«ABSTRACT Over the period 1972-1986, the correlations of GDP, employment and investment between the United States and an aggregate of Europe, Canada ...»
Figures 2 and 3 report the time paths from 1972 to 2000 for U.S. assets and liabilities relative to this group of countries. In each case we report the FDI position separately, as well as FDI plus equity holdings jointly. These ﬁgures illustrate dramatic growth in diversiﬁcation over the sample period. In particular, U.S. holdings of foreign stocks have grown strongly since the mid 19800 s, while the stocks of FDI and foreign owned equity in the U.S. have risen steadily over the entire period.
Overall, U.S. liabilities have grown faster than assets, reﬂecting the series of current account deﬁcits over the past two decades. Moreover inward and outward acquisitions of capital have taken diﬀerent forms: almost all of the increase in U.S.-owned capital abroad reﬂects an increase in the stock of equity portfolio investment, while most of the increase in foreign ownership of the U.S. capital stock reﬂects an increase in the direct investment position.
The observed growth in diversiﬁcation appears to be robust to a wider deﬁnition of the rest of the world, to broader classes of assets, and to alternative valuation methods. First, in addition to our benchmark measure described above, we examine assets and liabilities for the U.S. versus the entire rest of the world. Second, we use stock market capitalization instead of the current-cost replacement value for tangible assets to value capital stocks. These results are summarized in tables 7 and 8.8 Alternative methodologies generate diﬀerences in the measured level of international diversiﬁcation, but the ﬁnding that diversiﬁcation was much higher in the 19900 s than in the 19700 s clearly remains.
Note that Europe, Canada and Japan jointly account for almost all foreign holdings of U.S. assets Note that some assets are privately held. This is one reason why the level of diversiﬁcation appears higher when stock market capitalization rather than the BEA estimate for private non-residential assets is used to measure the capital stock.
and for the lions share of U.S. asset holdings abroad, though other countries are attracting an increasing share of U.S. equity portfolio investment. Growth in diversiﬁcation generally appears smaller when stock market capitalization rather than capital stock replacement cost estimates are used as a denominator (which is not surprisingly in light of surging stock markets), but even in this case we ﬁnd strong growth in the stocks of U.S. equity portfolio investment abroad and foreign direct investment in the U.S. Comparing the U.S. with the Europe / Canada / Japan aggregate, for example, U.S. holdings of foreign securities averaged 1.1 percent of total non-U.S. developed economies market capitalization over the ﬁrst half of the sample, while the corresponding ﬁgure for the second half of the sample was 5.5 percent.
Table 7. US foreign assets and liabilities as % of US capital stock
Table 9 provides yet another measure of global asset trade based on the fact that the current account is a measure of the change in a country’s net foreign asset position. Thus larger (positive or negative) values for the current account reﬂect more international asset trade. The table indicates that larger current account and net exports positions (as a fraction of GDP) have been observed in the second sub-period, indicating a signiﬁcant increase in the use of international borrowing and lending. The last three columns of table 9 report other interesting phenomena related to changes in international ﬁnancial markets. The third column shows that the volatility of the U.S. real exchange rate has signiﬁcantly declined over time; in the model we will present below less real exchange rate volatility is a natural consequence of increased ﬁnancial integration. The fourth and ﬁfth columns report correlations, at quarterly frequency, between U.S. stock market price indexes and returns (PU S, RU S ) and comparable indexes and returns for the rest of the world (PRW, RRW ).9 Real exchange rates and stock prices are logged and HP ﬁltered. Stock returns are computed as log ﬁrst diﬀerences of stock price indexes. See the data appendix for details.
Table 9 suggests that the decline in the correlation of business cycles has been associated with a decline in the correlation of international stock performance. This suggests that both greater international diversiﬁcation and less correlated investment may be driven by a weaker international correlation in the return to capital.10 Figure 4 ﬁnally provides more evidence on the link between ﬁnancial globalization and real regionalization by combining the evidence above on diversiﬁcation with the evidence presented earlier on the change in the international business cycle. The top line (right scale) is the United States foreign asset position (FDI plus equity) from ﬁgure 2. The bottom line (left scale) reports the rolling window correlation for investment from ﬁgure 1. The picture provides striking evidence that the two phenomena emerged at about the same time.
3. A simple model of ﬁnancial diversiﬁcation In this section we consider a simple model that is helpful for understanding potential interaction between the international correlation of shocks, the degree of international asset trade, and cross-country correlations in macro aggregates.
Note that other authors (for example Longin and Solnik 1995) have documented an increase in the international correlation of stock returns. However these authors have typically focussed on return correlations at frequencies higher than business cycles, and on the late 1990’s when correlation increased even at business cycle frequencies.
Consider an atemporal two-country exchange economy. A Lucas tree in each country produces some non-storable fruit. The quantity of fruit produced depends on the realization of the state of nature s. The endowment (of fruit) of the domestic tree is denoted X(s), while the foreign endowment is X ∗ (s). Prior to any trade, the representative domestic agent owns the domestic tree, while the foreign agent owns the foreign tree. Agents ﬁrst trade shares in their trees. Then the state of nature is revealed, contracts are honored, and agents consume any fruit to which they have claims. We now describe the representative domestic agent’s problem (the foreign agent’s problem is analogous).
At the start of the period, the domestic household buys a fraction λf of the foreign tree, subject to the constraint
where P is the price of the domestic tree, P ∗ the price of the foreign tree, and 1 − λ the fraction of the domestic tree sold.
An important assumption is that endowment income from abroad is subject to a proportional tax or shipping cost τ. Thus given a choice for λ, consumption in state s is given by
subject to eq. 2 and the short-selling constraint λ ≤ 1.11 We do not allow agents to go short in foreign shares, since for χ 0 this would allow agents to increase expected consumption. Agents may go short in domestic shares but opt not to in equilibrium.
To the extent that the domestic and foreign endowments are imperfectly correlated in some states, there is an incentive for households to diversify internationally, which amounts to choosing λ 1. However, the tax τ provides an incentive to bias portfolios in favor of domestic stocks.
The representative domestic household’s ﬁrst order condition is
and that endowments are drawn from a bivariate normal distribution, then it is possible to solve for the equilibrium λ analytically. In particular, assume that E[X] = E[X ∗ ] = µ, var[X] = var[X ∗ ] =
Note that the λ depends only on the tax τ, the correlation of shocks ρ, and the ratio µ/(Aσ 2 ).
Here we assume that tax revenues are wasted but results are similar if we assume that revenues are rebated lump sum to consumers.
We can also calculate the correlation of consumption across countries
Note that if χ = 0, then λ = 0.5 and corr(c, c∗ ) = 1 implying complete diversiﬁcation and perfect risk sharing regardless of the shock correlation.
We can also deﬁne net exports for the domestic country as domestic fruit produced minus
the sum of fruit consumed domestically plus imports of foreign fruit that is taxed:
Figures 5 and 6 illustrate how international diversiﬁcation (given by 1 − λ) and the crosscountry consumption correlation change as we reduce ρ, the shock correlation, from a high value (0.75) to a lower value (0.25). In this example, A = 1, µ = 2 and σ = 0.04. At a consumption level µ, these values translate to a coeﬃcient of relative risk aversion (corresponding to Aµ) of 2, and a percentage standard deviation of output (corresponding to 100 × σ/µ) of 2.
Figure 5 plots the equilibrium level of diversiﬁcation as a function of the tax τ. The ﬁgure shows that when τ = 0 and there is no cost to diversiﬁcation, agents always choose to be fully diversiﬁed (λ = 0.5). When the costs are suﬃciently high the constraint that agents cannot go short in foreign stocks is always binding and there is complete home bias (λ = 1). For intermediate values for the cost parameter, however, a reduction in the correlation of the shocks increases the marginal beneﬁt of diversiﬁcation and thus, holding constant the marginal cost τ, leads to an increase in international diversiﬁcation.
Figure 6 shows that the eﬀect of a reduction in the shock correlation on the consumption correlation is ambiguous. When costs are high there is little or no increase in diversiﬁcation and the consumption correlation closely follows the shock correlation. Thus reducing the shock correlation reduces the consumption correlation. When costs are low on the other hand, a decline in the shock correlation leads to a large increase in international diversiﬁcation which more than oﬀsets the direct eﬀect on the consumption correlation of less correlated endowments. Thus the consumption correlation actually increases. What is unambiguous, however, is that the decline in the consumption correlation is always smaller than the decline in the shock correlation.
Finally, a reduction in the shock correlation increases the volatility of net exports, both directly by reducing ρ, and indirectly by reducing λ (see eq. 5).
Thus this simple model is qualitatively consistent with some key features of the evidence presented in the data section of the paper. In particular, for intermediate values for the cost parameter, a decrease in the correlation of shocks is associated with (i) an increase in diversiﬁcation, (ii) a decline in the correlation of consumption that is smaller than the decline in the correlation of the shocks, and (iii) an increase in the volatility of net exports.
While the model economy described above is analytically and conceptually tractable, it abstracts from production and thus cannot address the observed changes in employment and investment correlations. We therefore proceed to consider a multi-period world economy in which the asset market structure is analogous to the one described above except that dividends are now determined endogenously by ﬁrms making capital investment and employment decisions. This model can be calibrated and simulated to assess whether the mechanism through which ﬁnancial globalization impacts the real business cycle is quantitatively as well as qualitatively consistent with the empirical evidence.
4. A model with investment The modelling framework is the one developed by Backus, Kehoe and Kydland, 1994. There are two countries, each of which is populated by the same measure of identical, inﬁnitely lived households. Firms in each country use country-speciﬁc capital and labor to produce an intermediate good.
The intermediate good produced in the domestic country is labeled a, while the good produced in the foreign country is labeled b. These are the only traded goods in the world economy. Within each country goods a and b are combined to produce country-speciﬁc ﬁnal consumption and investment goods. However, the ﬁnal goods production technologies are asymmetric across countries, in that they are biased towards using a larger fraction of the locally produced intermediate good. The only source of uncertainty in the model takes the form of country-speciﬁc productivity shocks to intermediate goods ﬁrms.
The only innovation in the model described here relative to previous work concerns the assumed asset market structure. Recall that the goal of the paper is to understand the potential role of asset market integration in accounting for observed changes to the real side of the international business cycle. The fact that it is diﬃcult to discuss asset market development in the context of a model with complete markets suggests that an incomplete markets framework is appropriate.
We therefore assume that the assets that are traded internationally are shares in the domestic and foreign representative intermediate-goods-producing ﬁrms. These ﬁrms make investment and employment decisions, and distribute any non-reinvested earnings to shareholders. Dividend income from abroad is potentially taxed. This is a natural framework to address growth in international diversiﬁcation, since purchases of foreign stocks in the model can be mapped to data on foreign direct investment and foreign equity portfolio investment. In particular, in the calibration section, we will set the foreign dividend tax rate so that the model reproduces the observed level of international portfolio diversiﬁcation for the U.S. in the ﬁrst half of our sample.