Further evidence on the determinants of regional stock market integration in Latin America.

AuthorGuesmi, Khaled
  1. Introduction

    Latin American countries have established a key economic region over the past twenty years. The regional economic dynamics is substantially driven by Brazil as the sixth largest economy in the world overtaking the United Kingdom and Italy as well as by Mexico as the second largest economy of the region and the 14th largest economy in the world (August 2012). For their parts, Argentina and Chile are ranked 27th and 41st largest economies respectively. With a combined GDP of nearly 4,400 billion, this group of four fastest-growing economies in the Latin American region is located between Germany (3,600 billion) and Japan (5,800 billion). A number of studies have been devoted to these countries given their important role in world's international trade and economic growth, but the main focus was extensively on the issue of trade integration. Indeed, while Mexico is part of the North American Free Trade Agreement (NAFTA), Argentina, Brazil, and Chile (associate member) are the source of inspiration behind the creation MERCOSUR, another free trade area.

    The issue of financial integration of these countries is seldom addressed, even though stock markets in Sao Paulo, Mexico City, Buenos Aires and Santiago have recently gained much attention from individual and professional investors, especially after respective governments undertook a series of structural economic and financial reforms to improve the transparency and attractiveness of their financial markets (Bekaert, 1995). In addition, existing attempts on Latin American market integration such as Bekaert and Harvey (1995, 1997), Adler and Qi (2003), and Hardouvelis et al., 2006) arbitrarily chose several financial and macroeconomic variables to model the dynamics of market integration without formally testing their economic relevance. For instance, Bekaert and Harvey (1995) propose a model allowing for time-varying world market integration of individual markets and find that increasing integration affects negatively and significantly the cost of equity. More recently, Bekaert and Harvey (2000) introduce changes in dividend yields to measure changes in the cost of equity and document that changes in dividend yields have an insignificant effect on the cost of equity. Carrieri et al. (2007) also use some arbitrary variables to model the dynamics of market integration measure which is then related to a number of factors that may explain the changes in the level of financial integration. Several studies have examined the issue of market linkages and integration outside the asset pricing frameworks. Using Geweke (1982)'s measures of feedback for different pairs of nine national equity markets, Bracker et al. (1999) show significant impacts of macroeconomic variables on the bilateral lead-lag market linkages. Chinn and Forbes (2004) find that direct trade with large economies is the only important factor explaining cross-market links whereas trade competition, bank lending and foreign investment have no significant effect. There is also evidence to suggest a weak role of macroeconomic fundamentals in explaining long-run cointegration of stock returns (Cheung and Lai, 1999).

    Although past studies have permitted a better understanding of Latin American equity market integration as well as its determinants, they mainly rely on the concept of market correlation that is not directly related to the true patterns of evolving market integration. As noted by Carrieri et al. (2007), correlations are informative for portfolio allocation and management, but they do not constitute an accurate measure of diversification benefits or overall integration. In particular, Pukthuanthong and Roll (2009) show the inappropriateness of correlation as a proper measure of integration and they argue that two highly integrated markets may have a low correlation. Indeed, if returns on the two markets are affected by the same common factors but do not have the same sensitivities to all of them, the two markets are highly integrated but only weakly correlated. Adler and Dumas (1983) also point out that the correlation between markets depends very much on their level of international trade. As a result, market co-movement reflects only sector linkages instead of market integration. Therefore, tests for market integration need to be built on asset pricing model frameworks which impose the similarity of systematic risks (Bekaert and Harvey, 1995; Bhattacharya and Daouk, 2002).

    This study contributes to the extant literature by exploring the determinants of regional market integration for emerging markets in Latin America. Given the exploratory nature of the empirical investigation, we attempt to encompass as much explanatory variables as possible. We consider a complete list of potential determinants from the past empirical literature on market integration. A partially segmented International Capital Asset Pricing Model (PS-ICAPM) in the spirit of Bekaert and Harvey (1995) is used to model the dynamics of expected returns. The model allows not only for the time-varying market integration but also for time-varying covariance risks. It considers the real exchange rates as a common source of systematic risk, in addition to the local and regional systematic risks. We adopt the Capiello et al. (2006)'s multivariate asymmetric DCC-GARCH process to accommodate the conditional variances and covariances of stock returns.

    Using monthly data from four largest markets in the Emerging Latin America over the period 1996-2008, we find that the number and nature of driving factors for regional integration are very sensitive to the exchange rate risk measures. In the meanwhile, the trade openness and local stock market development play a common and significant role in explaining the dynamics of regional market integration.

    The remainder of the article is organized as follows. Section 2 presents the empirical model. Section 3 describes the data. Section 4 discusses the obtained results. Section 5 concludes the article.

  2. The partially segmented ICAPM

    Under the assumption of purchasing power parity (PPP) and perfect integration, the international version of the CAPM of Sharpe (1964) and Lintner (1965) predicts that excess expected return on a security is priced with respect to the world market risk factor, usually represented by the stochastic fluctuations of a world market portfolio. When the regional integration is examined, the world market portfolio can be replaced by a regional market portfolio. In this article, we extend the well-known one-factor ICAPM to the case of partial integration with three sources of systematic risk that reflect changes in regional stock market, domestic stock market and real exchange rate. Formally, expected returns are priced according to the degree of regional market integration as follows


    where [R.sub.i,t], [R.sub.r,t] and [R.sub.k,t], represent respectively expected excess returns on the local market portfolio in the country i, the regional market portfolio and the exchange rate of currency t against the currency of the country under consideration. Returns on local and regional market portfolios are expressed in the currency of the considered country. [[lambda].sup.r.sub.t-1], [[lambda].sup.d.sub.i,t-1] and [[lambda].sup.k.sub.t-1] are the expected prices of a unit of risk, related to the regional market, the local market and the exchange rate, respectively. k denotes the currencies of the four countries that we consider in the Latin America: Argentina, Brazil, Chile, and Mexico. [[OMEGA].sup.i.sub.t-1] refers to a conditional measure of financial integration degree of market i with the regional market, which falls within the interval [0,1]. If [[OMEGA].sup.i.sub.t-1] = 1, only the covariance risks are priced and the strict segmentation hypothesis is rejected. Inversely, the model is reduced to the domestic CAPM when [[OMEGA].sup.i.sub.t-1] = 0 as only the national systematic risk is relevant for pricing financial securities.

    At the empirical stage, we estimate the following system of equations:


    where [r.sup.f.sub.it] = ([r.sup.f.sub.regt], [r.sup.f.sub.At], [r.sup.f.sub.Ct], [r.sup.f.sub.Mt] ,[r.sup.f.sub.Bt] [[??].sup.f.sub.At], [[??].sup.f.sub.Ct], [[??].sup.f.sub.Mt], [[??].sup.f.sub.Bt])' refers to the (9 x 1) vector of excess returns on the regional market, the four emerging markets and the four bilateral exchange rates, respectively. All the return series are assumed to be normally distributed. [X.sub.i,t-l] is the vector of information variables available at time t-1 that are likely to drive the integration degree of the market .. Expected prices of risk related to the regional market, to the four bilateral exchange rates and to the local market are allowed to vary through time. They reflect the risk aversion aggregated over all investors and should thus be positive (Adler and Dumas, 1983). [H.sub.t] is the conditional variance-covariance matrix of returns at time t with [h.sup.ii.sub.t] being the conditional variance of the market i, and [h.sup.i,j.sub.t] the covariance between two markets i and j. Following Hardouvelis and al. (2006) and Guesmi and Nguyen...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT