## spillovernote12 ## Source details **Canonical URL:** [spillovernote12](https://www.imf.org/-/media/files/publications/spillovernotes/spillovernote12.pdf) ## Other formats - [Markdown version](/-/media/files/publications/spillovernotes/spillovernote12.pdf.md) - [Structured JSON version](/-/media/files/publications/spillovernotes/spillovernote12.pdf.json) --- ### Introduction and Summary - After close to two decades of strong economic activity, overall growth in sub-Saharan Africa decelerated markedly in 2015–16, to its lowest level in more than 20 years at 1.4 percent. - Heterogeneity: largest economies (Nigeria and South Africa) experienced negative or flat growth in 2015–16; a third of countries continued to grow at 5 percent or more during the period. - Growth began to recover in 2017, prompting analysis of how trends in the largest economies spill over regionally. - Scope: focuses on trade, banking, financial, remittance, investment, fiscal, and security channels as principal transmission mechanisms. - Main conclusion: interdependence among sub-Saharan African countries is higher than generally assumed, implying a need for enhanced regional surveillance and spillover analysis in addition to bilateral surveillance. ### High-level quantitative findings - Intraregional trade rose from 6 percent of total exports (1 percent of GDP) in 1980 to 20 percent of total exports (4 percent of GDP) in 2016. - Ten countries account for 65 percent of total regional demand for intraregional exports. - Exports to the top 10 destinations represent between 5 percent and 10 percent of source-country GDP for some countries. - Econometric estimates attribute about half of growth in regional trade over 1980–2016 to subregional trade integration, particularly within the EAC and the SADC. - Growth spillovers via trade: a 5 percentage point increase in the export-weighted growth rate of intraregional partners is associated with about a 0.5 percent increase in the average sub-Saharan African country’s growth. ### Key policy implication (summary) - Need for enhanced regional surveillance and spillover analysis, monitoring subregional banking expansion and cross-border credit links, reducing remittance costs (including leveraging mobile money), addressing fiscal vulnerabilities tied to customs unions and revenue-sharing formulas, and coordinating policies to limit unintended cross-border effects such as fuel subsidies that encourage smuggling. ### Methodology and contributions - Uses multiple methodologies, draws on existing studies, identifies likely originator and recipient countries of spillovers, provides new empirical estimates of channel sizes, and documents new transmission channels. --- ### 1. Trade channel — Regional Trade Links Gaining Strength - Intraregional trade integration increased substantially over the past 35 years; regional trade was 6 percent of total exports in 1980 and 20 percent in 2016. - Small countries experienced faster increases in trade integration (faster growth in the simple average level). - Drivers: - Twofold increase in the relative price of commodity exports over 1995–2013. - Volumes of exported commodities increased by two and a half times over 1995–2013. - Strengthened macroeconomic policies and institutions, regional trade agreements, and bilateral tariff reductions. - Global comparisons: - As a share of total exports, sub-Saharan Africa exhibits the highest share of intraregional trade among emerging and developing regions. - Measured as percent of GDP, sub-Saharan Africa is mid-ranked globally. - Subregional concentration: - Subregional trade accounts for most intraregional trade. - SACU subregional trade alone represents half of total sub-Saharan Africa intraregional trade. - In SADC and SACU, subregional trade represents more than 80 percent of member countries’ intraregional trade. - SADC accounts for more than 70 percent of total sub-Saharan Africa intraregional trade; SACU accounts for more than 50 percent; WAEMU, EAC, and CEMAC each account for less than 10 percent. - Trade frictions: - Distance and socio-cultural differences hinder bilateral trade more in sub-Saharan Africa than elsewhere, explaining why most regional trade occurs within subregions. ### Intraregional exports: concentration and exposure - Ten sub-Saharan countries = 65 percent of regional demand for intraregional exports. - South Africa imports 15 percent of total intraregional exports. - Major bilateral shares (exporter GDP perspective): - South African imports from Swaziland, Lesotho, Zimbabwe, and Mozambique represent between 4 percent and 11 percent of those exporting economies’ GDP. - Zimbabwe’s total demand for goods from Zambia, Malawi, and Botswana constitutes between 1 percent and 4 percent of these countries’ GDP. - Nigeria, Mali, Ghana, and Burkina Faso import more than 1 percent of GDP of their subregional trading partners. - For four countries, exports to Nigeria and Mali represent more than 1 percent of their economy. - For 10 countries, exports to South Africa represent more than 1 percent of their GDP. - Role of economic structure: - Non-resource-intensive countries: intraregional exports = 7 percent of GDP and = 30 percent of total exports. - Oil-producing countries: exports to rest of world = 25 percent of GDP; intraregional exports = 1.5 percent of GDP. ### Trade–growth elasticity (1980–2016) - Panel regression: a 1 percentage point increase in the export-weighted growth rate of intraregional partners is associated with about a 0.11 percent increase in the average sub-Saharan African country’s growth (baseline). - A 1 percentage point increase in the growth rate of trading partners outside the region is associated with a 0.34 percent increase in average sub-Saharan African country growth (baseline). - Robustness: GMM, exclusion of largest economies, five-year averages preserve significance; excluding large economies does not materially change intraregional elasticity. - Cross-region comparison: SSA intraregional elasticity is slightly lower than Latin America and Asia, higher than MENA. --- ### 2. Banking channel — Banking Interdependence Becoming More Subregional - PABs and subregional banking groups’ share in sub-Saharan African financial systems is increasing; assets and deposits declined recently consistent with the regionwide economic deceleration. - Concentration of parent banks: - PAB parent banks concentrated in three countries; South Africa and Togo each home to about 40 percent of all PABs. - Subregional parent banks are about half as concentrated as PABs. - Host-country patterns: - PAB foreign subsidiaries spread across the region; subregional bank host countries are more geographically concentrated (South African banks in SADC, Kenyan in EAC, Nigerian in West Africa, Gabonese/Cameroonian in CEMAC). - Special case: Ecobank holding company headquartered in Togo; de facto economic headquarters in Nigeria. - Cross-border linkages and spillover channels: - Spillovers run both ways between parents and subsidiaries; channels include deposit placement, credit, governance, reputational concerns, liquidity-sharing, and syndicated loans. - Subsidiary model reduces but does not eliminate contagion risk. - Systemic importance measure: - Ratio: total deposits in foreign African subsidiaries/branches of PABs or subregional banks to total deposits by country; ratio highest in small countries. - Degree of systemic importance larger for PAB subsidiaries/branches than for subregional banks. - Financial deepening and growth: - Credit to private sector has deepened; deepening more pronounced in countries home to PABs and subregional banks. - Strong positive association between GDP growth and private credit growth. - Lower growth in parent-bank home countries could reduce credit and deposit growth in foreign subsidiaries/branches if parent supplies significant liquidity; evidence suggests bank funding is mostly local in the largest countries. - Commodity-related channels: - Countries severely hit by the commodity price decline experienced credit growth deceleration and a decline in deposits. - Reinforcing factors such as slowdown in economic activity and government arrears exacerbate deposit and credit declines. ### Deposits, loan-to-deposit ratios, and correspondent banking - PABs and subregional banks are relatively less active lenders in host countries: - Average loan-to-deposit ratio for PABs is about 34 percent less than country-level average. - Subregional banks have loan-to-deposit ratios about 22 percent less than country-level averages. - Selected country loan-to-deposit ratios (2015) — Percent (excerpt): - Kenya: Foreign-owned pan-African banks 61.3; Foreign-owned subregional banks 77.1; All banks* 88.9 - Tanzania: 57.4; 79.1; 73.6 - Ghana: 51.5; 63.8; 71.6 - Côte d’Ivoire: 58.2; 63.3; 80.9 - Cameroon: 57.3; 79.6; 90.6 - Uganda: 50.2; 83.4; 79.7 - Zambia: 48.8; 59.7; 69.9 - Mali: 41.1; 57.3; 95.0 - Botswana: 61.7; 73.7; 79.6 - Mozambique: 53.9; 68.3; 69.8 - Burkina Faso: 56.5; 64.3; 93.9 - Note: *Aggregate loan-to-deposit ratio measured using IFS bank credit-to-deposit ratio. - Correspondent banking relationships (CBRs): - Since 2011, sub-Saharan Africa saw a 4 percent decline in the number of active correspondent banks and a 9 percent decline in the number of counterparty countries. - CBR terminations affect banks’ ability to extend credit and transfer international payments; termination at a PAB or regional-bank parent can affect multiple countries. ### Sovereign spread spillovers — South Africa’s dominant role - Principal component analysis: 85 percent of co-movement in frontier market spreads explained by first common factor, strongly correlated with South African indicators (correlation 0.93 with SAVI; 0.94 with South African sovereign spread). - South African spread explains about 6 percent more of variation in frontier market spreads than domestic and global factors alone. - A 100 basis point change in the South African sovereign spread is estimated to be associated with a 20 basis point increase in the average frontier market spread. - Event analysis (December 2015): South Africa sovereign spread jumped within 24 hours of announcements (ratings downgrade, finance minister fired, new finance minister appointed); other countries’ spreads moved in concert. - Panel fixed effects results (January 2012–August 2017, N = 641): - South Africa Spread coefficient examples: 0.20*** (0.04) in early columns; 0.10* (0.05) when controlling for synthetic EMBIG. - SAVI: 0.06*** (0.01) where included. - VIX: 0.12*** (0.03) to 0.10** (0.03). - Oil Price: −0.25*** (0.07) to −0.03 (0.07). - Country spread (t−1): about −0.20* to −0.24** across columns. - Interpretation: South African and global emerging market factors both important in driving sub-Saharan frontier market yields. --- ### 3. Remittances channel — The Changing Pattern of Remittance Flows - Regional remittances account for a third of total remittance inflows and their share is growing as costs decline. - Total remittance inflows to sub-Saharan African countries have remained slightly over 2 percent of GDP over the past 10 years. - Regional remittances accounted for about 35 percent of the region’s total remittance inflows in 2015. - In 27 of the 45 sub-Saharan African countries, regional remittance inflows exceed interregional remittances. - Lesotho, Liberia, and Togo receive more than 5 percent of GDP in remittances from other sub-Saharan African countries. - Examples during commodity price shock: Liberia, Mali, and Nigeria had remittance inflows of 8, 4, and 2 percent of GDP, respectively. - Remittance outflows: - 31 out of 45 send more remittances to the region than to the rest of the world. - Three-quarters of total remittances from sub-Saharan African countries are sent to other countries in the region. - The four largest senders in 2015 accounted for 50 percent of total regional remittances. - Remittances from Chad, Cameroon, Côte d’Ivoire, and Ghana to Nigeria account for 50 percent of received remittances in the region. - Côte d’Ivoire and Ghana are important sources for West Africa; South Africa is main source for Southern and East Africa. - Costs and fintech: - Sub-Saharan Africa is the most expensive destination to send money to; remittance costs in 2017 were about 25 percent higher there than in the rest of world. - Mobile money transfers are two times less expensive than money transfer operators and post offices, and almost three times less expensive than transfers through commercial banks. - Box estimate: a decline in remittance costs to the world average (from 9.4 percent to 7.4 percent) could increase bilateral flows by up to 20 percent (assuming no corridor substitution). - Gravity model for remittances (2010–15): - Distance significantly reduces flows (distance coefficients around −0.25*** for 1000 km in baseline). - Common language, common colony, contiguous countries increase flows significantly. - Supply-side variables: median costs (% of amount sent) coefficient −0.10* (0.06) in Column (4). - Country-fixed-effects specifications show high R-squared (up to 0.93 in a small-sample specification). - Growth spillovers via remittances: - Panel evidence (annual 2010–15): a 1 percent increase in GDP growth in origin countries is associated with a 0.1 percent increase in growth in a receiving country (baseline coefficient 0.0917** (0.0365) in Box 5). - The 0.1 percent association holds for origin countries in the same region; interregional origin growth effects not significant. - Controlling for trade partner growth suggests trade and remittance channels each account for roughly half of the total estimated effect. - Caveats: remittance series short (2010–15), measurement issues, overlap of remittance and trade partners. --- ### 4. Foreign Direct Investment channel — South Africa as Dominant FDI Source - South Africa is the dominant source of regional FDI. - Sectoral composition: - About 75 percent of investment from South Africa to the continent is in services, trade, and the financial sectors. - Outward FDI stock: - Total stock of FDI from South Africa to sub-Saharan African countries was equivalent to 6.8 percent of South African GDP in 2015, up from 4.9 percent in 2010. - In receiving countries, South Africa’s investments represented as much as 3.2 percent of GDP (in Mauritius), with an average of 0.4 percent across all recipient sub-Saharan countries in which it invested in 2015. - Implications: - Countries with high stock of South African FDI converge more rapidly to South African per capita income levels; low bilateral FDI stocks show no evidence of convergence. - Deceleration of South African economy could spill over via lower FDI and lower GDP growth in recipient countries. --- ### 5. Fiscal and policy spillovers — The Fiscal Channel and Unintended Consequences - SACU revenue-sharing (2002 agreement) rules: - One hundred percent of customs revenue distributed on basis of intra-SACU imports. - Eighty-five percent of excise revenue distributed on basis of members’ GDP. - Fifteen percent of excise revenue distributed equally through a development component, with adjustment inversely proportional to member’s GDP per capita. - Formula presented verbatim in source: R_i = a_i C + 0.85 y_i E + 0.15 (1 _ 5) E ( 1– ( h_i – 1 _ 10 ) ) - Fiscal dependence and volatility: - In Swaziland and Lesotho, SACU revenues constitute more than 40 percent and 50 percent of total public revenues, respectively. - Standard deviation of receipts as a share of GDP was between 5 percent and 7 percent for period 2000–16 in smallest SACU members. - Fiscal deterioration in 2016: overall fiscal balance worsened by 9.5 percent of GDP in Lesotho and by 6 percent of GDP in Swaziland. - Lesotho: fiscal deficit excluding regional revenues was 34 percent of GDP in 2016. - Namibia: current account balance excluding SACU transfers at 31 percent of GDP. - Nigeria fuel pricing externalities: - Fuel subsidies are expensive: on average 2 percent of GDP per year. - Benin: only about 15 percent of fuel consumed is purchased on the formal (taxed) market (2008–2012). - Togo: correlation between formal market consumption and price differential vis-à-vis Nigeria is −0.85. - 2011 implied impact for Togo: an implicit subsidy of about 3 percent GDP to Togo, three-quarters captured by smugglers and one-quarter by Togolese consumers. - Policy recommendations: - First-best: institute automatic fuel pricing mechanism to follow international fuel prices more closely. - Second-best: greater cooperation to control borders, harmonize tax policies; caveat that automatic price adjustment may increase tax base erosion if smuggling large and borders porous. --- ### 6. Security and migration channel — The Rising Socioeconomic Impact of Forced Migration - Trends: - Share of forced migration declined through most of 1990s and 2000s but pace of decline has slowed or partially reversed; since 2013 intraregional refugees declined while internally displaced persons (IDPs) rose significantly. - Number of internally displaced persons has risen significantly because of conflicts and violence; terrorism-related events and civil conflicts have more than doubled since 2011 across the Sahel countries, Nigeria, and Cameroon. - Drivers: - Terrorism and civil conflict in the Sahel, Lake Chad area, eastern Democratic Republic of the Congo, Somalia, and South Sudan. - Boko Haram attacks leading cause of displacement in Niger, northeastern Nigeria, Chad, and northern Cameroon. - Collapse of Libyan government and rise of religious extremism increased availability of arms and frequency of terrorist attacks in the Sahel. - Humanitarian and macroeconomic impacts: - United Nations Office for the Coordination of Humanitarian Affairs estimates nearly 30 million people will suffer food insecurity due to the security situation, with almost 12 million at crisis or emergency levels. - Sahel region contended with approximately 4.9 million refugees and internally displaced persons in 2017. - Sahel region humanitarian and financial needs estimated at US$2.7 billion for 2017. - Fiscal costs of hosting displaced populations estimated to range between 1 percent and 5 percent of GDP. - United Nations Development Programme estimates cost of hosting refugees and asylum seekers in Uganda at about US$320 million, or about 1.3 percent of GDP. - Uganda currently hosts more than 1.2 million refugees and asylum seekers. - Forced migration reduces economic activity and imposes humanitarian and fiscal costs from fighting terrorism and hosting displaced persons. - Policy implications: - Address root causes of forced migration (economic and physical insecurity). - Implement systems to accommodate and integrate forced migrants sustainably. - Increased international aid would greatly facilitate accommodation and integration. - Recognize and mitigate fiscal costs of hosting displaced populations through policy planning and international support. --- ### Box and empirical details (selected) - Gravity estimation for bilateral trade and remittances includes controls for distance, common language, common ethnicity, common colony, common currency, origin/destination GDP per capita and population, FX rates, and regional-trend interactions. - Gravity regression excerpts (2010–16 trade flows): - Distance (log) coefficients reported as −1.60*** (0.09) and −1.60*** (0.09) in multiple specifications (Table 1.1 formatting preserved). - EAC trend: 0.04* (0.02); SADC trend: 0.02** (0.01). - Observations in gravity specifications: 92,132; 95,711; 556,476; 95,108 across columns. - Intraregional trade–growth elasticity: - Baseline SSA trading partners’ growth: 0.113* (0.0636) baseline; 0.133** (0.0610) in GMM; 0.0761** (0.0354) in 5-year GMM (Table 2.1 excerpts preserved). - Box 5 (remittance growth spillovers) summary: - Baseline regional remittance partners’ growth coefficient: 0.0917** (0.0365). - Conflict year large negative effects, e.g., −4.980** (1.941) in one specification (Table 5.1 excerpts preserved). - South African outward FDI: - South African outward FDI stock to sub-Saharan Africa = 6.8 percent of South African GDP in 2015 (up from 4.9 percent in 2010). - South African investments in recipient countries up to 3.2 percent of GDP (Mauritius) and average 0.4 percent across recipients in 2015. --- *International Monetary Fund | August 2018 — spillovernote12 (Introduction and Summary; sections 1–2; Box 1; Box 5; figures and tables as presented in source)* ### Introduction and Summary 1 ### Introduction and Summary ### Overview - After close to two decades of strong economic activity, overall growth in sub-Saharan Africa decelerated markedly in 2015–16, to its lowest level in more than 20 years at 1.4 percent. - Heterogeneity: while the largest economies (Nigeria and South Africa) experienced negative or flat growth in 2015–16, a third of countries in the region continued to grow at 5 percent or more during the period. - Growth began to recover in 2017, raising questions about how trends in the largest economies spill over to the rest of the region. - Scope: the note focuses on trade, banking, financial, remittance, investment, fiscal, and security channels as the most prominent transmission mechanisms for regional spillovers. - Main conclusion: the level of interdependence among sub-Saharan African countries is higher than generally assumed, implying a need for additional emphasis on regional surveillance and spillover analysis alongside traditional bilateral surveillance. ### Methodology and contributions - Uses several methodologies and draws on existing studies that identify transmission channels and mechanisms, and extends empirical analysis where literature is silent. - Identifies countries likely to originate spillovers and countries likely to receive them, and provides new empirical estimates of channel sizes. - Documents new channels of transmission not previously identified in the literature. ### High-level quantitative findings - Intraregional trade: rose from 6 percent of total exports (1 percent of GDP) in 1980 to 20 percent of total exports (4 percent of GDP) in 2016. - Concentration: Ten countries account for 65 percent of total regional demand for intraregional exports. - Exposure: Exports to the top 10 destinations represent between 5 percent and 10 percent of source-country GDP for some countries. - Trade integration drivers: econometric estimates suggest about half of the growth in regional trade over 1980–2016 stems from subregional trade integration, particularly within the EAC and the SADC. - Growth spillovers via trade: a 5 percentage point increase in the export-weighted growth rate of intraregional partners is associated with about a 0.5 percent increase in the average sub-Saharan African country’s growth. --- ### 1. Trade channel — Regional Trade Links Gaining Strength - Intraregional trade integration increased substantially over the past 35 years, amplifying potential intraregional spillovers. - Time series: Regional trade represented 6 percent of total exports in 1980 before taking off in the early 1990s and eventually reaching 20 percent in 2016. - Small countries experienced faster increases in trade integration, reflected in the faster growth in the simple average level of trade integration. - Drivers: - Twofold increase in the relative price of commodity exports over 1995–2013. - Volumes of exported commodities increased by two and a half times over 1995–2013. - Strengthening of macroeconomic policies and political and economic institutions over the past 20 years. - Establishment of regional trade agreements and bilateral tariff reductions. - Global comparisons: - Measured as a share of total exports, sub-Saharan Africa exhibits the highest share of intraregional trade integration among emerging and developing regions. - Relative to the size of the economy (percent of GDP), sub-Saharan Africa is in the middle of the pack. - Subregional concentration: - Subregional trade accounts for most intraregional trade in sub-Saharan Africa. - SACU subregional trade alone represents half of total sub-Saharan Africa intraregional trade. - In SADC and SACU, subregional trade represents more than 80 percent of member countries’ intraregional trade. - Trade frictions: - Bilateral trade is more likely to be hindered by distance and socio-cultural differences in sub-Saharan Africa than in the rest of the world, explaining why most regional trade occurs within subregions. --- ### 2. Banking channel — Banking Interdependence Becoming More Subregional - Regional banking links are strengthening, with rising importance of pan-African banks (PABs) and subregional banking groups. - Market participation: the share of PABs and subregional banking groups in sub-Saharan African financial systems is increasing, following a global trend of banking regionalization. - Geography: - Countries that are home to PABs and subregional banks tend to overlap, but recipient countries of banking spillovers are more widely dispersed. - Credit linkages: - Growth in countries that are home to PABs and subregional banks is associated with private sector credit growth in countries where these banks operate. - This private credit growth has reinforcing effects on growth in host countries. - Sovereign spread spillovers: - There are links from changes in the spread on South African sovereign debt to other sub-Saharan African frontier markets. - Global and emerging market financial cycles have a major impact on all issuing sub-Saharan countries, including South Africa, but evidence points to specific spillovers originating from South Africa. --- ### 3. Remittances channel — The Changing Pattern of Remittance Flows - Regional remittances have become relatively more important, outpacing growth of aid, FDI, and remittances from the rest of the world. - Concentration: - Remittance flows are concentrated in a few corridors. - Côte d’Ivoire and Ghana are important sources for West Africa; South Africa is an important source for Southern and East Africa. - Cost and technology: - Recent reductions in the cost to send money across borders are associated with the development of mobile money and explain part of the increase in regional remittances. - Remittance costs in sub-Saharan Africa remain the highest in the world, implying scope for further reductions and increases in flows. - Growth spillovers via remittances: - Growth in countries that send remittances is significantly associated with growth in receiving countries. - A 5 percent increase in the growth of remittance partners is estimated to raise recipient-country growth by 0.5 percent, although this effect is partially offset by trading partners’ growth spillovers. --- ### 4. Foreign Direct Investment channel — South Africa as Dominant FDI Source - South Africa is the dominant source of regional FDI in sub-Saharan Africa. - Corporate patterns: - South African firms seeking diversification into relatively faster-growing regional markets dominate outward investment. - The majority of investment is in services, trade, and the financial sector. --- ### 5. Fiscal and policy spillovers — The Fiscal Channel and Unintended Consequences - Customs unions and revenue-sharing: - The SACU revenue-sharing formula ties member countries’ fiscal revenues to economic developments in South Africa. - While it provides certainty for current revenue, the formula leads to high levels of volatility over the medium term and complicates fiscal management in the smallest members (Lesotho and Swaziland). - Fuel pricing externalities: - Subsidized fuel in Nigeria leads to widespread smuggling and erosion of the tax base in Benin and Togo. - Example statistic: for Benin, only about 15 percent of the fuel consumed is purchased on the formal (taxed) market. --- ### 6. Security and migration channel — The Rising Socioeconomic Impact of Forced Migration - Trends: - The share of forced migration across countries in sub-Saharan Africa declined significantly through most of the 1990s and 2000s, but the pace of decline has slowed or partially reversed. - Drivers: - Terrorism and civil conflict in the Sahel, the Lake Chad area, the eastern Democratic Republic of the Congo, Somalia, and South Sudan are main drivers of involuntary migration. - Spillover impacts: - Main negative spillovers include reduced economic activity, humanitarian damage, and fiscal costs of hosting displaced persons and fighting terrorism. --- ### Key takeaways and policy implications - Regional integration and interdependence are more extensive than generally assumed, with subregional integration moving faster than overall integration. - Spillover channels vary by country: - South Africa: spillovers operate via trade, banking, and remittance channels. - Nigeria: spillovers operate mainly through banking, fuel pricing policy, and trade (especially affecting neighboring countries). - Policy implications: - There is a need for enhanced regional surveillance and spillover analysis in addition to traditional bilateral surveillance to monitor and mitigate both positive and negative spillovers. - Policy attention should include: monitoring subregional banking expansion and cross-border credit links, reducing remittance costs (including leveraging mobile money), addressing fiscal vulnerabilities tied to customs unions and revenue-sharing formulas, and coordinating policies to limit unintended cross-border effects such as fuel subsidies that encourage smuggling. *International Monetary Fund | August 2018 — Introduction and Summary from spillovernote12* ### 1. Intraregional exports ### 1. Intraregional exports ### Overview - Regional demand for intraregional exports is highly concentrated: ten sub-Saharan countries represent 65 percent of total regional demand for intraregional exports. - South Africa alone imports 15 percent of total intraregional exports. - Intraregional exports are concentrated mainly in manufactures, food, and machinery, often in the context of regional or global value chains and re-exports. ### Concentration of Regional Demand and Key Links - Ten sub-Saharan countries = 65 percent of regional demand for intraregional exports. - South Africa imports 15 percent of total intraregional exports. - Intra-SACU trade is significant among members, with exports concentrated in manufactures, food, and machinery. - Major bilateral shares (exporter GDP perspective): - South African imports from Swaziland, Lesotho, Zimbabwe, and Mozambique represent between 4 percent and 11 percent of those exporting economies’ GDP. - Zimbabwe’s total demand for goods from Zambia, Malawi, and Botswana constitutes between 1 percent and 4 percent of these countries’ GDP. - Nigeria, Mali, Ghana, and Burkina Faso import more than 1 percent of GDP of their subregional trading partners (notable sources of subregional spillovers). - For four countries, exports to Nigeria and Mali represent more than 1 percent of their economy. - For 10 countries, exports to South Africa represent more than 1 percent of their GDP. ### Trade Vulnerabilities and Spillover Risk - An economic deceleration in any of the top importing countries (e.g., South Africa, Botswana, Namibia) can weaken demand for intraregional exports and be a source of wider negative spillovers. - Imports absorbed by the top 10 regional importers represent significant shares of the exporters’ economies, strengthening the potential spillover channel. ### Subregional Concentration and Integration - Subregional trade shares (2016): - SADC, SACU, and EAC: subregional trade accounts for more than 70 percent of their total trade with sub-Saharan Africa (member countries are mostly integrated within themselves). - CEMAC and WAEMU: subregional trade represents about half of their intraregional trade. - In absolute terms: - SADC accounts for more than 70 percent of total sub-Saharan Africa intraregional trade. - SACU accounts for more than 50 percent of total sub-Saharan Africa intraregional trade. - WAEMU, EAC, and CEMAC each account for less than 10 percent of total sub-Saharan Africa intraregional trade. - Drivers of subregional integration include geographic proximity, infrastructure constraints, regional trade agreements, and lower nontariff trade barriers within subregions. ### Role of Economic Structure in Exposure - Non-resource-intensive countries (average exposure): - Intraregional exports = 7 percent of GDP. - Intraregional exports = 30 percent of total exports. - Other resource-intensive countries (the “other” group) have the largest share of imports from the region, constituting 30 percent of total imports. - Oil-producing countries: - Exports to the rest of the world = 25 percent of GDP. - Intraregional exports = 1.5 percent of GDP. - Implication: oil-producing countries are relatively less likely to suffer intraregional spillovers through the trade channel. ### Gravity Model Findings (1980–2016; robustness for 2010–16 in Box 1) - Estimated specification highlights: - Distance significantly hampers trade flows across countries. - Exports increase significantly with both population and GDP per capita of origin and destination countries. - Exports are higher between partners that share a common language, ethnicity, and colonial heritage. - Bilateral exchange rates do not have a significant effect on bilateral trade flows. - Sub-Saharan Africa specific findings: - Distance is a greater hindrance when exporting to sub-Saharan destinations. - Having belonged to the same colony is a greater benefit within sub-Saharan Africa. - Sharing a language and colonial background increases exports in sub-Saharan Africa relative to other regions. - The distance between two sub-Saharan countries deters trade significantly more than between other countries (Wald test p-value = 0.02). - Effect of subregional economic unions (time-interacted corridor specification): - Trade among members of the EAC increased by an additional 4 percent per year on average. - Trade among members of the SADC increased by an additional 2 percent per year on average. - Using these estimates, average annual growth in regional trade would have been about 9 percent instead of 11 percent (implying trade levels without subregional integration would be about half as low as observed in 2015). - The interaction between distance and time suggests that in sub-Saharan Africa, distance has increasingly become a barrier, indicating infrastructure facilitating trade across unions has lagged relative to infrastructure within unions. *International Monetary Fund | August 2018* ### Box 1. Gravity Equation Estimation for 2010–16 Trade Flows (continued) ### Box 1. Gravity Equation Estimation for 2010–16 Trade Flows (continued) ### Determinants of Trade Flows (Table 1.1: dependent variable = logarithm of bilateral trade flows) - Contiguous countries: 1.57*** (0.21); 1.53*** (0.21); 0.48*** (0.11); 20.00 (0.01) - Distance (in log): 21.60*** (0.09); 21.60*** (0.09); 21.60*** (0.04); 0.18 (0.11); 20.40*** (0.11); 20.02*** (0.00) - Common language: 0.46*** (0.15); 0.45*** (0.14); 0.54*** (0.10); 0.17 (0.14); 20.11 (0.15); 0.01 (0.01) - Common ethnicity: 0.24* (0.14); 0.30** (0.13); 0.21** (0.09); 20.27** (0.14); 0.21* (0.12); 0.01 (0.01) - Belonged to common colony: 1.44*** (0.13); 1.37*** (0.12); 0.81*** (0.11); 1.16*** (0.20); 0.74*** (0.19); 20.00 (0.01) - Common religion: 0.14 (0.14); 0.14 (0.13); 0.25*** (0.06); 20.01 (0.16); 20.04 (0.13); 0.00 (0.01) - Common currency: 1.22*** (0.31); 1.28*** (0.31); 0.40 (0.29); 0.99** (0.39) - Origin GDP p.c.: 0.49*** (0.06) - Destination GDP p.c.: 0.48*** (0.13) - Origin population: 0.70** (0.30) - Destination population: 2.41*** (0.24) - Origin/destination FX rate: 0.00 (0.01) - WAEMU trend: 0.00 (0.01) - CEMAC trend: 0.00 (0.01) - EAC trend: 0.04* (0.02) - SADC trend: 0.02** (0.01) - SACU trend: 0.09 (0.07) - Observations: 92,132; 95,711; 556,476; 95,108 (across columns) - R-squared: 0.52; 0.57; 0.73; 0.77 (across columns) - Year FE: YES / NO / NO / NO (by specification) - Country FE: YES / NO / NO / NO (by specification) - Country-time FE: NO / YES / YES / YES (by specification) - Country-pair FE: NO / NO / YES (by specification) - Note on inference: Clustered standard errors in parentheses (Destination country). Significance: ***p < 0.01, ** p < 0.05, *p < 0.1. ### Intraregional Trade and Growth: Key Findings - A panel regression for sub-Saharan African countries for 1980–2016 indicates: - A 1 percentage point increase in the export-weighted growth rate of intraregional partners is associated with about a 0.11 percent increase in the average sub-Saharan African country’s growth (baseline; Table 2.1, column 1). - A 1 percentage point increase in the growth rate of trading partners outside the region is associated with a 0.34 percent increase in the growth rate of the average sub-Saharan African country (baseline). - Comparison with other regions: - Sub-Saharan Africa has a slightly lower intraregional elasticity of growth compared with Latin America and emerging and developing Asia, but a higher elasticity than Middle East and North Africa (Box 2 results). ### Model Specification and Controls (equation (1)) - Model: RealGDPgrowth_it = α + γ_i + β1 RealGDPgrowth_{t−1,i} + β2 RealGDPgrowth_{t−2,i} + β3 SSAtradingpartnergrowth_{t,i} + β4 Non−SSAtradingpartnergrowth_{t,i} + β5 X_it + e_it - Weights: export shares in the previous year (lagged, from IMF Direction of Trade Statistics). - Control vector X includes: - Lags of the dependent variable - Rate of investment to GDP - Inflation rate - Level of trade openness - Occurrence of conflict and war (Uppsala Conflict Data Program) - Fluctuations in the terms of trade - Degree of trade openness and share of regional exports in total exports - Change in the US Federal Funds rate (measure of international liquidity) - Country inflation rate - Change in the bilateral exchange rate with respect to the US dollar - Estimation sample: all sub-Saharan African countries over 1980–2016. - Fixed effects and inference: Country fixed effects γ_i included; standard errors clustered at the country level. ### Robustness Checks and Alternative Specifications (Table 2.1) - Robustness checks performed: - Excluding the 10 percent largest economies (Angola, Ghana, Kenya, Nigeria, South Africa) — Table 2.1, column 6. - Adding interaction between countries’ openness and partners’ average growth — Table 2.1, column 7. - Five-year averages specification including initial level of GDP per capita — Table 2.1, column 8. - Panel Generalized Method of Moments (GMM) to address endogeneity — Table 2.1, column 5. - Robustness results (selected coefficients and diagnostics from Table 2.1): - Real GDP growth (t–1): 0.318*** (0.0821) baseline; ranges to 0.410*** (0.0632) in 5-year GMM. - Real GDP growth (t–2): 20.0180 (0.0202) baseline; up to 20.0863* (0.0454) in exposure heterogeneity specification. - SSA trading partners’ growth: 0.113* (0.0636) baseline; 0.133** (0.0610) in GMM; 0.110* (0.0622) excluding largest; 0.0761** (0.0354) in 5-year GMM. - Non–SSA trading partners’ growth: 0.348* (0.184) baseline; becomes 0.0118 (0.0437) in 5-year GMM. - Conflict year (Uppsala): 24.548*** (0.900) baseline; consistently large and negative across specifications. - Terms of trade, percent change: 0.138*** (0.0457) baseline; generally positive and significant in several specifications. - Trade openness (t–1): 0.0524* (0.0305) baseline. - Share of regional exports in total exports (t–1): 3.108* (1.604) baseline; varying significance across specifications. - Observations: 1,345 baseline; 1,252 excluding 10% largest; 159 in 5-year GMM. - R2 (reported): 0.159 baseline; up to 0.188 with structural controls. - Number of countries: 45 baseline; 43 in some specifications. - Interpretation: - Results are robust to inclusion of structural and monetary controls. - GMM estimation supports robustness to dynamic-panel endogeneity. - Excluding large economies does not materially change the intraregional elasticity. - Interaction variables between openness and partner growth are insignificant, suggesting more open economies are not significantly more exposed to partner-demand variations. - Using five-year averages yields a smaller but still important coefficient for regional trading partner growth. ### Cross-Region Comparison (Table 2.2) - SSA trading partners’ growth coefficients by region (selected): - SSA: 0.133** (0.0610) - Latin America: 0.157** (0.0693) - MENA: 20.00140 (0.0629) - Asia: 0.157* (0.0921) - Europe and CIS: 0.204 (0.163) - Observations and sample sizes: - SSA: Observations 1,252; Number of countries 45 - Latin America: Observations 941; Number of countries 30 - MENA: Observations 560; Number of countries 20 - Asia: Observations 494; Number of countries 18 - Europe and CIS: Observations 420; Number of countries 18 - Interpretation: - Sub-Saharan Africa’s intraregional elasticity of growth is lower than Latin America and Asia, higher than MENA. - MENA results may reflect oil market developments. ### Additional empirical notes and diagnostics - Significance conventions: Robust standard errors in parentheses; ***p < 0.01, **p < 0.05, *p < 0.1. - Data sources cited in tables: IMF, Direction of Trade Statistics database, World Economic Outlook database; and IMF staff calculations. *Sources: IMF, Direction of Trade Statistics database, World Economic Outlook database; and IMF staff calculations.* ### 2. Deposits in host countries as a percent of host country deposits ### 2. Deposits in host countries as a percent of host country deposits ### Deposits and assets trends (2007–16) - Chart axes and labels appearing in the source: - Percent axis values: –60, 40, 140, 240, 340, 440, 540, 640, 740 - USD billion axis values: 0, 2, 4, 6, 8, 10, 14, 12 - Year span shown: 2007, 08, 09, 10, 11, 12, 13, 14, 15, 16 - Key descriptive finding: - The share of PABs and subregional banking groups in the total sub-Saharan African financial system is increasing, following a global trend of banking regionalization (IMF 2015b). - Nevertheless, assets and deposits have recently declined, consistent with the decline in regionwide assets that coincided with the economic deceleration in the region. ### Country exposure to pan-African banks (PABs) and subregional banks - Concentration of parent banks: - PAB parent banks are concentrated in three countries, with South Africa and Togo each home to about 40 percent of all PABs. - Subregional parent banks are only about half as concentrated as PABs. - Host-country patterns: - PAB foreign subsidiaries are spread across the region. - Countries that host subregional banks are notably more concentrated than PAB host sets. - Examples of geographic clustering by home country of subregional banks: - South African subregional bank subsidiaries: mostly in SADC countries. - Kenyan subregional banks: mostly in EAC countries. - Nigerian subregional banks: mainly in West African countries. - Gabonese and Cameroonian subregional banks: mostly in neighboring CEMAC countries. - Special case noted: - Ecobank in Togo: Only the holding company, Ecobank Transnational Incorporated, is headquartered in Togo; it has status and privileges of a nonresident supranational financial institution. The de facto economic headquarters is in Nigeria, where its largest subsidiary is located (IMF 2015c). ### Cross-border linkages and spillover channels - Direction and mechanisms of spillovers: - Spillovers run in both directions between parent banks and their subsidiaries, and across subsidiaries/branches of the same banking group. - Transmission channels include placement of deposits and credit, governance deficiencies, perceptions of mismanagement, or reputational concerns at the group level. - Parent banks may be exposed to risks in host countries depending on: - the systemic importance of foreign subsidiaries/branches to the local economy, - liquidity-sharing arrangements across the banking group, - the size of the foreign subsidiary or branch relative to the group. - Subsidiary model reduces but does not eliminate contagion risk; subsidiaries may still have exposure to parents or other group entities (Mecagni, Marchettini, and Maino 2015). - Syndicated loans between subsidiaries or branches can be important sources of cross-border risk. - Host countries of systemic PABs and subregional bank subsidiaries/branches may face risks from unilateral or uncoordinated actions by home authorities or parent banks, affecting host-country financial stability. ### Systemic importance of foreign-owned banks in host countries - Measure described: - Ratio of total deposits in foreign African subsidiaries or branches of PABs or subregional banks to total deposits by country (systemic importance measure). - Patterns and findings: - The ratio is highest in small countries (Figure 17). - The degree of systemic importance is larger for PAB subsidiaries/branches than for subregional banks. - Historically, spillovers from banking crises in African countries were often limited because: - foreign subsidiaries in host countries were mainly funded by local deposits and did not significantly depend on parent funding (for example, banks headquartered in Nigeria and South Africa), or - the foreign entities were not systemic (IMF 2012, 2015c). - Nonetheless, a parent bank shock transmitted across borders could have real effects on a host economy if the foreign subsidiary or branch is systemically important; such events may be difficult for host country policymakers to foresee. ### Cross-border banking, financial deepening, and growth - Long-run financial deepening: - Financial markets in sub-Saharan Africa have gradually deepened as measured by credit to the private sector (Figure 18.1). - Deepening is especially pronounced in countries that are home to PABs and subregional banks compared with those primarily hosting these groups. - Association with growth: - Evidence shows financial development and deepening have supported growth and reduced growth volatility in sub-Saharan Africa (IMF 2016a). - There is a strong positive association between GDP growth and private credit growth (Figure 18.2). - Potential for cross-border growth spillovers: - Depending on internal funding arrangements, lower growth in parent-bank home countries could reduce credit and deposit growth in their foreign African subsidiaries/branches if the parent supplies significant liquidity. - Evidence suggests bank funding is mostly local in the largest countries (IMF 2012, 2015c). - Conversely, lower growth in host countries limits cross-border expansion prospects for parents and constrains subsidiaries’ ability to repatriate excess liquidity to parents, which can limit credit growth in the parent country. - Commodity-related channels: - Countries severely hit by the commodity price decline experienced credit growth deceleration and a decline in deposits (Figure 19) (Agrawal, Duttagupta, and Presbitero 2017). - This partly reflects high exposure of African banks to the commodity sector (IMF 2017c). - Reinforcing factors such as a slowdown in economic activity and a buildup in government arrears to contractors can exacerbate deposit and credit declines. *Sources: Fitch Connect; IMF, International Financial Statistics; and IMF staff calculations.* ### 2. Private Credit Growth and GDP Growth in Sub-Saharan Africa, 2008–15 ### 2. Private Credit Growth and GDP Growth in Sub-Saharan Africa, 2008–15 (Average) ### African Bank Behavior and Nonbank Financial Interlinkages - PABs (foreign-owned pan-African banks) and foreign-owned subregional banks are relatively less active lenders in host countries: - Across hosting countries, the average loan-to-deposit ratio for PABs is about 34 percent less than the country-level average. - Subregional banks have average loan-to-deposit ratios about 22 percent less than country-level averages. - These patterns reflect supervisory limits, a preference to act as deposit-taking institutions with limited lending to the private sector, and greater exposure to sovereigns. - PABs and subregional banks have increasingly integrated nonbank activities, increasing spillover potential—particularly in the southern part of the continent where activities have a regional scope (insurance and securities dealing). - Country-specific examples of NBFI prominence: - Namibia: nonbank financial institutions have gross assets equivalent to 330 percent of GDP (four times those of traditional banks); shadow banking sector about 40 percent of the entire financial sector. - South Africa: NBFIs hold about two-thirds of all financial assets; pension funds hold assets equal to 110 percent of GDP (versus banking assets equal to 112 percent); long-term insurers hold assets equal to 64 percent of GDP. - PABs generally have large ownership shares in NBFIs, increasing the risk of spillovers from real economy developments to banking sectors; these risks are compounded by lack of regulation in the nonbank financial sector and low levels of compliance in the banking sector. ### Correspondent Banking Relationships (CBRs) and Spillover Channels - Withdrawal of CBRs in sub-Saharan Africa has been significant since 2011: - Since 2011, sub-Saharan Africa has seen a 4 percent decline in the number of active correspondent banks. - Since 2011, sub-Saharan Africa has seen a 9 percent decline in the number of counterparty countries. - Drivers of the decline include weaknesses in controls at respondent banks, inadequate supervision and regulation, country risk, and profitability. - Effects of CBR terminations: - A termination affects a bank’s ability to extend credit and transfer international payments, with direct effects on growth, trade, and internal and external stability. - If a CBR is terminated with a PAB or regional banking group parent, the impact can be felt across the entire group and multiple countries. - Table 1 (Loan-to-Deposit Ratios, Largest Sub-Saharan African Countries, 2015) — selected entries (Percent): - Kenya: Foreign-owned pan-African banks 61.3; Foreign-owned subregional banks 77.1; All banks* 88.9 - Tanzania: 57.4; 79.1; 73.6 - Ghana: 51.5; 63.8; 71.6 - Côte d’Ivoire: 58.2; 63.3; 80.9 - Cameroon: 57.3; 79.6; 90.6 - Uganda: 50.2; 83.4; 79.7 - Zambia: 48.8; 59.7; 69.9 - Mali: 41.1; 57.3; 95.0 - Botswana: 61.7; 73.7; 79.6 - Mozambique: 53.9; 68.3; 69.8 - Burkina Faso: 56.5; 64.3; 93.9 - Note: *Aggregate loan-to-deposit ratio measured using IFS bank credit-to-deposit ratio. ### Dominant Role of South Africa in Sovereign Spread Spillovers - Evidence of cross-country co-movement in frontier market sovereign spreads: - Principal component analysis: 85 percent of the co-movement in frontier market spreads is explained by their first common factor. - The first common factor is strongly correlated with South African indicators: - Correlation coefficient of 0.93 against the implied volatility index of the Johannesburg stock exchange (SAVI). - Correlation coefficient of 0.94 against the South African sovereign spread. - Estimated impact of South African spread on frontier markets: - The South African spread explains about 6 percent more of the variation in frontier market spreads than domestic and global factors (as measured by the CBOE Volatility Index (VIX)) alone. - A 100 basis point change in the South African sovereign spread is estimated to be associated with a 20 basis point increase in the average frontier market spread. - Global emerging market trends also influence sub-Saharan frontier market spreads: - When controlling for an index of emerging market bond spreads, the estimated impact of changes in South Africa’s spread remains positive and significant, though quantitatively smaller depending on the index used. - Event analysis (December 2015 South African news events) shows: - December 4: Ratings downgrade by Standard & Poor’s - December 9: Finance minister fired - December 13: New finance minister appointed - South Africa’s sovereign spread jumped within 24 hours of each announcement and continued for up to 72 hours; other countries’ spreads moved in concert with South Africa immediately following announcements. - Panel fixed effects model specification for sovereign yield spreads (first differences, monthly data January 2012–August 2017): - spread_it = α + γ_i + β1 spread_{t−1,i} + β2 ZAF_t + β3 Global_t + β4 X_it + e_it - ZAF corresponds to South African factors (SAVI or sovereign spread). Global includes VIX, oil prices, and—depending on columns—MSCI Emerging Markets index or a synthetic EMBIG. - X includes inflation, exchange rate relative to the US dollar, and an index of financial stress. - Estimation results (Table 3.1) — selected coefficients and significance (standard errors in parentheses): - Country spread (t−1): values range 20.20* (0.10) to 20.24** (0.09) across columns. - FSI: 0.02 (0.01) to −0.03 (0.02), with one column showing −0.03* (0.02). - Oil Price: −0.25*** (0.07) to −0.03 (0.07) across columns. - VIX: 0.12*** (0.03) to 0.10** (0.03). - Exchange Rate: 0.51** (0.18) to 0.17 (0.16) across columns. - South Africa Spread: 0.20*** (0.04) in early columns, 0.10* (0.05) when controlling for synthetic EMBIG. - SAVI: 0.06*** (0.01) where included. - MSCI: 0.03 (0.03) where included. - Synthetic EMBIG: 0.16*** (0.03) where included. - FE: Yes; N = 641 (626 in column with synthetic EMBIG); R2 (adj) ranges 0.25 to 0.32. - Interpretation: - South African factors combined explain 5 percent more of the variation in spreads (versus without them), while global emerging market factors explain up to an additional 2 percent of the variation. - Results indicate importance of both regional (South Africa) and global emerging market-specific factors in driving sub-Saharan African frontier market yields. ### Remittances: Composition, Trends, and Regional Exposure - Regional remittances among sub-Saharan African countries are relatively large: - Regional remittances account for a third of total remittance inflows, and their share is growing in parallel with declining costs. - Total remittance inflows to sub-Saharan African countries have remained constant at slightly over 2 percent of GDP over the past 10 years. - Remittances among sub-Saharan African countries have grown faster than those from the rest of the world in the past five years. - Regional remittances accounted for about 35 percent of the region’s total remittance inflows in 2015. - Measured as a share of GDP, total remittance inflows in sub-Saharan Africa are larger than those in other emerging and developing regions. - The relative importance of intraregional inflows in sub-Saharan Africa is the third highest globally, after the Commonwealth of Independent States (CIS) and the Middle East and North Africa. - Country-level exposure: - In 27 of the 45 sub-Saharan African countries, regional remittance inflows exceed interregional remittances. - Lesotho, Liberia, and Togo receive more than 5 percent of GDP in remittances from other sub-Saharan African countries. - Examples during the commodity price shock: Liberia, Mali, and Nigeria had remittance inflows of 8, 4, and 2 percent of GDP, respectively. - Remittance outflows and concentration: - Most remittance outflows from sub-Saharan African countries are sent to other countries in the region: 31 out of 45 send more remittances to the region than to the rest of the world. - Three-quarters of total remittances from sub-Saharan African countries are sent to other countries in the region. - Remittance outflows originate in a few countries: the four largest senders in 2015 accounted for 50 percent of total regional remittances. - Remittances from Chad, Cameroon, Côte d’Ivoire, and Ghana to Nigeria alone account for 50 percent of received remittances in the region. - Côte d’Ivoire and Ghana are important sources for West Africa; South Africa is the main source for Southern and East Africa. - Drivers and barriers: - Remittances are larger for geographically and culturally close countries (gravity equation results). - Geographical distance appears to be a greater barrier in sub-Saharan Africa because of higher travel and sending costs relative to migrants’ incomes. - Higher sending costs are associated with lower remittance flows, even after controlling for distance and origin/destination fixed effects. - Implication: - Because remittances tend to flow to poorer and more connected countries on a net basis, recipient countries can be relatively exposed to regional spillovers—both stabilizing (resource redistribution) and distress-transmitting—depending on the economic fortunes of remittance-sending countries. *Italic: Source: spillovernote12 - 2. Private Credit Growth and GDP Growth in Sub-Saharan Africa, 2008–15, spillovernote12 - 2. Private Credit Growth and GDP Growth in Sub-Saharan Africa, 2008–15 (PDF).* ### 1. Selected External Flows ### 1. Selected External Flows ### Remittance inflows: patterns and magnitudes - Figures and sources: - Figure 21 and Figure 22 summarize external flows and remittances for sub-Saharan Africa and remittance inflows in emerging and developing countries, 2010–15. - Sources: IMF, World Economic Outlook database; World Bank, World Development Indicators, Migration and Remittances database. - Note: Remittances in the World Bank databases are measured as the sum of three items in the IMF’s Balance of Payments Statistics Year Book: (1) personal transfers, (2) compensation of employees, and (3) migrants’ transfers. - Regional shares and corridors: - Intraregional and interregional shares and average inflows/outflows are reported for 2010–15 (country-level lists shown in figures). - The Gambia: diaspora about 5 percent of the population lives abroad; two out of three Gambians who graduate from foreign universities stay abroad. Large outflows essentially directed to Senegal. - Data coverage and measurement: - Only officially recorded remittances sent through formal channels are recorded, which explains omissions (for example, Zimbabwe and Chad not included in some graphs). ### Gravity model estimation for bilateral remittance flows (2010–15) - Model specifications: - Baseline (equation (1)): log F̄ij = α0 + β Xij + γ Ȳi + δ Z̄j + εij, where F̄ij is log average remittance flow, Xij corridor-specific variables (distance, exchange rate, common language, ethnicity, colonial origin, religion), Ȳi and Z̄j are log average GDP per capita and log average population of origin and destination. - Country fixed effects specification (equation (2)): log F̄i,j = αi + γj + β Xij + εij (country-level population and GDP dropped). - SSA interaction specification (equation (3)): log F̄ij = αi + γj + β Xij + θo Ii∈SSA Xij + θd Ij∈SSA Xij + εij to test differential distance effects for sub-Saharan Africa origin/destination countries. - Column 4 extends (2) by adding median cost of sending US$200 (% of US$200) and median number of days for transfer (supply-side variables) using the Remittance Price database subset. - Key estimation findings (Table 4.1, dependent variable: logarithm of average bilateral remittance flow): - Contiguous countries: - Column (1): 2.61*** (0.14) - Column (2): 2.40*** (0.16) - Column (3): 2.09*** (0.19) - Column (4): 0.56 (0.36) - SSA origin subcolumn: 0.94 (0.80) - SSA destination subcolumn: (no separate contiguous value reported beyond above) - Distance (1000 km): - Column (1): 20.25*** (0.01) - Column (2): 20.24*** (0.01) - Column (3): 20.22*** (0.01) - Column (4): 20.19*** (0.03) - SSA origin subcolumn: 20.11*** (0.02) - SSA destination subcolumn: 20.01 (0.05) - Common language: - Column (1): 1.52*** (0.14) - Column (2): 1.07*** (0.17) - Column (3): 0.93*** (0.21) - Column (4): 20.11 (0.33) - SSA origin subcolumn: 0.54** (0.24) - SSA destination subcolumn: 20.05 (0.41) - Common ethnicity: - Column (1): 0.64*** (0.14) - Column (2): 0.48*** (0.16) - Column (3): 0.65*** (0.19) - Column (4): 0.09 (0.34) - SSA origin subcolumn: 20.45* (0.23) - SSA destination subcolumn: 0.53 (0.51) - Belonged to common colony: - Column (1): 1.58*** (0.14) - Column (2): 1.50*** (0.22) - Column (3): 1.59*** (0.22) - Column (4): 20.94*** (0.31) - SSA origin subcolumn: 0.57* (0.31) - SSA destination subcolumn: 1.22*** (0.39) - Common religion: - Column (1): 0.11* (0.07) - Column (2): 0.61*** (0.10) - Column (3): 0.70*** (0.11) - Column (4): 20.14 (0.34) - SSA origin subcolumn: 20.11 (0.26) - SSA destination subcolumn: 20.20 (0.48) - Origin GDP p.c.: 1.04*** (0.02) (Column (1) only) - Destination GDP p.c.: 0.63*** (0.02) (Column (1) only) - Origin population: 0.67*** (0.01) (Column (1) only) - Destination population: 0.82*** (0.01) (Column (1) only) - Origin/destination FX rate: 20.02** (0.01) (Column (1) only) - Median costs (% of amount sent): 20.10* (0.06) (Column (4)) - Median completion time (days): 20.15 (0.12) (Column (4)) - Observations: - Column (1): 10,704 - Column (2): 10,814 - Column (3): 10,814 - Column (4): 220 - R-squared: - Column (1): 0.54 - Column (2): 0.81 - Column (3): 0.81 - Column (4): 0.93 - Country fixed effects included in Columns (2)–(4): YES - Note: Clustered standard errors in parentheses (Destination country). ***p < 0.01, **p < 0.05, *p < 0.1. - Interpretation of gravity model results: - Distance measures significantly reduce remittance flows across countries. - Remittance flows increase significantly with both population and GDP per capita of both origin and destination countries in the country-control specification; increase more with the GDP per capita of the origin country (implying net flows toward poorer countries increase with differences in GDP per capita). - Interaction results: distance is a greater hindrance for both origin and destination sub-Saharan countries, except for country pairs that belonged to the same colony. - Supply-side variables: median transfer costs significantly reduce bilateral flows (median costs coefficient 20.10*). ### Remittance costs, fintech, and potential increases in flows - Regional remittance costs and trends: - Sub-Saharan Africa is the most expensive destination to send money to (Remittance Price Database Report 2017). - Remittance costs in 2017 were about 25 percent higher there than in the rest of world. - Mobile money transfers are: - two times less expensive than those at money transfer operators and post offices, and - almost three times less expensive than transfers through commercial banks. - Figure 25: Percentage cost of sending US$200 across region and over time (series spanning 2011:Q1–2017:Q2). - Figure 26: Total average cost by remittance sending provider (Bank, Money transfer operator, Post office, Mobile operator). - Potential impact of cost reductions: - Based on Box 4 estimates and assuming no substitution across corridors, a decline in remittance costs to the world average (from 9.4 percent to 7.4 percent) could result in increases in bilateral flows of up to 20 percent. - Role of fintech: - Expansion of mobile money technology has contributed to decreasing costs over the past 10 years and can further increase remittance flows as coverage and usage increase across sub-Saharan Africa. ### Growth spillovers through remittances: panel evidence - Panel model specification (equation (6)): - RealGDPgrowthit = α + γi + β1 RealGDPgrowtht−1,i + β2 regionalremitpartners’growth + β3 extraregionalremitpartners’growth + β4 Xit + β5 Ii∈SSA * regionalremitpartners’growth + β6 Ii∈SSA * extraregionalremitpartners’growth + eit. - Growth averages weighted by share of lag remittance inflows. - Sample: annual data 2010–15 from World Bank Migration and Remittances database; includes all countries of the world to increase statistical power. - Controls include country-specific controls, country fixed effects γi, share of regional remittances in total remittance inflows, and share of remittance inflows in GDP. - Estimation approaches: fixed effects, Arellano-Bond for serial correlation and endogeneity where lagged dependent variable included. - Key empirical findings: - A 1 percent increase in GDP growth in origin countries is associated with a 0.1 percent increase in growth in a receiving country (Box 5). - The 0.1 percent association holds for origin countries that belong to the same region; interregional origin growth effects are not significant, possibly because interregional remittances are dominated by advanced economies with limited variation. - Regression results indicate growth spillovers through the remittance channel are significant for all countries in the world and are not different in sub-Saharan Africa (columns 1–3 of Table 5.1). - In specifications controlling for growth of trade partners (column 6), some spillovers captured by remittance variables may reflect trade spillovers; a Wald test yields p-value below 0.02 for joint significance of average growth of trade and remittance partners. - Estimated values suggest trade and remittance channels have similar magnitude, each accounting for half of the total effect identified in baseline estimation. - Caveats and limitations: - Remittance flow series are short (2010–15), reducing estimation efficiency. - Small sample sizes and imperfect measurement of remittance flows may weaken significance in some specifications. - For many countries, remittance and trade partners overlap; data limitations hinder precise disentanglement of channels. ### Foreign direct investment (FDI) channel: intraregional FDI and firm-level spillovers - Intraregional FDI shares: - For Togo, Rwanda, Guinea-Bissau, and Botswana, inward FDI from sub-Saharan Africa constitutes more than 40 percent of their total stock of FDI (Figure 27). - Firms from South Africa, Kenya, and Nigeria have large presences in other sub-Saharan markets; South African firms have more than 2,400 subsidiaries in other African countries. - Implications of multinational firms: - Potential positive spillovers: knowledge transfer, leveraging comparative advantage, diversification, economies of scale; foreign subsidiaries in high-growth countries can compensate for weaker performance at headquarters (example: foreign subsidiaries mitigated Nigeria slowdown effects). - Potential negative spillovers/risks: 1. If a firm is systemically important, performance of headquarters or foreign subsidiaries can have macroeconomic implications for host countries. 2. A firm’s borrowing ability can be affected by exposure to sovereign risk (surges in sovereign spreads can transmit to firm financing). 3. (Additional risk types are noted qualitatively in the source but not enumerated with numeric values in this excerpt.) *International Monetary Fund | August 2018* ### Box 5. Spillover Effects from Countries Sending Remittances ### Box 5. Spillover Effects from Countries Sending Remittances ### Empirical findings (Table 5.1: Dependent variable: Real GDP growth) - Regional remittance partners’ growth: - Baseline coefficient: 0.0917** (standard error 0.0365) - Other specifications: 0.0587 (0.0431); 0.0999** (0.0503); 0.0971* (0.0516); 0.109** (0.0450); 0.0588 (0.0692) - Extraregional remittance partners’ growth: - Coefficients in columns: 20.129 (0.179); 0.101 (0.240); 20.0114 (0.234); 20.0907 (0.261); 20.0378 (0.341); 20.0976 (0.234) - Conflict year (Uppsala database): large negative effects across specifications, e.g., 24.980** (1.941); 24.878** (1.899); 25.081*** (1.283); 25.327*** (1.104) - Share of regional remittances in total inflows (t-1): negative and sometimes significant, e.g., 20.0356 (0.0257); 20.0347 (0.0254); 20.0523** (0.0256); 20.0587** (0.0276); 20.0536** (0.0270); 20.0515** (0.0251) - Share of remittances inflows in GDP (t-1): mixed estimates, e.g., 20.0626 (0.0900); 20.0250 (0.0958); 0.127 (0.118); 0.136 (0.123); 0.0728 (0.178); 0.128 (0.119) - Region average growth: consistently positive and highly significant, e.g., 0.424*** (0.138); 0.485*** (0.164); 0.501*** (0.165); 0.500*** (0.172); 0.474*** (0.162) - Percent change in population: positive and sometimes significant, e.g., 0.524* (0.286); 0.615** (0.297); 0.751** (0.323); 0.726** (0.350); 0.708** (0.324); 0.687** (0.306) - Percent change in US Federal Funds rates: positive in some specifications, e.g., 0.0100** (0.00482); other columns: 0.00718 (0.00471); 0.00588 (0.00405); 0.00741 (0.00470); 0.00588 (0.00410); 0.00624 (0.00403) - Inflation: generally negative, e.g., 20.152 (0.127); 20.180 (0.125); 20.225* (0.121); 20.223* (0.135); 20.224* (0.124); 20.223* (0.121) - Trade channel controls (column 6): Regional trading partners’ growth 0.0497* (0.0302); Extraregional trading partners’ growth 0.480* (0.255) - Sample and fit: - Observations: 565; 565; 448; 393; 448; 448 (by column) - R-squared: 0.161; 0.181 (first two reported) - Number of countries: 117; 117; 117; 103; 117; 117 - Note: Robust standard errors in parentheses. ***p  0.01, **p  0.05, *p  0.1. ### South African outward FDI and growth spillovers - Sectoral composition: - About 75 percent of investment from South Africa to the continent is in the services, trade, and financial sectors. - Outward FDI stock: - Total stock of FDI from South Africa to sub-Saharan African countries was equivalent to 6.8 percent of South African GDP in 2015, up from 4.9 percent of GDP in 2010. - In receiving countries, South Africa’s investments represented as much as 3.2 percent of GDP (in Mauritius), with an average of 0.4 percent across all sub-Saharan African countries in which it invested in 2015. - Perceptions and growth effects: - Survey evidence: about 80 percent of sub-Saharan Africans who interact with South African firms find them to have a better reputation than local firms in the same industry (DNA Economics 2013). - Empirical evidence: countries with a high stock of South African FDI converge more rapidly to South African per capita income levels; countries with low bilateral FDI stocks vis-à-vis South Africa show no evidence of convergence (Dunne and Masiyandiam 2015). - Policy-relevant implication: - The deceleration of the South African economy could spill over to other countries that have large stocks and flows of South African FDI and could manifest as both lower FDI and lower GDP growth in these countries. ### SACU revenue-sharing mechanism and fiscal/external spillovers - Revenue-sharing rules (2002 agreement, as described): - One hundred percent of customs revenue is distributed on the basis of intra-SACU imports. - Eighty-five percent of excise revenue is distributed on the basis of members’ GDP. - Fifteen percent of excise revenue is distributed equally through a development component, with an adjustment inversely proportional to the member’s GDP per capita. - Formula as presented: - R_i = a_i C + 0.85 y_i E + 0.15 (1 _ 5) E ( 1– ( h_i – 1 _ 10 ) ) [formula shown verbatim in source text] - Explanation: C refers to custom duties and E to excises; a_i is the value at the border of imports to the country from all other SACU members, less re-exports, divided by the value of imports less re-exports for all SACU countries; y_i is the share of GDP of the country in the SACU GDP; h_i the level of GDP per capita in the country divided by the average across SACU members. - Fiscal dependence and volatility: - In Swaziland and Lesotho, SACU revenues constitute more than 40 percent and 50 percent of total public revenues, respectively. - Volatility example: the standard deviation of receipts as a share of GDP was between 5 percent and 7 percent for the period 2000–16 in the smallest SACU members. - Fiscal deterioration in 2016: overall fiscal balance worsened by 9.5 percent of GDP in Lesotho and by 6 percent of GDP in Swaziland. - Lesotho: fiscal deficit excluding regional revenues was 34 percent of GDP in 2016. - Namibia: current account balance excluding SACU transfers at 31 percent of GDP (importance of transfers to foreign exchange inflows and reserves). - Mechanism-related observation: - Transfers in any given year correspond to the forecasted value a year earlier; discrepancies between forecast and actuals are compensated the following year, which can increase short-term predictability but raise medium-term variance of SACU transfers beyond the variance of the underlying revenue pool. ### Unintended spillovers from Nigeria’s fuel pricing policies - Fuel subsidies and smuggling: - Fuel subsidies that lower domestic prices relative to neighbors tend to cause cross-border fuel smuggling. - Subsidies are expensive: on average 2 percent of GDP per year. - Benin and Togo case (2008–2012): - Benin: level of fuel sold on the formal (taxed) market declined to only 15 percent of total consumption. - Togo: formal market outcomes were also much lower than they should have been. - Togo: correlation between formal market consumption and the price differential vis-à-vis Nigeria is –0.85. - 2011 implied impact for Togo: an implicit subsidy of about 3 percent GDP to Togo, three-quarters of which was captured by smugglers and one-quarter by Togolese consumers. - Policy recommendations and trade-offs: - First-best: institute an automatic fuel pricing mechanism to follow international fuel prices more closely—reduces fiscal costs and negative spillovers to neighbors. - Second-best: greater cooperation to control borders to reduce smuggling and further harmonization of tax policies to avoid negative regional spillovers. - Caveat: if smuggling is large and borders porous, automatic price adjustment may increase tax base erosion; in that setting, lowering the tax rate (lowering domestic price) may be the best strategy despite substantial fiscal costs relative to a no-smuggling scenario. ### Socioeconomic impact of forced migration - Trends: - The number of internally displaced persons has risen significantly because of conflicts and violence, driven by religious extremism affecting the Sahel region and northeastern Nigeria. - Migration statistic cited: the ratio of refugees to the total migrant population declined from over 40 percent in 1990 to 10 percent (latest available migration data referenced). - Macroeconomic effects: - Forced migration developments hurt economic activity and weigh on public expenditures through higher welfare, security, and refugee-related spending. *Sources: French Centre d’Etudes Prospectives et d’Informations Internationales (CEPII) database; IMF, World Economic Outlook database; World Bank, World Development Indicators; and remittances and migration database.* ### 1. SSA IDPs, 2010–16 ### 1. SSA IDPs, 2010–16 ### Overview and recent trends - Since 2013 the number of intraregional sub-Saharan African refugees has declined while the number of internally displaced persons has risen significantly (referenced Figure 31, panel 1). - The Sahel countries, Nigeria, Democratic Republic of the Congo, South Sudan, and Central African Republic are among the countries most affected by internal displacement triggered by conflicts and violence (referenced Figure 31, panel 2). - While the number of refugees has been falling, the absolute number of migrants has risen considerably and is currently at record levels; the increase in migrants within sub-Saharan Africa is likely driven by individuals seeking greater economic opportunity and reflects reduced barriers to the movement of people. ### Drivers of displacement - Across the Sahel countries, Nigeria, and Cameroon, terrorism-related events and civil conflicts have more than doubled since 2011 (referenced Figure 32). - The collapse of the government in Libya and the rise of religious extremism increased availability of arms and frequency of terrorist attacks in the Sahel (notably Burkina Faso and Mali). - Boko Haram attacks are the leading cause of displacement in Niger, northeastern Nigeria, Chad, and northern Cameroon. - Domestic and neighboring political turmoil drives forced migration in other countries (example: Uganda hosts more than 1.2 million refugees and asylum seekers, mainly from South Sudan and Democratic Republic of the Congo). ### Humanitarian and economic impacts - Forced migration reduces economic activity and imposes humanitarian and fiscal costs from both fighting terrorism and hosting displaced persons (IMF 2016b). - The United Nations Office for the Coordination of Humanitarian Affairs estimates: - Nearly 30 million people will suffer food insecurity due to the security situation. - Almost 12 million of these are at crisis or emergency levels. - The Sahel region contended with approximately 4.9 million refugees and internally displaced persons in 2017. - The region’s humanitarian and financial needs for 2017 are estimated at US$2.7 billion. - Fiscal costs of hosting displaced populations vary; they are estimated to range between 1 percent and 5 percent of GDP, depending on the number of displaced persons. - The United Nations Development Programme estimates the cost of hosting refugees and asylum seekers in Uganda at about US$320 million, or about 1.3 percent of GDP. - Specific economic effects include shutdowns of sectors such as tourism in affected areas (example: Lake Chad area); some activities (for example, mining in remote regions) may be less immediately affected, but insecurity and higher costs deter domestic and foreign investment. ### Policy implications and recommendations - Address the main causes of forced migration, such as increased economic and physical insecurity. - Put in place systems that accommodate and integrate forced migrants in host countries in a sustainable way. - Increased international aid would greatly facilitate the accommodation and integration process. - Recognize and mitigate fiscal costs of hosting displaced populations through policy planning and international support. ### Key statistics and figures (preserved exactly as in source) - Uganda currently hosts more than 1.2 million refugees and asylum seekers. - United Nations Office for the Coordination of Humanitarian Affairs estimates nearly 30 million people suffering food insecurity, with almost 12 million at crisis or emergency levels. - Sahel region contended with approximately 4.9 million refugees and internally displaced persons in 2017. - Sahel region humanitarian and financial needs estimated at US$2.7 billion for 2017. - Fiscal costs of hosting displaced populations estimated to range between 1 percent and 5 percent of GDP. - United Nations Development Programme estimates cost in Uganda of about US$320 million, or about 1.3 percent of GDP. - Terrorism-related events and civil conflicts have more than doubled since 2011 (across the Sahel countries, Nigeria, and Cameroon). *Source: International Monetary Fund | August 2018 — 1. SSA IDPs, 2010–16.* --- _Source: https://www.imf.org/-/media/files/publications/spillovernotes/spillovernote12.pdf_