یکپارچگی بازار مالی: نظریه و نتایج تجربی

In this article, we introduce a new theoretical international asset pricing model which accounts for partial financial market segmentation. We show that if some investors do not hold all international assets because of implicit and/or explicit segmentation factors, the world market portfolio is not efficient and the classic ICAPM must be augmented by a new factor reflecting the local risk undiversifiable internationally. We test this model empirically for a sample of emerging markets. Our findings show that the degree of market integration is time-varying and that the premium associated with the domestic risk factors is the most important component of the total risk premium. However, our results also show that most of the emerging markets we study have become more integrated in the end of our sample period as a result of liberalization and reforms.

The extent to which a domestic stock market is integrated into the world market has a decisive impact on a number of issues addressed by international financial market theory. Indeed, expected gains from world portfolio diversification, the cost of capital, and other criteria for capital budgeting decisions are quite different under local, global, and mixed pricing. The evolution of market integration is affected by both institutional and behavioral factors. First, financial integration is a result of economic, institutional and political reforms. In particular, integration depends on the ability of global investors to access domestic securities as well as the ability of domestic investors to access foreign investment opportunities, legally and/or illegally. In fact, access to worldwide international investment opportunities, through legal means, homemade diversification and illegal means, increases the exposition of domestic assets to the global factor and therefore improves the national stock market integration level. Second, behavioral factors such as risk aversion, relative optimism, and information perception may also affect the desire to invest abroad and thus market integration. In recent decades, barriers to foreign investment have been removed, country funds have been introduced and American depository receipts (ADR) have been listed on mature markets in order to develop financially integrated stock markets. In fact, a move towards integrated stock markets should lead to a lower cost of capital, greater investment opportunities, and higher savings and growth made possible by international risk sharing (Stulz, 1999, Bekaert and Harvey, 2003 and Carrieri et al., 2006). This process of stock market integration is complex, gradual and takes years, with occasional reversals (Bekaert and Harvey, 1995). Most domestic stock markets should be between the theoretical extremes of strict segmentation (integration zero) and perfect integration; in other words they are partially integrated. Therefore, assessing the degree of market integration is a purely empirical question that can appropriately be addressed only within the context of an international capital asset-pricing model (ICAPM). In financial literature, we find theoretical domestic asset pricing models in which it is assumed that markets are strictly segmented (Sharpe, 1964 and Ross, 1976 among others) and theoretical international asset pricing models in which it is assumed that markets are perfectly integrated (Adler and Dumas, 1983 and Solnik, 1983 among others). However, there are no theoretical international asset pricing models for partially segmented markets, except those developed in the vein of Stulz, 1981a, Stulz, 1981b and Errunza and Losq, 1985.1 The main conclusion of these models is that assets to which all investors have access are priced as if the markets were integrated, whereas ineligible securities would generally command a super risk premium proportional to the differential risk aversion and the conditional market risk. In any case, the model of Stulz, 1981a and Stulz, 1981b reduces market segmentation to tax effects only, whereas the model of Errunza and Losq (1985) is empirically difficult to test because its main variables are not observable. Furthermore, these models postulate an overly simple world in which only two countries and two sets of securities are studied and numerous unrealistic hypotheses are made. There are, nevertheless, several empirical models of partial segmentation including Bekaert and Harvey, 1995, Adler and Qi, 2003, Hardouvelis et al., 2006 and Carrieri et al., 2006. These models offer a pure econometric combination of local and global risk factors and attempt ad hoc tests of market integration. The current article attempts to fill this gap by developing a testable theoretical ICAPM for partially integrated markets. Our model is based on more realistic assumptions and can be applied to all markets or individual assets. We assume simply that some global investors do not want and/or do not have access to foreign assets as a result of explicit and/or implicit barriers on inflows and/or outflows, barriers which may make markets partially segmented. Thus, our model enables us to price assets in dynamic intermediate market structures where markets are neither perfectly integrated nor strictly segmented. The available theoretical and empirical models for international asset pricing can be viewed as particular cases of our general ICAPM. More interestingly, we propose a unified theoretical and empirical framework for testing such issues in international finance as world market structure, international capital budgeting, and portfolio diversification. In addition, our model can be extended to address other questions such as exchange rate risk and regional integration dynamics. The rest of the article is organized as follows. Section 2 discusses the related previous works. Section 3 presents our model and its main implications. The econometric methodology is introduced in Section 4 and the data is described is Section 5. Section 6 discusses our major empirical results. Concluding remarks and future extensions are in Section 7.

In the absence of a testable theoretical international asset pricing model for partially segmented markets, previous works rely on pure econometric combinations of global and local risk factors for ad hoc tests of market integration. The results were very heterogeneous and mostly inconclusive. This article offers a theoretical testable partially integrated ICAPM. We assume simply that some global investors do not want and/or cannot have access to foreign assets as a result of explicit and/or implicit barriers on inflows and/or outflows, a situation that may make stock markets partially segmented. Thus, our model enables us to price assets in dynamic intermediate market structures where markets are not in the extreme states of perfect integration or complete segmentation. The available models studied in previous works can be seen as particular cases of our general model. In addition, we propose a unified theoretical and empirical framework for testing important issues in international finance such as integration dynamics, the cost of capital and world portfolio diversification. Econometrically, we introduce a suitable empirical framework using a multivariate GARCH in mean methodology and non-linear models. Our approach allows us not only to test our model but also to understand the nature of the integration process and assess the relative contributions of global and local risk factors to the total premium. We show that the degree of stock market segmentation is time-varying depending on a set of local and global variables related to national and international market structures. We show also that most studied markets have become more integrated as a result of liberalization and reforms. Finally, our findings suggest that because of market segmentation, the risk premium associated with the domestic risk undiversifiable internationally is the most statistically and economically significant component of the total risk premium for all the studied Asian and Latin American emerging markets and that the total risk premium reflects mainly local events. However, the share of the premium associated with global factors has increased in recent years indicating a higher degree of market integration and higher sensitivity to world events. By comparison with emerging markets, our two developed markets Canada and France have four main dissimilarities: the total risk premiums are significantly smaller, less volatile, dominated by global factors and reflect mostly international events.