For a Few Dollars More: Reserves and Growth in Times of Crisis

That’s the title of a paper coauthored by Matthieu Bussière (Banque de France), Gong Cheng(Sciences Po), Noëmie Lisack (European University Institute) and myself, in which we examine two key questions:

Fi rst, has the accumulation of reserves e ffectively protected countries during the 2008-09 financial crisis? And second, what explains the pattern of reserve accumulation observed during and after the crisis?

More specifically, the paper investigates the relation between international reserves and the existence of capital controls. We fi nd that the level of reserves matters: countries with high reserves relative to short-term debt suff ered less from the crisis, particularly if associated with a less open capital account. In the immediate aftermath of the crisis, countries that depleted foreign reserves during the crisis quickly rebuilt their stocks. This rapid rebuilding has, however, been followed by a deceleration in the pace of accumulation. The timing of this deceleration roughly coincides with the point when reserves reached their pre-crisis level and may be related to the fact that short-term debt accumulation has also decelerated in most countries over this period.

To place matters in context, recall that reserve accumulation has been proceeding apace over the past decade and a half, with the bulk of the accumulation undertaken by emerging market central banks. This is shown in Figure 1 from the paper.

Figure 1 from Bussière, Cheng, Chinn and Lisack.Reserve accumulation has taken different paths for advanced and emerging market/developing country groups. The paths have also diverged depending on the degree of financial openness. This is shown in Figure 2 from the paper.

Figure 2 from Bussière, Cheng, Chinn and Lisack.The capital openness variable is the inverted Chinn-Ito index, described here.

The accumulation of foreign exchange reserves must have some motivation; one motivation (among several possible) is that reserves are particularly helpful during times of financial stress, when liquid assets are needed (or their presence assuages investor fears of a currency collapse).

We find that growth is about 0.6 to 0.7 ppts higher than would otherwise occur if the reserves to short term debt ratio were to double (i.e., the regression coefficient on income growth deviation from baseline on log reserves to short term debt is about 0.6-0.7).

This coefficient is from a linear specification that omits any interaction effect from capital openness. We find that the lower the degree of capital openness, the greater the marginal impact of reserves/short term debt. This is shown in Figure 3.

Figure 3 from Bussière, Cheng, Chinn and Lisack.We have subjected our findings to a series of robustness tests (described at length in the paper, and documented in the appendices). One point that is notable is that dropping outliers and/or small countries typically increases the coefficient on log reserves/short term debt.

There is a question of why reserves matter. It could be that the reserve depletion itself (for defending a currency) mitigates economic distress; or perhaps the mere existence of large reserves deters speculation against a currency. We augment our regressions with a reserve depletion dummy variable and find the log reserves/short term debt coefficient retains its statistical significance, while the coefficient on reserve depletion is not statistically significant. This finding is suggestive of the deterrence channel.

We find that after the crisis, countries have sought to rebuild their stocks of foreign exchange reserves. Figure 5 from the paper illustrates this point.

Figure 5 from Bussière, Cheng, Chinn and Lisack.The paper concludes with some observations on the recent the recent stabilization in reserves-to-short-term-debt ratios.

Figure 8 from Bussière, Cheng, Chinn and Lisack.For several countries, the ratio has plateaued. China is one notable exception. (However, since the end of the graph’s sample, Chinese reserves (in levels) have stabilized, with a slow resumption of growth in 2013.)

The entire paper is here.

This piece is cross-posted from Econbrowser with permission.