عوامل تعیین کننده پیش فرض مستقل: تجزیه و تحلیل حساسیت

A vast and growing empirical literature aims at identifying key determinants of sovereign default. The literature is extensive and controversial. Can policy-makers use this body of research to learn anything that can help reduce the likelihood of sovereign default? We use a variant of Extreme Bound Analysis (EBA) to examine if any of the conclusions from the existing studies on the determinants of sovereign default is robust to small changes in the conditioning information set. Our EBA, spanning 190 countries over 1970–2010, upholds the robustness of the observed association between sovereign default and credit worthiness, growth, leverage on export earnings, debt service ratio, reserves, inflation, exchange rate, trade deficit, corruption, and democratic accountability. At the same time, our EBA reveals that the correlations between sovereign default and several of the controversial variables (namely, openness, central bank liabilities, interest payments, cost of borrowing, imports, exports, per capita GNP, and government stability) are highly sensitive to small alterations in the conditioning information set.

Concerns regarding sovereign defaults have been mounting through this millennium. While the current financial crisis often hits a raw nerve on probable causes and remedies of debt default, the first such recorded case dates back to the fourth century BC when more than three-quarters of the Greek Municipalities could not meet their debt obligations to the Delos temple (Winkler (1933)). Since then, over millennia, many countries have faced difficulties in servicing their debt. However, the most recent past decade has witnessed a phenomenal accumulation of global external debt, skyrocketing from $2 trillion in 2000 to $60 trillion in 2011. Consequently, the ratio of external debt to GDP jumped from 4.57% in 2000 to 72.5% in 2011 and per capita external debt increased from $329 in 2000 to $8700 in year 2011. During this period, the largest annual increase of external debt was in 2005, multiplying more than 5 times from 2004. As shown in Figs. 1and 2, despite improvements in the per capita output, default risk has concurrently been rising with external debt outpacing growth: the higher a country's external debt, the higher is its likelihood of not being able to make the payments which inflates accumulated arrears and raises the risk to default. Full-size image (19 K) Fig. 1. External debt per GDP and external debt growth. Figure options Full-size image (14 K) Fig. 2. External debt per capita. Figure options With the sharp rise in sovereign default risk of many countries in the world, it is not surprising that a large and growing body of research continues to look into the potential determinants of default risk of external debt. The lack of a general consensus over the conclusions reached by the wide range of empirical studies as to the relative importance and the direction of impact of the potential determinants of sovereign default can be explained, to some extent, in terms of the wide differences in perspectives, methodologies, sample selection and analytical tools. On top of the heterogeneity in approaches, these empirical studies, like many others in different fields of economics, rather unfortunately, form examples of “measurement without theory”: variables are searched for that show a significant influence on sovereign default, and the results are explained ex post. This has resulted in a diverse and somewhat unwieldy literature where most investigators have considered only a small number of explanatory variables at a time in an attempt to establish a statistically significant relationship between sovereign default and a particular variable or a set of variables of interest. Moreover, none of these studies meaningfully control, consciously or otherwise, for all the variables analyzed by early researchers as potential candidates explaining sovereign default. A wide range of variables has been studied and observed to be significantly correlated with sovereign default in different directions in different studies. All this not only points to obvious statistical and conceptual weaknesses prevalent in the empirical literature on the determinants of sovereign default, but also leaves the motivated reader rather perplexed, at the end, as to the confidence that should be placed in the findings of any particular study. It, therefore, becomes imperative to examine whether one can have any confidence in the conclusions reached by the existing studies on the determinants of sovereign default. To that end, we use a variant of Leamer's (1983) Extreme Bound Analysis (EBA) in determining which coefficients of the explanatory variables studied in the existing studies on the determinants of sovereign default are “robust” and which are “fragile” to small changes in the conditioning information set.1 The rest of this paper is organized as follows. In the next section, we describe the data and estimation technique and explain the methodology used in the sensitivity analysis. In Section 3, we discuss the role of the explanatory variables used in the EBA. This section may also be seen as a structured review of the relevant literature, cataloged in Table 1. In Section 4, we present key findings. In Section 5, we draw our conclusions. Table 1. The literature. Explanatory variables Observed effect on debt default Positive Negative Insignificant Reserves/debt Rahnama-Moghadam (1995) Georgievska et al. (2008) Taffler and Abassi (1984) Thacker (1999) Catao and Sutton (2002) Edwards (1984) Bird and Rowlands (2001) Afonso, Gomes, and Rother (2007) Hilscher and Nosbusch (2008) Total external debt/GDP (GNP) Peter (2002) Rivoli and Brewer (1997) Georgievska et al. (2008) Li (1992) Maltritz and Molchanov (2014) Thacker (1999) Balkan (1992) Karayalcin and Temel (1988) Kruger and Messmacher (2004) Oatley (2002) Detragiache and Spilimbergo (2001) Taffler and Abassi (1984) Kohlscheen (2010) Oatley and Yackee (2000) Aylward and Thorne (1998) Bird and Rowlands (2001) Feder and Uy (1985) Hilscher and Nosbusch (2008) Edwards (1984) Solberg (1988) Haque, Kumar, Mark, and Mathiesonm (1996) Celasun and Harms (2010) Lee (1991) Callier (1985) Reserves/imports Hajivassiliou (1987) Feder, Just, and Ross (1981) Balkan (1992) Rivoli and Brewer (1997) Frank and Cline (1971) Celasun and Harms (2010) Edwards (1984) Sargen (1977) Feder and Uy (1985) Saini and Bates (1984) Karayalcin and Temel (1988) Euh (1979) Elmore and McKenzie (1992) Detragiache and Spilimbergo (2001) Haque et al. (1996) Hajivassiliou (1989) Li (1992) Hajivassiliou (1994) McFadden et al. (1985) Aylward and Thorne (1998) Marashaden (1997) Kohlscheen (2010) Oatley and Yackee (2000) Bird and Rowlands (2001) Maltritz and Molchanov (2014) Outward orientation (openness) Maltritz and Molchanov (2014) Callier (1985) Celasun and Harms (2010) McFadden et al. (1985) Thacker (1999) Total debt service/exports Georgievska et al. (2008) Elmore and McKenzie (1992) Rahnama-Moghadam (1995) McFadden et al. (1985) Maltritz and Molchanov (2014) Taffler and Abassi (1984) Karayalcin and Temel (1988) Aylward and Thorne (1998) Detragiache and Spilimbergo (2001) Hajivassiliou (1989) Kohlscheen (2010) Oatley (2002) Snider (1990) Saini and Bates (1984) Thacker (1999) Frank and Cline (1971) Sargen (1977) Euh (1979) Edwards (1984) Berg and Sachs (1988) Solberg (1988) Feder et al. (1981) Balkan (1992) Marashaden (1997) Hajivassiliou (1987) Cline (1984) Rivoli and Brewer (1997) Catao and Sutton (2002) Bird and Rowlands (2001) Karagol and Sezgin (2004) Inflation rate Aylward and Thorne (1998) Hajivassiliou (1994) Gartner, Griesbach, and Jung (2011) Balkan (1992) Peter (2002) Celasun and Harms (2010) Sargen (1977) Panizza, Strurzenegger, and Zettelmeyer (2009) Saini and Bates (1984) Edwards (1984) Maltritz and Molchanov (2014) Money and quasi money/int. Hajivassiliou (1994) Cline (1984) reserves or GDP Callier (1985) Citron and Nickelsburg (1987) (money supply growth) Peter (2002) Thacker (1999) Saini and Bates (1984) Balkan (1992) Sargen (1977) Maltritz and Molchanov (2014) Real interest rate on Intl. lending Detragiache and Spilimbergo (2001) Hajivassiliou (1994) Lee (1991) Callier (1985) Peter (2002) Bird and Rowlands (2001) McFadden et al. (1985) Maltritz and Molchanov (2014) Imports/GDP Elmore and McKenzie (1992) Solberg (1988) Edwards (1984) Aylward and Thorne (1998) Saini and Bates (1984) McFadden et al. (1985) Frank and Cline (1971) Karayalcin and Temel (1988) Hajivassiliou (1987) Maltritz and Molchanov (2014) Exports/GDP Karagol and Sezgin (2004) Aylward and Thorne (1998) Feder et al. (1981) Cline (1984) Citron and Nickelsburg (1987) McFadden et al. (1985) Elmore and McKenzie (1992) Balkan (1992) Hajivassiliou (1987) Rivoli and Brewer (1997) Maltritz and Molchanov (2014) GDP growth rate Detragiache and Spilimbergo (2001) McFadden et al. (1985) Catao and Sutton (2002) Hajivassiliou (1987) Afonso et al. (2007) Karayalcin and Temel (1988) Kruger and Messmacher (2004) Lee (2009) Sargen (1977) Gartner et al. (2011) Euh (1979) Bird and Rowlands (2001) Hilscher and Nosbusch (2008) Edwards (1984) Feder and Uy (1985) Haque et al. (1996) Kohlscheen (2010) Lee (1991) Li (1992) Celasun and Harms (2010) Maltritz and Molchanov (2014) Total external debt/exports Peter (2002) Karayalcin and Temel (1988) Marashaden (1997) Oatley (2002) McFadden et al. (1985) Oatley and Yackee (2000) Kruger and Messmacher (2004) Berg and Sachs (1988) Hajivassiliou (1987) Lee (2009) Hajivassiliou (1989) Elmore and McKenzie (1992) Euh (1979) Hajivassiliou (1994) GDP (GNP) per capita Rahnama-Moghadam (1995) Aylward and Thorne (1998) Bird and Rowlands (2001) Oatley and Yackee (2000) Feder and Uy (1985) Sargen (1977) Celasun and Harms (2010) Berg and Sachs (1988) Elmore and McKenzie (1992) Thacker (1999) Frank and Cline (1971) McFadden et al. (1985) Edwards (1984) Karayalcin and Temel (1988) Saini and Bates (1984) Euh (1979) Hajivassiliou (1994) Hajivassiliou (1987) Balkan (1992) Gartner et al. (2011) Panizza et al. (2009) Maltritz and Molchanov (2014) Real effective exchange rate Catao and Sutton (2002) Edwards (1984) Afonso et al. (2007) Haque et al. (1996) Solberg (1988) Hajivassiliou (1994) Peter (2002) Bird and Rowlands (2001) Sargen (1977) Current account/GDP (exports) Hajivassiliou (1989) Kruger and Messmacher (2004) Detragiache and Spilimbergo (2001) Oatley (2002) Peter (2002) Hajivassiliou (1994) Edwards (1984) Bird and Rowlands (2001) Haque et al. (1996) Callier (1985) Marashaden (1997) Thacker (1999) Karayalcin and Temel (1988) Lee (2009) Panizza et al. (2009) Oatley and Yackee (2000)

This paper is a natural follow-up on the exploratory empirical investigations of the determinants of sovereign default. We systematically evaluate the robustness of the partial correlation between the sovereign default and a wide assortment of economic indicators. It marks the first attempt to assess the claims of a vast and growing body works on potential determinants sovereign default. Although we do not claim that our exercise constructs a “proof” to validate some of the findings of the existing empirical literature on the determinants of sovereign default, our EBA does uphold the robustness of the observed association between sovereign default and credit worthiness, growth, leverage on export earnings, debt service, and inflation. Such evidence should encourage policy makers, with intent on reducing sovereign debt, to focus on these variables. At the same time, our EBA reveals that the correlations between sovereign default and several of the controversial variables that have been reported in the empirical literature (namely, openness, central bank liabilities, interest payments, cost of borrowing, imports, exports, per capita GNP, and government stability) are highly sensitive to small alterations in the conditioning information set. It is important to interpret these findings with caution since some studies have imposed exclusion restrictions on certain variables for sound theoretical reasons and, therefore, continue to offer important lessons even when the empirical results have not been found to be robust in our EBA exercise.