Regional economic integration in Mercosur: The role of real and financial sectors

Basnet, Hem C.; Pradhan, Gyan Article Regional economic integration in Mercosur: The role of real and financial sectors Review of Development Finance Provided in Cooperation with: Africagrowth Institute, Bellville Suggested Citation: Basnet, Hem C.; Pradhan, Gyan (2017) : Regional economic integration in Mercosur: The role of real and financial sectors, Review of Development Finance, ISSN 2959-0930, Elsevier, Amsterdam, Vol. 7, Iss. 2, pp. 107-119, https://doi.org/10.1016/j.rdf.2017.05.001 This Version is available at: https://hdl.handle.net/10419/313580 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ Available online at www.sciencedirect.com ScienceDirect HOSTED BY Review of Development Finance 7 (2017) 107–119 Regional economic integration in Mercosur: The role of real and financial sectors Hem C. Basneta,∗, Gyan Pradhanb aMethodist University, 5400 Ramsey Street, Fayetteville, NC 28311, USA bEastern Kentucky University, 521 Lancaster Avenue, 106 Beckham Hall, Richmond, KY 40475, USA Available online 29 May 2017 Abstract This study explores economic interdependence in Mercosur by examining common trends and common cycles among key macro-variables representing both the real and financial sectors of the economy. The serial correlation common features test reveals that the key macroeconomic variables (real output, investment, and intra-regional trade) share common trends in the long run suggesting that macroeconomic interdependence in the Mercosur economies is strong. The exchange rates demonstrate co-movement in the long run as they share a single common trend. These finding suggests that these economies cannot swing away from long-run equilibrium for an extended duration; they will be brought together by their common trends. Similarly, each variable under consideration shares common cycles lending support to the notion of short-run synchronous movement. The trend-cycle decomposition results reveal that the cyclical movements of real output and trade are synchronized with a high degree of positive correlations. Our overall findings thus provide justification and optimism for deeper economic integration among Mercosur countries. © 2017 Africagrowth Institute. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). JEL classifications: F15; F42 Keywords: Economic integration; Co-movement; Business cycles; Mercosur; Exchange rates 1. Introduction Recent decades have witnessed regional economic integration gaining momentum in many parts of the world. Since 1990, there have been more than 14 agreements pertaining to free trade areas and custom unions. By lowering trade barriers and fostering greater mobility of human and physical capital, regional trading arrangements provide many benefits and may contribute to economic growth in member countries. These benefits include reduced transactions costs, lower prices for consumers, more efficient use of resources, scale economies, enhanced competition among firms, greater certainty and investment, technological improvements, and increases in productivity. Regional integration can also lead to deeper assimilation and may complement multilateralism by setting a precedent which other nations will follow (Carbaugh, 2015). Given these potential advantages, ∗Corresponding author. E-mail addresses: [email protected] (H.C. Basnet), [email protected] (G. Pradhan). it is not surprising that many countries around the world have continued to pursue greater economic integration. On the other hand, regional trade agreements are also discriminatory in that some nations are treated differently than others. Further, they may decrease incentives for nations to pursue multilateral agreements because trade bloc members may not gain additional economies of scale through multilateralism. Finally, as the recent experiences of some members of the European Union such as Greece and Spain have demonstrated, integration is no panacea. The loss of independent monetary and exchange rate policies can pose serious limitations in tackling economic crises. And as is also apparent from the recent exit of the United Kingdom from the European Union, non-economic factors, particularly the role of special interest politics, can be crucial. The main objective of this paper is to investigate the economic interdependence of the economies of Mercosur (Southern Common Market). Established in 1991 between Argentina, Brazil, http://dx.doi.org/10.1016/j.rdf.2017.05.001 1879-9337/© 2017 Africagrowth Institute. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 108 H.C. Basnet, G. Pradhan / Review of Development Finance 7 (2017) 107–119 Paraguay, and Uruguay,1Mercosur is one of the largest regional trade blocs in Latin America. The main goal of Mercosur is to eliminate barriers to facilitate the free movement of goods, people, and currency among member countries. The formation of Mercosur was inspired by the success of the European Union and represents a first step toward greater regional integration. The member countries decided to adopt a gradual approach toward deeper integration, starting with a free trade area to an eventual customs union, and from a contractual agreement to a structured international organization (UNCTAD, 2003). Although Mercosur has a long way to go to achieve its goals, member countries have agreed to set up an institutional framework to foster economic policy coordination. In 2000, a high level monitoring body (equivalent to the Economic and Financial Council in the European Union) was created to implement agreements and treaties among member countries with regard to the convergence of public deficit and debt ratios. The process of deeper integration in Mercosur appears to be steadily gaining momentum. In light of these efforts toward economic integration, this study examines the degree of macroeconomic synchronization (i.e., the co-movement of macroeconomic variables) among the member countries of Mercosur. For this purpose, we make an attempt to identify the number of common trends and common cycles. We also separate permanent and transitory components from the original variables which will allow us to identify the degree of co-movement in the long run and short run, and measure the degree of interdependence among the economies under consideration. For this purpose, we examine key macroeconomic variables in the Mercosur countries—real output, investment, intra-country trade, exchange rate, and interest rate, representing both real and financial sector of the economies. Studies suggest that a high degree of macroeconomic synchronization or business cycle co-movement is a necessary condition for promoting economic cooperation among countries involved in an economic integration process (Christodoulakis et al., 1995; Fiorito and Kollintzas, 1994). If business cycle fluctuations are synchronized, harmonized policies to cope with these cycles across countries can be effective (Sato and Zhang, 2006). Likewise, exchange rate dynamics often remain at the core of monetary policy discussions. In an open economy, the monetary authority needs to respond to exchange rate movements which work as shock absorbers. After the collapse of the Bretton Woods system, monetary authorities in developing countries began to emphasize exchange rate stability and correct exchange rate alignments to improve economic performance.2One of the reasons for establishing the European Monetary Union was to promote exchange rate stability among member countries and to encourage trade inside the European Union (Dell’Ariccia, 1999). Acknowledg- 1Venezuela is a recent member of Mercosur. Even though Venezuela signed the membership agreement with Mercosur in 2006, full membership was not granted until 2012. Therefore, we have opted to include only the founding members in our study. 2For details: Ann Krueger, Exchange rate determination, Cambridge University Press 1983. ing the importance of exchange rate movements, Basnet et al. (2015) examine exchange rate movements to assess monetary policy coordination in the ASEAN nations. The results of their study lend support for monetary policy coordination between some but not all ASEAN nations. The effect of exchange rates on macroeconomic stability is linked to interest rates and other macroeconomic variables. International shocks are also transmitted through, among other variables, interest rates. The existence of long-term common trends and short-term common cycles in a set of variables indicates that these variables do not swing for an extended period of time, ultimately move together, and share similar cyclical fluctuations in the short run. We submit that if member countries share synchronous long-term trends and short-term cycles in their key macroeconomic variables, these countries may find it mutually beneficial to strengthen their integration process. Eventually, these countries could potentially even move toward a monetary union, the highest level of economic integration. Such a union would be characterized by, among other features, a common currency, common fiscal and monetary policies, and free mobility of goods, services, labor, and capital. On the other hand, if the impact of a shock is not symmetric across countries seeking deeper integration, harmonized monetary and fiscal policies are unlikely to benefit these countries. That is, non-synchronized movements in macroeconomic variables may indicate weak interdependence which may require different policy prescriptions, and in turn, lower the prospects for integration. Therefore, an examination of the costs and benefits of integration must include a careful and rigorous investigation of the behavior of macroeconomic variables. The common feature analysis has been extensively used in the literature (e.g., Sato and Zhang, 2006; Abu-Qarn and Suleiman, 2008; Castillo Ponce and Ramirez, 2008; Adom et al. 2010; Weber, 2012; Basnet and Sharma, 2013), especially to assess the feasibility of higher levels of policy coordination involving an economic or monetary union. However, Mercosur has been largely exempt from this kind of analysis. Utilizing a variety of methodologies and hypotheses, studies have examined business cycle synchronization (Allegret and Sand-Zantman, 2009), labor market interdependence (Caceres, 2011), and convergence and inequality (Blyde, 2006) in Mercosur. To the best of our knowledge, no study has explicitly analyzed the real and financial sectors of Mercosur countries to explore the possibility of greater economic alliance, particularly from the perspectives of common trends and common cycles. We hope that our findings will provide helpful information as to how favorable the economic conditions are to expedite the process of economic integration in Mercosur. As a corollary, we hope to determine whether these countries require different policy adjustments to internal and external shocks. The rest of the paper is organized as follows. The next section provides a brief economic background of the four Mercosur countries. Next, we describe the data and methodology used to analyze macroeconomic interdependence and business-cycle synchronization. In the following section we discuss the empir- H.C. Basnet, G. Pradhan / Review of Development Finance 7 (2017) 107–119 109 ical results. The final section summarizes and concludes the paper. 2. Brief economic background3 Mercosur is the principal trade bloc in South America. It is the world’s fourth largest trading bloc after EU, NAFTA, and ASEAN. Mercosur is home to more than 250 million people and accounts for almost three-fourths of total economic activity in South America. Among the full member states in the bloc, Brazil and Argentina are the largest economies, and Paraguay and Uruguay are the smallest. Brazil, with a gross domestic product of $2.25 trillion in 2013, is the world’s seventh largest economy; it has large and well-developed agriculture, manufacturing, and service sectors, and is considered the leading voice in the alliance. With respect to trade patterns, Brazil is the largest trading partner for all three countries. Despite China’s growing presence in the Latin American region, each country in the Mercosur region is critically dependent on Brazil. In terms of the proportion of trade, Brazil and Argentina share a relationship of high mutual interdependence. Argentina’s trade with Brazil represented 21.16% of total exports and 26% of total imports in 2013. Brazil’s share of trade with Argentina represented 8.1% of exports and 6.87% of imports in 2013. It is interesting to note that Brazil’s major trade partners are distributed worldwide, led by the United States and China; none of the Mercosur countries is among Brazil’s top five trading partners. The trade shares of Paraguay and Uruguay with Brazil and Argentina are significantly higher, accounting for approximately 30–55% of total trade during the last 13 years. In 2013, Paraguay’s total export and import shares with Argentina and Brazil were 37.65% (7.61 and 30.04%) and 40.57% (14.21 and 26.36%). The same is true for Uruguay; its export and import shares with Argentina and Brazil were 24.32% (5.44 and 18.89%) and 30% (14.23 and 15.77%). It should be noted that there is growing trade reliance between China, Paraguay, and Uruguay as well. For instance, some 28.28% of Paraguay’s total imports in 2013 came from China and the corresponding number was 17% for Uruguay. Fig. 1 shows the percentage shares of total trade (exports plus imports) of each of the four countries in Mercosur. Brazil’s trade reliance within Mercosur is low; it is only about 10% for the study period. Paraguay and Uruguay, on the other hand, appear to be highly dependent on the Mercosur region. For instance, between the 2003 and 2005, Paraguay traded more with the Mercosur countries than with the rest of the world. The percentage share has decreased after 2005, due primarily to the growing commercial presence of China in the Central and South American regions. Fig. 2 shows the annual GDP growth of Mercosur countries and suggests that these countries have shared both good and bad economic times. In the late 1990s and the early 2000s when the 3The data in this section were obtained from the World Bank’s World Integrated Trade Solution (WITS) database. region was experiencing economic crises, all countries suffered significant losses, although some were hit harder than others (see Fig. 2). We also observe that all four countries enjoyed high rates of economic growth from 2003 through 2008, followed by a severe contraction during the global financial crisis (GFC) in 2009. The observed growth rates of these countries demonstrates significant similarities in their economic expansions and contractions, suggesting strong economic interdependence. As the leading economy in the alliance, Brazil recorded positive economic growth during the period, with the exception of 2009. After strong growth in 2007 (6.10%) and 2008 (5.17%), Brazil experienced a severe economic contraction in 2009 due to the GFC; the economy recorded a negative 0.33% annual growth. However, Brazil’s strong domestic and intra-regional markets proved to be less vulnerable to external crises, which made it one of the first emerging market economies to begin a recovery. In 2010, Brazil recorded the last decade’s highest rate of economic growth at 7.53%. Argentina, with a GDP of $610 billion in 2013, is the second largest economy in the coalition. It is endowed with rich natural resources and has benefitted from its export-oriented agricultural and industrial sectors. In recent years Argentina has experienced record economic growth (see Fig. 2). While all four countries show synchronous movement in their economic growth during the study period, Argentina certainly displays a greater degree of fluctuation, especially during the financial crisis in Brazil, and its own currency crisis period.4Argentina’s economy contracted by almost 11% in 2002. A huge spike in Fig. 2 corresponds to that period. However, Argentina has had robust growth thereafter (with the exception of the crisis period in 2009). Paraguay and Uruguay are the bloc’s smallest countries, with a population of 6.8 million and 3.4 million in 2013. Being a landlocked country, Paraguay is characterized by re-export of imported consumer goods to neighboring countries. Following the Argentine and Brazilian crises, other countries in the region were also hit hard by speculative attacks and capital outflows. As a consequence, the Paraguayan economy suffered from those episodes in Latin America in the early 2000s. Like Argentina and Brazil, Paraguay’s economy also grew rapidly between 2003 and 2008. The GFC in 2009 took a toll on Paraguay’s economy as well, causing the annual growth rate to fall by 4% in 2009. Growth, however, resumed at an impressive 13.09% in 2010, the highest growth not only in Mercosur, but in all of South America. Among the Mercosur countries, Paraguay experienced a negative growth of 1.24% in 2012 followed by another leap in 2013 (14%). Despite being the smallest country in the bloc in terms of size, Uruguay has an economy that is significantly larger than 4In the 1990s, Brazil suffered from record high inflation, ranging from 100% to nearly 3000 percent per year. Brazil’s economic situation deteriorated significantly, prompting it to owe almost 46% of GDP to foreign creditors. Fear and uncertainty among investors about the region’s largest economy escalated and resulted in massive capital flight. Higher inflation coupled with currency devaluation created a deep financial crisis in Brazil which engulfed the entire region. Argentina also faced massive speculative attacks on its currency from investors, causing its currency to depreciate by 255% in five months (from January 2002 to May 2002). Argentina defaulted on its debt in January 2002. 110 H.C. Basnet, G. Pradhan / Review of Development Finance 7 (2017) 107–119 0 10 20 30 40 50 60 00 01 02 03 04 05 06 07 08 09 10 11 12 ARG BRL PAR URG percent Fig. 1. Total trade in Mercosur. Source: UN COMTRADE. -15 -10 -5 0 5 10 15 00 01 02 03 04 05 06 07 08 09 10 11 12 ARG BRL PAR URG ARG BRL PAR URG percent Fig. 2. GDP growth. Source: World Bank. that of Paraguay ($56 billion versus $29 billion in 2013). Following the region’s crisis, Uruguay grew at an average rate of 8% annually during the period 2004–2008. Even though the GFC slowed down its rapid economic growth, which fell to 2.35% in 2009, it managed to avoid a recession and negative economic growth (see Fig. 2). Uruguay’s economic prosperity relies heavily on the economic health of the Mercosur giants Brazil and Argentina. Its total partner share of exports and imports with them is considerable; it was 58.13% in 2012. The Mercosur countries exhibit great similarities in their inflation rates as well. Among the member countries, Argentina has a long history of hyperinflation. With a few exceptions, it has typically experienced double-digit inflation during the study period, as is evident from Fig. 3. While Brazil, Paraguay, and Uruguay all suffered from relatively high inflation between 2002–2003, inflation remained in single digits thereafter (except for Paraguay in 2008, which was 10.2%). The average inflation rates during the past 13 years in Brazil, Paraguay, and Uruguay were 6.6%, 7.6%, and 8.3%, while it was 13.5% in Argentina. 3. Data and research methodology Our study is based on quarterly data from 2001 to 2012 on real gross domestic product (RGDP), domestic investment (INVT), intra-Mercosur trade (TRADE), nominal exchange rate (EX), and money market interest rates (INT) for the four member countries of Mercosur. Real gross domestic product is used as a measure of real output and gross capital formation is used as a proxy for domestic investment. The data for these two variables are obtained from the World Bank’s World Development H.C. Basnet, G. Pradhan / Review of Development Finance 7 (2017) 107–119 111 -4 0 4 8 12 16 20 24 28 32 00 01 02 03 04 05 06 07 08 09 10 11 12 ARG BRL PAR URG ARG BRG PAR URG pecentr Fig. 3. Inflation. Note: Inflation for Brazil, Paraguay, and Uruguay is measured by the consumer price index, whereas inflation for Argentina is measured by the annual growth rate of the GDP implicit deflator. Due to lack of data availability through standard data sources, the GDP implicit deflator is used for Argentina, which shows the rate of price changes in the economy as a whole. Source: World Bank. Indicators (WDI). Since the present study investigates the extent of macroeconomic interdependence among Mercosur countries, we choose to examine intra-Mercosur trade rather than trade flows in general. Intra-Mercosur trade includes exports plus imports of each country only within Mercosur. For example, Argentina’s trade refers to its exports to Brazil, Paraguay, and Uruguay, and its imports from the same three countries. Trade statistics are based on the United Nations Commodity Trade Statistics Database (UN Comtrade) and covers all products. Nominal exchange rates, obtained from International Financial Statistics (IFS) are expressed in terms of the domestic currencies per U.S. dollar. For the purpose of interest rates, money market rates are used. All data series except interest rates are normalized in logarithmic forms. The choice of time span is largely driven by the availability of data, especially on intra-Mercosur trade. 3.1. Methodology The empirical strategy consists of testing for common trends and common cycles. Prior to conducting these tests, all variables are tested for stationarity and their order of integration by employing the Dickey–Fuller test, the Augmented Dickey–Fuller test, the Phillips–Perron (PP) test, and the KPSS. Thereafter, following the Johansen (1988) and Johansen and Juselius (1991) maximum likelihood test, we estimate the following vector autoregressive model: yt= A0+ A1yt−1+ A2yt−2+ ... + Apyt−p+ εt(1) where ytis a (n × 1) vector of each variable (i.e., either real GDP, investment, intra-regional trade, exchange rate, or interest rate) of the countries under consideration; n = 4; A0is a (n × 1) vector of constants; Ai, i = 1, 2,. . .p, is a (n × n) matrix of coefficients to be estimated; p is the selected lag length, and εtis the vector of error term which is expected to be serially uncorrelated with zero mean. We then rewrite Eq. (1) in the following Vector Error Correction (VEC) form when all series are I(1): Δyt= A0+ Γ1Δyt−1+ Γ2Δyt−2+ ... + Γp−1Δyt−p+1+ Πyt−p+ εt(2) where yt= yt− yt−1and Γi= −⎡ ⎣I − p  j=i+1 Aj⎤ ⎦, Π = −I − p  i=1 Aiare n × n matrices of coefficients and contain information about the long-run relationship between the variables. Two likelihood ratio tests used in Johansen (1988) to test the rank of  matrix are the maximum eigenvalue test statistics, λmax, and the trace test statistics, λtrace: λtrace = −T n  i=r+1 ln(1 − λi) (3) λmax = −T ln(1 − λr+1) (4) where λiis the estimated value of the characteristic roots (also called eigenvalues) obtained from the estimated П matrix and T is the number of usable observations. If 0 < rank () < n, the n variables are cointegrated and thus share long-run common trend. The number of common trends is determined by the number of independent cointegrating vectors. Johansen (1988) shows that given an (n × 1) vector yt, there can exist r < n linearly independent co-integrating vectors (r), which implies (n − r) common trends. Vahid and Engle (1993) propose a test for determining the number of common cycles given the presence of common trends. The number of common cycles is determined by the co-feature vectors, which are identified by testing the significance of the canonical correlations between ytand 112 H.C. Basnet, G. Pradhan / Review of Development Finance 7 (2017) 107–119 W = (αyt−1,Δyt−1,Δyt−2,, ..., yt−p+1), where α is a (n × r) matrix. The test points out that given r linearly independent co-integrating vectors, if a series ythas common cycles, there can, at most, exist s = (n − r) co-feature vectors that eliminate common cycles. The presence of co-feature vectors represents a form of convergence in the short run. To investigate common cycles, we apply the test suggested by Vahid and Engle (1993) for determining the significance of the smallest canonical correlation: C(k∗, s) = −(T − k ∗ −1)Σ ln(1 − ρ2 i) (5) where ρ2 i(i = 1,2. . .s) are the s smallest squared canonical correlations between ytand W = (αyt−1,Δyt−1,Δyt−2,, ..., yt−p+1), T is the number of observations, and k* is the lag length in the VAR system. Under the null hypothesis, this statistic has a χ2distribution with (nk*s − rs − ns + s2) degrees of freedom. Engle and Issler (1993), however, use the F-test5approximation proposed by Rao (1973) to test the significance of canonical correlations. If a system contains s independent co-feature vectors then there are (n − s) common cycles. A dimension of (n × s) matrix ˜ α and of (n × r) matrix α are referred to as the co-feature and co-integrating vectors, respectively. Vahid and Engle (1993) decompose the permanent (trend) and transitory (cyclical) components of the original series when the sum of co-integrating vectors (r) and the number of co-feature vectors (s) is equal to the total variables (n) i.e., r + s = n. They further note that when r + s = n then an (n × n) matrix A =˜ α αis of full rank and thus A−1exists. The trend and cycle decomposition can be obtained by partitioning the columns of A−1such as A−1=˜ α−|α−, where ˜ α and β are the matrices of the dimension of (n × s) and (n × r) for co-feature and co-integrating vectors, respectively. Finally, the trend and cyclical components are recovered as follows: yt= A−1Ayt=˜ α−˜ αyt+ α−αyt(6) Eq. (6) is used to decompose the trend-cycle in a series to analyze long-term and short-term co-movement among the exchange rates. 4. Results The unit root test results, reported in Table 1, suggest that all series are stationary in the first difference. While the ADF and the PP tests fail to reject the null hypothesis of unit roots in the real output of Brazil at the conventional level, the KPSS test rejects the null at the five percent significance level. Therefore, we proceed with our analysis under the assumption that all series are integrated of the first order, denoted as ∼I(1). 5Engle and Issler (1993) claim that the F-statistic yields superior results; we present the results of both χ2 and the F-test. 4.1. Common trend analysis Johansen’s (1988) and Johansen and Juselius’s (1991) cointegration test is used to identify the common trend(s) in the Mercosur zone that link macroeconomic variables together in the long run. Keeping the sensitivity of lag in the VAR structure, the present study selects the lag length by utilizing the AIC and LR tests. Both test results are reported in Table 3. Panel A indicates that the appropriate lag length for all models except for the exchange rate is two; a lag length of 3 is identified for the exchange rate. Panel B shows that the LM test assures that the selected lags do not suffer from serial autocorrelation. The results of the cointegration tests are reported in Table 2. From these results, we can safely reject the null hypothesis of no cointegrating vector (r) in real outputs. Both λtrace and λmax statistics indicate the presence of at least two cointegrating vectors in real output. This means that there exist two common trends (i.e., n − r: 4 − 2 = 2) in real outputs. Likewise, the cointegration results indicate that investment and intra-Mercosur trade are also cointegrated in the long run. Investment and trade have one and two cointegrating vectors, implying that these variables share three (n − r: 4 − 1 = 3) and two (n − r: 4 − 2 = 2) common trends in the long run, respectively. The test results show that the λtrace and λmax statistics do not produce conflicting cointegrating vectors. With regard to the financial sector, the test results denote at least three cointegrating vectors among the four exchange rate series, implying a common trend (i.e., 4 − 3 = 1). While the existence of one or more cointegrating vectors is sufficient to establish the long run relationship between variables, the n − 1 cointegrating vectors ensures a stable long run relationship. Likewise, the null hypothesis of no cointegrating vector is rejected at the 1% significance level for interest rate as well. The test results in Table 2 indicate that there is evidence of at least one cointegrating vector that establishes the long run relationship among interest rates. The results further imply that there are three (n − r: 4 − 1 = 3) common trends, suggesting that the series moves together in the long run. Note that a common trend implies that permanent shocks eventually affect all the countries in the same way (Engle and Issler, 1993) whereas common trends ensure that the series moves together in the long run. The long run synchronous movement among a number of macroeconomic variables often begins with the countries facing similar external conditions. Since the exchange rate series shares a common trend, exchange rate forecasts of one country in Mercosur may be improved by taking the forecasts of other countries into consideration. Our results suggest, based on the movements of their macroeconomic variables, that the economies of the four Mercosur countries cannot swing for a long period of time, and that they eventually move together. Note that the existence of a long-run relationship does not imply that these countries do not differ in their policy implementation over time. It simply means that any deviation in the short run will be corrected by internal dynamics within the system that corrects the misalignment and pushes these economies back toward the equilibrium path in the long run (Darrat and Al-Shamsi, 2005). To support this conclusion, vari- H.C. Basnet, G. Pradhan / Review of Development Finance 7 (2017) 107–119 113 Table 1 Unit root tests. Variables ADF (first difference) PP (first difference) KPSS (first diff) Constant Constant + trend Constant Constant + trend Argentine RGDP −3.35** −3.26*−3.23** −3.15*0.29** INVT −3.25** −3.12 −3.15** −3.00 0.18** TRADE −3.11** −2.98 −3.13** −2.93 0.13** EX −4.56** −4.72** −5.73** −5.68** 0.11** INT −7.49** −7.42** −7.49** −7.42** 0.24** Brazil RGDP −2.28 −2.23 −2.28 −2.16 0.18** INVT −2.85*−2.80 −2.98** −2.85 0.14** TRADE −3.05** −2.89 −3.07** −2.78 0.13** EX −6.29** −6.24** −6.21** −6.15** 0.13** INT −4.57** −4.83** −3.39** −3.35 0.22** Paraguay RGDP −3.14** −2.93 −3.19** −2.97 0.18** INVT −4.77** −4.61** −3.67** −3.53** 0.10** TRADE −4.08** −5.00** −3.97** −3.86** 0.09** EX −4.17** −4.34** −5.99** −6.12** 0.30** INT −11.05** −10.93** −21.73** −21.95** 0.25** Uruguay RGDP −3.40** −2.36 −2.70*−2.27 0.39* INVT −3.60** −4.36** −3.27** −3.18*0.27** TRADE −3.14** −3.74** −3.16** −2.90 0.18** EX −5.47** −5.59** −5.46** −5.49** 0.27** INT −4.85** −4.81** −5.97** −5.92** 0.18** RGDP is real gross domestic product, INVT is private investment, TRADE is intra-Mercosur trade, EX is nominal exchange rate, and INT is interest rate (money market rate). *Indicates significance at the 5% level. ** Indicates significance at the 1% level. Table 2 Cointegration test results. Variables Eigenvalues Null hypothesis λ-trace λ-max Critical values (5%) λ-trace λ-max RGDP 0.489 r = 0 57.11*30.18*40.17 24.16 0.311 r ≤ 1 26.93** 16.77** 24.28 17.80 0.167 r ≤ 2 10.16 8.22 12.32 11.22 0.042 r ≤ 3 1.94 1.94 4.13 4.13 INVT 0.53 r = 0 72.24*33.66*63.87 32.12 0.36 r ≤ 1 38.58 20.15 42.91 25.82 0.28 r ≤ 2 18.43 14.72 25.87 19.38 0.08 r ≤ 3 3.72 3.72 12.52 12.52 TRADE 0.575 r = 0 62.59*38.48*40.17 24.15 0.306 r ≤ 1 24.10** 16.40** 24.27 17.80 0.143 r ≤ 2 7.69 6.96 12.32 11.22 0.016 r ≤ 3 0.72 0.73 4.12 4.12 EX 0.57 r = 0 99.40*43.96*63.87 32.11 0.39 r ≤ 1 55.43*26.40*42.91 25.82 0.30 r ≤ 2 29.03*18.85** 25.87 19.38 0.17 r ≤ 3 10.17 10.17 12.51 12.51 INT 0.55 r = 0 62.41*43.35*40.17 24.15 0.19 r ≤ 1 19.06 11.34 24.27 17.79 0.12 r ≤ 2 7.72 7.25 12.32 11.12 0.01 r ≤ 3 0.46 0.46 4.12 4.12 *Indicates significance at the 1% level. ** Indicates significance at the 5% level. 114 H.C. Basnet, G. Pradhan / Review of Development Finance 7 (2017) 107–119 Table 3 Test statistics for lag length selection and serial autocorrelation. Panel A: Lag length selection criteria Panel B: Serial autocorrelation LM test χ2(49) LR AIC SC HQ LM test p-Value Lags RGDP 0 NA −4.85 −4.70 −4.80 – – 1 483.92 −16.54 −15.73 −16.24 11.63 0.77 2 52.43a−17.31a−15.85a−16.77 16.97 0.39 3 8.87 −16.87 −14.76 −16.09 8.71 0.93 4 6.03 −16.36 −13.61 −15.34 76.65 0.00 Lags INVT 0 NA −2.72 −2.56 −2.66 – – 1 417.07 −12.69 −11.88 −12.39 5.97 0.98 2 53.42a−13.49a−12.03a−12.95a15.09 0.52 3 7.99 −13.02 −10.91 −12.24 4.88 0.99 4 5.06 −12.48 −9.72 −11.46 83.04 0.00 Lags TRADE 0 NA −6.36 −6.20 −6.30 – – 1 477.17 −17.87 −17.06 −17.57 24.76 0.07 2 66.34a−19.04a−17.58a−18.49a19.90 0.22 3 16.02 −18.82 −16.72 −18.04 16.55 0.42 4 23.27 −18.96 −16.20 −17.94 95.55 0.01 Lags EX 0 NA −2.83 −2.68 −2.77 – – 1 324.85 −9.26 −8.51a−8.97 45.50 0.00 2 40.13 −9.59 −8.23 −9.07 35.05 0.00 3 40.97a−10.04a−8.07 −9.29 15.44 0.49 4 21.34 −10.04 −7.47 −9.06 19.43 0.24 Lags INT 0 NA 21.66 21.82 21.72 – – 1 239.76 16.37 17.18 16.67 14.21 0.58 2 64.25a15.30a16.75a15.84a13.50 0.63 3 13.20 15.60 17.69 16.38 24.36 0.08 4 20.86 15.56 18.29 16.58 35.88 0.00 Autocorrelation LR test Ho: no serial correlation at the selected lag. aIndicates lag order selected by the criterion. ous hypotheses as they relate to the cointegrating relation (β) and the speed of adjustments (α) are tested. Table 4, Panel A reports the test results on β, which examines whether a particular variable in the model can be excluded from the long-run relationship. In order to establish the individual significance of each variable we conduct the likelihood ratio (LR) test for the null hypothesis that each variable in the model does not contribute to the long-run relationship, i.e., H0= βk= 0, where, k = 1, 2. . .4. The test results (Table 4, Panel A) indicate that all of the variables are significant in the cointegration terms, suggesting an equal contribution in moving the system toward long-run equilibrium. Table 4, Panel B reports the results of the weak exogeneity test. A variable is said to be weakly exogenous with respect to the long-run parameter β if that variable does not respond to the discrepancy from the long-run equilibrium (Enders, 2004). In other words, if the speed of adjustment parameter αiis zero, the variable in question is weakly exogenous. We test the null hypothesis that each variable in the system is weakly exogenous, i.e., αi= 0, where i = 1, 2. . .4. While we reject the null hypothesis for all three variables, we fail to reject the null hypothesis for the investment variable for Uruguay6at the conventional level (Table 4, Panel B). Results of all the tests to establish the long-run convergence relationship indicate that macroeconomic variables in Mercosur have a long-run link; they move together in the long run and any short-run deviation from equilibrium tends to be transitory. Note that the more cointegrating vectors there are the more stable the system (Dickey et al., 1991). In other words, it is desirable for an economic system to have n − 1 cointegrating vectors which ensures long run stability from as many directions as possible. Standard investment has more common trend than cointegrating vectors, suggesting a relatively less stable long run relationship. It corroborates with the speed of adjustment coefficient for Uruguay (Table 4, Panel B), which indicates that the investment variable for Uruguay is weakly exogenous. 6These results are not robust with regard to different lag length. However, for the sake of consistency, we used the lag selected by the AIC and LR tests for all three variables. Further, the test for exclusion of variables rejects the insignificant role of any of the variables under consideration. Therefore, we proceed with our analysis with all four countries.