How to measure country risk?

Transcription

1 How to measure country risk? Produced by: Cross-country Emerging Markets Unit For the Occasion of: Second BBVA Resarch Emerging Market Seminar Madrid, July 13, 211

2 Road map to the presentation 1. Previous work on country risk (CR) 2. Our methodology 3. Results 3.a Key determinants of CR 3.b Differences across regions 4. Conclusions

3 1. Previous work on country risk

4 Previous Economic Policy Research Dependent variable Type of Model Global Variables Idiosyncratic Variables IMF 1 EMBI Spread (Levels) Static Long run (panel) 3 month Fed Funds Volatility Fed Funds VIX Credit Rating Index IADB 2 EMBI Spread (Differences) Dynamic & Static (panel ECM) 1 year US bond High Yield US corporate paper Credit Rating Index Bank of England 3 EMBI Spread (Differences) Dynamic & Static ( PMG) 1 and 3year US bond US Baa-Aaa spread SP5 index Fiscal Budget, Openness Amortization to Reserves Current Account Short term Ext Debt /Reserves ECB 4 Sovereign Debt Ratings (Levels) Static ( Ordered Probit) GDP per capita, GDP growth, Unemployment, Inflation,Gov. Debt, Govt.Balance, Govt Effectiveness External Debt, Curr.Account, Reserves, Default History (1) Hartelious et al (28): Emerging Market Spread Compression:Is it Real or is it Liquidity?. IMF WP 8/1 (2) Gonzalez Rozada and LevyYeyati (26): Global Factors and Emerging Market Spreads. IADB WP 552 (3) Ferruci (23) Empirical determinants of emerging market economies sovereign bond spreads BOE WP 25 (4) Afonso et Al (27). What HIDES Behind Sovereign Debt Ratings. ECB WP 711 Page 4

5 2. Our methodology

6 Sample coverage Country risk (CR) proxied by CDS spreads Sample period: lifespan of CDS (24 to 21) Number of countries: determined by data availability (mainly CDS but also CR determinants) Developed Countries Emerging Countries Europe Asia EMEA Asia Latam Austria Belgium Denmark France Germany Greece Ireland Iceland Italy Norway Portugal Sweden Spain Australia Japan Bulgaria Croatia Czech Rep Hungary Poland Romania Russia Slovenia Slovakia Turkey South Africa China Indonesia Malaysia Philippines S.Korea Thailand Argentina Brazil Chile Colombia Mexico Peru Venezuela

7 Variables chosen CDS Spreads Global Factor Idiosyncratic Fundamentals Global Common factor General Extracted with Kalman filter GDP Consumer Prices Fiscal Public Debt (% GDP) External External Debt (% GDP) International Reserves to imports Institutional PCA Rule of law Corruption Gov. Effectiveness Political Stability Investor ProtectionPage 7 Days to start business

8 A quick look at the data 2 CD Swaps (spreads in bp, average ) 5 GDP (% yoy, average ) 7 Inflation (Cons. Prices % yoy, average ) Total Developed Emerging Total Developed Emerging Total Developed Emerging Gross Public Debt (% GDP, average ) External Debt (% GDP, average ) Reserves to Imports ( average ) Total Developed Emerging Total Developed Emerging Total Developed Emerging

9 Extracting the Global Component Measurement Equation CDSwaps i, t = μ 1, t + β j, t x i, j, t + v i, t, State Space Model State Equation β μ j, t = β j, t 1 + ω β j, t 1 = + 1, t μ1, t 1 ωμ1, t 1, Unobserved CD Swaps Global Component Where, v t ~ N(, t ) and ω ~ t N(, Ω t ) j ( 1, r), i (1, n) We use the State Space Model to isolate the Global from the idiosyncratic model in the most possible orthogonal way Page 9

10 Our global component vs other measures 2 15 Our measure vs alternatives Measures of Global Risk Aversion (spreads in bp) CD Sw ap Global Component US Corporate Aaa Spread US Corporate Baa Spread US Corporate Junk Bond Spread (ML) CD Swap Global component vs alternative measures (spreads in bp) US Aaa Corporate Bond Spread CD Swap Global Component US Baa Corporate Bond Spread CD Swap Global Component 1 2, 1 5 US Junk Bond Spread (ML) 1,6 1,2 8 4 SP5 VIX Index Jan-7 Apr-7 Jul-7 Oct-7 Jan-8 Apr-8 Jul-8 Oct-8 Jan-9 Apr-9 Jul-9 Oct-9 Jan-1 Apr-1 Jul-1 Oct-1 Jan-11 Apr ,2 1,6 2, CD Swap Global Component CD Swap Global Component The extracted Global Component moves close to other alternative Global Risk Measures Specially the US Baa Corporate Bond Spread. Junk bond too volatile Page 1

11 Panel Data Model to explain SR Δ log( CDSwap) Panel Data Dynamic Error Correction Model (ECM) i, t = Δ log( CDSwap) i, t 1 + φδ log( Global) t 1 + λ( CDSwap) i, t 1 γx i, t 1 FEffi ) β + υ i, t Change In Spreads Short run Dynamics Short run Effect of Global Component Deviation from Long Run Global & Idiosyncratic Fundamentals including Fixed Effects Speed of Adjustment to Long Run Equilibrium (<) Page 11

12 3.Results 3.a Key determinants of CR

13 Global risk aversion matters in the long run CD Swaps Long & Short Run Elasticities: Global Factor Total Developed Emerging Long Run Short Run Global developments (global risk aversion) do matter even in the long run with a higher than unitary elasticity Page 13

14 Long term results relevant CD Swaps Speed of Adjustment to Long Run Equilibrium Total Developed Emerging * Grey bars stands for non significant variables at 1% level This is because the stimated speed of adjustment to the equilibrium is quick (75% of the shock adjusts in 1 year). Page 14

15 Relevance of GDP& inflation CD Swaps Long Run Elasticities: GDP CD Swaps Long Run Elasticities: Consumer Prices Total Developed Emerging Total Developed Emerging * Grey bars stands for non significant variables at 1% level * Grey bars stands for non significant variables at 1% level Economic growth does not seem relevant for emerging markets but it is for the developed world. Inflation appears to be more relevant for the developed world (or the absence of it) Page 15

16 Relevance of Public Debt & Institutions CD Swaps Elasticities: Public Debt (% GDP) CD Swaps Elasticities: Institutional (kauffman) Total Developed Emerging Total Developed Emerging * Grey bars stands for non significant variables at 1% level * Grey bars stands for non significant variables at 1% level The increase in public debt worsens country risk in developed countries. The result is inconclusive for emerging economies Institutional factors affect country risk, especially for developed economies Page 16

17 Relevance of external liquidity measures CD Swaps Elasticities: External Debt to GDP Total Developed Emerging CD Swaps Elasticities: Reserves to Imports Total Developed Emerging * Grey bars stands for non significant variables at 1% level External debt matters but especially in emerging markets A confortable liquidity position in terms of international reserves to imports helps to reduce the spreads Page 17

18 3.Results 3.b Differences across regions

19 Developed Europe: Actual vs Equilibrium CR Actual CD Swaps and estimated equilibrium CR from panel ECM (median CDSwaps by region) 14 Core Europe (Germany, France, Austria, Norway, Sweden) 25 Europe Periphery I (Belgium, Italy, Spain) 8 Europe Periphery II (Portugal,Ireland,Greece) Actual Estimated Actual Estimated Actual Estimated Jan 4 May 4 Sep 4 Jan 5 May 5 Sep 5 Jan 6 May 6 Sep 6 Jan 7 May 7 Sep 7 Jan 8 May 8 Sep 8 Jan 9 May 9 Sep 9 Jan 1 May 1 Sep 1 Jan 11 May 11 Jan 4 May 4 Sep 4 Jan 5 May 5 Sep 5 Jan 6 May 6 Sep 6 Jan 7 May 7 Sep 7 Jan 8 May 8 Sep 8 Jan 9 May 9 Sep 9 Jan 1 May 1 Sep 1 Jan 11 May 11 Jan 4 May 4 Sep 4 Jan 5 May 5 Sep 5 Jan 6 May 6 Sep 6 Jan 7 May 7 Sep 7 Jan 8 May 8 Sep 8 Jan 9 May 9 Sep 9 Jan 1 May 1 Sep 1 Jan 11 May 11 Core Europe remains above but near safe long run equilibrium levels Portugal, Ireland and Greece current CDS level is similar to the equilibrium one Spain, Belgium and Italy s country risk is clearly above the equilibrium: Contagion exists Page 19

20 Emerging markets: Actual vs Equilibrium CR Actual CD Swaps and estimated equilibrium CR from dynamic ECM (median CDSwaps by region) Emerging Europe Emerging Asia Latam Actual Estimated Actual Estimated Actual Estimated Jan 4 May 4 Sep 4 Jan 5 May 5 Sep 5 Jan 6 May 6 Sep 6 Jan 7 May 7 Sep 7 Jan 8 May 8 Sep 8 Jan 9 May 9 Sep 9 Jan 1 May 1 Sep 1 Jan 11 May 11 Jan 4 May 4 Sep 4 Jan 5 May 5 Sep 5 Jan 6 May 6 Sep 6 Jan 7 May 7 Sep 7 Jan 8 May 8 Sep 8 Jan 9 May 9 Sep 9 Jan 1 May 1 Sep 1 Jan 11 May 11 Jan 4 May 4 Sep 4 Jan 5 May 5 Sep 5 Jan 6 May 6 Sep 6 Jan 7 May 7 Sep 7 Jan 8 May 8 Sep 8 Jan 9 May 9 Sep 9 Jan 1 May 1 Sep 1 Jan 11 May 11 Emerging Markets seem to be somehow overvalued by the market: Equilibrium CR appers to be above current CDS level Page 2

21 Global Risk Aversion vs idiosyncratic CR across regions Average Contribution of different determinants of CR (based on coefficients estimated in Panel ECM) Western Europe Core Europe Peripheral Europe Emerging Europe Emerging Asia Latam exargven Latam World Global Idiosyncratic + Fixed Effect Global risk aversion explains as much as half of CR across regions More relevant the riskier the region: Latam and Pheripheral Europe Page 21

22 Fiscal dominating idiosyncratic CR across regions Average Contribution of different determinants of CR (based on coefficientes estimated in ECM) Western Europe Core Europe Peripheral Europe Emerging Europe Emerging Asia Latam exargven Macro Fiscal External Institutional Fiscal factors clearly dominate: specially important in Peripheral Europe and Latam External debt most relevant for Europe Quality of institutions specially important for Latam, followed by Asia Macro variables hardly relevan Latam World Page 22

23 In most recent period fiscal continues dominating Contribution of different determinants of country risk (based on coefficientes estimated in ECM) Western Europe Core Europe Peripheral Europe Emerging Europe Emerging Asia Latam exargven Macro Fiscal External Institutional External liquidity factors become more relevant for pheripheral Europe Institutions less relevant for Latam and Asia Latam World Page 23

24 4. Conclusions

25 On our methodology Our CR aims at: Estimating true CR of a country, compared to the market one (CDS) Key is to separate global factors from idiosyncratic ones: strong methodology - Robustness tests to be conducted with other indicates Focus on long-term CR because - Easier to separate global market factors from - Estimated speed of adjustment relatively short so long term relevant for policy making/risk assessment Determining which are the key factors determining country risk This allows to have different early warning indicators for different groups of countries Model can also be used to predict CR Using our forecasts of idiosyncratic variables Forecast of global risk aversion harder - Work to be done on the latter

26 On our findings Global risk aversion is a key component of country risk even in long term In Western Europe, global risk aversion worsening CR beyond equilibrium level The opposite is true for the Emerging Word: markets seem to be underestimating risk Among the idiosyncratic determinants of CR: The fiscal situation has and continues to be the most relevant Important warning for countries conducting reckless fiscal policies External debt more relevant for European countries Quality of institutions for emerging countries