Exchange Rate Determination in Low and Lower-Middle Income Countries: The Case of Sub-Saharan - PowerPoint Presentation

1. Exchange Rate Determination in Low and Lower-Middle Income Countries: The Case of Sub-Saharan AfricaAnnina Kaltenbrunner, Leeds University Business School, a.kaltenbrunner@leeds.ac.ukDaniel Perez Ruiz, Leeds University Business School, bndapr@leeds.ac.ukAnjelo Okot, Leeds University Business School and Central Bank of Uganda, bn15ao@leeds.ac.uk

2. BackgroundEIB Starebei Project on Nominal Exchange Rate Determination in Low and Lower-Middle Income Countries (LLMICs) Increased importance of nominal exchange rate dynamics in the context of floating exchange rate regimes and non-resident investments in LLMICs local currency assets Very limited panel data literature on nominal exchange determination in LLMICsLittle knowledge of micro-structural characteristics of LLMICs foreign exchange markets > importance for understanding exchange rate determination and central bank operations (feasibility of floating exchange rates)

3. Extensive mixed-method study of nominal exchange rate determination in (Sub-Saharan African) LLMICs 13 semi-structured interviews with 17 foreign exchange experts in 6 case study countries and the City of London Main flows, market-microstructure, and central bank operationsMacro-Panel Econometrics for 15 LLMICs from 1997-2019Statistical and economic significance of Main Research Hypothesis: Exchange rate dynamics in SSA LLMICs fundamentally shaped by the specific nature of their productive and monetary/financial integration in world economy Productive: export concentration in limited set of mineral and/or food commoditiesMonetary/Financial: “weak” currencies and sensitivity to foreign financial flows and international market conditionsBackground

4. Outline2 Working Papers Working Paper 1: A microstructural analysis of foreign exchange markets and exchange rate management in SSA LLMICs (Daniel and Nina)Working Paper 2: A mixed-method study of exchange rate determination in African LLMICs (Anjelo; panel results) Brief literature reviews and motivations > Focus on Results Policy Recommendations

5. Working Paper 1: Microstructure and Exchange Rate Management Motivation and Contribution Microstructure Literature on Exchange Rate Determination (e.g. Lyons 2001, Sarno and Taylor 2001; de Grauwe and Grimaldi, 2006)Exchange rates determined by context and time specific buying and selling decisions of heterogeneous agents in foreign exchange market Buying and selling decisions shaped by institutional structure (organisation of FX market) and might be non-rational (behavioural finance)But: Focus on developed foreign exchange markets and aggregate price dynamics Very limited and outdated literature on foreign exchange market microstructure in Low and Middle income countries (Canales-Kriljenko (2004) and Azam (2006) Not specifically on Africa nor LLMICs Survey-based Don’t consider interaction between macro-structures (productive and monetary/financial integration in world economy) and micro-structures, and exchange rate management

6. 13 semi-structured interviews with 17 foreign exchange market experts 6 case study countries (Ghana, Kenya, Malawi, Sierra Leone, Uganda, and Zambia) and City of London Africa: 8 representatives from central banks, 3 representatives from commercial banks, one asset managers, and 3 experts from research institutions and development agencies London: One frontier market asset manager and one chief economist at a major international bank specialising in frontier markets in the City of London Three areas of investigation: (1) Foreign exchange determinants and main flows; (2) Microstructure (organisation, instruments, actors); (3) Exchange rate management ZoomCoded with NvivoQualitative study: Focus on uncovering unique features of selected African foreign exchange markets rather than statistical significance Key issues and themes, and develop concepts and theorize based on the concrete realities of economic agents Working Paper 1: Microstructure and Exchange Rate Management Methodology

7. All respondents pointed to the most significant balance of payments flows.For all countries traditional agricultural and mining exports continue to represent an important source of foreign exchange. Some also export oil. Most countries’ exports remain concentrated in a few export commodities. Remittances and tourism were mentioned as an important source and foreign donor flows (which have decreased over recent years). Financial flows. In countries like Ghana and Uganda, offshore players such as portfolio investors, foreign multinationals, and international banks have an active role. Ghana-> non-resident investors and institutional investors play an important role as these actors are particularly active in the Eurobond market and the local currency bond market. Uganda -> offshore investors play a key role in the local bond market. Other important drivers of the exchange rate mentioned in the case studies include: the fiscal situation and government debt and borrowing, external debt, inflation, monetary policy, and the political cycle.Working Paper 1: Microstructure and Exchange Rate Management Results 1: Exchange Rate Fundamentals and Drivers

8. Most countries functioning interbank marketDevelopment of electronic trading in most countries (with some variation).Existence of brokers still very limited Seasonal, volatile, and “lumpy” foreign exchange market liquidity Dependence on volatile and seasonal export commodities undermines development of interbank market (Sierra Leone, Malawi)Lumpiness and uncertainty about flows creates hoarding behaviour by foreign exchange market actors Parallel market Working Paper 1: Microstructure and Exchange Rate Management Results 2: Microstructure - Organisation

9. Dominated by US Dollar spot operations in particular in less liquid and highly import dependent marketsSome development of derivatives market (forwards and FX swaps)Forwards for hedging; swaps for fundingSwaps allows banks to act as counterparties to foreign investors (Uganda) All trade and many financial transactions denominated in Dollar which maintains dominance of American currency (latent risk or actual currency substitution) Working Paper 1: Microstructure and Exchange Rate Management Results 2: Microstructure – Instruments

10. As in developed foreign exchange markets, banks are key actors in foreign exchange markets In contrast to developed FX markets, frequently dominated by foreign banks Key points of access for foreign institutional investors and often invest for their clients across the African continent.Bring expertise but also strengthen link to international funding conditions (contagion)Recent decline in corresponding banking and rise of regional/pan-African banksImportance of client rather than dealer flows Working Paper 1: Microstructure and Exchange Rate Management Results 2: Microstructure – Actors

11. Low and erratic interbank liquidity, as well as the relatively large size and lumpiness of client flows, makes these flows relatively more important and gives those access to FX significant market, and power and ability to influence price dynamics For countries with low level of financial integration (Malawi, Sierra Leone, Zambia) -> large commodity and mineral exportersIn markets with a high participation of foreign investors (Ghana, Uganda), the foreign players can have significant impact on exchange rate movements(Foreign) Banks Most of the trading activities occur among the major commercial banks. Working Paper 1: Microstructure and Exchange Rate Management Results 2: Microstructure – Inefficiencies

12. Peculiar feature of all case study foreign exchange markets is existence of foreign exchange market bureaus Provide access to spot/cash foreign exchange to the local populationNo access to central bank liquidity and frequently not allowed to hold accounts offshore Further reduce availability of foreign exchange to interbank market (?)Collect cash from parallel market (?)Customer services Working Paper 1: Microstructure and Exchange Rate Management Results 2: Microstructure – Inefficiencies

13. All case study countries have moved to a de-jure floating exchange rate regimeHowever, given limited and volatile liquidity in FX markets, central banks remain key actors in African LLMIC foreign exchange markets > De-facto, all central banks continue to intervene significantly in the foreign exchange marketAccording to interviews, interventions are mainly aimed at smoothing exchange rate volatility and avoiding large and sudden changes in foreign exchange liquidity Some controversy over exchange rate level target in Kenya Working Paper 1: Microstructure and Exchange Rate Management Results 3: Central Bank Operations – Exchange Rate Regime

14. Variegated reasons for interventions shaped by specific microstructures:Deal with acute periods of foreign exchange illiquidity and securing vital imports, such as food staples and healthcare products (e.g. Sierra Leone)Manage the cycle of global liquidity and smoothing the impact of large foreign financial flows on the exchange rate (e.g. Uganda and Ghana) Avoid excessive depreciations to limit flight in dollar and for political reasons (currency strength as signal of economic strength) (e.g. Kenya)Asymmetric interventions: Generally, less concern for appreciations than depreciations > real appreciations Import dependent Commodity exports priced in dollar Contain appreciation to avoid excessive depreciation and accumulate reserves Working Paper 1: Microstructure and Exchange Rate Management Results 3: Central Bank Operations – Reasons for Interventions

15. At the same time though, microstructural weaknesses (e.g. FX hoarding outside the official banking system; actors with significant market power) undermine central banks’ ability to intervene and manage the exchange rateReflected in instruments SSA LLMICs central banks use in addition to market-based interventionsMoral suasion (e.g. Kenya, Malawi)Agreements with banks (Kenya)At times, direct pressures on the banking system Foreign exchange purchases from government owned institutions (e.g. Ghana)Approach actors with significant inflow directly (Uganda)Working Paper 1: Microstructure and Exchange Rate Management Results 3: Central Bank Operations – Instruments of Intervention

16. Very limited (no) panel literature that investigates nominal exchange rate determinants in LLMICs > Existing studies focus on larger cross-sections and the real exchange rate No panel data studies on nominal exchange rate determination (level/return, volatility, crash risk) in AfricaThomas (2012): monthly panel for 12 SSA. 2003-2010; level only Main Research Hypothesis: Exchange rate determination in SSA LLMICs fundamentally shaped by the specific nature of their productive and monetary/financial integration in world economy Productive: export concentration in limited set of mineral and/or food commoditiesMonetary/Financial: “weak” currencies and sensitivity to foreign financial flows and international market conditionsWorking Paper 2: Panel-data estimation Motivation and Contribution

17. Different Panel Data Techniques have been adopted on annual series for 15 African LLMICs from 1997-2019 For the estimation of for the exchange rate level and the volatility, two techniques are adopted:the Augmented Mean Group model (AMG). The AMG estimator was chosen because of its ability to account for potential non-stationarity, slope heterogeneity, and cross-sectional dependence in the data. Moreover, it works well in panels with moderate-T and moderate-N.(b) Panel Dynamics Ordinary Least Squares (DOLS) method for robustness check. DOLS also has ability to deal with cross-sectional dependence problem and can work well with variables which are either I(0) or I(1). For the estimation of crash risk, Panel quantile regression method is adopted. quantile estimator is typically employed on different quantiles of the conditional distribution.Quantile regression is more robust in nature and is able to capture outliers effectively unlike the OLS which loses its effectiveness particularly in the extremes of the distributions or tailed analysisWorking Paper 2: Panel-data estimation Methodology – Panel techniques

18. Two types of pre-tests are conducted to guide the selection of the above panel approaches. Cross-sectional dependence tests - the Pesaran (2003) cross-sectional dependence (CD) test, and Pesaran’ (2004/2015) cross-sectional dependence (CD) test were conducted. Panel unit root tests, including both the first-generation and the second-generation approach. In the first-generation approach, we implement the Im-Pesaran-Shin (IPS) and Fisher- Type unit root tests. In the second-generation approach, which accounts for cross-sectional dependence in the variables, we implement Pesaran’s (2003) unit root test which is based on the Dickey–Fuller regression augmented with the cross-section averages of lagged levels and first differences of the individual series. Working Paper 2: Panel-data estimation Methodology – Pre-Tests

19. Dependent Variables: (i) Log exchange rate-normalized ; (ii) Volatility, measured as the annualised standard deviation from monthly exchange rate changes, and (iii) crash risk measured by depreciation of the nominal exchange rateExplanatory Variables (informed by research hypothesis and interviews): Core Model Productive factors: (i) terms of trade, (ii) export concentration index, (iii) Commodity Export Price Index, Individual Countries, (iv) climatic vulnerability Financial Factors: (i) yield differential, (iii) portfolio investments, and (iii) other investmentsControls (Macroeconomic Factors): (i) GDP growth, and (ii) inflationOther Variables (added alternatingly): Other macroeconomic variables and balance of payments flows: (i) current account, (ii) FDI Public Financial Flows: (i) External debts, (ii) remittances, (iii) foreign reserves and (iv) official development assistance Political and Institutional factors: (i) political stability index, (ii) corruption control index, (iii) trade openness, and (iv) financial markets development index Global factors: (i) VIX, (ii) commodity price index, and (iii) oil price index.Working Paper 2: Panel-data estimation Data and Estimation Strategy

20. Productive factors: prices of country’s national export commodities, export concentration, terms of trade and lag of climatic vulnerability play crucial role on the level of the exchange rate. Financial factors: Interest rate differential and other investment (mainly banking flows). Macroeconomics factors: the current account and FDI flows appreciate the exchange rate. Inflation also plays a significant role Public financial flows: official development assistance and remittances Political and instructional factors: trade openness, corruption, and financial development, have impact on the exchange rate Global factors: The interaction of VIX with openness is significant and depreciates the exchange rateWorking Paper 2: Panel-data estimation Results: AMG and Exchange Rate Level

21.  Model1Model2Model3Model4Model5 Model6Model7Model8Model9Model10 GDP-0.003 0.0060.009-0.0010.002 -0.001-0.0010.002-0.0020.004  (0.559)(0.187)(0.409)(0.824)(0.599) (0.762)(0.765)(0.703)(0.710)(0.221) Inflation0.018**0.018**0.030**0.019***0.021***0.018***0.016***0.016***0.020***0.017***  (0.002)(0.018)(0.010)(0.001)(0.000) (0.002)(0.000) (0.001)(0.000)(0.004) Yielddiff-0.009***-0.001-0.016***-0.009**-0.010***-0.008***-0.007**-0.006-0.010***-0.006***  (0.003)(0.900)(0.005)(0.014)(0.001) (0.000)(0.032)(0.104)(0.000)(0.052) ToT (log)0.196***0.188***0.187**0.125**0.180***0.178***0.184***0.182***0.190***-0.006  (0.000)(0.000)(0.034)(0.017)(0.000) (0.002)(0.000)(0.000)(0.001)(0.901) ECI0.455***0.544***0.5190.381**0.420*** 0.549***0.550***0.586***0.603***0.576*** (0.001)(0.001)(0.182)(0.014)(0.007) (0.000)(0.000)(0.000)(0.001)(0.000) ECPI (log)-0.427***-0.269*0.065-0.370***-0.405***-0.316**-0.349***-0.345***-0.355***-0.192*** (0.004)(0.069)(0.763)(0.003)(0.000) (0.012)(0.000) (0.000)(0.003)(0.002) Portfolioinv0.0760.166-0.0040.009-0.012 0.015-0.009 0.0010.0030.067  (0.589)(0.141)(0.930)(0.685)(0.662) (0.733)(0.634)(0.956)(0.891)(0.620) Otherinv0.024*0.009*0.0040.015***0.013*** 0.003-0.004-0.0050.008-0.029***  (0.039)(0.066)(0.762)(0.000)(0.007) (0.755)(0.439)(0.467)(0.281)0.004 Cab-0.006*          (0.059)         FDI -0.011**          (0.046)        remmitance  0.047***          (0.000)       ODA   -0.006***          (0.003)      Debts    -0.001* *           (0.040)      financialdev     0.994**          (0.012)    openness       0.006***          (0.000)   vixoppennes       0.325***          (0.000)  Corruption control        0.120**          (0.038) climaticvul (lag1)         11.739***

22. Productive factors: the concentration index, the terms of trade, current account and climatic vulnerability play crucial role in driving exchange rate. Financial factors: other investments reduces exchange rate volatilityMacroeconomic factors: GDP growth reduces exchange rate volatility and inflation increases itPublic financial factors: External debts increases FX volatilityPolitical and Instructional factors: financial development index and control of corruption index which are all found to have positive impacts on exchange rate volatilityWorking Paper 2: Panel-data estimation Results: AMG and Exchange Rate Volatility

23. .  Model1Model2Model3Model4Model5Model6Model7Model8Model9Model10 GDP-0.821*-0.738-1.038**-0.631-0.794*-0.707*-0.736-0.755-0.838-0.560* *  (0.072)(0.116)(0.023)(0.239)(0.085)(0.094)(0.120)(0.111)(0.114)(0.020) Inflation0.374**0.293*0.3430.3890.441**0.371*0.443*0.470**0.445**0.344*  (0.046)(0.079)(0.219)(0.120)(0.038)(0.050)(0.027)(0.028)(0.032)(0.073) Yielddiff-0.062-0.119-0.222-0.035-0.153-0.140-0.178-0.174-0.165-0.188  (0.613)(0.258)(0.172)(0.824)(0.238)(0.296)(0.148)(0.174)(0.191)(0.265) ToT (log)-3.415*-2.138-2.1352.010-2.516-0.832-1.802-1.809-2.502-2.396  (0.050)(0.154)(0.362)(0.394)(0.162)(0.574)(0.152)(0.131)(0.268)(0.247) ECI13.701*16.227**20.571*17.947***16.678**16.497***13.921**14.313**14.134*14.343***  (0.078)(0.026)(0.079)(0.005)(0.023)(0.005)(0.046)(0.037)(0.092)(0.005) ECPI (log)-5.735-4.629-4.452-3.927-5.394-0.612-4.208-4.564-5.8902.799  (0.236)(0.325)(0.483)(0.535)(0.295)(0.915)(0.385)(0.345)(0.234)(0.479) Portfolioinv-2.2320.211-13.039-2.492-0.054-1.7772.1232.1477.250-1.254  (0.840)(0.986)(0.186)(0.841)(0.997)(0.924)(0.874)(0.874)(0.666)(0.968) Otherinv-0.639*-0.622*-0.679-0.669*-0.603*0.010-0.789**-0.786**-0.4971.099  (0.071)(0.083)(0.184)(0.059)(0.092)(0.992)(0.022)(0.022)(0.196)(0.373) Cab0.184**          (0.012)         FDI -0.079          (0.531)        remmitance  -0.315          (0.485)       ODA   0.327**          (0.044)      Debts    0.039**          (0.038)     financialdev     52.055**          (0.026)    Openness      0.029          (0.320)   vixoppenness       1.700          (0.376)  Corruption control        5.451***          (0.000) Climaticvu L1         106.031***          (0.000)

24. Productive factors: improvement in current account balance, FDI and export commodity price index for individual countries reduces the potential of currency crash risks. Export concentration increases crash risk in most quartilesFinancial Factors: Rising yield differential is associated with increasing currency crash risk. other investments reduces crash risk though only significant at slightly above the conventional confidence limit. increase VIX leads to a rise in currency crash risk in LLMICMacroeconomics factors: GDP and inflation become significance from 75th percentile. While GDP reduces crash risk, inflation rises its possibility. Political and institutional factors: improvement in political stability and corruption control has the impact of lowering currency crash risk in most quantiles. Working Paper 2: Panel-data estimation Results: Panel Quantile Regressions of Crash Risk

25. Panel Estimation of Currency Crash Risk- 25th QuantileVariable NameModel1Model2Model3Model4Model5Model6Model7Model8Model9Model10 Cab-0.203*-0.218**-0.213*-0.172-0.198*-0.204*-0.154-0.160-0.235**-0.252**  (0.069)(0.042)(0.061)(0.109)(0.081)(0.075)(0.131)(0.111)(0.029)(0.032) yielddiff0.343***0.358***0.277**0.356***0.374***0.425***0.324***0.385***0.363***0.380*** (0.000)(0.000)(0.003)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000) portfolioinv0.0580.018-0.130-0.0230.0130.128-0.044-0.0770.0010.017  (0.864)(0.953)(0.694)(0.943)(0.971)(0.715)(0.886)(0.800)(0.997)(0.963) VIX-0.012-0.027-0.0690.024-0.014-0.047-0.048-0.0020.002-0.021  (0.895)(0.754)(0.477)(0.793)(0.888)(0.626)(0.579)(0.985)(0.979)(0.830) otherinv0.0530.003-0.153-0.0150.0110.0350.057-0.1670.0870.092  (0.790)(0.987)(0.435)(0.935)(0.957)(0.861)(0.755)(0.348)(0.651)(0.662) FDI-0.403**-0.356**-0.389**-0.274*-0.348**-0.122-0.270*-0.218-0.316**-0.361**  (0.012)(0.017)(0.019)(0.097)(0.032)(0.500)(0.062)(0.129)(0.040)(0.032) ToT (log)2.1721.7284.530**2.7062.0312.9381.6652.4903.922**3.659**  (0.202)(0.272)(0.022)(0.123)(0.245)(0.115)(0.300)(0.107)(0.021)(0.046) ECI5.7837.907*11.436**9.626**5.4805.1653.8854.9564.2794.443  (0.199)(0.061)(0.020)(0.036)(0.267)(0.261)(0.357)(0.225)(0.326)(0.350) ECPI-0.022-0.033-0.047-0.048*-0.035-0.040-0.037-0.0280.0390.028  (0.525)(0.210)(0.154)(0.098)(0.211)(0.153)(0.167)(0.285)(0.371)(0.518) climaticvul-30.76**-25.75*-29.50*-35.83**-22.69-27.116*-23.735-20.154-15.820-18.961  (0.050)(0.098)(0.060)(0.021)(0.266)(0.089)(0.106)(0.164)(0.298)(0.254) inflation0.0390.0840.1350.0460.0440.0190.0560.1090.0790.046  (0.718)(0.405)(0.218)(0.651)(0.688)(0.864)(0.581)(0.277)(0.456)(0.687) GDP-0.082-0.074-0.113-0.181-0.058-0.099-0.118-0.005-0.022-0.069  (0.557)(0.572)(0.445)(0.192)(0.682)(0.489)(0.361)(0.970)(0.870)(0.643) debts0.012          (0.564)         ODA -0.068          (0.477)        remmitance  -0.017          (0.948)       reserves   -0.006          (0.965)      financialdev    3.137          (0.813)     openness     -0.089**          (0.043)    Corruption control      2.381          (0.143)   Political stability       -1.516*          (0.080)  Commodity price        -0.060**          (0.016) Oil price         -0.037**           (0.045)

26. Panel Quantile Estimation of Crash Risk- 75th QuantileVariable NameModel1Model2Model3Model4Model5Model6Model7Model8Model9Model10 Cab-0.096-0.099-0.224-0.202-0.099-0.105-0.109-0.1160.0070.053  (0.659)(0.647)(0.325)(0.392)(0.660)(0.632)(0.597)(0.585)(0.970)(0.786) yielddiff0.687***0.699***0.527**0.622***0.677***0.655***0.708***0.671***0.522***0.594*** (0.000)(0.000)(0.004)(0.001)(0.000)(0.000)(0.000)(0.000)(0.001)(0.000) portfolioinv-2.194**-2.190***-2.019**-2.082**-1.629*-2.188**-2.140***-2.109**-1.920**-2.057*** (0.001)(0.001)(0.003)(0.003)(0.019)(0.001)(0.001)(0.001)(0.001)(0.001) VIX0.349*0.347*0.363*0.468**0.357*0.377**0.300*0.2940.293*0.290*  (0.059)(0.050)(0.062)(0.020)(0.064)(0.042)(0.089)(0.107)(0.073)(0.083) otherinv-0.426-0.429-0.455-0.332-0.286-0.421-0.397-0.376-0.452-0.470  (0.274)(0.249)(0.249)(0.423)(0.487)(0.276)(0.286)(0.322)(0.189)(0.183) FDI-0.090-0.089-0.105-0.187-0.0870.040-0.102-0.1230.1220.186  (0.772)(0.769)(0.750)(0.607)(0.787)(0.908)(0.728)(0.685)(0.656)(0.507) ToT (log)2.1252.0662.4880.5541.3421.6791.8791.3961.9870.678  (0.523)(0.517)(0.529)(0.886)(0.700)(0.636)(0.564)(0.671)(0.511)(0.825) ECI4.8265.31412.39412.1632.0135.1234.0205.3316.9579.226  (0.583)(0.534)(0.207)(0.227)(0.838)(0.559)(0.639)(0.539)(0.370)(0.247) ECPI-0.013-0.016-0.063-0.015-0.009-0.014-0.012-0.0140.1460.133  (0.846)(0.770)(0.338)(0.809)(0.869)(0.795)(0.819)(0.804)(0.059)(0.066) climaticvul-40.002-39.767-16.517-32.603-52.828-31.337-41.606-44.058-19.322-25.554  (0.191)(0.208)(0.599)(0.339)(0.194)(0.302)(0.162)(0.152)(0.476)(0.358) inflation0.3100.3080.489**0.2990.2520.3400.3040.3050.369*0.282  (0.139)(0.131)(0.027)(0.181)(0.250)(0.111)(0.141)(0.154)(0.053)(0.145) GDP-0.364-0.366-0.297-0.426-0.440-0.432-0.464*-0.408-0.345-0.337  (0.183)(0.167)(0.317)(0.164)(0.123)(0.115)(0.078)(0.134)(0.154)(0.174) debts0.002          (0.969)         ODA -0.005          (0.980)        remmitance  0.392          (0.455)       reserves   0.083          (0.798)      financialdev    -22.128          (0.403)     openness     -0.034          (0.688)    Corruption control      0.111          (0.973)   Political stability       0.741          (0.687)  Commodity price        -0.103**          (0.021) Oil price         -0.066**           (0.030)

27. Panel DOLS model is estimated on FX level and panel limited dependent variable model is applied on crash risk. First, the results confirm the important impact of African LLMICs’ weak productive structures, approximated by the current account balance, FDI, export concentration and terms of trade,Second, private financial factors, especially interest rate differential and portfolio flows depreciate exchange rate, while other investments appreciate itThird, macroeconomic factors - inflation and GDP growth play significant role in determining exchange rate dynamics in African LLMCs. Finally, political, and institutional factors show significant effect on exchange rate. Working Paper 2: Panel-data estimation Robustness Checks

28. ConclusionThis paper investigates the exchange rate determination and currency crash risk in African LLMICs.Panel date econometrics estimations show four broad resultsthe key role of African LLMICs productive structure, approximated by their concentration of exports, terms of trade, prices of their main export commodities and exposure to climatic vulnerability, for exchange rate determinationFinancial factors have gained importance (yield differential, other investments and VIX)Macroeconomics factors (GDP growth and inflation) continue to play important rolePolitical and institutional factors (political stability, corruption, openness and financial markets development) also play importance role in driving exchange rate. Public financial flows also show importance roleMore generally, FX drivers in African countries greatly differ from advanced economies. Their Africa’s distinct integration in the global economy, both on the productive and the financial side are key in shaping their exchange rate and degree of crash risk.

29. Policy RecommendationsDevelopment Institutions: Efficiency of Lending Clear guidance for onwards lendingStringent selection of intermediate agenciesUse of public and development banksCost of LendingConsider presence of short-term financial flows and vulnerability to international market conditions (crash risk)Counter-cyclical dispersion of funds

30. Policy RecommendationsDomestic Policy Makers: Develop mechanisms to ensure the transfer of foreign exchange revenues into the official foreign exchange marketSupporting the careful development of local derivatives market though with a focus on instruments which support hedging rather than speculationReducing the dependence on the dollarAttraction of stable financial resources Reconsider the suitability of floating exchange rates for SSA