A government can build a large reserve stock in two ways: it can earn the dollars or borrow them. On the central bank’s balance sheet, the two look identical. Sovereign spreads do not treat them the same.

The distinction maps onto the two standard motives for holding reserves. Precautionary self-insurance — a buffer against the sudden stops and runs of the 1990s (Aizenman and Lee 2007, Obstfeld et al. 2010) — is typically debt-financed, through foreign-currency bonds or International Monetary Fund (IMF) drawings. Leaning against the wind (LAW) is not: when capital inflows or export booms push the currency up, the central bank buys the excess foreign exchange and sterilises the liquidity it creates (Blanchard et al. 2015, Levy Yeyati and Gómez 2022), paying for the reserves out of private inflows.

Both motives end with a larger stock. They do very different things to the consolidated public sector’s net foreign asset position — which is what spreads and crisis risk respond to.

Why the net foreign asset position is the channel

From the perspective of the consolidated public sector — Treasury plus central bank — the two motives have opposite balance-sheet implications. Leaning-against-the-wind accumulation raises foreign assets without adding foreign-currency liabilities, so the net position improves. Debt-financed accumulation raises assets and liabilities in step: for every dollar added to reserves, a dollar is owed abroad, and the net position is unchanged. Since spreads and crisis risk depend on the net external position, only the first type directly improves creditworthiness.

The central bank’s balance sheet alone can be misleading: if the Treasury borrows abroad and deposits the proceeds at the central bank, reserves and domestic liabilities both rise, and the operation looks indistinguishable from a sterilised leaning-against-the-wind intervention even though its ultimate source is public borrowing. Sosa-Padilla and Sturzenegger (2023) classify reserve changes by the currency composition of central bank liabilities and find that only locally financed reserves lower spreads; we step back to the balance of payments to capture the ultimate source of the foreign exchange. When the sample is reclassified using the central-bank approach, the estimated leaning-against-the-wind effect is attenuated, consistent with debt-financed operations being misread as leaning against the wind.

Tracing reserves through the balance of payments

The novelty is in the accounting. Rather than treating reserves as a single stock, we trace each monthly change — net of valuation and interest — back to its origin in the balance of payments: current account surpluses and private financial flows on the private side; Treasury and central-bank borrowing, with IMF lending tracked separately, on the public side. The panel covers 44 emerging economies and around 7,000 monthly observations from 2000 to 2024, excluding the Global Crisis (2008–09), Covid-19 (2020), and episodes with spreads above 1,000 basis points, so the estimates describe pricing under normal market access.

Private flows are the dominant and most variable driver of reserve changes, but Treasury borrowing plays a clear secondary role, while IMF lending is substantial but infrequent. Public borrowing also tends to move against private flows, as governments borrow abroad to cushion private outflows, and that offset weakens after the Global Crisis, pointing less to precautionary accumulation than to leaning-against-the-wind intervention in recent years. The mix varies by country: Brazil, Peru, the Philippines, and Vietnam track private flows, while Argentina, Colombia, Indonesia, and South Africa lean on public borrowing.

Figure 1 Balance of payments contributions to the change in international reserves (in % of GDP)

Privately financed reserves compress spreads and lower crisis risk

We regress the log of the Emerging Market Bond Index (EMBI) spread on global factors, ratings, debt, reserve stocks, a lagged dependent variable, and the reserve-financing components, with country fixed effects and standard errors clustered by country and date. Total reserve accumulation is associated with lower spreads, in line with earlier work (Levy Yeyati 2008). Once the sources are separated, the contrast is clear. A one-percentage-point increase in privately financed reserves as a share of GDP is associated with spreads roughly 1.3% lower, while the public-borrowing component carries a smaller coefficient significant only at the 10% level, and IMF lending is insignificant.

Within private flows, current account surpluses do most of the work: a one-point increase is associated with spreads about 1.8% lower, against roughly 1.0% for financial inflows. Translated into levels at the sample-average spread of roughly 350 basis points, a one-point-of-GDP leaning-against-the-wind accumulation tightens spreads by about 4.4 basis points, rising to about 6.4 points from the current account and falling to about 3.6 points for financial inflows. The public-borrowing effect is about 1.4 basis points, and statistically weaker. This pattern holds in the more homogeneous post-crisis sample, where private flows remain strongly significant and public flows turn insignificant with point estimates near zero.

Figure 2 Effect on the sovereign spread, by financing source

The same ordering appears in the probability of losing market access — defined in the paper as the spread crossing 1,000 basis points — using a logit with the same controls plus a lagged crisis indicator to capture persistence. The marginal effects mirror the spread results. A one-point-of-GDP increase in privately financed reserves is associated with a fall in crisis probability of about 1.1 percentage points, and the current account component with about 1.3 points, while public borrowing is not significant at any conventional level. The reserve stock itself lowers crisis probability by about 0.7 points per point of GDP.

IMF lending goes the other way, raising estimated crisis probability by about 2.9 points per point of GDP. As with spreads, this reflects the reverse causality: countries turn to the IMF once they are in distress. The ranking is robust to alternative distress thresholds between 800 and 1,200 basis points and to a quarterly specification.

The headline is that the quality of reserves matters alongside the quantity. Accumulating reserves by leaning against the wind during inflow or export booms strengthens the external balance sheet and is associated with cheaper borrowing and lower distress risk. Piling up reserves through external borrowing reshuffles the balance sheet without improving the net position and buys little market confidence. IMF programmes are best seen as countercyclical safety nets rather than substitutes for ex ante accumulation.

For inflation-targeting floaters this leaves room for a measured, macroprudential form of intervention rather than habitual currency management. Sterilisation can carry a quasi-fiscal cost when domestic rates exceed reserve returns, but that cost has been modest for most countries in recent years and is partly offset by the spread savings, which implies that the standard cost-of-reserves calculation overstates the net burden when accumulation follows the leaning-against-the-wind pattern (Levy Yeyati and Gómez 2020). Occasional, rules-based intervention need not undermine a float or the credibility of the inflation target. 

Reserves earned through leaning against the wind strengthen the public sector’s external position and travel with cheaper borrowing and lower distress risk. The numbers look the same on the balance sheet; markets treat them differently.

References

Aizenman, J and K Lee (2007), “International reserves: precautionary versus mercantilist views, theory and evidence”, Open Economies Review 18(2): 191–214.

Blanchard, O, G Adler and I de Carvalho Filho (2015), “Can foreign exchange intervention stem exchange rate pressures from global capital flow shocks?”, NBER Working Paper 21427.

Gómez, J F, E Levy Yeyati and P Temperley (2025), “The source matters: reserve financing and sovereign credit risk”, working paper.

Levy Yeyati, E (2008), “The cost of reserves”, Economics Letters 100: 39–42.

Levy Yeyati, E and J F Gómez (2020), “The Cost of Holding Foreign Exchange Reserves”, in Asset Management at Central Banks and Monetary Authorities, Springer.

Levy Yeyati, E and J F Gómez (2022), “Leaning-against-the-wind intervention and the ‘carry-trade’ view of the cost of reserves”, Open Economies Review 33(5): 853–877.

Obstfeld, M, J C Shambaugh and A M Taylor (2010), “Financial stability, the trilemma, and international reserves”, American Economic Journal: Macroeconomics 2(2): 57–94.

Sosa-Padilla, C and F Sturzenegger (2023), “Does it matter how central banks accumulate reserves? Evidence from sovereign spreads”, Journal of International Economics 140: 103711.