Sovereign Bond Yield Spreads: An International Analysis Giuseppe Corvasce

Transcription

1 Sovereign Bond Yield Spreads: An International Analysis Giuseppe Corvasce Rutgers University Center for Financial Statistics and Risk Management Society for Financial Studies 8 th Financial Risks and INTERNATIONAL FORUM March 31 st, 2015

2 Agenda Motivations An overview of the literature Econometric Methodology Pseudo out-of-sample forecasting methodology Data and sample selection criteria Estimation of the Model and Empirical Results Conclusions

3 Motivations The recent Sovereign Debt crisis is the result of the rising private and government debt levels around the world together linked to the impossibility for some countries in the eurozone area to re-finance their government debt, provocating a downgrade of these economies and an increase of the global risk aversion. Limitations of the primary capital market indicator of the sovereign bond default (such as the CDS) to provide warning signals regarding the conditions of a country (Altman et al. 2011). Indicators of implied default probability (higher PDs based on CDS) for Greece, Spain, Portugal, Ireland and Italy shown a warning signal only in The 5 years implied PDs (based on a recovery rate of 40% or 20%) for Greece, Spain, Portugal, Ireland and Italy peaked a level of (for Greece), (for Portugal), (for Ireland), (for Spain), (for Italy). (Sources: NYU Salomon Center and Bloomberg) The need to build time-varying indicators for modeling and simulating the evolution of the uncertainty related to the Sovereign Bond Yield Spreads.

4 Overview of the literature The Role of macroeconomic fundamentals for explaining the uncertainty related to the evolution of the sovereign bond yield spreads: Babbel (1996), Duffie et al. (2003), Uribe et al. (2006), Hilscher et al. (2010), Bernoth et al. (2012); Exchange rate expectations: Favero (1997); International factors: Codogno et al. (2003), Manganelli et al. (2009), Baek et al. (2005), Gerlach et al. (2010); Common time-varying factor: Sgherri et al. (2009); Contagion effects caused by the sovereign debt crisis: Favero (2012), Favero et al. (2012), Amisano et al. (2012); Methodologies for predicting sovereign defaults: Frank et al. (1971), Merton (1974), Gray et al. (2006, 2007), Zhang et al. (2011); Credit ratings and Sovereign health: Pomerleano (1998), Remolona et al. (2008), Altman et al. (2011)

5 Econometric Methodology r r i, t i i i, t 1 i, t r r US, t US, t 1 US, t log log 2 2 log z E z z i, t i i i, t 1 i i, t 1 i, t 1 i i, t log vol c s vol f innov E innov m innov US, t US, t 1 US, t 1 US, t 1 US, t 1

6 Simulation procedure s s i, t 1 US, t 1 r r s 1,..., S 1 z innov s 1,..., S s s i, t 1 US, t 1 h h 1 vol _ spread sigma sigma 2 cov h h h h i, US, t 1 i, t 1 US, t 1 i, US, t 1 cov sigma sigma h h h h i, US, t 1 i, US, t 1 i, t 1 US, t 1 Christoffersen et al. (1998), Figlewski (2004), Andersen et al. (2005)

7 Data and Summary Statistics COUNTRY % Var. Bond Yield Jan 00 to June 07 Median Sov. Bond Yield % Var. Bond Yield July 07 to Dec. 08 Median Sov. Bond Yield % Var. Bond Yield Jan 09 to Dec. 11 Median Sov. Bond Yield % Var. Bond Yield Jan 12 to May 12 Median Sov. Bond Yield AUSTRALIA AUSTRIA BELGIUM BRAZIL CANADA CHINA CZECH. REP. DENMARK FINLAND FRANCE GERMANY GREECE HONG KONG HUNGARY INDIA INDONESIA ITALY JAPAN MEXICO NETHERLANDS NEW ZEALAND NORWAY POLAND PORTUGAL RUSSIA SINGAPORE SOUTH AFRICA SPAIN SWEDEN SWITZERLAND THAILAND UK U.S.A 17.40% 9.23% 8.84% % -5.32% 34.07% 5.61% 7.78% 9.61% 11.96% 12.87% 0.54% % 7.15% % % 13.19% -2.72% % 10.45% 40.96% 13.58% % 12.28% % % 10.02% -3.50% 14.65% % 16.00% % 3.85% 4.76% 4.79% 10.71% 3.54% 0.41% 4.97% 0.65% 4.73% 4.70% 4.62% 4.99% 0.58% 0.00% 0.16% 0.00% 4.87% 0.01% 0.08% 4.71% 3.57% 0.07% 1.88% 4.80% % 1.22% 4.76% 0.54% 2.02% 0.01% 7.71% 4.59% % % % 0.76% % % -8.46% % % % % 12.18% % 18.98% % 7.93% -7.47% % % % % % -9.13% % % % % % % % % % 5.40% 6.22% 6.37% 6.09% 3.73% 0.59% 6.58% 3.90% 6.21% 6.23% 6.04% 6.65% 0.41% 0.04% 0.20% 0.00% 6.57% 0.00% 0.73% 6.22% 4.81% 0.08% 2.36% 6.46% % 1.20% 6.31% 0.66% 2.62% 0.14% 9.34% 3.91% % % 0.75% % 40.03% 44.27% 13.92% % % -9.88% % % % % 54.20% % 2.88% 8.77% -2.94% % 55.53% 20.60% 11.22% 65.56% 11.52% 39.82% 57.27% 33.48% 51.38% -9.77% 60.72% 22.34% 55.91% 4.72% 4.94% 5.35% 4.77% 3.09% 0.52% 5.54% 2.06% 4.63% 4.76% 4.18% 12.86% 0.30% 0.03% 0.02% 0.00% 5.98% 0.00% 0.01% 4.68% 3.93% 0.06% 1.99% 6.76% 5.54% 1.75% 1.14% 5.74% 0.04% 1.84% 0.01% 5.42% 3.19% % % % % % -8.74% 17.95% % % % % % % -4.17% % % 14.55% % % % % % % 67.15% % % % 11.14% % % -2.41% % % 4.01% 3.77% 4.55% 3.54% 2.03% 0.56% 4.51% 0.32% 2.94% 3.90% 2.36% 40.63% 0.17% 0.04% 0.17% 0.00% 7.35% 0.01% 0.47% 2.97% 3.27% 0.41% 1.73% 16.45% 4.10% 1.23% 1.00% 7.11% 0.27% 0.83% 0.12% 3.31% 1.97%

8 Estimation Results Country μ θ k α φ γ Brazil Canada Mexico Median US δ v c s f m

9 Estimation Results (cont.) Country m θ k α φ γ Austria Belgium Czech Rep. Denmark Finland France Germany Greece Hungary Ireland Israel Italy Netherlands Norway Poland Portugal Russia South Africa Spain Sweden Switzerland Thailand Turkey UK Median

10 Estimation Results (cont.) Country μ θ k α φ γ Australia China Hong Kong Indonesia Japan New Zealand Singapore Thailand India Median

11 DCC(1,1) Estimation results across geographical areas GEO AREAS α_dcc β_dcc Average Median Average Median AMERICAS EMEA ASIA-PACIFIC

12 Summary Results of the dynamic correlation for a group of economies COUNTRY July 07 to Dec. 08 % Variation of the correlation between 10 Years Sovereign Bond YTMs July 07 to Dec. 11 Jul 07 to May 12 AUSTRALIA AUSTRIA BELGIUM BRAZIL CANADA DENMARK FINLAND FRANCE GERMANY GREECE INDIA ITALY NEW ZEALAND NETHERLANDS NORWAY PORTUGAL SINGAPORE SPAIN SWEDEN UK % -8.78% -9.14% % -4.89% -3.71% -9.11% -7.17% -5.14% % % -7.45% % % 7.35% % 6.60% % 10.35% 6.21% 13.53% -9.81% % % 16.56% 2.09% -0.76% -6.27% 4.71% % 43.09% % 12.42% 4.14% 23.37% % 63.13% % 15.42% 27.56% 4.78% -9.38% % % 6.46% 2.44% 0.52% % 5.20% % 2.88% % % 3.87% 8.45% % 59.46% % 14.44% 25.27%

13 AMERICAS Spreads

14 EMEA Spreads

15 Dynamics of the correlation for the Sovereign Bond YTM Spreads of the GIPS economies

16 ASIA-PACIFIC Spreads

17 Dynamics of the Volatility for the Sovereign Bond YTM Spreads of the G7 economies

18 Conditional Volatility of the Spread for a group of economies COUNTRY % Variation Sovereign Bond Yield Volatility Spread Jan 00 to June 07 July 07 to May 12 AUSTRALIA AUSTRIA BELGIUM BRAZIL CANADA CHINA CZECH. REP. DENMARK FINLAND FRANCE GERMANY GREECE HONG KONG HUNGARY JAPAN INDIA INDONESIA ITALY MEXICO NEW ZEALAND NETHERLANDS NORWAY POLAND PORTUGAL RUSSIA SINGAPORE SOUTH AFRICA SPAIN SWEDEN UK % % % % % 16.06% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % 69.36% % % % % % % % % % 28.64% 55.93% 87.24% % % %

19 Dynamics of the Volatility for the Sovereign Bond YTM Spreads of the BRICS economies

20 Dynamics of the Volatility for the Sovereign Bond YTM Spreads of the GIPS economies

21 Simulating the 10 Years Sovereign Bond YTM annualized volatilities via Bootstrapping procedure COUNTRY Simulated YTM Annualized volatilities 4 weeks 8 weeks 12 weeks 16 weeks 32 weeks AUSTRALIA AUSTRIA BELGIUM BRAZIL CANADA CHINA CZECH REP. DENMARK FINLAND FRANCE GERMANY GREECE HONG KONG HUNGARY INDONESIA ISRAEL ITALY JAPAN MEXICO NEW ZEALAND NETHERLANDS NORWAY POLAND PORTUGAL SINGAPORE SOUTH AFRICA SPAIN SWEDEN SWITZERLAND THAILAND TURKEY UK INDIA 41.13% 56.02% 38.56% 28.40% 36.08% 25.85% 25.41% 49.41% 47.56% 42.76% 47.46% 71.06% 46.03% 20.01% 33.67% 19.83% 26.75% 30.10% 17.40% 34.06% 47.59% 61.59% 18.16% 48.17% 28.52% 17.72% 25.82% 56.03% 54.54% 20.25% 15.23% 43.58% 16.69% 39.74% 52.95% 37.47% 28.63% 35.53% 25.22% 25.18% 48.89% 46.31% 41.63% 46.59% 67.84% 45.67% 20.16% 32.71% 19.69% 26.32% 30.26% 17.39% 33.22% 46.27% 58.21% 18.21% 47.29% 28.61% 17.78% 25.49% 54.51% 53.14% 20.80% 15.29% 42.66% 16.72% 38.53% 50.30% 36.42% 28.80% 35.03% 24.81% 24.95% 48.25% 45.18% 40.67% 45.76% 64.80% 45.39% 20.24% 31.96% 19.55% 25.91% 30.38% 17.34% 32.45% 45.03% 55.32% 18.25% 46.36% 28.74% 17.83% 25.17% 53.21% 51.77% 21.30% 15.32% 41.77% 16.77% 37.48% 47.99% 35.51% 28.91% 34.55% 24.55% 24.77% 47.81% 44.13% 39.73% 45.02% 62.12% 45.06% 20.33% 31.33% 19.46% 25.51% 30.44% 17.28% 31.75% 43.98% 52.66% 18.26% 45.60% 28.79% 17.87% 24.91% 51.92% 50.55% 21.78% 15.35% 40.99% 16.77% 34.27% 41.29% 32.50% 28.99% 32.90% 23.99% 24.11% 45.88% 40.57% 36.57% 42.32% 53.98% 44.02% 20.50% 30.18% 19.24% 24.22% 30.51% 17.17% 29.55% 40.07% 45.26% 18.35% 42.77% 29.07% 17.98% 23.93% 47.53% 46.63% 23.19% 15.33% 38.23% 16.76%

22 Summary Results of the Simulated YTMs across GEOGRAPHICAL AREAS 4 weeks 8 weeks 12 weeks 16 weeks 32 weeks GEO AREAS Mean Median Mean Median Mean Median Mean Median Mean Median AMERICAS 27.99% 28.40% 27.18% 28.63% 27.06% 28.10% 26.91% 28.91% 26.35% 27.88% EMEA 39.68% 43.58% 38.66% 42.66% 37.72% 41.77% 36.88% 40.99% 34.16% 38.23% ASIA PACIFIC 30.70% 30.10% 30.33% 30.26% 30.04% 30.38% 29.77% 30.44% 29.06% 29.55%

23 Simulating annualized volatilities of the Spread for each ECONOMY COUNTRY Simulated YTM Annualized volatilities 4 weeks 8 weeks 12 weeks 16 weeks 32 weeks AUSTRALIA AUSTRIA BELGIUM BRAZIL CANADA CHINA CZECH REP. DENMARK FINLAND FRANCE GERMANY GREECE HONG KONG HUNGARY INDONESIA ISRAEL ITALY JAPAN MEXICO NEW ZEALAND NETHERLANDS NORWAY POLAND PORTUGAL SINGAPORE SOUTH AFRICA SPAIN SWEDEN SWITZERLAND THAILAND TURKEY UK INDIA 40.44% 50.79% 38.57% 55.11% 34.46% 66.93% 41.94% 36.93% 40.06% 37.81% 38.78% 68.44% 39.54% 43.10% 51.56% 63.20% 36.01% 42.43% 56.05% 41.15% 39.61% 60.89% 36.29% 47.03% 36.14% 38.21% 35.03% 47.69% 48.70% 41.41% 37.38% 37.52% 39.28% 39.64% 49.22% 38.03% 54.92% 34.09% 66.56% 41.62% 36.64% 39.32% 37.18% 38.29% 66.01% 39.38% 42.96% 50.88% 61.76% 35.74% 42.38% 55.47% 40.64% 38.82% 58.61% 36.10% 46.61% 36.11% 38.01% 34.82% 46.64% 47.70% 41.35% 37.35% 37.01% 39.06% 38.93% 47.84% 37.46% 54.60% 33.73% 66.19% 41.31% 36.25% 38.63% 36.66% 37.81% 63.57% 39.26% 42.75% 50.28% 60.57% 35.51% 42.29% 54.89% 40.13% 38.08% 56.68% 35.89% 46.04% 36.04% 37.78% 34.60% 45.77% 46.71% 41.28% 37.30% 36.48% 38.88% 38.29% 46.63% 36.97% 54.24% 33.37% 65.85% 41.00% 36.00% 38.00% 36.13% 37.38% 61.44% 39.11% 42.58% 49.68% 59.48% 35.21% 42.15% 54.33% 39.67% 37.46% 54.90% 35.69% 45.56% 35.94% 37.59% 34.38% 44.88% 45.85% 41.21% 37.20% 36.02% 38.66% 36.30% 43.07% 35.24% 52.72% 31.97% 64.44% 39.95% 34.83% 35.84% 34.31% 35.79% 54.96% 38.53% 41.80% 48.25% 56.21% 34.18% 41.54% 52.21% 38.06% 35.13% 49.92% 35.01% 43.61% 35.62% 36.73% 33.46% 41.84% 43.02% 40.85% 36.95% 34.35% 37.81%

24 Summary Results of the Simulated Annualized Volatilities of the Spread across GEOGRAPHICAL AREAS 4 weeks 8 weeks 12 weeks 16 weeks 32 weeks GEO AREAS Mean Median Mean Median Mean Median Mean Median Mean Median AMERICAS 48.54% 55.11% 48.16% 54.92% 47.74% 54.60% 47.31% 54.24% 45.63% 52.21% EMEA 44.00% 39.61% 43.26% 38.82% 42.56% 38.08% 41.92% 37.59% 39.82% 36.73% ASIA PACIFIC 44.81% 41.28% 44.55% 41.00% 44.29% 40.71% 44.03% 40.44% 43.14% 39.69%

25 Conclusions The recent policy debates regarding crisis-resolution tools, the monetary policy interventions undertaken by many central banks around the world as well as the decisions of the Troika (IMF, ECB and EU) to provide unconventional support in favor of many economies have re-pointed out the need to understand the causes of the recent Sovereign Debt Crisis as well as propose econometric indicators for monitoring and predicting the sovereign health of a country; Conditional Volatility of the Spread: the co-movement between Sovereign Bond YTMs is an unstable component during financial crisis periods; The conditional volatility of the spread started to increase from the third quarter of 2008; whereas, indicators of sovereign default probabilities, such as the CDS market, shown warning signals (higher PDs) for Greece, Italy, Spain and Portugal only in 2010; In line with the conventional wisdom, this methodology complements well known bottomup approaches (Altman et al. 2011) for monitoring the global sovereign bond markets and provide useful information for markets participants concerned about the sovereign health.

26 Insights for thinking? New estimator with higher frequency data, in order to simulate the dynamics of the vol_spread for shorter horizons and so decreasing the SMAPE (Armstrong 1985, Flores 1986) between the benchmark model and the new methodology. Sovereign Bond YTMs are traded on different stock exchanges with different currencies. (Source: Bloomberg LP) Econometric Model with asynchronicity (Schwert 1977, Scholes and Williams 1977, Engle et al. 1998). Estimating non linear time-varying metrics of dependence via a t-copula metrics (Engle et al. 2014), in order to better depict the degree of dependence between Sovereign Bond YTMs. Modeling the dynamics of the exchange rates (Favero 1997) in order to provide a methodology able to consider shocks of the exchange rates. Is the pseudo out of sample performance better than a model with an exogenous exchange rate both for short and long horizons?

27