VIDEO
Euro Crisis is Just Beginning, Global Collapse is Coming
Legal Barriers to Exit
Even if there is agreement that the transition would be smoothed by redenominating all Italian debt contracts into lira, there is the question of what exactly constitutes an Italian debt contract. Not all such contracts are between Italian debtors and Italian creditors, are issued inItaly, and specify Italian courts for adjudicating disputes. Italian companies issue bonds abroad and borrow from foreign banks. Foreign multinationals sell bonds in Italy.
Foreigners hold the bonds of Italian governments. A further complication is that contracts are not simply being redenominated from one Italian currency to another; rather, they are being redenominated from a European currency to an Italian currency. Foreign courts might therefore take EU law as the law of the currency issuer (Italy) and invalidate the redenomination of certain contracts. Mann (1960) argues that when a case involves two competing currencies, the courts should apply the law specified in the contract.For instruments such as Italian government bonds issued domestically, this is Italian law. But foreign laws govern a variety of other Italian financial instruments, such as corporate bonds issued abroad. And in some cases, no explicit choice of law is specified in the contract. For example, this is the case for loans by German banks to Italian corporations or for purchases of parts inGermanyby Italian manufacturing firms. Italian courts would presumably rule in favor of the redenomination of all loans to Italian borrowers, including those from German banks, but German courts might rule against redenomination. And there are few precedents to guide the courts’ decision in such circumstances.28 This opens the door to litigation and to an extended period of uncertainty.
Still,Argentina’s dealings with its creditors suggest that the government of a country altering its currency arrangements is in a relatively strong position. While that case also gave rise to litigation in a variety of venues, it did not force the redollarization of previously pesified contracts or force other compensation to aggrieved creditors. But cases involving suits against Italian debtors in the courts of other European countries and in the European Court of Justice could be messier. And the Italian government would be loath to disregard their judgments insofar as it attached value to the country’s other links with the European Union.
Evidence
Since 2002, the Euro barometer has conducted annual surveys of public opinion regarding the euro in the participating member states. Here, I analyze answers to the following question: “In your opinion, for [COUNTRY], is the adoption of the euro advantageous overall and will it strengthen us for the future, or rather the opposite, disadvantageous overall and will it weaken us?” Evidently, the euro is least popular, as measured here, in low- income euro area member states (Greece, Portugal) and in slowly growing economies (Italy and again Portugal) but also in the Netherlands (where concerns are disproportionately over inflation). 2006 shows regressions of the share of the population, by country and year, that views the euro as disadvantageous. The dependent variable, a log it transformation of this share, is regressed on inflation and growth in the current year.29 The results are consistent with the notion that higher inflation raises dissatisfaction with the euro, while higher growth reduces dissatisfaction. In the basic regression on pooled data, in column (1), the growth term is statistically signifi cant at conventional levels, while the inflation term is not quite significant. When year effects are added in column (2), the coefficients on both the inflation and growth terms differ significantly from 0 at standard confidence levels. When we estimate the same equation with random country effects in column (3), it is the inflation term but not the growth term that is statistically significant.
Thus, while there are not enough data to obtain precise point estimates, there are indications that slow growth and high inflation could fan dissatisfaction with membership in the euro area.30 The second empirical exercise has in fact been undertaken by Hallerberg and Wolff (2006), although they do not draw out the implications for exit from the euro area.
They test whether both membership in the monetary union and fiscal reforms that reduce deficit bias have a negative impact on sovereign borrowing costs. Thus, they speak at least obliquely to the hypothesis that a country could minimize any adverse impact on debt- servicing costs of abandoning the euro by strengthening its fiscal institutions. They estimate panel regressions with country fixed effects for ten EU member states, where the dependent variable is the yield on ten- year government bond rates relative to the corresponding German yield, and the period covered is 1993 to 2005.
This spread is regressed on the difference in the budget deficit between country andGermanyand the difference in the public debt/GDP ratio between country i andGermany. Control variables include a measure of market liquidity and a measure of global risk aversion. The key explanatory variables are then dummy variables for membership in the euro area and for the strength of fiscal institutions, which are entered by themselves and interacted with the deficit measure.31
The authors follow Von Hagen (1992) in arguing that deficit bias reflects a common- pool problem: special interests benefiting from additional public spending fail to internalize the implications for the deficit and therefore for the government’s borrowing costs. They argue that this bias can be minimized by assigning authority over the budget to a single individual, the finance minister, who will have a greater tendency to internalize such effects.
They operationalize this idea by constructing an index measuring the ability of the finance minister to affect the budget. They also consider a survey based measure of the structure of the budget process and a synthetic measure that relies not on delegation but on fiscal targets for countries where the ideological distance between coalition partners is large and therefore where delegation is unlikely to be effective.32 Results are similar for the alternative measures, so I discuss the most straightforward ones-those for delegation
of authority to the finance minister-here.
Higher debts and deficits increase spreads, although the effects are small. The effect of EMU is also evident: an increase in the deficit by 1 percent of GDP raises the spread by 4 basis points for a no euro area country but only by 1.5 percent for a euro area member. An increase in the finance minister’s powers from Portuguese to Austrian levels reduces the spread by 2 to 4 basis points; it also reduces the impact of an increase in the deficit by 1 percent of GDP by 2 basis points. These results are consistent with the hypothesis that EMU and strengthened budgetary procedures are alternative ways of strengthening fiscal discipline.33 They suggest that countries exiting the monetary union can avoid higher interest costs if they put in place efficient budgetary procedures that mitigate common- pool problems.
At the same time, the size of the effects is small. Just 4.5 additional basis points for a euro area country whose deficit grows from 0 to 3 percent of GDP makes one wonder whether these estimates are picking up the full effect or if something else is going on. One explanation for why economic policies and institutions do not have a larger impact on spreads is that the ECB carries out open market operations in the bonds of all its members, regardless of the strength of their policies and institutions; this does not force spreads to equality but may limit differentials.34
I further investigated the robustness of these results by analyzing the impact of EMU and fiscal institutions on sovereign credit ratings. This involves analyzing their impact on three credit rating measures: Fitch’s, Standard and Poor’s, and an average of the two rating agencies. In the interest of space, here I report the results using the average of the two ratings as the dependent variable.35 The country sample and period are essentially the same as in the Hallerberg and Wolff study, as the analysis is constrained by the availability of their indices of fiscal measures. One difference here is the use of quarterly data: the fiscal measures are available at a quarterly frequency, and the credit ratings can be sampled at the end of each quarter. Another difference is that I look at the absolute level of credit ratings, not ratings (or spreads) relative to Germany (and not the strength of fiscal institutions relative to Germany).36
I start with a simple panel regression of the credit rating(s) on the measure of fiscal institutions. Year fixed effects are then added, and if these are jointly significant, they are then included in the remaining regressions. Considering adds country effects (using the Hausman test to choose between fixed and random effects). Considering adds the entire vector of macroeconomic and financial variables. The empirical specification follows Christensen and Solomonsen (2007), who estimate empirical models of credit ratings; the main difference here is the addition of interaction effects for euro area countries, plus the use of total debt rather than public debt (following Hallerberg and Wolff [2006]).
Finally, I incorporate improvements in the measures of fiscal arrangements developed by the authors since the appearance of their earlier working paper.37 Specifically, – I employ three measures of fiscal arrangements: “Strong finance minister” (a measure of the power of the finance minister during budget negotiations in the cabinet and with Parliament), “Index S2” (the authors’ synthetic measure that relies not on delegation to a strong finance minister but on fiscal targets for countries where the ideological distance between coalition partners is large), and “Fiscgov” (the authors’ survey- based measure of the degree of centralization of the budgetary process). All three measures are scaled so as to vary from 0 to 1, with larger values indicating arrangements better suited for resolving common- pool problems.
The results found are broadly consistent with those using spreads as the dependent variable.38 All three measures of the centralization of fiscal policymaking are positively associated with the rating agencies’ measures of credit quality. This remains the case, except for Index S2, when a wide range of controls are included in the estimating equation.
Macroeconomic and financial conditions generally affect ratings in the expected direction, although their effects are not always significant at conventional confidence levels. Inflation, unemployment, large current account deficits, and high debts lower ratings. So far, so good.
Evidence on whether adopting the euro attenuates the impact of macroeconomic and financial imbalances on credit ratings is mixed. Consistent with the hypothesis, the negative effects of inflation and unemployment on credit ratings are attenuated by participation in the monetary union. Countries with large current account deficits suffer less in terms of credit rating if they are members of the monetary union. The one uncomfortable result is that the interaction of the EMU dummy with the debt ratio (general government- consolidated gross debt as a percentage of GDP) is negative, not positive as anticipated under the maintained hypothesis. This coefficient is 0 in the final, where the lagged dependent variable is included (as seems to be preferred by the data), which makes the result somewhat less perplexing. Sensitivity analysis-dropping countries one by one-reveals that these anomalous results are driven byBelgium. Without the observations for this one country, one obtains a negative and significant coefficient on the debt/ GDP ratio and a smaller positive and significant coefficient on the debt/ GDP ratio interacted with EMU. This is not entirely surprising in thatBelgium has long had a relatively high credit rating, despite its very high government debt, for reasons that are not entirely clear.
One interpretation of these results is that any increase in debt-servicing costs experienced by a country like Portugal that is abandoning the euro can be neutralized by reforming fiscal institutions to delegate more authority to the prime minister, by addressing concerns over the common- pool problem, and by reassuring investors that exit will not result in a loss of fiscal discipline.
The financial disincentive may not, therefore, be an insurmountable obstacle to abandoning the euro. One reason for questioning these results is that the impact of debts and deficits-euro adoption and fiscal institutions notwithstanding-are suspiciously small in these regressions, as in the earlier work of Hallerberg and Wolff on interest rate spreads.39 One worries that for whatever reason, these results are not picking up the entire effect of fiscal conditions, current and prospective, on credit ratings. But the fact that the rating agencies do not dramatically differentiate between fiscally messy Belgium and Italy and fiscally responsible Finland and Ireland is widely commented on-just as it is noted that markets differentiate between them relatively little in terms of interest rate spreads. If there is an anomaly, in other words, it would appear to be in the behavior of investors and rating agencies rather than in the econometrics.
In addition, one worry that ratings fail to reflect differences in current fiscal conditions among euro area countries, not because the euro represents a commitment to get one’s fiscal house together in the not- too- distant future, but rather because fiscally profligate governments can expect a debt bailout from their euro area partners. At the same time, the prospects for a post security can be questioned. And even if the mechanism making for rosier future prospects is a bailout rather than fiscal reform, this does not change the argument that a potential benefit of euro area membership is an easier fiscal ride. One worries that in a more turbulent environment (out of sample), the results might differ-although it is not entirely clear why the Lucas critique would apply in this context.
Finally, to the extent that fiscal rules are endogenous (to the extent that they reflect the same political pressures that lead to large observed deficits), it may be naive to think that a country abandoning the euro because of chronic deficit problems will then be able to turn around and strengthen its policy- making institutions. That said, it is interesting to observe thatItalysucceeded in significantly strengthening the ability of the finance minister to affect the budget following the 1992 crisis that ejected it from the Exchange Rate Mechanism of the European Monetary System and presumably weakened the disciplining effect of EMU on its budget.40
Finally, it is possible to compare these results with Standard & Poor’s own exercise (S&P; 2005).
Standard & Poor’s considered the impact of a country leaving the euro area in 2006 using its own proprietary model (which similarly regresses ratings on a range of indicators intended to capture political, economic, and financial conditions). It was assumed that a country leaving the euro area was able to successfully depreciate the real exchange rate, restoring it to the average level prevailing in the 1990s-something that had the effect of improving ratings, other things equal. But it was also assumed that interest rates on government debt rose by 100 basis points.
Thus, the conclusion was that leaving the euro area would have relatively little effect on ratings for lightly indebted countries that had suffered significant deteriorations in competitiveness but would have a significant negative effect on heavily indebted members whose competitiveness losses had been limited (Greece, Italy, Portugal, Spain, and Belgium). The main difference from the exercise is that S&P assumed no further change in current or expected future fiscal policies and procedures. Its analysis does not contradict the point that significant fiscal reform could offset the impact on ratings of abandoning the euro; it simply does not consider the possibility.
28. Technically, the country in which delivery is physically taken (where the transaction is physically completed) should be the one whose law governs international contracts. In the present instance, this would be German law if the Italian company’s truck drives toStuttgart to pick up parts at the German factory, but it would be Italian law if the German company’s truck is used to transport the parts to the Italian assembly plant.
29. One can imagine more sophisticated specifications, but the limited amount of data available does not really permit their estimation.
30. It is also possible to analyze the individual survey responses in order to see how sentiment toward the euro varies with education, gender, urbanization, and so forth. See Jonung and Confl itti (2008).
31. In addition, the EMU variable is interacted with the measure of market liquidity and with the debt ratio.
32. In addition, they consider a measure of the degree of the legislature or the parliament over the budget (Lienert’s [2005] parliamentary index). However, it is possible to raise questions about the relevance of this particular measure to the issues at hand. Hence, I do not consider it further in what follows.
33. The assumption underlying this interpretation is that the smaller impact of deficits on spreads in euro area countries reflects the disciplining effect of the monetary union-that deficits will not persist or that larger deficits now will be followed by smaller deficits later-rather than assuming myopia on the part of governments or that the latter will receive a debt bailout from their partners in the event of fiscal difficulties.
34. More precisely, the ECB assigns the short- term sovereign debt instruments of all euro area member governments to the same (highest) liquidity category, implying the lowest haircut when accepting them as collateral. Because the ECB mainly accepts short- term instruments in its market operations, it is these on which spreads should show the strongest tendency to converge. Spreads on the longer- term instruments considered by Hallerberg and Wolff are then freer to vary, although they will still be affected by the term- structure relationship. See Buiter and Sibert (2005).
35. The additional results for Fitch and S&P separately are available on request. The Fitch and S&P letter scores are both converted to a numerical score ranging from one to twenty- one.
36. As a result, I have an additional set of country observations forGermanyitself.
37. And that were kindly made available by Mark Hallerberg.
38. I adopt the same variable names as Hallerberg and Wolff for ease of comparison, except that I refer to the squared deviation of real GDP per capita from trend as “trend deviation” (or simply “deviation”) as opposed to “sustainability” to avoid confusion with debt sustainability.
39. Thus, an increase in the debt ratio from 50 to 100 percent of GDP is expected to lower a country’s credit rating by just one notch, say, from A to – A. This small effect is a widely commented-on phenomenon (see, for example, Buiter and Sibert [2005]), although here it applies not just to euro area but also to non euro area countries.
40. The same was true, inter alia, of Spainand Finland, according to the indices of Hallerberg and Wolff (2006).
