The collapse of external finance to developing countries

Published on: 10/02/24

By: Ishac Diwan,  Brendan Harnoys-Vannier,

What were the main trends in debt flows to developing countries in 2022?

The most recent information on sovereign debt was released by the World Bank in late 2023 (World Bank 2023), and it covers developments up to the end of 2022. In this note, we examine aggregate trends, with a special focus on the financial situation of the low-income countries and the lower-middle-income countries (LLMICs), and the movement of funds from/to the official sector and the financial markets.

Photo by Ma He, Unsplash

The central variable of interest is net transfers (NTs) on external debt – defined as loan disbursements by creditors, minus the repayment by the debtor of interest and principal on past debts. NT measures the actual movement of funds on account of external debt. For LLMICs, this variable has traditionally played an important role in financing the balance of payments. In 2019, NT on external debt was $105b, a figure comparable to net flows of FDI ($136b). By 2022, it was down to 20b, while net FDI declined only marginally. As such, this sizable drop in NTs had a great influence on a LLMICs’ level of international reserves, and their ability to pay for essential imports.

This note updates FDL’s previous analysis, which looks at data at end-2021 (Cohen and Harnoys-Vannier, 2023). The same trends are found to be at work, but in an exacerbated fashion. There are five key findings:

  • Total NTs have collapsed in 2022. In 85% of the 65 countries we study, they were substantially lower in 2022 than in 2019; and in 70%, they were close to zero or negative.
  • In the aggregate, positive NTs from MDBs and (non-China) bilateral donors leaked out completely to private sources and, to a lesser extent, to Chinese creditors (public and private).
  • More precisely, it is for countries that we deem insolvent that the change over time was the most abrupt. These countries received on average low or negative NTs from all sources in 2022. They ended up in worse shape than countries that we deem illiquid, or with no debt problem.
  • In countries that we classify as illiquid but solvent, the NTs coming from IFIs remained high, but these were on average completely offset by outflows to private and Chinese creditors.
  • Countries with no debt problems continued receiving sizable positive NTs. They did experience a fall in LT borrowings, but this was offset by increased ST debt.

Developing countries were hard hit by a series of negative shocks since 2019.

Their finances, already hurting from declining lending by China, were further weakened by the rise in global interest rates starting in 2022, which made it harder to refinance the large maturities coming due as a result of the borrowing boom of the late 2010s. But these losses were partly compensated by a surge of support by the IFIs after 2020. The movements we unveil for 2022 continues these trends, but in an exacerbated way: we find that the efforts of the official sector became completely neutralized by outflows to private and Chinese lenders.

While NTs on long-term external debt going to LLMICs have started to fall since 2019, they collapsed in 2022. [1] Total NTs declined from $106b in 2019 to $21b in 2022 (Table 1). At the same time, NTs on long term (LT) debt declined from +$84b in 2019 to $-16b in 2022 It is unusual for such a critical driver of development to flow out of developing countries rather than in. It is historically a rare event. The previous period that witnessed negative LT NTs (as apparent in the IDS database) to the same group of countries was around 20 years ago, in 2005.

In relation to the size of the economy, the average total NT received by LICs and LMICs stands at 0.5 and 0.4 % of GDP, respectively, in 2022. The change between 2019-2022 was -1.8% GDP for the median country in both groups. In the upper MICs, by contrast, NTs improved over the period by an average of +1% of GDP.

In terms of regions, net outflows of capital were negative in 2022 in MENA and South Asia (Table 2). The most affected regions were MENA, South Asia, and Sub-Saharan-Africa – these lost respectively 3.7, 2.5, and 1.4 % GDP in NT between 2019 and 2022. In MENA, the negative NT is largely due to developments in Egypt and Tunisia. Sri Lanka and Pakistan are the main drivers of the situation in South Asia. In SSA, the largest cases of negative NT are Lesotho, Angola and Mozambique.

Surging official financing and collapsing private flows

How did this dramatic transformation take place, leading so many poor developing countries to become providers instead of recipients of capital? The data indicates that in the aggregate, the main reason for the fall in NT in 2022 is the massive shift in private finance. In 2019, NT from the private sector (adding both the guaranteed and non-guaranteed private flows) on long term debt was +$54b. In 2022, there was instead -$51b moving out to private creditors -- a huge deterioration. This was due to a combination of rising principal due and decline in new lending (Properzi 2023). The capital market has moved from exuberance to gloom, with this shift constituting by far the most dominant factor in the recent evolution of the external finances of LLMICs. But in addition, long-term NT from China swung from +$5b in 2019 to -$6b going out in 2022.[1]

The financial situation of LLMICs could have been much worse were it not for the positive NTs received from the IFIs. Indeed, the large fall in NT happened despite historically large net disbursements by IFIs. These surged from $26b in 2019, to $68 in 2020, and then back to a lower but still high level of $32b in 2022. This has helped countries smooth the shocks they were experiencing. As part of this, IMF funding also initially shot up, with largest NTs in 2020, at $25b. By 2022 however, IMF’s NTs were close to zero.[2] In addition, bilateral donors have played a supporting role. Excepting China, bilateral NTs went up from $2b in 2019 to $10b in 2022.

Adding the effects, 2022 saw a large LT inflow of $42b from the (non-China) official sector, offset by an extraordinary large outflow of $58b to Chinese and private creditors -- ending up with a net outflow of $16b of long-term capital from the LLMIC group to the rest of the world.

Variations across debt-risk categories

The trends in the data described above are aggregated and average figures for the LLMIC group as a whole. There are country variations around these trends. A question important from a policy perspective is whether there are more precise characteristics of NTs when grouping countries according to the gravity of their debt problem. In a recent paper, Albinet et al (2023), based on DSA-like projections over 2023-2028, classify countries in three groups: those with no debt problems, those that are only illiquid, and those that are insolvent – using the flow and stock thresholds of the DSA-LIC.[3] Among the 65 countries covered here, they find that 14 countries are insolvent (and most of these are also illiquid), and that 18 are illiquid but solvent (see list in the annex).The preponderance of illiquidity is a characteristic of the current situation and it calls for solutions other than deep debt restructuring (Diwan, Kessler, and Songwe 2024).

There are two important questions to take to the data: is there evidence that it is especially in liquidity-constrained countries that the inflows from the IFIs are leak out as repayment to other creditors? And is there evidence that illiquidity is related to some inefficiencies connected to creditor heterogeneity?[4]

The data reveals a clear correlation between the type of debt challenges and how NTs evolved.  Table 3 below reports simple group averages. They show clear trends:

  • Insolvent countries experienced on average a large decline in NTs between 2019 and 2022 - 6 percentage points of GDP; they also have the lowest NTs in 2022– an average of -1.3% GDP.
  • Illiquid countries were affected, but less: they have moved from receiving a positive NT of 2.1% GDP to about zero.
  • In sharp contrast, countries not exposed to debt risk experienced only a minor change in NT over the period, and they keep receiving a sizable NT of 1.3% GDP on average in 2022.[5]

This pattern of NTs bolsters leads to a first discovery: groups with graver debt problems have suffered larger financial outflows, and they now receive smaller capital flows.

Figure A1 (see Annex) depicts NT in 2019 and 2022 for all LLMICs, where each country has been colored by the debt risk group it currently belongs to. In 2019, 10 countries had sizable negative NTs (over 0.5% GDP). In 2022, the number shot up to 25, 40% of our sample; another 20% had NT close to zero. All countries that ended with negative NT in 2022 were illiquid (yellow) or insolvent (red).[1]

Where are leakages out of inflows from international financial institutions largest?

Are the leakages out of IFIs’ transfers largest in illiquid countries as speculated above? To start, one must be careful when talking about “leakages” in two dimensions. First, there may be composition effects: IFIs sending positive NTs to some countries, but negative NTs to private creditors leaving from other countries. Second, we are speaking of leakages from foreign reserves, with implication for countries external balances.[2] They may or not involve leakages from public budgets.[3]

Table 3 shows that the positive NTs from IFIs are, on average, the largest for the group of illiquid countries, and that this flow is close to the NTs going out to private and Chinese creditors (around 1% GDP). In contrast, in insolvent countries, NTs coming from the IFIs, and the outflows going to other creditors are both much smaller. Finally, the no-risk group receives relatively large NTs from the IFIs, but these do not leak out. What is driving these regularities?

The behavior of private flows explains in part for these dynamics. We would expect private flows to slow down more in more indebted countries, as their outlook becomes riskier. Indeed, NTs collapsed in insolvent countries, on average, by 6.5% GDP. Private NTs also fell in illiquid countries but only by 0.7% GDP on average. And NTs remained flat in no-risk countries. Importantly, by 2022, private LT NT were not positive, on average, in any group. And while private creditors managed to extract more than 1% GDP from illiquid countries, they extracted very little from insolvent ones, presumably because these countries were in default, running arrears, or about to default.

Going beyond averages, Table A2 reveals that only 10/65 countries received in 2022 sizable positive NT from private plus Chinese creditors.[4] Interestingly, there are many “green” countries with negative NTs (8/50) – this reflects the generalized flight to quality that started in the capital market during 2022.

The other part of the story comes from the behavior of the IFIs. It is less easy a priori to guess how MDBs would deal with countries at risk. On the one hand, their poverty alleviation mandate should push them to play a countercyclical role. On the other, they also worry about the risk of default as they strive to maintain their credit rating. [5] In dollar terms, MDBs have scaled up their NT in no-risk countries, and have maintained them in the other two risker categories (Table 4). But when expressed as shares of GDP, the picture is clearer (Table 3). MDBs’ disbursements have been maintained not just in no-risk but also in illiquid countries. This shows their commitment, on average, despite the risks of leakages introduced by illiquidity. However, MDBs have reduced their NTs in insolvent countries, showing that the constraints they face in lending to countries at risk do have a restraining effect.

Figure A2 shows that there is less variation here than in private lending behavior. Six countries experienced negative NTs: Tunisia (LMIC, illiquid), Sierra Leone (LIC, insolvent), Mauritania (LMIC, insolvent); Central African Republic (LIC, illiquid), Guinea-Bissau (LIC, illiquid), and Lao (LMIC, insolvent). All other countries, including illiquid and insolvent countries, receive positive NTs. The red countries with positive NTs include: Ghana (a darling of development agencies until it defaulted at the end of 2022), Zambia (to support its debt restructuring deal), but also Cape Verde and Congo (both LMICs). These cases suggest that the line between IDA and IBRD countries is blurred with respect to lending by MDBs to red countries.

The bottom right panel of Figure A2 contains countries that experienced positive NTs from the official sector, and negative NTs to private and Chinese creditors. These represent two thirds of the countries that have received positive NTs from the official sector. In these countries, repayments to the private and Chinese creditors would have been more difficult, if not impossible, in the absence of positive official transfers. Aid effectiveness was lowered in these countries by the diversion of FX away from necessary imports to repayment to external creditors.

We end by asking if NTs to private and Chinese creditors appear to move in unison. Figure A3 seems to support this claim: except for a handful of off-diagonal countries, flows to/from these creditors move together – in or out. This supports a more speculative presumption, that of the existence of a “bank-run” phenomena - where creditors withdraw in reaction to the other withdrawing, leading to a bad equilibrium. A “bank run” seems to be taking place in particular in several large debtor countries that are illiquid–Pakistan, Kenya, and Angola. Among red countries, Mozambique and Congo stand out.

The rise of short-term debt

Lastly, the switch in financial flows from long to short-term is worth noting. The decline in total NT is smaller than the decline in long-term NTs. The difference is due to a rise of short-term debt, from $16b in 2019 to $36b in 2022. This suggests that some countries managed to refinance some long-term debts with shorter loans. Here, absolute figures tell the story more clearly (Figure 4). In insolvent countries, both LT and ST debt went down. On the other hand, in no-risk countries, NT on LT debt came down by $62b, but a rise in short-term borrowings managed to partly compensate, rising by $14b. Interestingly, this is also the case for illiquid countries. Their NT on LT debt came down by $28b between 2019-2022, but this was partly compensated by a rise in ST debt of around $10b. So while the “flight to quality” in the Eurobond market affected all LLMICs, short-term debt, but also bailout funds from China (Park et al 2023) and GCC lenders (Alhasan and Lons 2023), in addition to IMF funding, made the difference for the countries that could assess such sources.[6]

FIgure 4: Net Transfers by maturity and risk group


The outlook may be improving but remains worrisome

We have documented a steep rise in the outflow of financial resources from LLMICs during 2022. There are many signs that these trends grew worse in 2023, for two reasons. First, the bond market closed down entirely after March 2023 and not a single bond was raised afterwards by an LLMIC (Properzi 2023). Second, interest payments on variable rate debt rose together with interest rates. This includes in particular IBRD loans, for which we estimate interest payments to have risen by over $5b. Financial tightness could deteriorate further in 2024 and 2025 if no remedial action is undertaken. On the other hand, there may be some improvement in the medium-term, if global interest rates continue to decline. At the time of writing, Ivory Coast was able to reenter the market, and Benin as well as Kenya may follow suit. New bond issues will ease the liquidity problem by allowing for some roll-over of maturities. But these are happening at high interest rates, and moreover, only the least risky countries will be able to benefit in the near term.


Ahmed, Masood. 2023. “Defaulting on Africa’s Future”. Project Syndicate. June.

Albinet, C., Kessler, M. & Brancher, M. (2023). Mapping external debt vulnerabilities – an update. Finance for Development Lab.

Alhasan, Hasan, and Camille Lons (2023). Gulf Bailout Diplomacy: Aid as Economic Statecraft in a Turbulent Region. ISS Research papers, October.

Baqir, R., Diwan, I., & Rodrik, D. (2023). A framework to evaluate economic adjustment-cum-debt restructuring packages. Finance for Development Lab Working Paper2.

Cohen, Daniel and Brendan Harnoys-Vannier (2023): The resistible rise of external debt in the global south (2011-2021). FDL working paper no 3, March.

 Diwan, Ishac, Martin Kessler, and Vera Songwe (2024). A bridge to climate action: A tripartite deal for times of illiquidity. Finance for Development Lab, Policy Note 14.  

Parks, B. C., Malik, A. A., Escobar, B., Zhang, S., Fedorochko, R., Solomon, K., Wang, F., Vlasto, L., Walsh, K. & Goodman, S. (2023). Belt and Road Reboot: Beijing’s Bid to De-Risk Its Global Infrastructure Initiative. Williamsburg, VA: AidData at William & Mary.

Properzi, E. (2023). Closing the spigots: the rise and collapse of bond markets for developing countries. Finance for Development Lab.

Ramos, L., Ray, R., Bhandary, R.R., Gallagher, K.P., and W.N. Kring (2023). Debt Relief for a Green and Inclusive Recovery: Guaranteeing Sustainable Development. Boston University Global Development Policy Center.

Songwe, Vera, Nicholas Stern and Amar Bhattacharya (2022). “Finance for Climate Action: Scaling Up Investment for Climate and Development.” London: Grantham Research Institute on Climate Change and the Environment, London School of Economics and Political Science.

World Bank (2023), “International Debt Report”, December

[1] Our sample includes 65 countries with an aggregate GDP of $9.3 trillon and total external debt of $2.6Trillion.

[2] Because of uncertainties on the classification of public and private institutions in China, we group their flows together. Balance of payment support to some countries, when accounted for, is classified as ST debt.

[3] We have removed Ukraine from the sample, even though it is classified as an LMIC, because the massive flows it has received in 2022 would distorts the overall picture. In 2022, there was a NT of +$13.6 from MDBs and bilateral creditors, and a -$7.4b NT out to private creditors. If added to the group in the table, the leakage from the official to the private sector would seem considerably magnified.

[4] The projection used 2022 IDS data, and is in the process of being updated with the new 2023 release.

[5] A main specificity of the current debt crisis is that creditors are highly heterogeneous–traditional donors from the Paris Club, creditors from China, lenders from the private sector – who have little experience working together.

[6] It is interesting to note that the group that ended up insolvent had the highest level of NT in 2019, followed by the group that ended up illiquid, and then by the group that ended up with no debt problem. This seems to indicate that it is countries that overborrowed that were more vulnerable to a market reversal, as opposed to countries that were mismanaged. The issue of the genesis of the current crisis remains in need of more research.

[7] Only two insolvent (red) countries received positive NT in 2022: Mongolia, and Vanuatu. Both are closely aligned with and financed by China.

[8] Large negative net transfers out of reserves reduce foreign reserves, can lead to devaluation, imply that important import need to be curtailed – and all these effects can increase poverty, hurt fiscal balances, and reduce growth.

[9] In most cases, there are inflows from IFIs to the state, and outflows to private creditors, coming from the domestic private or public enterprise sectors. These flows do not always belong to the same budget constraint. But they all come in/out of international reserves, and as such, they all affect external balances.

[10] Again, this includes only two insolvent countries, Mongolia, and Vanuatu.

[11] This should especially be the case in LICs, as the IDA country allocation formulae has low sensitivity to debt risk. In IBRD countries however, country risk is taken into consideration openly in the lending decision.

[12] From a debt management perspective, the move to short-term debt poses disadvantages and advantages: short-term lending carries additional rollover risks, but it is also a natural reaction in the face of a steep yield curve as it avoids locking-in long term loans with high-interest rates.