RDP 2009-09: Volatility in International Capital Movements 2. Variability of the Capital Account

There are several ways of measuring the variability of the capital account and its components. We take our lead from Classens et al (1995) who employ a number of simple statistical techniques to test whether some types of capital flows can be reliably identified as being inherently more volatile than others. Contrary to the commonly held view, their findings suggest that the composition of capital flows is endogenously determined by domestic factors and that little can be learned from looking at individual flows in isolation when the source of instability tends to be aggregate shocks to the determinants of the capital account. Our work is somewhat different in the way it measures volatility and offers some explanations for why industrialised and emerging economies have different experiences when it comes to capital flows.

Throughout the paper we use standard balance of payments data sourced from the International Monetary Fund's International Financial Statistics (IFS), on a quarterly basis. Except where explicitly stated, the data are then converted into local currency terms and scaled by nominal GDP (for details see Appendix B) for analysis. The balance of payments identity imposes the constraint that the current account balance (CAB) is equal to the capital account balance (KAB), and the two concepts can be used more or less interchangeably. The capital account refers to what has become more conventionally termed the financial account and consists of foreign direct investment (FDI), portfolio equity (PFE), portfolio debt (PFD), bank and money market (BMM)[4], and official reserve (RES) flows. We use this disaggregation of the data for the remainder of the paper. Our sample of six industrialised economies comprises the United States, Japan, Germany, the United Kingdom, Australia and Sweden. The six emerging economies for which data are readily available are South Korea, Thailand, the Philippines, Brazil, Mexico and Argentina. In the interest of brevity we present most results in terms of the simple unweighted average for industrialised and emerging economies. Where interesting results are apparent on a by-country basis these are discussed in their own right. The sample period spans the first quarter of 1980 to the fourth quarter of 2005, inclusively.

2.1 Volatility of Capital Flows

We measure the volatility of flows as the standard deviation of the ratio of the flows to GDP. Scaling by GDP is important because we are most interested in large swings in the flows from each country's perspective. To capture how volatility among the flows has evolved over time we calculate the standard deviations of quarterly data over a one-year rolling window for each country. For expositional purposes we average the results across emerging economies and across industrialised economies as depicted in Figure 2.

Figure 2: Volatility of Capital Flows

The overall volatility of the capital account in emerging economies has generally been around double that experienced by industrialised economies. Furthermore, emerging economies have more discrete episodes when volatility rises markedly, in part reflecting the fact that they have more frequent balance of payments crises.

These outcomes are in line with what we know about emerging economies. More insights are to be found in the developments in the volatility of the different types of flows, and the contrast between the experiences of emerging and industrialised economies.

One of the most noteworthy findings is that while there has been little change in the average degree of volatility observed in total flows, the pattern of volatility has evolved very differently among the various components of the capital account. In industrialised economies, the component flows are generally more volatile than the total and have virtually all exhibited a trend rise in volatility. Given no such trend in the volatility of the overall capital account, there must be a degree of negative correlation between various components. Indeed, there is a sizeable negative correlation between portfolio debt and bank and money market flows. This suggests a degree of substitutability between different forms of capital that allows industrialised economies to accommodate variability in the mix of different types of capital flows without significant adverse consequences for overall flows.[5]

Another interesting feature of the data for industrialised economies is the sharp movement in the volatility of foreign direct investment and portfolio equity investment flows earlier this decade. This principally reflects the increase in mergers and acquisitions among European countries, which was financed through stock swaps. Under these deals, direct investment is financed by exchanging stocks between companies. This results in offsetting portfolio equity and foreign direct investment flows, with little if any effect on the overall capital account.[6]

Among emerging economies, the trend rise in the standard deviation of the constituent flows is less pronounced, and the constituent flows within the capital account are often less volatile than the total. Importantly, this suggests not only that the substitutability between the flows evident in industrialised economies may not be present in emerging economies but also that at least some flows tend to move in the same direction. The volatility of bank and money market flows is relatively high for emerging economies, in which there is typically also a greater reliance on bank-intermediated finance and local currency debt markets remain relatively underdeveloped.

The behaviour of reserve flows is quite different for emerging and industrialised economies. Not surprisingly, reserves are considerably more volatile among the emerging economies where monetary authorities are typically more active in foreign exchange markets. With the exceptions of the Plaza Accord and the European Exchange Rate Mechanism crisis, the less activist role of central banks among industrialised economies, with their typically floating exchange rates, results in reserve flows that are, on average, less than half as volatile as those of emerging economies. Japan is an obvious exception to this, with a jump in volatility in 2003–2004.

2.2 Persistence of Capital Flows

A complementary measure of the stability of capital flows is their degree of persistence over time. ‘Cold’ flows that are perceived to be relatively stable should also display evidence of strong positive correlation with their own past values. To assess persistence we calculate autocorrelation coefficients for each flow in each country over the sample. The data are quarterly ratios of flows to GDP and the correlations are calculated for 16 lags (Figures 3 and 4).

Figure 3: Autocorrelation Coefficients – Industrialised 
Figure 3: Autocorrelation Coefficients – Industrialised 
Figure 4: Autocorrelation Coefficients – Emerging 
Figure 4: Autocorrelation Coefficients – Emerging 

Total capital flows are found to exhibit a high degree of persistence in most industrialised economies. The autocorrelation coefficients are typically large, positive, and gradually decay as the lags increase. This suggests that there is a high degree of persistence in the overall capital account for at least one to two years.[7] The capital account in emerging economies is typically less autocorrelated and only two of the emerging economies examined have autocorrelation coefficients of greater than two-thirds for four or more lags, compared to five out of the six industrialised economies.

The autocorrelation coefficients for the components of the capital account suggest that these flows generally display little if any persistence for industrialised economies. The coefficients are small and change sign frequently. There are, however, a number of notable exceptions. For the United States we find that portfolio debt flows are highly persistent. This is not surprising given that the United States is home to the largest debt markets and the US dollar is the world's main reserve currency. Japanese foreign direct investment flows are also shown to be highly persistent. This may reflect the structural ‘hollowing out’ of Japanese manufacturing as Japanese companies undertook direct investment to set up plants in other Asian countries where labour costs were lower.

Other than these exceptions, there is no evidence among industrialised economies to support the view that some types of capital flow are inherently more persistent than others. Foreign direct investment is typically not as persistent as might have been expected and can hardly be distinguished from the bank and money market or portfolio flows, which are often thought of as being relatively temporary.

The evidence is somewhat different for emerging economies. Foreign direct investment flows are relatively persistent for a number of these economies. This can probably be attributed to emerging economies being natural destinations for such investments, with inflows typically dominating this category.[8] There are also several other examples of persistence for some components of the capital account, but there are no consistent patterns across emerging economies.

2.3 Composition of Cross-border Finance

Policy-makers may be interested in understanding how capital account volatility is related to the composition of cross-border flows. More specifically, as a flow assumes a more prominent share in the capital account, a systematic relationship between its volatility and that of the total may be discernible. If some flows are inherently more or less stable, then as these flows become more important as a source of cross-border financing we should be able to observe whether or not this exerts an influence on the overall capital account.

To test whether such statistical regularities are observable we disaggregate the quarterly country data by type of flow, as defined earlier. We then calculate the importance of each type of capital flow in the overall capital account. The importance of each type of flow in a particular economy's capital account is calculated as the ratio of the absolute value of the net flow to the sum of the absolute value of all flows in the capital account.[9] This is done over five-year blocks from 1981 to 2005. The changing importance of each flow for every country is measured as the difference in this ratio from one five-year block to the next. A positive number indicates that a flow has become more important in the overall capital account of the country in question.

To measure the variability of total net capital flows, we first scale the quarterly capital account balance for each country by GDP and then calculate the standard deviation of the data over the same five-year blocks (Table 1). Our gauge of how the variability of the capital account has changed is then given by the difference in the standard deviations from one five-year block to the next.

Table 1: The Volatility of Capital Account Flows
Standard deviation in the quarterly ratio of the capital account to GDP
1981–1985 1986–1990 1991–1995 1996–2000 2001–2005
US 1.3 0.9 0.9 1.1 0.8
Japan 1.5 1.1 0.5 0.6 0.7
Germany 1.7 1.0 0.6 0.9 2.1
UK 1.4 2.0 1.0 1.2 1.0
Australia 1.0 1.1 1.3 1.2 1.7
Sweden 2.0 1.4 2.1 0.8 2.0
Korea 2.7 3.6 1.5 5.7 1.5
Thailand 2.8 4.8 3.2 8.6 4.4
Philippines 3.7 3.5 2.7 3.6 3.7
Brazil 3.6 1.9 1.7 1.2 2.5
Mexico 3.8 2.4 2.3 1.3 1.0
Argentina 1.7 3.0 1.8 1.3 4.2

Source: authors' calculations

Figure 5 plots the relationship between changes in the importance of each of the flows and capital account volatility for industrialised and emerging economies. The changing importance of the flows lies along the vertical axis and the changing volatility in the capital account on the horizontal axis.

Figure 5: Composition and Volatility of Capital Flows

A positive relationship in the scatter plots would be consistent with more volatile flows becoming more important, thereby raising the average volatility of the capital account. Conversely, a negative relationship would be consistent with less volatile flows becoming more important.

Few statistically significant relationships emerge when regression lines are fitted to the data (not shown). For industrialised economies there is a statistically significant positive relationship between the increasing importance of reserve flows and increasing capital account volatility. For emerging economies the increasing importance of portfolio equity flows tend to be associated with decreasing capital account volatility, contrary to the conventional wisdom.[10]

In addition, a number of specific cases can be cited that are clearly at odds with the conventional wisdom. For example, in Australia from the 1980s to the 1990s there was a rise in the importance of foreign direct investment and a decline in the importance of bank and money market flows. While preconceived ideas about the former being cold and the latter being hot flows would imply lower variability in overall capital inflows, the opposite occurred (refer also to Table 1). Similarly, at the same time, the United States experienced a decline in the importance of foreign direct investment at the expense of bank and money market flows, but, on average, volatility of overall flows declined.

What is evident from Figure 5 is that for industrialised economies the observations are clustered in an ellipse around the vertical axis. This implies that the composition of finance changed noticeably over time, but the capital account remained relatively stable. The observations for emerging economies are more widely dispersed and spread out along the horizontal axis, indicating that overall capital account volatility changed more noticeably over time. Overall, these results suggest that volatility of the capital account is not systematically related to its composition.


In the balance of payments these flows appear under the category of ‘other’. As bank loans and money market transactions are the main components of this category, we refer to these flows as ‘bank and money market’ flows to lend them a more meaningful label. [4]

Levchenko and Mauro (2007) also find that while the overall capital account of emerging economies is more volatile than that of industrialised economies, portfolio flows of industrialised economies are two to five times more volatile than those of emerging economies. [5]

The merger of Vodafone Plc in the United Kingdom with Mannesmann AG in Germany is a prominent example of this phenomenon. For more details see Becker (2003). [6]

Further evidence of this persistence is presented when we investigate the forecastability of the flows in Appendix A. For more persistent flows, the expected future value will be closer to the current value of that flow. There is some evidence to suggest that for industrialised economies the current account is endogenous to domestic economic fundamentals such as growth, saving and investment, and that current account deficits do not by themselves precipitate sudden stops that cause adjustment in other variables (Debelle and Galati 2005). [7]

In contrast, for industrialised economies foreign direct investment typically flows in both directions. [8]

We take the absolute value of the quarterly flows and the capital account to avoid the problems of interpretation associated with either the numerator or denominator switching sign. [9]

Comparing the composition of the capital account to its overall volatility in levels, rather than changes, also fails to reveal any compelling patterns. Although we find that a high share of portfolio debt flows tends to be associated with relatively low capital account volatility for industrialised economies, the reverse is true for emerging economies. We also find that high capital account volatility is associated with a high share of bank and money market flows for emerging economies, but not for industrialised countries. Note that this analysis does not account for possible unobserved country fixed effects. [10]