RDP 2007-11: Global Imbalances and the Global Saving Glut – A Panel Data Assessment 6. Implications for Regional Current Account Balances

In this section we examine the ability of the model to explain various elements of the global saving glut hypothesis, including the widening of the US current account deficit and the increased capital outflows from east Asia, China and the oil-producing nations.

The model estimates that the US current account deficit should have narrowed slightly from 1997 onwards, rather than the widening that actually occurred (Figure 4).^[21] This result was also obtained by Gruber and Kamin (2007), and while the model of Chinn and Ito (2007) predicts an increase in the current account deficit over this period, the estimated level is around 2 per cent of GDP, instead of the actual deficit of 5 per cent. Nevertheless, it is still instructive to look at the contributions of the various factors, which are shown in Figure 4.^[22] The large role attributed to the quality of institutions is immediately obvious; ignoring the interaction terms, as institutions in the US were above the GDP-weighted average for the entire sample, they contributed more than 3 percentage points to the current account deficit, although, interestingly, their contribution has decreased since the late 1990s. More generally, this supports Bernanke's (2005) argument that, in order for developing countries to move to having net capital inflows, they should ‘… improve their investment climates by continuing to increase macroeconomic stability, strengthen property rights, reduce corruption, and remove barriers to the free flow of financial capital’.

The estimated contribution of the widening of the fiscal deficit does not appear to have been the primary reason for the widening current account, despite being considerably larger than that in Gruber and Kamin (2007). Similarly, the influence of demographic factors is small, although this may in part reflect the inclusion of fixed effects.^[23] The impact of the increase in the growth forecasts since 2000 is reduced by the interaction with the institutions variable (included in ‘other’ in Figure 4).

It is possible that the model does not capture the widening of the US current account deficit as it does not adequately ‘channel’ the impact of financial crises to the US.^[24] However, even if all of the increases in current account balances due to countries undergoing financial crises are allocated to the US it would not be of a sufficient magnitude to explain its current account deficit (Figure 5). The estimated outflows from crisis-affected countries in recent years are particularly small as the last crises in the sample were in 2001, which the model estimates to be too distant in the past to be of importance. If the most recent financial crises are actually influencing the destination of international capital flows in a sizeable way their impact must be considerably greater in magnitude and more prolonged than that of previous crises.

The model captures the broad trends in the current account balances for some of the other major regions (Figure 6), with the performance particularly good for the major oil-exporting nations. The notable exception is Japan, which the model predicts should have shifted from running current account surpluses to deficits, whereas actually the surplus increased. This prediction partially reflects the relatively rapid ageing of the Japanese population and the widening of the fiscal deficit.

Interestingly, the model considerably overpredicts east Asia's current account balance prior to the Asian crisis, which might suggest that there were excessive capital inflows during this period.

The performance of the model appears to deteriorate from around 2003, with divergences of around 2 percentage points for both east Asia and the major oil exporters. It also fails to explain the sizeable widening of China's current account surplus in 2005.

Footnotes

To construct the estimate of the fixed effect, we take the parameters from our estimated equation (which is the difference of Equation (1)) and apply them to Equation (1). The fixed effect for each country is then the average residual. [21]

This figure was motivated by Figure 7 in Gruber and Kamin (2007). [22]

In Figure 4, the demographic factors are in the ‘other’ aggregate, which also includes the interaction terms and the lags of the current account, but excludes the fixed effect. [23]

Recall that, in the model, the impact of financial crises are ‘channelled’ to the non-crisis countries as the dummy variables enter as deviations from the GDP-weighted average. [24]