Using the consumption correlation-based criterion, this paper analyzes international capital mobility for both advanced and developing countries. We provide evidence that global capital markets are imperfectly integrated for both advanced and developing countries. However, a clear difference between these groups of countries emerges when their consumption growth has stagnated; in developing countries at such times, the opportunity to smooth their consumption drops dramatically.
International capital mobility is an important research topic in international finance since most countries are now engaged in exchanges of, not only economic goods and services, but also financial assets. Furthermore, different theoretical assumptions regarding the level of a country's integration to the rest of the world lead to different policy implications. For these reasons, much research has been conducted in this area in the past.
There are broadly two categories of methodology when assessing international capital mobility. One is based on macroeconomic variables; the investment-savings (Feldstein and Horioka, 1980) and consumption correlation (Obstfeld, 1994) criteria. Among them the former approach, which suggests no investment-savings correlation in perfectly integrated markets, dominates the literature. However, there is no definitive conclusion reached in previous studies from this approach. Although international market integration has been advancing over the years and higher integration is observed at the intra-country level rather than in the cross-country context (Atkeson and Bayoumi, 1993), it is not clear as to whether this is appropriate for assessing capital mobility.2 For example, there is still strong evidence against perfect capital mobility even for advanced countries during a period with minimal regulation (Obstfeld and Rogoff, 2000), and furthermore it suggests a higher level of integration for developing countries than advanced countries (Sinha and Sinha, 2004).3
International capital mobility can be also examined using interest parity conditions, and it is probably fair to say that in the long-run there is more evidence of global financial market integration using interest parity conditions than from the investment-savings criterion. For example, Camarero et al. (2010) showed that the real interest parity condition holds for a panel of advanced countries in the long-run when structural breaks are considered. Lothian and Wu (2011) instead used the uncovered interest parity (UIP) condition using a long history of data and provided evidence in favor of this condition when the sample period of 1980s is dropped from the analysis. Similarly, Chinn and Meredith (2004) provided support for the UIP for a longer maturity. Furthermore, Taylor (1987) used contemporaneously sampled data to test the covered interest parity (CIP) condition and overwhelmingly supported this condition for advanced countries. In contrast, these interest parity conditions seem to be less supported in the short-run because of the presence of transaction costs, expectations errors, risk premiums, among many other factors (e.g., Sarno, 2005).
Against this background, we study global capital market integration based on the consumption correlation criterion for advanced and developing countries while at the same time considering regime-shifts in the data. This criterion has been argued as having a more solid theoretical foundation than the investment-savings criterion (Obstfeld, 1986 and Taylor, 1994) and is viewed as a second best approach since our data set has wide country coverage and includes data from developing countries which often do not possess long historical data on interest rates.4 Furthermore, the importance of shifts is underlined in our analysis since they have been discussed as one reason for the poor performance of the consumption function (e.g., Koedijk and Smant, 1994, Hall et al., 1997 and Dufrenot and Mignon, 2004). Finally, note that our main focus on a cross-country consumption correlation is closely related to the consumption correlation puzzle (Backus et al., 1992) which asserts that consumption should be more highly correlated across countries than with domestic output since country-specific income risks are insured in a perfect world.
We assessed international capital mobility in the panel data context for advanced and developing countries based on Obstfeld's theoretical model (1994). The distinguishing features of this paper are (1) to utilize the data from a wide range of countries and (2) to consider the regime-sensitive relationship in the consumption correlation across countries. Given the high consumption correlation and the high correlation between consumption and domestic resources for advanced countries, we could not provide clear evidence that they have more/easier access to international capital markets than developing countries. However, there is a sharp difference between these two groups of countries when their consumption growth slows down. In particular, the opportunity of risk-sharing is rather limited when consumption growth is low in developing countries; a consumption correlation between developing countries and the rest of the world becomes negative. While we acknowledge that our model is open to criticism due to its simplicity, empirical results from the correlation based approach yield statistical results more consistent with conventional expectations than the investment-savings approach and give useful information particularly when identifying heterogeneities between advanced and developing countries.