EARLY WARNING MODELS FOR DEBT CRISES CASE STUDY FOR ROMANIA, CZECH REPUBLIC AND HUNGARY

Transcription

1 Assistat Rodica-Oaa IONITA, PhD The Bucharest Uiversity of Ecoomic Studies Professor Dumitra STANCU Techical Uiversity of Civil Egieerig of Bucharest EARLY WARNING MODELS FOR DEBT CRISES CASE STUDY FOR ROMANIA, CZECH REPUBLIC AND HUNGARY Abstract. This paper aims to achieve those early warig idicators of debt crises i Romaia, Czech Republic ad Hugary i the period 1999, 4th quarter up to 2013, 4th quarter. Startig from three type of idicators: exteral idicators related to capital accout, exteral idicators related to curret accout ad domestic idicators, we have computed a database of potetial leadig idicators cotaiig twelve idicators for Romaia, as follows: Real effective exchage rate from tred, Relative chage of exports, Relative chage of imports, Terms of trade, Curret accout balace as % of GDP, The relative chage of foreig reserves, Capital accout balace as % of GDP, Net exteral debt as % of GDP, Govermet debt as % of GDP, Uemploymet rate, Idustrial productio growth, Net Iteratioal ivestmet positio as % of GDP. This research aim to observe which of the potetial leadig idicators used i the aalysis are sigificat i explaiig the icidece of currecy istability periods ad give us a warig regardless ay egative treds i the macroecoomic or fiacial activity, affectig the atioal or the global situatio. Usig ecoometrics techiques, we have determied those idicators which are ecoometrically sigificatly i explaiig the appearace of debt crises. Weightig their role i explaiig currecy, we have composed the early warig idex of debt crises. The evaluatio results suggest that there is a rage of leadig idicators which gave us a warig sigal regardig ay istability periods of debt crises which occur i the ecoomy. Keywords: warig, currecy crises, leadig idicators. JEL classificatio: G01, E5 1. Itroductio The purpose of this paper is to capture those idicators whose behavior ca reveal some iformatio about the occurrece of a debt.

2 Rodica Oaa Ioita, Dumitra Stacu Therefore, we computed a database cotaiig three types of idicators: exteral idicators related to curret accout, exteral idicators related to capital accout ad domestic idicators. The database was collected for the period 1999 fourth quarter util 2013 fourth quarter for Romaia. Aalysig their behaviour ad takig otes about the literature review i the field, we could assess the role of each idicator i determiig a positive or egative role i the appearace of a debt. Therefore, we idetified those variables which gave us a warig regardig the appearace of a debt ad we computed a idicator of debt warig for Romaia. This idicator was validated through the fact that it was appropriate for detectio of the last debt of Romaia. The paper is orgaized as follows: Secod sectio presets us the Literature review. Third sectio presets the Database. Fourth sectio presets the Methodology used ad the results. Fifth sectio presets the Coclusios. 2. Literature review The aalysis of Stephe G. Cecchetti, Mario Kohler ad Christia Upper based o 40 systemic bakig crises emphasizes that last fiacial crises is computed from a wide rage of ecoomic factors. The paper This Time is Differet: A Paoramic View of Eight Ceturies of Fiacial Crises icludes Africa, Asia, Europea, Lati America coutries, North America ad Oceaia, i the period ad threat exteral debt, domestic default, bakig crises, currecy crashes ad iflatio excesses, by costructig a composite idex of fiacial istability that is multidimesioal, cocludig that fiacial crises are more a way of life affectig all. Aother paper of Carme M. Reihart ad Keeth S. Rogoff is focused o bakig crises ad highlights that crises are more severe for the fiacial ceters like UK, USA ad Frace. They aalyze for the first time i the literature the role of housig prices variables ad fid similarities of the behavior of frequecy ad duratio of bakig crises betwee developed ad middle-icome coutries, highlightig that most coutries experiece a surge i debt i the wake of a fiacial, with real cetral govermet debt icreasig 86% o average durig the three years followig the. Some Romaia ecoomists used a set of prudetial idicators ad the aggregate moetary balace sheet to fid out that the level of risk was maageable, eve that accelerated durig 2009 ad 2010 ad that the exposure of Romaia s baks to foreig fuds costituted a importat source of risk. Professor Albulescu Claudiu Tiberiu build a early warig system based o the bakig ratigs deterioratio usig CAAMPL approach for the period ad cocluded that ratig dowgrade ad calculatio of probability of bakig fiacial distress ca be determied through a early warig system. Last revealed importat aspects i the curret global architecture ad official mechaism that facilitate global fiacial istability. Curret crises emphasizes that the surveillace for crises prevetio must be more rigorous, with a better icorporatio of fiacial sector ad regulatory issues, with better iformatio

3 Early Warig Models for Crises Case Study for Romaia, Czech Republic ad Hugary regardig cross-border spillover ad systemic risks. If prior to the curret crises were cosidered just vulerabilities i emergig market ecoomies because they are supposed to be more fragile, ow there are take i cosideratio the vulerabilities i advaced ecoomies too, because they could create broader distress through cross border likages. 3. Database The database cotais three types of idicators: exteral idicators related to curret accout, exteral idicators related to capital accout ad domestic idicators. It was collected from Eurostat Database for Romaia aalysig the period 1999 fourth quarter util 2013 fourth quarter, o a quarterly basis. The ratioale for which the idicators where quarterly collected it was that they could give us a warig at least few quarters before the evet happeed. Firstly, we started the model by collectig mothly idicators, but because ot all the idicators were available i mothly frequecy, thus could make us arrow our view obtaied through the database collected. We have also cosidered that a aually database would t help us, because it is very difficult to fid those idicators which gave us a warig with years before the evet happeed, existig the possibility to omit those idicators which really gave us warig with moths or quarter before. Therefore, we chose to use quarterly data for a better efficiecy of the issued sigals cosistig i the fact that allows timely reactio from the decisio factors, as policy makers. The idicators used are preseted i the bellow table. Table 1.Idicators Type of idicator Idicator Exteral idicator (curret accout) Real effective exchage rate from tred Exteral idicator (curret accout) Relative chage of exports Exteral idicator (curret accout) Relative chage of imports Exteral idicator (curret accout) Terms of trade Exteral idicator (curret accout) Curret accout balace as % of GDP Exteral idicator (capital accout) The relative chage of foreig reserves Exteral idicator (capital accout) Capital accout balace as % of GDP Domestic idicator Net exteral debt as % of GDP Domestic idicator Govermet debt as % of GDP

4 Rodica Oaa Ioita, Dumitra Stacu Domestic idicator Uemploymet rate Domestic idicator Idustrial productio growth Domestic idicator Net Iteratioal ivestmet positio as % of GDP Source: Eurostat Database Aalyzig their behavior ad takig otes about the literature review i the field, we could assess the role of each idicator i determiig a positive or egative role i the appearace of a debt. 4. Methodology 4.1 Hypothesis of research The mai hypothesis of the research represets the fact that there are potetial leadig idicators which could gave us warig sigals regardig the appearace of a debt crises. If the behaviour of some idicators is aalysed i the cotext of the specific ecoomy, we ca idetify some vulerabilities ad threats which could have a egative effect agaist the ecoomy of a coutry. Therefore, i the followig pages we have review the behaviour of the potetial leadig idicators: a. Real effective exchage rate from tred represets a measure of iteratioal competitiveess of each coutry. It is used a proxy i order to determie the over or uder evaluatio. Hypothesis1.1: A overvalued real exchage rate represets a high probability of appearace of currecy. Therefore, we have cosidered that a icrease of real effective exchage rate represets a higher probability of appearace of currecy, therefore, eters with positive sig i the debt crises debt idex. Hypothesis1.2: A udervalued real exchage rate represets a lower probability of appearace of currecy.

5 Early Warig Models for Crises Case Study for Romaia, Czech Republic ad Hugary Figure 1. Real effective exchage rate deviatio from tred Source: Eurostat database, ow calculatios a. Relative chage of exports represets a measure of iteratioal competitiveess of a coutry. A decrease i exports growth may be determied by a overvaluatio of atioal (domestic) currecy. If the decrease of exports growth is produced by other reaso that exchage rate, thus ca put pressure o depreciatio of atioal currecy. Hypothesis2: I both cases, the decrease of exports growth is cosidered a potetial leadig idicator for depreciatio of the atioal currecy, therefore it eters with egative sig i the debt crises idex. Figure 2. Export growth Source: Eurostat database, ow calculatios

6 Rodica Oaa Ioita, Dumitra Stacu b. Relative chage of import growth represets a measure of iteratioal depedece of a coutry. Hypothesis3: A icrease of this idex ca determie the depreciatio of atioal currecy ad therefore ca determie a high probability of occurrece of currecy ad debt. Figure 3. Import growth Source: Eurostat database, ow calculatios c. Terms of trade idex are determied as a fractio betwee exports of a coutry ad imports of the same coutry. Hypothesis4.1: The icrease of this idex leads to the improvemet of the balace of paymets of a coutry, havig a decreased probability of occurrece of a. Hypothesis4.2: A deterioratio of this idex ca determie a icrease probability of occurrece of currecy ad debt Figure 4. Terms of trade Source: Eurostat database, ow calculatios

7 Early Warig Models for Crises Case Study for Romaia, Czech Republic ad Hugary d. A icrease of curret accout balace as % of GDP is associated with massive iflows of capital which are itermediated by atioal fiacial system ad which ca facilitate the appearace of asset price booms ad credit booms. Hypothesis5: The curret accout surplus reveals a low probability of depreciatio ad therefore a low probability of occurrece of debt. Figure 5. Curret accout balace as percetage of GDP Source: Eurostat database, ow calculatios e. A icrease of capital accout balace as % of GDP is associated with massive iflows of capital which are itermediated by atioal fiacial system ad which ca facilitate the appearace of asset price booms ad credit booms. Hypothesis6: The capital accout excedet reveal a low probability of depreciatio ad therefore a low probability of occurrece of debt.

8 Rodica Oaa Ioita, Dumitra Stacu Figure 6. Capital accout balace as percetage of GDP Source: Eurostat database, ow calculatios f. The relative chage of foreig reserves. The decrease of foreig exchage reserves is a idicator which maifests pressure o depreciatio of atioal currecy. The total value of foreig exchage reserves reveals the ability of a coutry to fulfil its foreig debts obligatio. Hypothesis7: Therefore a icrease i foreig exchage reserves eters with egative sig i debt equatio. (I some cases, Natioal Baks ca take measures i order to support atioal currecy). Figure 7. Growth of foreig exchage reserves Source: Eurostat database, ow calculatios

9 Early Warig Models for Crises Case Study for Romaia, Czech Republic ad Hugary g. The high leverage degree of a coutry s et exteral debt (as % of GDP) idicates higher vulerabilities of the ecoomy, outflows of capital ad therefore the icrease probability of occurrece of. Hypothesis8: Therefore it eters with positive sig i the debt occurrece. Figure 8. Idustrial productio growth Source: Eurostat database, ow calculatios h. The share of govermetal debt i GDP. The high leverage degree of a coutry s govermet idicates higher vulerabilities of the ecoomy, outflows of capital ad therefore the icrease probability of occurrece of. Hypothesis9: Therefore it eters with positive sig i the debt occurrece.

10 Rodica Oaa Ioita, Dumitra Stacu Figure 9. Govermetal debt as percetage of GDP Source: Eurostat database, ow calculatios i. GDP per capita/ Idustrial productio growth. We have used idustrial productio growth as proxy for GDP per capita. Hypothesis10: A icrease of idustrial productio growth determies positive effects of the ecoomy. Figure 10. Net exteral debt as percetage of GDP Source: Eurostat database, ow calculatios j. Icreased uemploymet rate may be associated with macroecoomic threats which ca affect the ecoomy through they re cross effects o the other macroecoomic variables. Hypothesis11: Therefore, it eters with positive i debt idex.

11 Early Warig Models for Crises Case Study for Romaia, Czech Republic ad Hugary Figure 11. Uemploymet rate Source: Eurostat database, ow calculatios k. Chage i et iteratioal ivestmet positio as % of GDP. A icrease i a coutry s ivestmet positio determies a decrease probability of occurrece of. Hypothesis12: Therefore, it eters with egative sig i the debt idex. Figure 12. Chage i et iteratioal ivestmet positio as % of GDP Source: Eurostat database, ow calculatios

12 Rodica Oaa Ioita, Dumitra Stacu 4.2 Assessmet Takig ito cosideratio the above evolutios of each idicator, we assessed their role i the occurrece of a debt, as follows. Therefore, we idetified those variables which gave us a warig regardig the appearace of a debt ad we computed a idicator of debt warig for Romaia. This idicator was validated through the fact that it was appropriate for detectio of the last debt crises of Romaia. Table 2. Crt. No. Idicator Assessmet 1 Real effective exchage rate from tred A overvalued real exchage rate represets a high probability of appearace of currecy. Therefore, we have cosidered that a icrease of real effective exchage rate represets a higher probability of appearace of currecy, therefore, eters with positive sig i the debt crises debt idex. A udervalued real exchage rate represets a lower probability of appearace of currecy. (+) 2 Relative chage of exports A decrease i exports growth may be determied by a overvaluatio of atioal (domestic) currecy If the decrease of exports growth is produced by other reaso that exchage rate, thus ca put pressure o depreciatio of atioal currecy. I both cases, the decrease of exports growth is cosidered a potetial leadig idicator for depreciatio of the atioal currecy; therefore it eters with egative sig i the debt crises idex. (-)

13 Early Warig Models for Crises Case Study for Romaia, Czech Republic ad Hugary 3 Relative chage of import growth 4 Terms of trade 5 6 Curret accout balace as % of GDP Capital accout balace as % of GDP A icrease of this idex ca determie the depreciatio of atioal currecy ad therefore ca determie a high probability of occurrece of currecy ad debt. The idex is determied as a fractio betwee exports of a coutry ad imports of the same coutry. The icrease of this idex leads to the improvemet of the balace of paymets of a coutry, havig a decreased probability of occurrece of a. A deterioratio of this idex ca determie a icrease probability of occurrece of currecy ad debt. A icrease of curret accout balace as % of GDP is associated with massive iflows of capital which are itermediated by atioal fiacial system ad which ca facilitate the appearace of asset price booms ad credit booms. The curret accout surplus reveals a low probability of depreciatio ad therefore a low probability of occurrece of debt. A icrease of capital accout balace as % of GDP is associated with massive iflows of capital which are itermediated by atioal fiacial system ad which ca facilitate the appearace of asset price booms ad credit booms. The capital accout surplus reveals a low probability of depreciatio ad therefore a low probability of occurrece of debt. (+) (-) (-) (-)

14 Rodica Oaa Ioita, Dumitra Stacu The relative chage of foreig reserves Net exteral debt as % of GDP The share of govermetal debt i GDP GDP per capita/ Idustrial productio growth Uemploymet rate The decrease of foreig exchage reserves is a idicator which maifests pressure o depreciatio of atioal currecy. The total value of foreig exchage reserves reveals the ability of a coutry to fulfil its foreig debts obligatio. Therefore a icrease i foreig exchage reserves eters with egative sig i debt equatio. I some cases, Natioal Baks ca take measures i order to support atioal currecy. The high leverage degree of a coutry s et exteral debt idicates higher vulerabilities of the ecoomy, outflows of capital ad therefore the icrease probability of occurrece of. Therefore it eters with positive sig i the debt occurrece. The high leverage degree of a coutry s govermet idicates higher vulerabilities of the ecoomy, outflows of capital ad therefore the icrease probability of occurrece of. Therefore it eters with positive sig i the debt occurrece. We have used as proxy for GDP per capita idustrial productio growth. A icrease of idustrial productio growth determies positive effects of the ecoomy. Icreased uemploymet rates may be associated with macroecoomic threats which ca affect the ecoomy through they re cross effects o the other macroecoomic variables. Therefore, it eters with positive i debt idex. (-) (+) (+) (-) (+)

15 Early Warig Models for Crises Case Study for Romaia, Czech Republic ad Hugary 12 Chage i et iteratioal ivestmet positio as % of GDP A icrease i a coutry s ivestmet positio determies a decrease probability of occurrece of. Therefore, it eters with egative sig i the debt idex. (-) 4.3 Computatio of the debt idex Therefore we have computed a idex which cotais each of the above idicators, with positive or egative sig depedig o their role i explaiig the appearace of a debt, accordig to the table assessmet. We expect that a icrease of this value to reflect a cost agaist the ecoomy, thus represetig a distress situatio, therefore a debt period. The equatio of Crisis Idex (which is defied i the same way for the three coutries i research) is preseted below: DCI = Where: REER EXP IMP TT CAGDP TAGDP i 1 I 1 i 1 i 1 i 1 i 1 i 1 NDEBT GOVGDP i 1 i 1 i 1 UNEMP i 1 INDPROD DCI = Crisis Idex i=1, = time (expressed i quarters) REER = Real effective exchage rate from tred EXP = Relative chage of exports IMP = Relative chage of imports TT = Terms of trade CAGDP = Curret accout balace as % of GDP TAGDP = Capital accout balace as % of GDP GRTRES =Relative chage of foreig reserves NDEBT = Net exteral debt as % of GDP GOVGDP = Govermet debt as % of GDP UNEMP = Uemploymet rate INDPROD = Idustrial productio growth NINTINV = Net Iteratioal ivestmet positio as % of GDP i NINTINV GRTRES

16 Rodica Oaa Ioita, Dumitra Stacu As it is visible from the bellow table, all of the three coutries experieces icreases of debt pressure as follows: for Romaia startig with 2 d quarter of 2007, for Hugary ad Czech Republic startig with 4 th quarter of The highest poit of pressure was reached i 1st quarter of 2009 for Hugary, i 1 st quarter of 2011 for Czech Republic ad i 4 th quarter of 2011 for Romaia. The differece i the momet of appearace of debt pressure as well as the differet momet i which the highest poit of the pressure is reached is explaied by other factors which ifluece the ecoomy of those coutries, as well as the measures of the Govermet to dimiish the egative effects, which are also differet from oe state to aother. Figure 13. idex Source: Eurostat database, ow calculatios 4.4 Computatio of the warig We have estimated a regressio for each coutry i order to see if all of those potetial leadig idicator ifluece the warig debt idex. Accordig to the estimatio results, the hypothesis of the research that there are potetial leadig idicators which ca brig us a warig regardig the debt period are cofirmed. I the bellow table there are preseted the coefficiets with which the variables ifluece the Crisis Idex ad their p-value (as the p-value is closer to 0, the sigificace of that variable i explaiig the model is higher). The results of the ecoometric estimatio also cofirms us that we have take ito accout that the variable ifluece a debt idex i right directio (either positive or egative), aother hypothesis of the model which was cofirmed.

17 Early Warig Models for Crises Case Study for Romaia, Czech Republic ad Hugary Figure 14. Comparative aalysis of the regressio models betwee the DCI ad the 12 potetial leadig idicators Coutry Czech Republic Hugary Romaia Coeff. P- P- P- Coeff. Coeff. value value value Itercept REER Deviatio from tred 2. Exports of goods ad services 3. Imports of goods ad services 4. Terms of trade Curret accout balace i % of GDP 6. Capital accout i % of GDP 7. Growth of foreig exchage reserves 8. Volume idex of productio Govermet debt i % of GDP 10. Net exteral debt i % GDP Net iteratioal ivestmet

18 Rodica Oaa Ioita, Dumitra Stacu positio i % of GDP - quarterly data 12. Uemployme t rate * Semificat at 1% ** Semificat at 5% ***Semificat at 10% Please see the results of regressio performed for all coutries icluded i the research. Adjusted R2 is if 97,67% for Czech Republic ad Hugary ad 100% for Romaia, this meaig that the models are described i high proportio by the potetial leadig idicators. Thus is o oe side explaied by the fact that, we have aalysed at this step the ecoometric ifluece of the potetial variables which have bee already icluded i the defiitio of the Crisis Idex. I a prior research, where other few variables (macroecoomic ad fiacial variables) were added, the explaatory power of the models decreased at about 70%, because ot all of them where importat i explaiig the occurrece of a debt. Figure 15. Comparative aalysis of the explaatory power of the models Czech Hugary Romaia Coutry Republic Regressio Statistics Multiple R R Square Adjusted R Square Stadard Error Observatios Source: Eurostat Database, ow calculatios Based o the coefficiets provided by the ecoometric estimatio, we have composed the early warig idex of debt for each coutry, as follows: DCW= REER EXP IMP TT CAGDP TAGDP GRTRES i 1 I 1 i 1 i 1 i 1 i 1 i 1 i 1 2 GOVGDP i 1 i 1 Where, UNEMP i 1 INDPROD i NINTINV NDEBT

19 Early Warig Models for Crises Case Study for Romaia, Czech Republic ad Hugary DCW= Crisis Warig I this case the debt warig surprise a highest cost agaist the ecoomy, thus beig icreased by the coefficiet of the govermet debt, revealig the fact that the govermet debt idicator is oe of the most importat i revealig a warig regardig the probable occurrece of a debt. Figure 16. Crisis Warig Source: Eurostat Database, ow calculatios We have also composed for each coutry a VAR betwee EWI ad each potetial idicator ad a VAR for FWI ad each potetial idicator i order to allow for bidirectioal causality. I observed the idexes respose to a shock i each potetial leadig idicator. Usig orthogoal impulse respose fuctios I set the lag for each idicator equal to the lead where the respose fuctio reaches its maximum, without takig ito cosideratio its sig or its statistical sigificace. I allow for a miimum lag legth of twelve moths assumig that a potetial leadig idicator provides a early warig oly if it predicts icidece at least oe year ahead so that timely policy actio ca be take. Takig i cosideratio the fact that the lag must be eough for the early warig, but also avoidig excessive parameterizatio I cosidered miimum lag of 4 quarters, because oly if it is idetified i time a egative aspect i the ecoomy ca be treated. After completig the aalysis of VAR the key idea is that eve if the potetial idicators are sigificat i explaiig the icidece, ot all of them give importat sigals regardig its predictio. Ad also aother importat aspect is that this model has t the same efficiecy for all the coutries i the sample. This ca be caused by the fact that the

20 Rodica Oaa Ioita, Dumitra Stacu maifested differet i those coutries both at idicators levels ad direct cosequeces at the ecoomy level. The mai importat leadig idicators are GDP growth rate, real effective exchage rate from tred ad exteral debt. Accordig to the aalysis, there are some sigificat variables which are very importat i explaiig the occurrece of a debt which give importat sigals to policy makers i order to limit potetial damages which ca appear through such a. I the future, a special attetio should be grated to permaetly moitorig potetial leadig idicators, reactig to the sigals received ad iclude those idicators i the policy measures udertake, i order to cover all the areas of the ecoomy ad to protect agaist potetial risks. 5. Coclusios This aalysis reveals the hypothesis of the research, that there are potetial leadig idicators which gave us a warig i idetifyig the momets of debt pressure of the ecoomy. Summarizig the mai results, we foud that debt / istability warig sigals come at various horizos. Aother importat aspect to metio is that this model as it is defied - has t the same efficiecy for all the coutries i the sample. There are variables which are sigificat i explaiig the debt of oe coutry, eve that for the other coutry they have ot such impact. Oe explaatio for that fact ca be the decisios take i each coutry, which ca differ both as measures take, but also as effects eeded to be achieved. REFERENCES [1]Arellao, C. ad N.R. Kocherlakota (2008), Iteral Crises ad Sovereig Defaults; Workig Paper 13794; [2]Babecy, J., T. Havraek, J. Mateju, M. Rusak, K. Smidkova, B. Vasicek (2011), Early Warig Idicators of Ecoomic Crises-Evidece from a Pael of Developed Coutries. Workig Paper Series, Natioal Bak of Czech; [3]Berg,A., E. Boresztei ad C. Pattilo (2004), Assessig Early Warig Systems: How Have They Worked i Practice?; IMF Workig Paper/04/52; [4]Cecchetti, S.G., M. Kohler ad C. Upper (2009),Fiacial Crises ad Ecoomic Activity; NBER Workig Paper Series; [5]DeLog,J.B. (2009), The Fiacial Crisis of : Uderstadig its Causes, Cosequeces-ad its Possible Cures; [6]

21 Early Warig Models for Crises Case Study for Romaia, Czech Republic ad Hugary [7]Isarescu, Mugur (2006), Cotributios to the Theory of Macroecoomic Stabilizatio, Reflectios of Ecoomic Stabilizatio; Goveror of the Natioal Bak of Romaia; [8]Iteratioal Moetary Fud [9]Ioita, Rodica, Oaa (2012), Disertatio : Early warig idicators of ecoomic crises; [10]Krugma, P. (1979), A Model of Balace-of-Paymets Crises; Joural of Moey,Credit ad Bakig 11(3), pp ; [11]Krugma,P. (2000), The Retur of Depressio Ecoomics; [12]Kamisky, G., S. Lizodo ad C. Reihart (1998), Leadig Idicators of Currecy Crises. [13]Meegatti,C.ad N.Roubii (2006), Vulerabilities i Cetral ad Souther Europe; [14]Natioal Bak of Romaia, mothly reports [15]Reihart, C.M. ad K.S. Rogoff(2008), This Time is Differet: A Paoramic view of eight Ceturies of Fiacial Crises; NBER Workig Paper.