RDP 2009-09: Volatility in International Capital Movements 3. Interactions Between Flows

The results in the preceding Sections suggest that the co-movement of different types of capital flows seems to be central to understanding the overall variability of the capital account.

To provide a more comprehensive view of the data and how the flows interact we estimate cross-correlation coefficients for all capital flows. Importantly, we distinguish between how different types of flows are correlated ‘within’ each country's capital account and with the flows of ‘other’ countries. The quarterly data are summed to annual totals (and expressed as a ratio to GDP) for this purpose, thereby shifting focus away from high-frequency changes in the flows.

3.1 Within Capital Account Correlations

Table 2 summarises within country cross-correlation results by showing the average of the correlation coefficients across industrialised economies and across emerging economies. It also shows, in brackets, the number of countries with negative correlation coefficients for a given pairing of capital account flows.

Correlations between various types of flows within each country's capital account are generally negative. For industrialised economies, around 75 per cent of the flows are negatively correlated within their own capital account. For emerging economies the degree of negative correlation is smaller at 47 per cent.

For industrialised economies, on average there is a sizeable negative correlation between foreign direct investment and portfolio equity flows. Bank and money market and portfolio debt flows also show a negative correlation on average for industrialised economies.

There is a strong positive correlation between portfolio debt flows and the capital account for both industrialised and emerging economies. This positive relationship is particularly strong for Australia, Japan and the United States. For emerging economies, the capital account is also always positively correlated with bank and money market flows, and typically more so than for industrialised economies. This strong link is consistent with a greater degree of bank dependence among emerging economies.

For emerging economies there is also a consistent pattern of negative correlation between reserves and private flows. In particular, bank and money market flows stand out in the country data as always being negatively correlated with official reserve flows. There are two possible explanations for this outcome. First, domestic monetary authorities may aim to offset the effects of bank and money market flows on the overall capital account and the ensuing consequences for the exchange rate. Indeed, intervention during the Asian financial crisis was squarely aimed at mitigating the sudden reversal of bank and money market flows. Second, it is also feasible that since the capital account, current account and the exchange rate are jointly determined, actions to maintain a given exchange rate through variations in reserves may cause disturbances in the other types of flows within the capital account. Such disturbances may manifest themselves in such a way that it is typically the bank and money markets component of the capital account which is most accommodating. The scope of this paper is too limited to adequately address this interesting aside in the research.

3.2 Cross-country Correlations

The correlation in capital flows between countries may also be useful in understanding volatility. Several interesting links across countries are evident and worth highlighting, although there is no overwhelmingly regular pattern among the flows (results not shown).

There is a negative correlation between total net flows for Japan and those of other industrialised economies that are net capital importers, such as Australia and the United States. There is less evidence of correlation between the capital accounts of industrialised and emerging economies. This is indicative of a relatively high degree of financial integration among industrialised economies, while emerging economies are less integrated into global financial markets.

The capital accounts tend to be positively correlated among the emerging economies. Given that these balances are also relatively volatile, this result is consistent with evidence that these countries are subject to the same balance of payments shocks. That is, they tend to experience crises at the same time, which also reflects a degree of contagion (Broner and Rigobon 2006).

Another interesting aspect of the correlations is that they reflect foreign exchange intervention that, to varying degrees, attempts to limit currency variations. A priori, if reserves are accumulated (a capital outflow) to stem an incipient appreciation of the exchange rate (from a capital inflow) and are invested in fixed-income assets, this would lead to a negative correlation between reserves in the intervening country and portfolio debt inflows for the recipient country, or countries. For both Japan and Korea, reserves are significantly negatively correlated with portfolio debt flows for the United States, the United Kingdom and Australia. Holding reserves in US dollars and investing the proceeds in Treasury securities is common to most central banks around the world.