VoxEU Column International Finance

For a few dollars more: Reserves and growth in times of crises

The financial crisis that swept the global economy at the end of 2008 provides a natural experiment to test the proposition that international reserves are useful during crises. This column presents cross-country evidence based on a panel of 112 emerging and developing countries. Countries with more reserves relative to short-term debt fared better.

In the decade preceding the 2008 global financial crisis (GFC), emerging market economies accumulated large stocks of international reserves (see Figure 1). The unprecedented pace of reserve accumulation was at least partly a response to the lessons drawn from previous financial crises, which predominantly affected emerging markets. Most research on emerging-market crises suggests that countries with an insufficient level of reserves, measured against appropriately chosen benchmarks, suffered more from crises in the 1990s.1

Figure 1. World international reserves

Source: IMF COFER database and authors’ calculations.

A natural question arising from this observation is to what extent the accumulation of international reserves has protected countries from the negative shock of the latest crisis.

Have countries with more reserves fared better, in terms of output growth performance, than countries with fewer?

In addition, are there other policy tools that can strengthen or dampen the effects of reserves on growth performance?

One of these tools is capital account restrictions: emerging market economies tend to hold more reserves and to have tighter capital controls than advanced countries; the gradual liberalisation of capital accounts in emerging and developing economies could also influence their resilience to external shocks (see Figure 2).2

Figure 2. International reserves vs. capital controls

How do we assess the role of reserves in a cross-section?

Our database includes 112 countries composed of 31 emerging market economies (EME) and 81 developing countries. The details about our country coverage can be found in the Appendix of Bussière et al. (2014).

The analysis of this paper is based on cross-section regressions (taking the timing of the crisis as an exogenous shock for all EMEs in the sample). Our benchmark specification is described below:

yi,09 = β0+β1 rsvi,07 + βXi,07 + ϵi,09 

where rsvi,07 stands for one of the four commonly used reserve adequacy ratios (reserves to GDP, reserves to imports, reserves to M2 and reserves to short-term debt). Xi,07 corresponds to additional control variables (including capital controls, trade openness, an exchange rate regime dummy and an oil exporter dummy). Note that all the explanatory variables are lagged two periods. Taking lagged independent variables allows us to have a snapshot of the situation of the country before the start of the crisis, and to use it to explain its performance during the crisis.

The explained variable captures the impact of the GFC on countries’ real GDP growth. To compare the extent of the crisis across countries, two measures are used:

  • ‘Purged real GDP growth’ measures the difference between the realised real GDP growth rate in 2009 and a linear prediction based on a 6-year historical mean;
  • ‘Unexpected real GDP growth’ measures the difference between the realised real GDP growth rate in 2009 with the IMF WEO forecast in April 2008 (i.e. before the bankruptcy of Lehman Brothers).
Do reserves matter for economic growth during the GFC?

Among the four reserve adequacy ratios, we find that the reserves to short-term debt ratio is the most useful indicator to explain the real output growth during the crisis (see Table 2 in the companion paper, Bussière et al. 2014). The stock of foreign reserves scaled by the level of short-term debt two years prior to the crisis is positively and significantly correlated with the real GDP growth deviation from the trend; the coefficient from the full specification with control variables (Table 3) is 0.73. Using the deviation from the WEO forecast, we obtain a similar estimate of 0.62. Hence, a doubling of the reserves to debt ratio is associated with a 0.4 to 0.5 percentage point faster growth rate. This result is robust to the exclusion of outliers and small countries.

Once controlling for the interaction between the reserve ratio and capital controls, the marginal effect of the reserve ratio on real GDP remains positive and significant; it is even amplified by capital controls (Figure 3) if the full sample is considered. If outliers and small countries are removed, the amplifying effect of capital controls disappears. However, the marginal effect of the reserve ratio itself remains positive and significant.

Figure 3. Marginal effect of reserves

Foreign reserves might be held by a central bank in anticipation of a future negative shock to the national economy; both foreign reserves and higher GDP growth might also be by-products of a mercantilist exchange rate policy. In both cases, the OLS coefficient associated to reserves may be biased. To control for endogeneity, we propose several variables to instrument the reserves to short-term debt ratio based on the idea of regional peer pressure. The first one is an inverse distance weighted average of the reserves to GDP ratio of all countries in the world except the country we are considering for instrumentation. As a second possibility, we use the sum of the reserves to GDP ratios of the neighbouring countries (here again, excluding the country considered for instrumentation). Although not significantly different from zero, the signs of the 2SLS estimates are consistent with the OLS estimates (Table 5 in the companion paper). Considering that the corresponding Hausman test fails to reject the null hypothesis of exogenous right-hand-side variables, we feel more confident on relying on our OLS estimates.

Foreign reserve accumulation after the global financial crisis

A further question regards the post-crisis behaviour of non-advanced countries in terms of reserves accumulation. Do we see any pattern emerging in the last few years?

We find that after the crisis, countries sought to rebuild their stocks of foreign reserves. This rapid rebuilding has, however, been followed by a deceleration in the pace of accumulation (Figure 4).

Figure 4. Evolution of foreign reserve accumulation

Several features of post-crisis reserve accumulation are noteworthy. First, a significant rebuilding is more pronounced in countries that had a relatively low pre-crisis reserve adequacy ratio (see Figure 5). This might reflect an increasing demand for reserves in countries that were insufficiently self-insured before the GFC.

Figure 5. Rebuilding vs. Pre-crisis level of reserves

We test this proposition empirically. We find that in most cases the post-crisis reserve rebuilding rate is significantly and negatively correlated with pre-crisis reserve adequacy ratio. The coefficient of the pre-crisis adequacy ratio loses statistical significance once we add the oil country dummy. Oil countries did not seem to recover their reserve stock after the GFC; this however might be due to the collapse of world oil demand. This result is robust to controlling for outliers.

Moreover, we observe that a larger depletion of reserves during the crisis is associated with a stronger rebound (Figure 6). This seems once again to confirm countries' increasing appetite for reserve assets as a means of self-insurance.

Figure 6. Rebuilding vs. Depletion of reserves

Finally, the pace of reserve accumulation has slowed down in recent years, as is summarised in Figure 4.

There are several competing stories about the recent ‘flattening-out’ in reserve accumulation. First, it is possible that, once a country reaches its pre-crisis level of reserves, it slows the pace of foreign reserve accumulation, since holding large reserves incurs opportunity costs and possibly large risks associated with valuation effects. Second, the deceleration of foreign reserve accumulation might reflect a change of policy priority with regard to monetary autonomy, exchange rate stability and financial openness in the wake of the 2008-2009 financial crisis (Aizenman et al. 2010). Last, if foreign reserve accumulation tails off, it might be because of the stabilisation of the underlying macroeconomic variable – short-term debt as we argue here – that foreign reserves are accumulated to cover. Results obtained from estimating Vector Error Correction Models lend support to this last interpretation. That is, with the `flattening-out' of short-term debt after the financial crisis (the reasons why short-term debt diminishes after the GFC are multiple, e.g. Great Retrenchment), the demand for foreign reserves must fall.


In the late 1990s and early 2000s, a consensus developed that reserves were useful in averting, or at least mitigating, the occurrence of crises in emerging market and developing countries. Policymakers from these countries have apparently absorbed the lessons from this literature, as the level of international reserves dramatically increased throughout the 2000s (bearing in mind of course that other motives have played a role). The results presented in this paper suggest that the Great Financial Crisis has further demonstrated the usefulness of reserves: empirically, the countries that held more reserves as a percentage of short-term debt have been less negatively impacted, ceteris paribus. The results also suggest that this effect is especially strong when the capital account is less open.

Given that reserves seem to have played a role in offsetting the effect of the crisis, it is not surprising that the countries that depleted reserves to a greater extent are also those that rebuilt them more quickly in the direct aftermath of the crisis. One possible factor is that policymakers in emerging market and developing countries have concluded from the experience of the GFC that reserves are indeed very useful in protecting countries against crises.


Aizenman, Joshua, Chinn, Menzie D., and Ito, H. (2010), “The emerging global financial architecture: Tracing and evaluating new patterns of the trilemma configuration”, Journal of International Money and Finance, 29(4), 615-641.

Berg, Andrew, and Catherine Pattillo (1999), “Are currency crises predictable? A test”, IMF Staff Papers 46(2), 1.

Berkmen, S. Pelin, Gaston Gelos, Robert Rennhack and James P Walsh (2012), “The global financial crisis: Explaining cross-country differences in the output impact.”, Journal of International Money and Finance 31(1), 42-59.

Blanchard, Olivier J., Mitali Das, and Hamid Faruqee (2010) “The initial impact of the crisis on emerging market countries”, Brookings Papers on Economic Activity, 41- 1(Spring), 263-323.

Bussière, Matthieu, and Christian B. Mulder (1999), “External vulnerability in emerging market economies - how high liquidity can offset weak fundamentals and the effects of contagion”, IMF Working Papers 99/88, International Monetary Fund, July.

Bussière, Matthieu, Gong Cheng, Noëmie Lisack and Menzie Chinn (2014), “For a Few Dollars More: Reserves and Growth in Times of Crises”, NBER Working Paper 19791, January 2014. http://www.lafollette.wisc.edu/publications/workingpapers/chinn2013-014.pdf

Catao, Luis, and Gian-Maria Milesi-Ferretti (2013), “External liabilities and crises”, IMF Working Papers 13/113, International Monetary Fund, May.

Dominguez, Kathryn M E, Yuko Hashimoto and Takatoshi Ito (2012) “International reserves and the global financial crisis”, Journal of International Economics 88(2), 388-406.

Flood, Robert, and Nancy Marion (1999), “Perspectives on the recent currency crisis literature”, International Journal of Finance and Economics 4(1), 1–26.

Gourinchas, Pierre-Olivier, and Maurice Obstfeld (2012), “Stories of the twentieth century for the twenty-first”, American Economic Journal: Macroeconomics 4(1), 226–65.

Obstfeld, Maurice (2013), “Never say never: commentary on a policymaker’s reflections”, Manuscript, UC Berkeley, November.

Reinhart, Carmen M and Graciela L Kaminsky (1999), “The twin crises: The causes of banking and balance-of-payments problems”, The American Economic Review 89(3), 473–500.

1 For a review of this literature, see for instance Flood and Marion (1999), Berg and Pattillo (1999), Reinhart and Kaminsky (1999), Bussière and Mulder (1999), Gourinchas and Obstfeld (2012), Catao and Milesi-Ferretti (2013) and Obstfeld (2013).

2 As there is no space to review the relevant literature here, see Berkmen et al. (2012), Blanchard, Das and Faruqee (2010), Dominguez, Hashimoto and Ito (2012) and more references in Bussière et al. (2014) for recent discussions.

1,995 Reads