- Email: [email protected]
Comment on “Fundamentally Wrong: Market Pricing of Sovereigns and the Greek Financial Crisis”
Journal of Macroeconomics 39 (2014) 420–423
Contents lists available at ScienceDirect
Journal of Macroeconomics journal homepage: www.elsevier.com/locate/jmacro
Comment on ‘‘Fundamentally Wrong: Market Pricing of Sovereigns and the Greek Financial Crisis’’ Thanassis Kazanas, Elias Tzavalis ⇑ Department of Economics, Athens University of Economics and Business, Greece
a r t i c l e
i n f o
Article history: Available online 21 September 2013 JEL classification: C22 C52 E43 E62 G12 Keywords: Credit spreads Financial crisis Euro area Credit ratings
a b s t r a c t Gibson’s et al. (2013) provide evidence that credit ratings have exerted an independent influence on credit (sovereign) spreads for Greece beyond that implied by economic fundamentals. Based on the Markov Regime-switching model of Hamilton (1989), we show that this happens during the recent financial crisis regime, characterized by a higher mean and volatility of credit spreads. It is also true for Ireland and Portugal, also bailed out by their EU partners and IMF. We show that, for Greece and Portugal, the shift of credit spreads to their higher mean-volatility regime occurred before the collapse of Lehman brothers, thus discounting a higher price of sovereign credit risk for these two countries. In contrast to Ireland, this regime shift has not been triggered by a rating downgrades for Greece and Portugal. In this higher volatility regime, credit ratings seem to significantly influence future changes in credit spreads independently of economic fundamentals, for Greece and Portugal. For Ireland, they constitute the main factor of determining credit spreads. Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction There is recently growing interest in investigating the determinants of credit (sovereign) spreads of EMU countries relative to Germany and, in particular, of Greece, Ireland and Portugal bailed out by their EU partners and IMF (see, e.g., Arghyrou and Kontonikas, 2012; De Santis, 2012; Bernoth and Erdogan, 2012; Afonso et al., 2013). Most of these studies show that the factors affecting the EMU credit spreads are associated with aggregate (mainly international), country-specific and contagion sources of risk. The country-specific risks are related to economic fundamentals such as fiscal and/or other macroeconomic imbalances, which increase the likelihood of a country to default on its sovereign debt. Gibson’s, Hall and Tavlas paper (2013) examines if credit ratings, announced by credit rating agencies, exert an independent impact on credit spreads, over-and-above that of the above economic fundamentals for Greece. This is an interesting question given that credit ratings are determined by movements in the above economic fundamentals such as fiscal imbalances, competitiveness, debt sustainability and economic growth. As aptly argued by De Santis (2012), credit ratings can also bring contagion risk to the force. The paper finds that, indeed, credit ratings have a significant impact on the EMU credit spreads for Greece, beyond that implied by economic fundamentals. Based on Kalman filter estimation procedure (or recursive least squares, see an earlier version of the paper), the authors indicate that this result is even stronger after the outbreak of the recent international financial crisis in year 2008.
⇑ Corresponding author. Address: Department of Economics, Athens University of Economics and Business, 76 Patission Street, 104 34 Athens, Greece. Tel.: +30 2108203332. E-mail address: [email protected] (E. Tzavalis). 0164-0704/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jmacro.2013.09.002
421
T. Kazanas, E. Tzavalis / Journal of Macroeconomics 39 (2014) 420–423
In this short note, we examine how robust is Gibson’s et al. (2013) above result with respect to explicitly modeling a regime shift in the relationship between credit spread and economic fundamentals. To this end, we re-estimate this relationship for Greece allowing for a regime shift in its conditional mean and volatility functions, based on Hamilton’s (see Hamilton (1989)) Markov regime-switching model (MRS). This is also done for Portugal and Ireland, the two other countries bailed by their EU partners and IMF. The MRS model can reveal if the large in magnitude, time-varying changes of the coefficients of the credit spread and economic fundamental relationship of the above three EMU periphery countries, found by Gibson’s et al. paper for Greece, can be explained by economic fundamentals or credit ratings in the different regimes of our sample, identified by applying the MRS model to our data. 2. Estimation of the credit spread – economic fundamentals relationship allowing for regime-switching Table 1 presents estimates of the following credit spread relationship for the above three EMU-periphery countries, without allowing for regime-switching:
sprjt ¼ constj þ bj2
gdjt1 gbjt1 cajt1 þ bj3 þ bj4 ipg jt1 þ bj5 þ bj6 crjt1 þ ejt ; gdpjt1 gdpjt1 gdpjt1
ð1Þ
for j = {Greece, Ireland and Portugal}, where sprjt ¼ rjt r GE t is the credit spread between the 10-year government bond yield of country j and that of Germany (GE),
gdjt1 gdpjt1
ca
; gdpjt1 and jt1
gbjt1 gdpjt1
stand for the government deficit (gd), current account (ca) and
government debt (gb) as ratio to GDP, respectively, IPGjt1 is the annual growth rate of industrial production (ipg) and, finally, cr jt1 ¼ cr jt1 av erageðcr jt1 Þ is a variable capturing the impact of a new rating at time t 1 (denoted as crjt) compared to the average (average) of those of the last twelve-months. To construct credit ratings variable crjt, we use the ratings assigned to each country j by Moody’s, S& P and Fitch. We assign values of 22-1 to different rating of the above three agencies and we extract their common factor, based on principal components analysis. This approach of measuring crjt mitigates any small, indiosyncratic differences of ratings across the three agencies on our results. Our frequency of our data is monthly and covers period 2001:02–2012:12. The three ratio to gdp variables are given in quarterly basis, and thus have been interpolated, as in Gibson’s et al. (2013). gb ca The results of Table 1 indicate that there are economic fundamentals, such as gdpjt1 , or gdpjt1 and ipgjt1 for Greece at 10% jt1
jt1
level, which influence future credit spread changes one-period ahead. However, with the exception of variables have the incorrect sign. Apart from
gbjt1 , gdpjt1
gbjt1 , gdpjt1
the other two
the results of Table 1 also indicate that the credit rating news variable,
cr jt1 , has also an important effect on credit spread sprjt. Its sign is also negative and it is consistent with the theory, meaning that a rating downgrade of a country’s sovereign will lead to an increase of sprjt. Note that, in terms of magnitude, the effects of cr jt1 on sprjt are found to be larger for Greece. The results of the MRS specification of model (1) are given in Table 2. This specification assumes that all parameters of model (1) switch between two regimes, denoted as st = ‘‘0’’ and st = ‘‘1’’, respectively. The first regime is characterized by a lower level of the volatility and conditional mean functions of spread sprjt, while the second by a higher. Note that the table presents estimates of the slope coefficient of cr jt1 only in regime ‘‘1’’, given that these are found to be insignificant in regime ‘‘0’’. The results of Table 2 and Fig. 1 lead to a number of interesting conclusions. First, the very small values of the transition probabilities between regimes ‘‘0’’ and ‘‘1’’, denoted as p01 and p10, reported in the table indicate that there is a small transition probability across them. This result implies a high degree of persistency of each regime, during our sample. This can be also confirmed by the inspection of Fig. 1, which presents smoothed over-the-whole-sample estimates of the prob-
Table 1 Estimates of model (1). constj
gdjt1 gdpjt1
gbjt1 gdpjt1
cr jt1
R2
Greece 12.77 (2.54)
0.04 (0.49)
0.13 (1.75)
0.27 (1.87)
0.14 (2.69)
4.08 (2.57)
0.82
Portugal 1.36 (2.17)
0.03 (0.85)
0.013 (0.34)
0.03 (0.96)
0.03 (1.98)
2.94 (6.11)
0.90
Ireland 0.49 (1.28)
0.014 (0.76)
0.03 (1.10)
0.030 (1.31)
0.023 (2.01)
1.87 (4.70)
0.90
cajt1 gdpjt1
ipgjt1
Notes: Sample period: 2001:02–2012:12, t statistics are in parentheses correct for White-heteroscedasticity and Newey-West standard errors allowing for one lag.
422
T. Kazanas, E. Tzavalis / Journal of Macroeconomics 39 (2014) 420–423
Table 2 Estimates of the Markov regime-switching specification of model (1).
Greece st = ‘‘0’’ st = ‘‘1’’ Portugal st = ‘‘0’’ st = ‘‘1’’ Ireland st = ‘‘0’’ st = ‘‘1’’
constj
gdjt1 gdpjt1
cajt1 gdpjt1
ipigjt1
gbjt1 gdpjt1
0.63 (3.00) 7.68 (1.20)
0.001 (0.56) 0.13 (0.81)
0.0000 (0.04) 0.22 (2.20)
0.0099 (1.92) 0.39 (1.44)
0.009 (4.18) 0.13 (2.39)
0.65 (8.54) 6.46 (2.93)
0.004 (1.44) 0.10 (2.62)
0.007 (2.13) 0.10 (1.79)
0.030 (6.28) 0.11 (1.98)
0.009 (6.95) 0.09 (3.12)
1.22 (23.88) 1.59 (1.64)
0.003 (1.36) 0.004 (0.31)
0.04 (7.41) 0.05 (1.12)
0.008 (1.90) 0.022 (0.39)
0.004 (6.78) 0.008 (0.55)
cr jt1
pj01
pj10
r2j0
r2j1
0.02 (7.01)
0.01 (9.12)
0.06 (11.40)
4.05 (11.33)
0.02
0.01
0.06 (10.58)
0.90 (10.59)
0.08 (6.32)
0.02 (8.23)
0.15 (15.01)
0.83 (8.64)
3.63 (3.64)
2.33 (8.37)
1.74 (8.36)
Notes: Sample period: 2000:12–2012:12, t statistics are in parentheses based on Quasi ML standard errors.
Fig. 1. Smoothed probabilities of regime ’’1’’ (Pstar) versus credit rating news (CR).
ability that the economy is in regime ‘‘1’’ in each point of time of our sample. This probability is labeled as PSTAR in Fig. 1. Second, by allowing for regime-switching in model (1), economic fundamentals seem to affect future changes in credit spread sprjt. Note that some of them are significant even in the low volatility regime ‘‘0’’ (i.e., and
cajt1 gdpjt1
for Ireland and Portugal). Some of these fundamentals (i.e.,
gbjt1 gdpjt1
for Greece, and also
gbjt1 , gdpjt1
ipgjt1, for all countries,
gdjt1 gdpjt1
and ipgjt1 for Portugal)
can still predict changes of sprjt in regime ‘‘1’’. Third, credit rating news cr jt1 clearly influences sprjt in the high-volatility credit spread regime ‘‘1’’, for all three countries examined. Note that, for Ireland, cr jt1 constitutes the main variable determining future changes in sprjt for regime ‘‘1’’. This can be attributed to the fact that the sovereign debt crisis of Ireland was mainly due to the insolvency of the banking system triggered by the recent financial crisis and not on the government and/or current account imbalances of this country.
423
T. Kazanas, E. Tzavalis / Journal of Macroeconomics 39 (2014) 420–423
Finally, inspection of the graphs of Fig. 1, which also presents the values of variable cr jt through our sample, indicates that the shift of credit spread sprjt to regime ‘‘1’’ for Greece and Portugal was occurred before the collapse of Lehman brothers in September of year 2008. This shift can be taken to reflect a higher price of Greek and Portuguese sovereign credit risk than gd
regime ‘‘0’’ due to the steady deterioration of the economic fundamentals of these two countries, mainly gdpjt1 , started before jt1
year 2008. Credit rating news cr jt1 have appeared affecting these countries almost a year after the collapse of Lehman’s brothers. However, for Ireland, cr jt1 seems to play a critical role in determining the regime shift of sprjt to regime ‘‘1’’, immediately after the collapse of Lehman brothers. This is consistent with our interpretation, given above, that the origin of the Irish debt sovereign crisis may be attributed to the insolvency of this country’s banking system, mainly triggered by the start of the recent international financial crisis. 3. Conclusions Based on the Markov regime-switching model, we show that credit spreads of Greece and Portugal have moved to a new regime of higher mean and volatility even before the start of the recent financial crisis. For these countries, credit ratings are found to critically determine the level of credit spreads in this regime independently of the underlying economic fundamentals, as argued by Gibson’s et al. (2013) for Greece. However, there are not found to trigger this regime shift. In contrast, for Ireland credit ratings played an important role in determining the above regime shift. We argue that this may happen due to the insolvency problems of the Irish banking system, faced immediately after the collapse of Lehman brothers. Acknowledgement The authors would like to thank Michael Argyrou and George Tavlas, as well as participants of the Bank of Greece conference on ‘‘the crisis of euro area’’ held in Athens from 23–24 of May 2013 for useful comments. References Afonso, A., Arghyrou, M.G., Bagdatoglou, G., Kontonikas, A., 2013. On the Time-Varying Relationship Between EMU Sovereign Spreads and their Determinants. Mimeo. Cardiff Business School, UK. Arghyrou, M.G., Kontonikas, A., 2012. The EMU sovereign-debt crisis: fundamentals, expectations and contagion. Journal of International Financial Markets Institutions and Money 22, 658–677. Bernoth, K., Erdogan, B., 2012. Sovereign bond yield spreads: a time-varying coefficient approach. Journal of International Money and Finance 31, 639–656. De Santis, R., 2012. The euro area sovereign debt crisis: safe haven, credit rating agencies and the spread of the fever from Greece, Ireland and Portugal. ECB Working Paper 1419. Gibson, H.D., Hall, S.G., Tavlas, G.S., 2013. Fundamentaly wrong? The economic fundamentals and sovereign spreads during the Greek financial crisis. Journal of Macroeconomics (this volume). Hamilton, J., 1989. A new approach to the economic analysis of nonstationary time series and business cycle. Econometrica 57, 357–384.
Contents lists available at ScienceDirect
Journal of Macroeconomics journal homepage: www.elsevier.com/locate/jmacro
Comment on ‘‘Fundamentally Wrong: Market Pricing of Sovereigns and the Greek Financial Crisis’’ Thanassis Kazanas, Elias Tzavalis ⇑ Department of Economics, Athens University of Economics and Business, Greece
a r t i c l e
i n f o
Article history: Available online 21 September 2013 JEL classification: C22 C52 E43 E62 G12 Keywords: Credit spreads Financial crisis Euro area Credit ratings
a b s t r a c t Gibson’s et al. (2013) provide evidence that credit ratings have exerted an independent influence on credit (sovereign) spreads for Greece beyond that implied by economic fundamentals. Based on the Markov Regime-switching model of Hamilton (1989), we show that this happens during the recent financial crisis regime, characterized by a higher mean and volatility of credit spreads. It is also true for Ireland and Portugal, also bailed out by their EU partners and IMF. We show that, for Greece and Portugal, the shift of credit spreads to their higher mean-volatility regime occurred before the collapse of Lehman brothers, thus discounting a higher price of sovereign credit risk for these two countries. In contrast to Ireland, this regime shift has not been triggered by a rating downgrades for Greece and Portugal. In this higher volatility regime, credit ratings seem to significantly influence future changes in credit spreads independently of economic fundamentals, for Greece and Portugal. For Ireland, they constitute the main factor of determining credit spreads. Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction There is recently growing interest in investigating the determinants of credit (sovereign) spreads of EMU countries relative to Germany and, in particular, of Greece, Ireland and Portugal bailed out by their EU partners and IMF (see, e.g., Arghyrou and Kontonikas, 2012; De Santis, 2012; Bernoth and Erdogan, 2012; Afonso et al., 2013). Most of these studies show that the factors affecting the EMU credit spreads are associated with aggregate (mainly international), country-specific and contagion sources of risk. The country-specific risks are related to economic fundamentals such as fiscal and/or other macroeconomic imbalances, which increase the likelihood of a country to default on its sovereign debt. Gibson’s, Hall and Tavlas paper (2013) examines if credit ratings, announced by credit rating agencies, exert an independent impact on credit spreads, over-and-above that of the above economic fundamentals for Greece. This is an interesting question given that credit ratings are determined by movements in the above economic fundamentals such as fiscal imbalances, competitiveness, debt sustainability and economic growth. As aptly argued by De Santis (2012), credit ratings can also bring contagion risk to the force. The paper finds that, indeed, credit ratings have a significant impact on the EMU credit spreads for Greece, beyond that implied by economic fundamentals. Based on Kalman filter estimation procedure (or recursive least squares, see an earlier version of the paper), the authors indicate that this result is even stronger after the outbreak of the recent international financial crisis in year 2008.
⇑ Corresponding author. Address: Department of Economics, Athens University of Economics and Business, 76 Patission Street, 104 34 Athens, Greece. Tel.: +30 2108203332. E-mail address: [email protected] (E. Tzavalis). 0164-0704/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jmacro.2013.09.002
421
T. Kazanas, E. Tzavalis / Journal of Macroeconomics 39 (2014) 420–423
In this short note, we examine how robust is Gibson’s et al. (2013) above result with respect to explicitly modeling a regime shift in the relationship between credit spread and economic fundamentals. To this end, we re-estimate this relationship for Greece allowing for a regime shift in its conditional mean and volatility functions, based on Hamilton’s (see Hamilton (1989)) Markov regime-switching model (MRS). This is also done for Portugal and Ireland, the two other countries bailed by their EU partners and IMF. The MRS model can reveal if the large in magnitude, time-varying changes of the coefficients of the credit spread and economic fundamental relationship of the above three EMU periphery countries, found by Gibson’s et al. paper for Greece, can be explained by economic fundamentals or credit ratings in the different regimes of our sample, identified by applying the MRS model to our data. 2. Estimation of the credit spread – economic fundamentals relationship allowing for regime-switching Table 1 presents estimates of the following credit spread relationship for the above three EMU-periphery countries, without allowing for regime-switching:
sprjt ¼ constj þ bj2
gdjt1 gbjt1 cajt1 þ bj3 þ bj4 ipg jt1 þ bj5 þ bj6 crjt1 þ ejt ; gdpjt1 gdpjt1 gdpjt1
ð1Þ
for j = {Greece, Ireland and Portugal}, where sprjt ¼ rjt r GE t is the credit spread between the 10-year government bond yield of country j and that of Germany (GE),
gdjt1 gdpjt1
ca
; gdpjt1 and jt1
gbjt1 gdpjt1
stand for the government deficit (gd), current account (ca) and
government debt (gb) as ratio to GDP, respectively, IPGjt1 is the annual growth rate of industrial production (ipg) and, finally, cr jt1 ¼ cr jt1 av erageðcr jt1 Þ is a variable capturing the impact of a new rating at time t 1 (denoted as crjt) compared to the average (average) of those of the last twelve-months. To construct credit ratings variable crjt, we use the ratings assigned to each country j by Moody’s, S& P and Fitch. We assign values of 22-1 to different rating of the above three agencies and we extract their common factor, based on principal components analysis. This approach of measuring crjt mitigates any small, indiosyncratic differences of ratings across the three agencies on our results. Our frequency of our data is monthly and covers period 2001:02–2012:12. The three ratio to gdp variables are given in quarterly basis, and thus have been interpolated, as in Gibson’s et al. (2013). gb ca The results of Table 1 indicate that there are economic fundamentals, such as gdpjt1 , or gdpjt1 and ipgjt1 for Greece at 10% jt1
jt1
level, which influence future credit spread changes one-period ahead. However, with the exception of variables have the incorrect sign. Apart from
gbjt1 , gdpjt1
gbjt1 , gdpjt1
the other two
the results of Table 1 also indicate that the credit rating news variable,
cr jt1 , has also an important effect on credit spread sprjt. Its sign is also negative and it is consistent with the theory, meaning that a rating downgrade of a country’s sovereign will lead to an increase of sprjt. Note that, in terms of magnitude, the effects of cr jt1 on sprjt are found to be larger for Greece. The results of the MRS specification of model (1) are given in Table 2. This specification assumes that all parameters of model (1) switch between two regimes, denoted as st = ‘‘0’’ and st = ‘‘1’’, respectively. The first regime is characterized by a lower level of the volatility and conditional mean functions of spread sprjt, while the second by a higher. Note that the table presents estimates of the slope coefficient of cr jt1 only in regime ‘‘1’’, given that these are found to be insignificant in regime ‘‘0’’. The results of Table 2 and Fig. 1 lead to a number of interesting conclusions. First, the very small values of the transition probabilities between regimes ‘‘0’’ and ‘‘1’’, denoted as p01 and p10, reported in the table indicate that there is a small transition probability across them. This result implies a high degree of persistency of each regime, during our sample. This can be also confirmed by the inspection of Fig. 1, which presents smoothed over-the-whole-sample estimates of the prob-
Table 1 Estimates of model (1). constj
gdjt1 gdpjt1
gbjt1 gdpjt1
cr jt1
R2
Greece 12.77 (2.54)
0.04 (0.49)
0.13 (1.75)
0.27 (1.87)
0.14 (2.69)
4.08 (2.57)
0.82
Portugal 1.36 (2.17)
0.03 (0.85)
0.013 (0.34)
0.03 (0.96)
0.03 (1.98)
2.94 (6.11)
0.90
Ireland 0.49 (1.28)
0.014 (0.76)
0.03 (1.10)
0.030 (1.31)
0.023 (2.01)
1.87 (4.70)
0.90
cajt1 gdpjt1
ipgjt1
Notes: Sample period: 2001:02–2012:12, t statistics are in parentheses correct for White-heteroscedasticity and Newey-West standard errors allowing for one lag.
422
T. Kazanas, E. Tzavalis / Journal of Macroeconomics 39 (2014) 420–423
Table 2 Estimates of the Markov regime-switching specification of model (1).
Greece st = ‘‘0’’ st = ‘‘1’’ Portugal st = ‘‘0’’ st = ‘‘1’’ Ireland st = ‘‘0’’ st = ‘‘1’’
constj
gdjt1 gdpjt1
cajt1 gdpjt1
ipigjt1
gbjt1 gdpjt1
0.63 (3.00) 7.68 (1.20)
0.001 (0.56) 0.13 (0.81)
0.0000 (0.04) 0.22 (2.20)
0.0099 (1.92) 0.39 (1.44)
0.009 (4.18) 0.13 (2.39)
0.65 (8.54) 6.46 (2.93)
0.004 (1.44) 0.10 (2.62)
0.007 (2.13) 0.10 (1.79)
0.030 (6.28) 0.11 (1.98)
0.009 (6.95) 0.09 (3.12)
1.22 (23.88) 1.59 (1.64)
0.003 (1.36) 0.004 (0.31)
0.04 (7.41) 0.05 (1.12)
0.008 (1.90) 0.022 (0.39)
0.004 (6.78) 0.008 (0.55)
cr jt1
pj01
pj10
r2j0
r2j1
0.02 (7.01)
0.01 (9.12)
0.06 (11.40)
4.05 (11.33)
0.02
0.01
0.06 (10.58)
0.90 (10.59)
0.08 (6.32)
0.02 (8.23)
0.15 (15.01)
0.83 (8.64)
3.63 (3.64)
2.33 (8.37)
1.74 (8.36)
Notes: Sample period: 2000:12–2012:12, t statistics are in parentheses based on Quasi ML standard errors.
Fig. 1. Smoothed probabilities of regime ’’1’’ (Pstar) versus credit rating news (CR).
ability that the economy is in regime ‘‘1’’ in each point of time of our sample. This probability is labeled as PSTAR in Fig. 1. Second, by allowing for regime-switching in model (1), economic fundamentals seem to affect future changes in credit spread sprjt. Note that some of them are significant even in the low volatility regime ‘‘0’’ (i.e., and
cajt1 gdpjt1
for Ireland and Portugal). Some of these fundamentals (i.e.,
gbjt1 gdpjt1
for Greece, and also
gbjt1 , gdpjt1
ipgjt1, for all countries,
gdjt1 gdpjt1
and ipgjt1 for Portugal)
can still predict changes of sprjt in regime ‘‘1’’. Third, credit rating news cr jt1 clearly influences sprjt in the high-volatility credit spread regime ‘‘1’’, for all three countries examined. Note that, for Ireland, cr jt1 constitutes the main variable determining future changes in sprjt for regime ‘‘1’’. This can be attributed to the fact that the sovereign debt crisis of Ireland was mainly due to the insolvency of the banking system triggered by the recent financial crisis and not on the government and/or current account imbalances of this country.
423
T. Kazanas, E. Tzavalis / Journal of Macroeconomics 39 (2014) 420–423
Finally, inspection of the graphs of Fig. 1, which also presents the values of variable cr jt through our sample, indicates that the shift of credit spread sprjt to regime ‘‘1’’ for Greece and Portugal was occurred before the collapse of Lehman brothers in September of year 2008. This shift can be taken to reflect a higher price of Greek and Portuguese sovereign credit risk than gd
regime ‘‘0’’ due to the steady deterioration of the economic fundamentals of these two countries, mainly gdpjt1 , started before jt1
year 2008. Credit rating news cr jt1 have appeared affecting these countries almost a year after the collapse of Lehman’s brothers. However, for Ireland, cr jt1 seems to play a critical role in determining the regime shift of sprjt to regime ‘‘1’’, immediately after the collapse of Lehman brothers. This is consistent with our interpretation, given above, that the origin of the Irish debt sovereign crisis may be attributed to the insolvency of this country’s banking system, mainly triggered by the start of the recent international financial crisis. 3. Conclusions Based on the Markov regime-switching model, we show that credit spreads of Greece and Portugal have moved to a new regime of higher mean and volatility even before the start of the recent financial crisis. For these countries, credit ratings are found to critically determine the level of credit spreads in this regime independently of the underlying economic fundamentals, as argued by Gibson’s et al. (2013) for Greece. However, there are not found to trigger this regime shift. In contrast, for Ireland credit ratings played an important role in determining the above regime shift. We argue that this may happen due to the insolvency problems of the Irish banking system, faced immediately after the collapse of Lehman brothers. Acknowledgement The authors would like to thank Michael Argyrou and George Tavlas, as well as participants of the Bank of Greece conference on ‘‘the crisis of euro area’’ held in Athens from 23–24 of May 2013 for useful comments. References Afonso, A., Arghyrou, M.G., Bagdatoglou, G., Kontonikas, A., 2013. On the Time-Varying Relationship Between EMU Sovereign Spreads and their Determinants. Mimeo. Cardiff Business School, UK. Arghyrou, M.G., Kontonikas, A., 2012. The EMU sovereign-debt crisis: fundamentals, expectations and contagion. Journal of International Financial Markets Institutions and Money 22, 658–677. Bernoth, K., Erdogan, B., 2012. Sovereign bond yield spreads: a time-varying coefficient approach. Journal of International Money and Finance 31, 639–656. De Santis, R., 2012. The euro area sovereign debt crisis: safe haven, credit rating agencies and the spread of the fever from Greece, Ireland and Portugal. ECB Working Paper 1419. Gibson, H.D., Hall, S.G., Tavlas, G.S., 2013. Fundamentaly wrong? The economic fundamentals and sovereign spreads during the Greek financial crisis. Journal of Macroeconomics (this volume). Hamilton, J., 1989. A new approach to the economic analysis of nonstationary time series and business cycle. Econometrica 57, 357–384.