mercosul trade partners are reported in Table 1. We used the two-step System GMM and limited the number of explanatory variables to three, with no time dummies, and collapsed the number of instruments. The Hansen and Hansen-Diff statistics suggest that the set of instruments is valid and that there is no excessive use of instruments; moreover, the AR(2) test reveals no second-order autocorrelation. The estimated coefficients for lagged exports and commodity prices are statistically significant, regardless of model specification. The coefficient for trade-weighted GDP is positive and significant in model 5; the coefficient for lagged trade-weighted real exchange rate is negative and significant in model 6, in both cases, with the expected signs. The coefficients for current trade-weighted real exchange rate and state size (state GDP) were not significant. The estimated values of the coefficients for price (real exchange rate) and foreign income indicate that demand is inelastic for both price and income: lower than for lagged and current trade-weighted real exchange rate, and lower than 0.23 for trade-weighted GDP. Table 2 reports the results for Mercosul trade partners, but with the addition of a dummy variable for southern states (Paraná, Santa Catarina, and Rio Grande do Sul) since these have a long history of trading with the Mercosul countries due to geographic proximity and cultural links. The results are similar to the previous estimation; therefore, the main conclusion

9 TABLE 1 State Export Performance: System GMM (Two-Step) Dependent Variable: Log of Exports Mercosul Export Partners Models Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Dealing with Instrument Proliferation Collapse Collapse Collapse Collapse Collapse Collapse Collapse Collapse LEXPORT t (17.89) (6.64) (11.76) (7.55) (22.21) (3.69) (10.32) (9.69) LTWRER ( 0.32) (0.06) ( 0.76) ( 1.36) LTWRER t ( 0.13) LTWGDP ( 0.1) 0.23 (2.46) LPCOM (3.06) (2.58) LGDPSTATE ( 0.36) LGDPSTATEt (1.30) ( 2.46) ( 1.56) AR(2) Hansen Hansen-Diff Number of Groups Number of Instruments ( 0.50) ( 0.77) ( 0.65) Source: MDIC ALICEWEB and WDI (2010). Note. i),, and indicate significance at 10%, 5%, and 1%. Robust standard errors. AR(2) is the test of no second order autocorrelation. Hansen and Hansen-Diff are overidentification tests. Maximum of three explanatory variables and no time dummies. 260

10 TABLE 2 State Export Performance: System GMM (Two-Step) Dependent Variable: Log of Exports Mercosul Export Partners With Dummy for Southern States Models Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Dealing with Instrument Proliferation Collapse Collapse Collapse Collapse Collapse Collapse Collapse Collapse LEXPORTt (17.56) (5.93) (11.57) (5.55) (21.43) (3.94) (9.35) (8.75) LTWRER ( 0.33) (0.07) ( 0.66) ( 1.04) LTWRER t ( 0.11) LTWGDP 0.02 (0.12) (2.45) LPCOM (3.15) (2.76) LGDPSTATE (0.32) (2.43) ( 1.36) LGDPSTATEt (1.81) ( 0.37) DU South (1.54) (3.06) (0.09) ( 1.26) ( 0.16) (2.26) (3.35) (0.51) AR(2) Hansen Hansen-Diff Number of Groups Number of Instruments ( 0.60) (0.40) Source: MDIC ALICEWEB and WDI (2010). Note. i),, and indicate significance at 10%, 5%, and 1%. Robust standard errors. AR(2) is the test of no second order autocorrelation. Hansen and Hansen-Diff are overidentification tests. Maximum of four explanatory variables and no time dummies. DU South: Paraná, Santa Catarina, and Rio Grande do Sul. 261

11 TABLE 3 State Export Performance: System GMM (Two-Step) Dependent Variable: Log of Exports Non-Mercosul Export Partners Models Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Dealing with Instrument Proliferation Collapse Collapse Collapse Collapse Collapse Collapse Collapse Collapse LEXPORTt (26.04) (6.49) (6.47) (7.22) (35.97) (9.04) (9.04) (9.48) LTWRER ( 2.74) ( 2.83) ( 3.33) ( 5.13) LTWRER t ( 2.37) ( 3.75) ( 4.33) ( 4.26) LTWGDP (0.43) LPCOM (2.02) (1.15) LGDPSTATE (1.39) LGDPSTATEt (0.88) (0.69) (1.98) AR(2) Hansen Hansen-Diff Number of Groups Number of Instruments (1.18) Source: MDIC ALICEWEB and WDI (2010). Note. i),, and indicate significance at 10%, 5%, and 1%. Robust standard errors. AR(2) is the test of no second order autocorrelation. Hansen and Hansen-Diff are overidentification tests. Maximum of three explanatory variables and no time dummies. 262

12 Export Performance of Brazilian States 263 stands, even after taking into account the specificities of those three states. The Hansen and Hansen-Diff statistics suggest that the set of instruments are valid and there is no excessive use of instruments, while the AR(2) test reveals no second order autocorrelation. Models 5 and 6 are the ones with the better fit to explain state exports, with model 6 presenting the best fit. This indicates that lagged exports, lagged trade-weighted real exchange rates, and commodity prices are relevant variables affecting state export performance. STATE EXPORT PERFORMANCE: NON-MERCOSUL PARTNERS The results for the non-mercosul trade partners are reported in Table 3. Lagged exports and current and lagged trade-weighted real exchange rates are statistically significant, regardless of model specification. There is some evidence that commodity prices (model 2) and states GDP (model 7) are relevant. There is no statistical significance for trade-weighted GDP (foreign income) or lagged state GDP (size). 4 The estimated coefficients for tradeweighted exchange rate (LTWRER) and foreign income (LTWGDP) have the expected signs, negative and positive, respectively. The trade-weighted price and foreign income estimated coefficients for the non-mercosul trade partners reveal that demand is price and incomeinelastic; however, the estimated elasticities are higher for the trade-weighted real exchange rate (lower than 0.177) and lower for trade-weighted GDP (lower than 0.158), as compared to the Mercosul estimation. As in the previous two tables, there are no problems of invalid set of instruments or too many instruments, and there is no second order autocorrelation. A striking difference between the groups of partners is the relevance of the price effect (real exchange rate) for the non-mercosul trade partners. One possible explanation for this result is that state export flows to Mercosul are subject to specific rules (common external tariff and trade agreements), which may overcome the effect of the real exchange rate. CONCLUDING REMARKS Although the development of the Mercosul since mid-1990s has been a focus of much debate and analysis in terms of its trade and economic impact for Brazil, Argentina, Uruguay, and Paraguay, there is still no consensus on these issues. Our work seeks to develop an empirical investigation on the Brazilian state-level export performance, comparing the results for Mercosul and non-mercosul export partners, considering the influence of trade-weighted real exchange rate, trade-weighted foreign income, commodity prices, and state sizes. The estimation for the Mercosul export partners reveals that lagged exports and commodity prices are statistically significant, but there is only

13 264 F. V. Vieira et al. scarce evidence of the relevance of trade-weighted GDP and lagged trade-weighted real exchange rate. The results do not reveal statistical significance for current trade-weighted real exchange rate and state size (state GDP). The estimated coefficients for trade-weighted real exchange rate and foreign income indicate that state exports are price and income inelastic. Including a dummy variable for the southern States (Paraná, Santa Catarina, and Rio Grande do Sul) does not change these results. The second step was to compare the results with those for the non- Mercosul export partners. The results reveal that lagged exports as well as current and lagged trade-weighted real exchange rate are statistically significant; there is partial evidence in favor of the role of commodity prices and state GDP; and there is no statistical significance for trade-weighted GDP or lagged state GDP. The estimated coefficients for the trade-weighted price and foreign income for the non-mercosul export partners indicate that demand is inelastic both to price and income, but the estimated elasticities are higher for the trade-weighted real exchange rate and lower for tradeweighted GDP, when compared to the Mercosul estimation. One crucial difference for the non-mercosul export partners is the relevance of the relative price (real exchange rate) effect. Among other possible explanations, we can mention the fact that state export flows to Mercosul are subject to specific rules (common external tariff and trade agreements), which might overcome the usual effect of relative competitiveness associated to movements in the real exchange rate. NOTES 1. Availability of data for nominal exchange rate and domestic and foreign Consumer Price Index (CPI) restricts the number of trade partners to The number of trade partners (j) equals three (Argentina, Uruguay, and Paraguay) for Mercosul and 151 for the non-mercosul. 3. The average state export share to the United States dropped from 18.81% in 1996 to 9.41% in 2009, while to China, the average share increased from 1.35% to 11.24%. 4. Such evidence of no statistical relevance for the trade-weighted GDP was also partially verified in Tables 1 and 2. REFERENCES Arellano, M., & Olympia, B. (1995). Another look at the instrumental variable estimation of error-components models. Journal of Econometrics, 68(4), Azzoni, C., & Porto, P. S. (2007). How international integration affects the exports of Brazilian states. Revista de Economía Mundial, 17, Batista, P. N. (1994). O Mercosul e os interesses do Brasil. Estudos Avanc ados, 8(21), Baumann, R. (2011). Mercosul aos vinte anos: uma avaliac ão econômica [Texto para Discussão n 1627]. IPEA, Rio de Janeiro. Retrieved from br/portal/index.php?option=com_content&view=article&id=9809.

14 Export Performance of Brazilian States 265 Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87(1), Carlino, G. A., Voith, R., & Cody, B. (1994). The effects of exchange rate and relative productivity changes on US industrial output at the state level. Urban Studies, 31, Cavalcanti, M. A. F. H., & Ribeiro, F. J. (1998). As exportac ões brasileiras no período : desempenho e determinantes [Texto para Discussão n 545]. IPEA, Rio de Janeiro. Cronovich, R., & Gazel, R. (1998). Do exchange rates and foreign incomes matter for exports at the state level? Journal of Regional Science, 38(4), Daumal, M., & Zignago, S. (2010). Measure and determinants of border effects of Brazilian states. Papers in Regional Science, 89(4), Dupeyron, B. (2009). Perspectives on Mercosul borders and border spaces: Implications for border theories. Journal of Borderlands Studies, 24(3), Erickson, R. A., & Hayward, D. J. (1991). The international flows of industrial Exports from US regions. Annals of the Association of American Geographers, 81, Gazel, R. C., & Schwer, R. K. (1998). Growth of international exports among the states: can a modified shift-share analysis explain it? International Regional Science Review, 21, Leusin Jr., S., & Azevedo, A. F. Z. (2009). O efeito fronteira das regiões brasileiras: uma aplicac ão do modelo gravitacional. Revista Economia Contemporânea, 13(2), Markwald, R., & Puga, F. P. (2002). Focando a política de promoc ão de exportac ões. In: Pinheiro, A. C., Markwald, R., & Pereira, L. V. (Orgs.), O desafio das exportac ões. Rio de Janeiro: BNDES. Neves, P., Cristina, A., & Lélis, M. T. C. (2007). Exportac ões estaduais no Brasil: estimativas para as elasticidades prec o e renda. Revista de Economia Política, 27(2), Roodman, D. (2009a). How to do xtabond2: An introduction to difference and system GMM in Stata. Stata Journal, 9(1), Roodman, D. (2009b). A note on the theme of too many instruments. Oxford Bulletin of Economics and Statistics, 71(1), Sá Porto, P. C. (2002a). Mercosul and regional development in Brazil: A gravity model approach. Revista de Estudos Econômicos, 32(1), Sá Porto, P. C. (2002b). Os impactos dos fluxos de comércio do Mercosul sobre as regiões brasileiras. Doctoral Dissertation, Universidade Estadual de Campinas, Campinas, SP. Sá Porto, P. C., & Canuto, O. (2002). Mercosul: gains from regional integration and exchange rate regimes. Economia Aplicada, 6(4). Sá Porto, P. C., & Canuto, O. (2004). Uma avaliac ão dos impactos regionais do Mercosul usando dados em painel. Pesquisa e Planejamento Econômico, 34(3). Scarpelli, M. C. (2010). Hysteresis nas exportac ões manufaturadas brasileiras: uma análise de cointegrac ão com dados em painel. Dissertac ão de Mestrado, Departamento de Economia, Universidade de São Paulo, Ribeirão Preto, SP. Silva, O. M., Almeida, F. M., & Oliveira, B. M. (2007). Comércio internacional x intranacional no Brasil: medindo o efeito-fronteira. Nova Economia, 17(3),

15 266 F. V. Vieira et al. APPENDIX: REGIONAL DEFINITIONS AND DESCRIPTIVE STATISTICS TABLE A1 Brazilian States Acre Alagoas Amapá Amazonas Bahia Ceará Distrito Federal Espírito Santo Goiás Maranhão Mato Grosso Mato Grosso do Sul Minas Gerais Pará Paraíba Paraná Pernambuco Piauí Rio de Janeiro Rio Grande Norte Rio Grande Sul Rondônia Roraima Santa Catarina São Paulo Sergipe Tocantins

16 TABLE A2 Export Share of Selected Foreign Countries Brazilian States (%) 1996 and 2009 Argentina Uruguay Paraguay USA China ROW Export Partners States Acre Alagoas Amapá Amazonas Bahia Ceará Distrito Federal Espírito Santo Goiás Maranhão Mato Grosso Mato Grosso do Sul Minas Gerais Pará Paraíba Paraná Pernambuco Piauí Rio de Janeiro Rio Grande Norte Rio Grande Sul Rondônia Roraima Santa Catarina São Paulo Sergipe Tocantins Average Note. ROW ¼ rest of the world. 267