We collect daily, currency prices (against the US$) from Datastream. Specifically we use the WM/Reuters time series.
Interest rate data
In order to calculate excess currency returns, we collected interest rates from a number of sources. For developed markets we used euro-currency deposit rates from Datastream. For emerging markets the task is far more challenging. These rates are compiled from a number of sources including Bridge Information Systems, DRI, and the International Monetary Fund.
The Behavior of Portfolio Flows
Table 1 provides general information about the data. Total transactions (buys plus sells) come to over $845 billion—$854 million per day—from over 3.2 million transactions during the sample period. The largest number of these cross-border transactions took place in Japan and the UK followed by Hong Kong and France. While there are 72 transactions on average per day per country, some countries, such as Zimbabwe and Morocco, average one or fewer transactions per day.
Overall, the transactions account for a net average daily inflow of $130 million into our 46 countries, $30 million of which went into emerging markets (predominantly Latin America and East Asia), and $100 million of which went into developed countries. The average trade size ranges between about $100,000 (Venezuela, Peru, and Turkey) to about $500,000 (Switzerland and the Netherlands). The standard deviation of trade size is very large for Brazil, for which we have a small number of very large transactions. But for most countries, the average trade size and standard deviation of average daily trade size are a few hundred thousand dollars. We did not exclude or censor any data in our analysis.
Factor analysis and the cross-correlation of flows
We begin by looking at the correlation matrix of the daily flows. A “heat map” of the correlations efficiently summarizes over 1,000 correlation coefficients16 and is shown in Figure 1. It is evident from the figure that the flow correlations are on average slightly positive. They are more positive within some regions, particularly Asia and somewhat positive in Latin America and Developed Countries. The data reject the hypothesis that the cross-correlations are zero.
Average correlation coefficients for the world and regions are shown at the bottom of Figure 1. Note how the flow correlations are larger in Asia, and how they have risen in Asia and elsewhere during the Asian crisis period. It is useful to compare Figure 1 with similar heat maps of dollar stock and currency return correlations. These are shown in Figures 2 and 3, respectively. The regional character of stock returns is evident in Figure 2, but this is less so for the currency returns in Figure 3.