versus $75.70. Moreover, unemployment rates are more than double in the South compared to the West, as delineated by the vast difference between 14.29% and 6.88% on average. This has been an underlying problem in the South, especially, in Spain, after the 2008 global financial crisis, where unemployment grew at accelerated rates, as indicated by the raw data on unemployment.
market, which can spill over to the financial option markets.76 On the other hand, Western CDS spreads exhibited a statistically significant (5%) negative relation to the MOVE index of -1.5475.
This contends economic intuition; however, a plausible explanation might be that contagion or spillover between the bond and financial derivatives market is less prevalent in the West, allowing investors to maintain confidence. Naifar (2020) derived similar results, and his explanation to the inverse relation was due to “bullish markets”. This conforms with our theory, as bullish markets are characterized by confidence. Hence, further information on investor sentiments within the bond market can shed light on their respective reactions’ volatilities.
The effects of GVZ, the gold market volatility index, induces noteworthy positive reactions within Southern European CDS spreads, in contrast with the inverse relation with Western CDS spreads. This can also be justified by: (i) the magnitude of contagion between gold volatilities and the financial market, and (ii) investor reactions. The implication of elevated gold volatility creates inflated borrowing costs, and institutes directionally adverse pressures on the aggregate market mechanisms of domestic currency, which imposes fluctuations on inflation. Theoretically, this can be affiliated with their distinct gross debt levels and international reserves, which accumulate a ratio of 1.652 debt-to-reserves in the Southern region, compared to the West, who endure an efficient ratio of 0.896 approximately, enabling volatilities in the gold price to entail relatively more intensified inverse effects in Western Europe. The global interest rate affects the South at comprehensively higher rates than the West, as depicted by the excessive inverse reaction of -102.66 and -46.38 respectively. This potentially suggests that the US treasury rate plays a stronger role in less-advanced economies. A justifiable explanation might be that the more developed economies of the West, generally retain interest rates that are within the same wavelength as the global, hence minimizing volatility, uncertainty, and investment planning errors, while those in less-advanced economies face a greater hit.
Furthermore, the current account balance displays the least effect on CDS spreads across all variables consistently with an 0.0005 coefficient in the South and 0 in the West. Due to the trivial nature of their effect, we will conclude that CDS spreads are independent of current account balances for both groups. This implies that a country’s foreign trading activity and foreign investment settlements are not interrelated with credit risk, which is not incorrect, as economically one does not stimulate nor hinder the other, and there are no past literatures that dispute this
76Naifar, “What explains sovereign CDS spread changes in the GCC region”
finding. Similarly, terms of trade manifested notable negative coefficients for Southern nations, while Western nations display a trivial positive relation to CDS spreads. Whereas a 1-unit increase in the terms of trade index diminishes CDS spreads by 8.24 bps in the Southern region and increases CDS spreads in the Western region by 2.30 bps approximately. Next, the coefficient for gross external debt has been found to be inversely associated with CDS spreads, which makes unquestionable economic sense, as greater accumulated levels of debt hinder credit ratings, which implements upward pressure on CDS spreads to insure against debt in cases where the economic environment fails to foster required growth, and where higher interest rates and investor risk aversion prevails. Gross domestic product (GDP) was found to be consistent with economic philosophy, as both groups displays inverse coefficients, meaning that higher GDP rates drive down CDS spreads, however, at a less intense rate in the West relative to South. The Western GDP coefficient was found statistically significant at 0.1%. This result is verified by Brandorf and Holberg’s findings (2010) that implied higher GDP figures were associated with lower CDS spreads in the PIIGS block. Likewise, since GDP and inflation rate are directly associated as described in the previous section, inflation coefficients were found to entail negative effects on CDS spreads as well, with both coefficients significant at 1%. Another measure that instrumentalizes economic activity, industrial production, yielded negative effects on CDS premiums, as advancements in aggregate output signal that an economy is cultivating economic growth, which will provide a foundation on which it can repay its debt efficiently.
The TED spreads, which provide an insight on the interbank credit market liquidity, exhibit negative, but insignificant effects in the South, whilst provides positive significant (5%) effects in the West. This conforms with the empirical findings of Naifar (2020), where increases in TED spreads stimulate growth in CDS spreads in bullish markets with confident environments, opposed to bearish markets with pessimistic atmospheres. Hence, the imbalanced impact of TED spreads signifies that Southern markets are essentially associated with narrower liquidity, contrary to Western markets with wider liquidities. The notion of international reserves manifested inverse associations with CDS spread levels, with -1.027 bps and -0.281 bps in the Southern and Western European regions respectively. This is logically verifiable as greater amounts of reserves in the form of gold, foreign currency, deposits, and securities provide countries with armor to which they can repay debt, avoid default, and alleviate credit ratings. Nevertheless, both were statistically significant at 5%.
Moreover, the S&P credit ratings were found to be independent of Western CDS spreads, whilst demonstrated a comprehensively negative effect on Southern countries. This finding empirically makes sense, as higher credit ratings signify risk-minimal, investment-grade sovereigns, which systematically implies credit risk, and thus lower CDS spreads to compensate.
The rate of unemployment was found to have positive impacts on CDS spreads, as predicted, since lower employment rates generally translates to lower economic growth rates, which as demonstrated frequently, is inversely related to CDS spreads. The Southern countries displayed higher sensitivity rates, as delineated by the larger and statistically significant (5%) coefficient value of 31.26 relative to 5.68.
Furthermore, when assessing the global covariates solely, in regression (3) and (4), the coefficient signs remained unchanged for all six determinants, however, increased in absolute value, suggesting that macroeconomic factors do encompass amplifying effects on global covariates. Other noticeable changes are that Western coefficients displayed enhanced statistical significance: VIX and MOVE were significant at 1%, and US10Y at 5%. Moreover, in regression (5) and (6), which evaluated the impact of macroeconomic factors solely, statistical significance changed for the Southern group this time, where inflation became significant at 0.1%, credit ratings and unemployment at 1%. All coefficients retained the same sign, however, also, decreased in absolute value, except for debt, terms of trade, credit rating, and inflation, which increased in value, delineating that, in the absence of global covariates, their magnitude on CDS spreads widens, this may be justified by the theory that global uncertainty and market volatilities deviates a portion of the effects away from these factors and to the indices. Finally, regressions (7) and (8), which account for all variables in crisis periods, exhibited similar results to regressions (1) and (2), with constant coefficient signs, constant statistical significance, but coefficient values dropped in absolute value, possibly alluding to risk-shifting from the covariates to the crisis period.
Two important figures that must be evaluated are the R-squared statistic of the regressions, and the standard errors. Firstly, the R-squared in regression (1) and (2) were 0.8994 and 0.8668, respectively. This implies that, despite utilizing the exact same regressing variables, the covariates perform a slightly better job in explaining CDS spreads in the Southern countries, relative to the Western ones, however, they are technically identical. This suggests that either the choice of variables were best-fitting for both groups, due to their indistinguishable R-squared value, and the effective magnitude of the explanatory power. This miniscule disparity between western and
southern R-squared values is present in all regressions, however, as mentioned, it is insignificant.
It is potentially connected to the fact that Southern countries entail larger standard errors, across all variables and time, signifying that the estimates are less representative of the population, as opposed to the West, who retain little to no standard errors, hence are closely distributed around the population mean. So, the focus will shift on comparing the R-squared between the global covariates and the country-specific macroeconomic factors. Regressions (3) and (4) produce an squared value of 0.4157 and 0.4016 respectively, while regressions (5) and (6) produce an R-squared of 0.8437 and 0.7825. Hence, concluding that country-specific factors play a much larger role in explaining CDS spreads, to a substantial extent, compared to global financial variables, as exhibited by the doubled R-squared values. This corresponds to the findings of Hilschier and Nosbuch (2010), who similarly deduced that macroeconomic factors have substantially superior explanatory power in determining sovereign risk relative to global factors.77 Panel regressions (7) and (8) generated explanatory power worth 0.9139 and 0.8726 accordingly, indicating that R-squared values have risen only slightly from regressions (1) and (2). This suggests that the crisis variable does provide trivial, disputable input into explaining CDS spread levels, but not as comprehensive as expected.
In the context of our four hypotheses, these empirical findings fully reinforce and support two propositions (hypothesis 2 and 4), reject one proposition (hypothesis 1), and is neutral with hypothesis 3. Firstly, hypothesis 1 proposes that global factors will effectuate larger response rates (elasticity) in the Western European CDS spreads through larger beta coefficients in absolute value due to their elite and controlling status in the global economy. This hypothesis is unanimously rejected by all regression outcomes, as Southern European CDS spreads depict more intense coefficients as delineated by panels (3) and (4) in Table 6. Moreover, hypothesis 2 suggests that country-specific macroeconomic factors will impose intensified response rates within the Southern regions relatively, due to their less stable macroeconomic conditions, which exposes them to shocks and fluctuations. This hypothesis was supported by our empirical findings, as deduced by panels (5) and (6) that demonstrated larger coefficient absolute values for Southern European CDS spreads. Hypothesis 3 proposed that the crisis variables in panels (7) and (8) would contribute effective explanatory power, which did occur, only on an extremely trivial level, hence, this hypothesis is semi-accepted, yet not fully supported to the insignificant difference in R-squared.
77Hilscher and Nosbusch, “Determinants of sovereign risk”, 247-253
Finally, hypothesis 4 theorized that country-specific macroeconomic factors would be more competent and sufficient in explaining CDS spread levels relative to global financial market variables, which is supported by the tremendous disparity in their R-squared values, which sides with the macroeconomic factors.