«ABSTRACT Over the period 1972-1986, the correlations of GDP, employment and investment between the United States and an aggregate of Europe, Canada ...»
A. Shock correlation and diversiﬁcation In table 11 we report the equilibrium levels of international diversiﬁcation in the models with restricted and unrestricted stock trade, and compare both with international diversiﬁcation in the data.
Table 11. International Diversiﬁcation
To understand the equilibrium determination of λ, it is helpful to consider ﬁgure 7. The curves plot equilibrium levels of diversiﬁcation given particular tax rates τ on foreign dividends.
Recall that the tax rate in the restricted stock trade economy is set so that the model reproduces the average level of diversiﬁcation observed in the ﬁrst sub-sample of data. The picture shows that in the second period (characterized by less correlated shocks) more international diversiﬁcation is observed in equilibrium; the amount of foreign assets held by domestic consumers increases from Diversiﬁcation in the data is measured as the average across asset and liabilities of the ratio of FDI plus equity to the capital stock for the U.S. v/s Europe plus Canada and Japan (see table 7).
5.5 percent to 15 percent of total asset holdings. Since the tax rate τ is held constant across the two periods, this suggests that the correlation of shocks is a quantitatively important factor in determining the extent of international diversiﬁcation.
Alternatively, if trade in foreign stocks is assumed to be costless (τ = 0) then ﬁgure 7 conﬁrms the results of proposition 1. In this case, the equilibrium share of foreign assets does not depend on the correlation of the shocks and thus it is the same in both periods. The share of foreign assets is approximately 20%, which is the value obtained by plugging the parameters of the model into equation 25.
Thus our ﬁrst conclusion from ﬁgure 7 and table 11 is that the model with restricted stock trade can be used to relate the observed increase in diversiﬁcation to the change in the correlation of shocks, while the model with unrestricted stock trade has nothing to say about the trend towards ﬁnancial globalization.
Figure 7 also shows that for certain tax rates the model has two equilibria corresponding to two diﬀerent levels of diversiﬁcation. We conjecture that this feature is due to a diversiﬁcation externality. The speciﬁcation of the ﬁrms’ objective implies that the value to households of diversifying their asset holdings depends on the aggregate level of diversiﬁcation, since when aggregate diversiﬁcation is higher, ﬁrms place a higher weight on the preferences of foreign shareholders. If this eﬀect is suﬃciently strong, it is possible to have a low diversiﬁcation equilibrium in which agents do not diversify because foreign ﬁrms do not consider them when deciding dividends, and a high diversiﬁcation equilibrium in which agents do diversify because foreign ﬁrms now pay suﬃcient attention to their preferences when making investment, employment and dividend decisions. The picture suggests that for the calibrations corresponding to both sub-periods there is a (small) range of taxes for which this phenomenon arises. To verify the conjecture regarding the source of multiplicity, we consider an alternative speciﬁcation in which we eliminate the diversiﬁcation externality by assuming that domestic ﬁrms care only about domestic consumers (regardless of the level of diversiﬁcation).
In this case, we ﬁnd only one equilibrium for each level of the tax (see ﬁgure 8). Notice also that when ﬁrms only care about domestic consumers the value of international diversiﬁcation is reduced, and for any given tax rate less diversiﬁcation is observed in equilibrium. Naturally, the two varieties of the model coincide when there is perfect home bias (λ = 1) and when there is perfect risk sharing (λ is given by eq. 25).
B. Shock correlation and the international business cycle In table 12 we report selected empirical business cycle statistics along with the predictions of the calibrated model economies with restricted and unrestricted stock trade. Statistics for the models are averages across 200 simulations, each of which is 58 periods long. The equilibrium levels of diversiﬁcation for each period are those reported in table 11.
Table 13 reports the changes in the empirical and model simulation statistics across the two time periods (for example, the output correlation for period two minus the output correlation for period one). The last two lines of this table report results from two additional experiments that we conduct in the restricted stock trade economy. In the ﬁrst experiment (labeled constant diversiﬁcation) we change the correlation of the real shocks (as in the the other models) but we do not let agents reoptimize their portfolios. In the second experiment (labeled constant shock correlation) we keep the correlation of the shocks ﬁxed (at the value ρ estimated over the whole sample) but we change the tax on foreign dividends across the two periods to obtain the same increase in international diversiﬁcation as in the benchmark model.
The model with restricted stock trade predicts international correlations that are quite close to those observed empirically. This is the case in both sub-periods, and applies to all variables (though the model slightly over-predicts the consumption correlation). On the negative side, in the All numbers in the table are diﬀerences between the statistic in the ﬁrst period (72.1-86.2) and the statistic in the second period (86.3-00.4). For example, the data number for the output correlation is 0.26 − 0.76 = −0.50.
data output is more strongly correlated across countries than consumption, while the model predicts the reverse. Moreover, the real exchange rate is too smooth, and there is too little intertemporal trade. These are common shortcomings in this class of models, and have been noted by many authors beginning with Backus, Kehoe and Kydland 1994.
The model with unrestricted stock trade generally underpredicts international correlations in the ﬁrst period. This is due to the fact that the market structure enables large capital ﬂows from the less to the more productive country. These ﬂows tend to lower the international correlation of investment and thus of employment and output. The real exchange rate volatility is even lower than in the restricted stock trade economy. On the positive side the model does predict that output should be more strongly correlated across countries than consumption, as is the case empirically.26 The diﬀerence between the two economies that we want to emphasize is how the predicted business cycle statistics change when the correlation of the shocks is reduced (see table 13). The restricted stock trade economy predicts an increase in intertemporal trade and a reduction in exchange rate volatility, as we observe in the data. In addition, the predicted declines in international correlations are very similar in magnitude to those observed empirically. In the unrestricted stock trade economy, by contrast, the size of the decline in co-movement predicted by the model is too small. Another failing of the unrestricted stock trade economy is that it counterfactually predicts an increase in the volatility of the real exchange rate.
To better understand why the model with restricted trade is better able to account for the observed decline in international business cycle correlations, consider the experiments reported in the last two lines of table 13. When diversiﬁcation is held constant and the shock correlation is reduced, the model-implied correlations fall but not by as much as in the data.27 The same thing happens This feature of models with complete risk sharing has been noted by Arvanitis and Mikkola (1996). It does not survive for higher values of the elasticity of substitution between domestic and foreign goods.
One could imagine that at the same time that the shock correlation falls, the tax rate on foreign dividends rises by an amount such that the optimal level of diversiﬁcation remains unchanged.
when the correlation of the shocks is held constant and diversiﬁcation is increased. Thus these experiments indicate that both less correlated shocks and increased diversiﬁcation are required to match the magnitude of observed declines in business cycle correlations. The model with restricted stock trade provides a simple mechanism through which less correlated shocks endogenously induce an increase in international diversiﬁcation.
Why does increasing portfolio diversiﬁcation reduce international co-movement in investment and employment? For simplicity, consider a situation of no international diversiﬁcation and imagine that domestic productivity rises while foreign productivity is constant. Domestic ﬁrms would like to reduce dividends to increase investment, but they recognize that the lower are dividends, the lower will be the income and consumption of domestic shareholders. This eﬀectively limits the size of the domestic investment boom. If agents are diversiﬁed, however, domestic (and foreign) owners of the domestic ﬁrm receive dividend income from abroad. Thus each additional dollar of domestic investment has a smaller negative eﬀect on domestic income and consumption than in the no diversiﬁcation economy, and the increase in domestic investment is consequently larger.
In addition, with positive diversiﬁcation the value of foreign agents’ holdings of domestic stocks increases following a positive domestic shock. Thus the wealth of foreign agents increases, which tends to reduce labor supply and consequently investment abroad.28
8. Conclusion Financial markets are becoming increasingly integrated internationally. A trend towards portfolio diversiﬁcation has left asset holders less exposed to country-speciﬁc risk, and the ﬂow of capital to its most productive location is increasingly unhindered by restrictions on international Note that this diversiﬁcation eﬀect is also apparent in table 12. In particular, the second and third lines of the table indicate that increasing diversiﬁcation in the restricted stock trade economy from the equilibrium level for period 1 to the level that supports perfect risk sharing (the level deﬁned by eq. 25) implies large declines in business cycle correlations borrowing and lending.
This paper explores the implications of the ongoing growth in international asset trade for the real economy. We ﬁnd that empirically the trend towards ﬁnancial globalization has been accompanied by a trend towards real regionalization. In particular, while output, employment and investment in the United States were strongly correlated with their foreign counterparts in the immediate post Bretton-Woods period, these correlations have since fallen dramatically.
We then develop a model in which stocks are traded internationally subject to certain frictions which limit risk-sharing. Within this model we are able to simultaneously account for both the trend towards ﬁnancial globalization and the trend towards real regionalization. When stocks may be traded freely, we cannot account for either trend.
Our conclusions are threefold. First, there is evidence of increasing country-speciﬁc risk, which is consistent with observed growth in international asset trade. Second, in models which quantitatively capture this growth in asset trade, ﬁnancial integration has large implications for the real side of the international business cycle. Third, observed changes in the international business cycle are diﬃcult to account for when the extent to which countries are linked via international ﬁnancial markets is assumed constant, but are readily reconciled in a model in which international ﬁnancial integration increases endogenously in response to increased country-speciﬁc risk.
References  Arvanitis, A.V. and A. Mikkola, 1996, Asset market structure and international trade dynamics, AEA Papers and Proceedings 86, 67-70.
 Backus, D.K. and P.J. Kehoe, 1992, International evidence on the historical properties of business cycles, American Economic Review 82, 864-888.
 Backus, D.K., P.J. Kehoe, and F.E. Kydland, 1992, International real business cycles, Journal of Political Economy 101, 745-775.
 Backus, D.K., P.J. Kehoe, and F.E. Kydland, 1994, Dynamics of the trade balance and the terms of trade: the J-curve?, American Economic Review 84, 84-103.
 Baxter, M. and M.J. Crucini, 1995, Business cycles and the asset structure of foreign trade, International Economic Review 36, 821-854.
 Baxter, M. and U. Jermann, 1997, The international diversiﬁcation puzzle is worse than you think, American Economic Review 87, 170-180.
 Cantor, R. and N.C. Mark, 1988, The international transmission of real business cycles, International Economic Review 29(3), 493-507.
 Cole, H.L. and M. Obstfeld, 1991, Commodity trade and international risk sharing, Journal of Monetary Economics 28, 3-24.
 Cooley, T. and E. Prescott, 1995, Economic growth and business cycles, in T. Cooley ed., Frontiers of Business Cycle Research, Princeton University Press, Princeton, 1-38.
 Diamond, P.A., 1967, The role of a stock market in a general equilibrium model with technological constraints, American Economic Review 57, 759-776.
 Drèze, J., 1985, (Uncertainty and) the ﬁrm in general equilibrium theory, in J. Drèze ed., Essays on economic decisions under uncertainty, Wiley.
 Engle, R., 2000, Dynamic conditional correlation - a simple class of multivariate Garch models, Mimeo, Stern School of Business, NYU.
 Forbes, K. and R. Rigobon, 2002, No contagion, only interdependence: Measuring stock market co-movements, Journal of Finance, 57, 2223-2261.
 Grossman, S.J. and O. Hart, 1979, A theory of competitive equilibrium in stock market economies, Econometrica 47, 293-330.
 Heathcote, J. and F. Perri, 2002, Financial autarky and international business cycles, Journal of Monetary Economics 49 (3), 601-627.
 Kehoe, P. and F. Perri, 2002, International business cycles with endogenous incomplete markets, Econometrica 70 (3), 907-928.
 Kollmann, R., 2001, Macroeconomic eﬀects of nominal exchange rate regimes: new insights into the roles of price dynamics, Mimeo, University of Bonn.