The Geography of Remittances: An Updated Global Bilateral Remittances Matrix

Introduction

Remittance flows to low- and middle-income countries are larger than international aid and foreign direct investment (FDI) combined. In 2024, for 59 countries, remittances were equal to more than 5 percent of gross national income (GNI). As aid budgets shrink and the foreign investment landscape remains volatile, remittances are becoming an ever more important source of development finance for poor households.

However, remittance flows follow migration networks. This means two things: first, remittance-receiving nations are often highly dependent for inflows upon a small number of remittance-sending countries—in some cases upon one country. When those flows are macroeconomically significant, this can create exposure to risk. (In 2024, 16 bilateral remittance corridors were worth more than 10 percent of recipient GNI.)

Second, access to remittances remains limited for low-income countries, whose people face barriers to emigration. Given the earnings gains available via labour migration to higher-income countries, development policymakers should seek to increase opportunities for productive emigration and monitor remittance channels as part of macroeconomic risk management.

Yet we know frustratingly little about the current geography of bilateral remittance flows. The most recent (and still often cited) data is from 2021, produced by the World Bank’s Global Knowledge Partnership on Migration and Development (KNOMAD). KNOMAD has since been closed due to funding cuts. Given the importance of updating this data, in this piece we replicate their methodology for the latest year with data: 2024. The resulting bilateral remittance flows should be interpreted with caution, but still offer a valuable update. In this note we model and visualise some of the most interesting points and trends from the new corridor-level matrix.

Remittance flows, 2021-2024

Remittances are a major source of external finance for many countries. In the context of aid cuts, trade disruptions, and challenges to FDI, the relative importance of remittances is rising. At the household level remittances allow resilience to shocks and a wide range of key investments—such as in education, healthcare, or businesses—at the macroeconomic level remittances are a key supply of foreign currency, and are also often valuable for their countercyclicality.

Global remittances received rose by US$160 billion (22 percent) from 2021 to 2024 (despite the COVID-19 pandemic). Remittance inflows increased for nearly three-quarters of countries; only 40 countries saw decreased remittance receipts. Some, however, saw significantly higher remittance growth than others: 28 countries (15 percent) saw an increase of more than 50 percent; 12 countries saw more than 100 percent; and four saw an increase of more than 500 percent. Ethiopia, following a deliberate policy push to encourage regular remittances, saw a staggering 1,500 percent increase (US$6.7 billion) over the three years. (A similar effort is now bearing fruit in Nigeria, and should show up in future updates.) India saw remittances growth of nearly 54 percent (US$50 billion).

Some countries saw major decreases in modelled remittance inflows during the period. Kuwait, for example, saw a 97 percent decrease; Russia’s inflows fell by 80 percent (possibly due to sanctions). It should be noted, however, that the World Bank’s remittance data is derived from International Monetary Fund (IMF) balance of payments data, and that big fluctuations—as for Kuwait—may be the result of data artifacts and should be interpreted cautiously. Figure 1 shows changes in total remittance inflows over the period.

Figure 1.

The geography of remittance flows

Geographically, the distribution of remittance flows has—unsurprisingly—changed relatively little from 2021 to 2024. In Figure 2 we show a map of remittance flows at the global, country, and intra-regional levels. A few things stand out. Notably, several corridors are enormous, even at the global level: the flow from the United States to Mexico, prominently, accounts for 7.3 percent of all remittances. The Gulf comes through in the model as a predictably important remittance-sending hub— although its importance is understated due to gaps in migrant stock data. India stands out as a major remittance recipient, accounting for 15.3 percent of all country inflows. The top five recipients account for over 35 percent of all inflows. The US is by far the largest modelled source of remittances, accounting for 26.1 percent of all outflows; Saudi Arabia accounts for 6.3 percent, and the UAE for 5.5 percent. Europe is a tangle of intra-regional flows.

Figure 2.

At the regional level (Figure 3), the largest remittance-receiving region is Europe and Central Asia with 26 percent of flows, followed by South Asia with 24.3 percent—which also saw the largest increase in modelled remittances received, rising 39.2 percent (US$61.5 billion) from 2021 to 2024. Europe and Central Asia, North America, and the Middle East and North Africa (MENA) cumulatively accounted for 81.4 percent of all remittance outflows in 2024. The modelled data indicates high levels of intra-regional remittance flows within Europe and Central Asia, and large inter-regional flows from North America to Latin America and Caribbean (migrants to the US) and MENA to South Asia (migrants to the Gulf). This is unsurprising given labour migration networks. The MENA region was the only one for which remittance receipts fell; flows from MENA to South Asia grew the most in absolute terms, however, rising by US$29.2 billion. For inter-regional flows, this was followed by North America—Latin America and Caribbean (rising by US$24.0 billion), and North America—South Asia (US$14.0 billion).

Total remittance outflows from North America rose by 21.6 percent (US$47.7 billion), and from Europe and Central Asia by 17.7 percent (US$43.5 billion). Outflows from countries in East Asia and Pacific (30 percent), Latin America and the Caribbean (30.8 percent), South Asia (28.3 percent), and sub-Saharan Africa (32.9 percent) grew faster in proportional terms, but less in absolute terms.

Figure 3.

When flows are sliced by country income level, the largest flow by far is from high-income countries to lower-middle-income countries: this accounts for 40.2 percent of all bilateral flows, and lower-middle-income countries accounted for 56 percent of total growth in remittance receipts from 2021 to 2024. High-income countries account for 85.6 percent of all modelled outflows (US$768.9 billion), but only 22.6 percent of inflows. By contrast, low- and middle-income countries receive 77.4 percent of all remittance flows (US$695.1 billion).

Of all flows sent from high-income countries, those to low-income countries are the smallest, totalling just US$14.7 billion in 2024 (see Figure 4). This is consistent with scholarship observing the difficulties faced by people from low-income countries in accessing migration. This cumulative-level corridor saw the highest growth from 2021, however, rising by US$5.7 billion (63 percent).

Figure 4.

Access to remittances is ultimately determined by access to migration opportunities. In Figure 5, we show the stock of migrants sliced by income group, and the average remittances per capita sent between income groups. While there were 64.5 million migrants from lower-middle-income countries in high-income countries in 2024, and 70 million from upper-middle-income countries, there were only 8.6 million people from low-income countries in high-income countries. The median low-income origin country’s migrants were hosted in countries of destination with an average GNI per capita (in current international dollars) of US$20,700 in 2021 and US$24,900 in 2024; migrants from the median lower-middle-income country were hosted in countries of destination with a much higher 2024 average GNI per capita of around US$48,800, while migrants from upper-middle-income origin countries were hosted in countries of destination with average GNI per capita of US$64,200. During the same time, the total stock of migrants from low-income countries of origin increased by 20.5 percent. But migrants from low-income countries remain far less represented in high-income countries of destination. Without access to high-earning opportunities, their countries of origin have less access to remittance flows.

Figure 5.

Concentration and dependence at the corridor level

Most remittance-receiving countries obtain a high proportion of their remittances from a relatively small number of remittance-sending countries. This is probably unsurprising, given the importance of networks and path dependency to migration patterns: success in accessing migration and remittances naturally begets success. At the macro level, however, dependence on a small number of countries (and still more when on just one country) may introduce structural risks, as was demonstrated in 2025 when the US announced its intention to tax remittances (now in implementation in a reduced form).

More than a quarter—51 countries—receive more than half their remittances from one remittance-sending country; 21 countries receive more than 75 percent. The median country received 38 percent of modelled remittances from its single largest corridor in 2024. This suggests that for many remittance-receiving countries, changes in labour-market access, exchange rates, a recession, conflict, or other major hazards in a single other country could seriously disrupt their access to remittances. Several countries are extremely dependent on a single remittance-sending country: Mexico, in particular, is estimated to receive nearly 97 percent of its remittances from the US. Other major dependencies include that of the Marshall Islands on the US; Papua New Guinea on Australia; El Salvador on the US; and Tajikistan on Russia.

A higher proportion of countries are highly reliant on three remittance-sending countries. More than a third of countries receive more than 75 percent of their modelled remittance inflows from their top three remittance source countries; 85 percent receive over half their remittances from them. Dependence on the largest corridor and largest top three corridors is similar across income levels, but varies considerably across regions.

The model shows little change in concentration between 2021 and 2024 (Figure 6). The average share of remittances coming from the largest corridor fell from about 44 percent to 43 percent; the average share from the top three corridors fell from about 70 percent to 69 percent. Similarly, there was very little change in the identities of countries’ top source corridors: of 181 remittance-receiving countries, only 9 had a different top remittance source in 2024 versus 2021.

Figure 6.

On the flip side, several remittance-sending countries also demonstrate high levels of importance. These divide broadly into two categories: some, such as the US or Saudi Arabia, send very large volumes of remittances to multiple recipients and are important contributors on average. Others may be smaller remittance-senders overall, but may be crucial for one or more recipients, such as South Africa for Lesotho; Australia and New Zealand for Pacific states; or Côte d’Ivoire for Burkina Faso.

A small number of remittance sources are exceptionally important. In Figure 7, we show the largest share contributed by each remittance-sending country and its “average importance” score, measuring how important a source country is to its recipients. (For each recipient, we calculate the share of its remittances coming from that source, then take an average weighted by the size of the source’s remittance corridors. A high score can therefore reflect either importance across several recipients or very high importance to one recipient that receives most of the source’s outflows.)

The median remittance-sending country has an average-importance score of just 2.2 percent (Figure 7). Only 41 remittance source countries (migrants’ countries of destination) have average importance scores over 10 percent, 12 above 25 percent, and just one above 50 percent: the US. Six remittance-sending countries contribute more than 80 percent of the remittances received by their highest-proportion recipient. This indicates that for these countries, changes to labour market and other policy areas may have significant externalities for particular countries of origin.

Figure 7.

The importance of source countries was broadly stable in the modelled results from 2021 to 2024, with little change. The median average-importance score rose slightly, from 2.1 percent in 2021 to 2.2 percent in 2024.

Remittances in the context of other financial flows

Globally, remittances are well recognised to be a crucial financial flow. This is especially the case for low- and middle-income countries: in 2024 these country groups received around US$695.1 billion in remittances, compared with US$95.8 billion in gross Official Development Assistance (ODA) and US$420 billion in net FDI. In 2024, remittances to low- and middle-income countries were more than a third larger than gross ODA and net FDI! Even if one uses the “headline” (net) ODA figures —which are less useful for comparing annual flows—, setting 2024 ODA at US$227 billion, remittances remain larger than both combined.

For low-income countries, however, remittances are only modestly larger than ODA (at US$24.5 billion and US$19.7 billion in 2024 respectively), likely following the difficulty experienced by people from these countries in accessing productive emigration. Across low- and middle-income countries, remittances rose by US$140.7 billion from 2021 to 2024, while ODA rose by only US$13 billion and net FDI fell by US$283 billion. Of 70 low- and lower-middle-income countries receiving both remittances and ODA, total remittances exceeded ODA for 48 in 2024.

In Figure 8 we show remittances versus gross ODA and net FDI from OECD countries at the level of the bilateral corridor. (Note that ODA and FDI figures are only for OECD countries: the chart lacks non-OECD flows.) Where corridors contained both remittances and ODA, in 2024 remittances exceeded gross ODA for 36 percent of corridors to low-income countries (from 481 corridors), and 68 percent of corridors to lower-middle-income countries (from 1,118).

Figure 8.

Remittances are thus highly significant at the macroeconomic level for many recipient countries: in 2024 they totalled the equivalent of more than five percent of GNI for 59 countries. In Figure 9 we show total remittances relative to the recipient countries’ GNI. Total remittances equal or exceed 30 percent of GNI for two countries—Tonga (39.9 percent) and Tajikistan (36.5 percent)—and are greater than 20 percent for 11 countries. Broadly, remittances’ contribution to GNI declined slightly from 2021 – 2024; the average recipient country saw modelled remittances fall from 5.6 percent to 5.3 percent of GNI, while the median fell from 2.5 to 2.4 percent. However, while the typical movement was small, the tails are quite large. Seven countries saw remittances as a percentage of GNI increase by more than four percentage points; nine saw it decline by more than the same amount.

The average ratio of remittances to GNI was broadly unchanged for low- and lower-middle-income countries, but the upper-middle-income average fell by about 0.9 percentage points.

Figure 9.

For many remittance-receiving countries, single remittance corridors are also macroeconomically important. In 2024, 16 corridors were greater than 10 percent of recipient GNI; 27 corridors were worth over 5 percent (see Figure 10). Three corridors were worth more than 20 percent. These corridors are not, however, necessarily the largest in absolute terms. The US-Mexico corridor holds the largest absolute flow, but is equal to (only) 3.6 percent of Mexico’s GNI. By contrast, the France-Comoros corridor is worth only US$265 million, but equals 18.3 percent of Comoros’ GNI.

The corridors with greatest relative importance generally fall into several clusters: remittances from the US to Latin America and the Caribbean; from Russia to former Soviet countries; from Pacific and other economies to small island states; and from larger to smaller economies within Southern Africa. No modelled remittance flows from any individual Gulf state are greater than five percent of recipient GNI in 2024. This is largely because countries receiving large remittance volumes from the Gulf normally receive from multiple states. (And, in all likelihood, because migrant stock data for the Gulf is very patchy, meaning the model understates the importance of Gulf remittances). On aggregate, however, four countries—Nepal (7.0 percent); Pakistan (6.0 percent); Jordan (5.5 percent); and Egypt (5.3 percent)—received the equivalent of more than five percent of GNI from Gulf states cumulatively. Nine countries received remittances from the Gulf equivalent to more than 1 percent of GNI, suggesting that a serious disruption to flows from the region (as the US – Iran war has threatened to cause) would have macroeconomic consequences.

Figure 10.

At the corridor level, changes from 2021 to 2024 are mostly very small, but there are again some large tails. Across all corridors, the mean change is roughly zero. Among corridors that are macroeconomically relevant (here defined as over one percent of recipient GNI in either year) there is on average a downward trajectory, falling by 0.3 percentage points from roughly 3.5 percent of GNI to 3.3 percent. The average corridor’s contribution as a percentage of GNI remained roughly static 2021 to 2024. Among the 35 corridors that exceeded five percent of recipient GNI in either year, however, the average fall was much larger at 0.8 percentage points. The number of very large corridors fell slightly: in 2021 there were 183 corridors equal to more than one percent of recipient GNI, versus 166 in 2024; those equal to more than 20 percent fell from five to three (falls in the Russia – Kyrgyzstan and France – Comoros corridors).

Methodology and caveats

This section briefly details the methodological approach followed and its limitations. A more detailed overview is available in the accompanying methodology document.

To construct the new bilateral remittances matrix, we follow KNOMAD’s previous methodology. We allocate each recipient economy’s total personal remittances received across likely source economies, weighting corridors by the number of migrants in countries of destination and disparities in GNI. The key data sources are World Bank data on remittance inflow totals; Eurostat country-of-birth data; the United Nations Department of Economic and Social Affairs (UN DESA)’s migrant stock data; and World Bank data on GNI per capita for income scoring. Where UN DESA’s 2024 release omits countries of origin that its 2020 release named, we fill them from the 2020 release and, for the United States, United Kingdom and Germany, from national census and register data. The World Bank’s personal-remittances measure combines personal transfers (money sent home by migrants remaining abroad and repatriated savings) with compensation of employees. It can therefore include earnings of short-term and cross-border workers as well as transfers between households. It should be noted that some UN DESA and Eurostat population estimates draw on information from before 2021 or 2024, so changes in modelled corridor shares may not fully capture recent changes in migration geography.

The remittance allocation methodology used by KNOMAD and this update is that developed by Ratha and Shaw (2007). It weights each potential corridor by the number of migrants from the remittance-receiving country living in the remittance-sending country and by an income-based score (GNI per capita) proxying migrants’ likely remitting capacity. It then normalises corridor scores within each recipient economy, so that the estimated bilateral flows for each remittance-receiving country sum to the World Bank’s remittance inflow total. Following KNOMAD’s approach, which blocked remittance corridors in which sending had been disallowed for political reasons, we add the Russia – Ukraine corridor (for 2024) to a list of previously blocked corridors including India – Pakistan and Armenia – Azerbaijan.

The model output is heavily sensitive to available migrant stock data. Over 70 percent of possible bilateral corridors are not covered (see Figure 11). In many cases this lack of coverage is likely to be legitimate: many countries are small, and have a limited number of potential migrants who will anyway largely follow networks to known countries of destination. It is probably unsurprising, for example, that UN DESA records no Tongans in Togo, and thus also unsurprising that that corridor is not “covered”. In other cases, absences are a problem: for example, UN DESA records no Kenyans in the Gulf, whereas it is estimated that 300,000 may be working there. In some, although far fewer, cases coverage is missing due to a lack of GNI data. In other cases there are surprising inclusions. Are there really more than 1,400 Andorrans in Mali?

Figure 11. 

The model’s outputs also vary considerably against the World Bank’s data on remittances paid (see Figure 12). This is unsurprising given that the World Bank’s remittances data itself does not reconcile: it records US$738 billion and US$898 billion received in 2021 and 2024 respectively, versus just US$478 billion and US$622 billion paid— the outflows data thus accounting for just 65 and 69 percent of inflows. The additional allocations made by the model mean that remittance-sending countries’ outflow totals are generally much higher than the World Bank’s data currently suggests: this is the case for 72 percent of countries (154), with an average difference of US$2.6 billion. This is somewhat skewed by the fact that the World Bank’s data reports no remittance outflows for 55 countries, including Singapore (with a modelled value of US$7.8 billion; other modelling suggests remittance outflows of over US$8 billion). A high proportion of the difference is attributable to the US, for which the model estimates remittance outflows US$131.5 billion higher than in the World Bank’s data. In relative terms there are several countries with vast differences: the modelled estimate for Spain (at US$28.3 billion), for example, is more than 4,700 percent higher than the World Bank’s (US$0.6 billion)— versus the Banco de España’s estimate of roughly US$13.2 billion.

Figure 12.

This said, the model does not always result in higher remittance-sending figures: for 28 percent of countries it suggests smaller sums, with an average difference of US$2.2 billion. This is notably the case for Switzerland, which has a very high total —US$40 billion, the fourth-largest—sent in 2024 per the World Bank’s data, ascribed by the International Organization for Migration (IOM) to payments made to workers commuting from France. These cross-border workers appear as French-domiciled in UN DESA data, and thus a lower estimate for remittances paid is allocated to Switzerland by the model. (This lower total, of US$11.1 billion sent in 2024, is in fact much closer to the Swiss government’s own estimate of US$10.3 billion.) A similar dynamic is present in the case of Luxembourg. For other reasons, modelled outflows for China (US$17.7 billion; 88 percent), Indonesia (US$10.3 billion; 92 percent), and the United Arab Emirates (UAE, US$9.0 billion; 15 percent) are also significantly lower.

Above all, the figures presented should be interpreted as corridor estimates modelled using data that is not always reliable. As KNOMAD points out, both bilateral migration stock data and remittance inflow data are often dicey; and the GNI proxy used is only rough. The corridor estimates can therefore be used to indicate broad likely patterns of remittance-sending activity, the reliance of remittance-receiving countries on source countries, and trends— but they are indicative.

Conclusion: Policy and data

Remittances are receiving growing attention in the context of aid cuts and wider macroeconomic turmoil. On the basis of this analysis, this attention is deserved: from 2021 to 2024, remittance flows rose substantially. In 2024 remittances exceeded five percent of GNI for 59 countries; for 11, they exceeded 20 percent. They are increasingly crucial to both household-level development and macroeconomic policy.

Importantly, they are (generally) intra-household transfers, and “redirecting” them into investment or public goods is both easier said than done and ethically fraught. Policies targeting remittance-receiving households or migrants should thus be considered with care. At the macro level, however, this analysis suggests that remittance flows should be carefully monitored, and that the risks associated with high dependence on a small number of remittance-sending countries should, where possible, be hedged against. Remittance-dependent states, especially when a high proportion of remittances are received from one country, are highly exposed to externalities from policy choices elsewhere. Migrant-sending countries could, for example, attempt to diversify the countries to which they send workers. Development-minded high-income countries, addressing the other side of the table, could proactively work to recruit from countries currently struggling to diversify or to send workers at all— such as, for example, via targeted agricultural worker schemes.

This analysis also demonstrates why better migration data is needed. This update to the global bilateral remittances matrix is valuable in large part because so little bilateral remittance data is available. It performs a useful role, but still has to make do with a clunky proxy and depend on patchy and unreliable data on both migration and financial flows. Improvements to remittance flow data would be a public good and could, if it allows even slightly better macro policy, have a high return on investment.

Who should manage this data? With the demise of KNOMAD, there isn’t a clear answer to this question. But there are several options. The World Bank is still formally the custodian of Sustainable Development Goal 10.c.1 on reducing remittance costs, and still hosts KNOMAD’s work and the Remittances Price Worldwide datasets (albeit without having updated them since August 2025). The International Fund for Agricultural Development (IFAD) has a substantial Financing Facility for Remittances and, for the past several years, the RemitSCOPE platform, working with central banks to piece together a global repository of remittance-related data. The UN Capital Development Fund, like IFAD, has pooled data to provide detailed country-specific studies of remittance flows to inform policymakers; but like KNOMAD, its Migrant Money initiative has also been shut due to funding cuts. The best suited organisation, however, may be the International Monetary Fund, given the macroeconomic importance of remittances and the IMF’s existing data provision role.

Remittances are increasingly important, but their information landscape has sadly only worsened in recent years: as a public good and for better policy, better data is sorely needed.

The authors thank Dany Bahar, Helen Dempster, and Çağlar Özden for being willing to review the paper.