APPENDIX: Assessment of Country Credit Risk: Methodologies Used by Rating Agencies
Basu, K., “The International Debt Problem, Credit Rationing and Loan Pushing: Theory and Experience,” Princeton Studies in International Finance No. 70, October 1991.
Brewer, T. L. and P. Rivali, “Politics and Perceived Country Creditworthiness in International Bombing,” Journal of Money. Credit and Banking, August 1990.
Calvo, Guillermo, Leonardo Leiderman, and Carmen M. Rainhart, “Capital Inflows to Latin America: The Role of External Factors,” IMF Staff Papers, March 1993, p.108–51.
Cooper, J., “Country Creditworthiness: The Use of Cluster Analysis and Discriminant Analysis,” Crlasgow Department of Economics Discussion Paper, No.6, September 1987.
Dooley, M., Eduardo Fernandez-Areas and Kenneth Kletzer, “Is the Debt Crisis History? Recent Private Capital Inflows to Developing Countries” draft, World Bank, 1995.
Eaton J. and M. Gerouritz, “Debt with Potential Republiation: Theoretical and Empirical Analysis,” Review of Economic Studies, April 1981.
Feder, A., and L. Uy, “The Determinants of International Creditworthiness and their Policy Implications,” Journal of Policy Modelling, October 1984.
Frankel, Jeffrey, “Why Haven’t Interest Rates in Latin American and Asian Countries Converged to World Levels? Recent Portfolio Capital Inflows and Stabilization,” Mimeo. November 1994.
Froot, K., “Buybacks, Exit Bonds, and the Optimality of Debt and Liquidity Relief,” International Economic Review, February 1989.
Gibbons, Jean Dickinson (1993) “Nonparametric Measures of Association” Sage University Paper series on Quantative Applications in the Social Sciences, 07-091. Newbury Park, CA: Sage.
Guidotti, Pablo and Manmohan S. Kumar, Domestic Public Debt of Externally Indebted Developing Countries, IMF Occasional Paper 80, June 1991.
Kharas, H. “The Long Run Creditworthiness of Developing Countries: Theory and Practice,” Quarterly Journal of Economics, August 1984.
Krugman, Paul, “International Debt Strategies in an Uncertain World,” in Gordon Smith and John Cuddington, eds. International Debt and the Developing Countries, World Bank, 1985.
Lee, S.H., “Relative Importance of Political Instability and Economic Variables on Perceived Country Creditworthiness,” Journal of International Business Studies. Fourth Quarter, 1993.
McDonald, D., “Debt Capacity and Developing country Borrowing: A Survey of the literature,” International Monetary Fund Staff Papers, December 1982.
Powell, James L., “Symmeterically Trimmed Least Squares Estimation for Tobit Models,” Econometrica, Vol. 54, November 1986, p.1435–60.
Razin, A., and L. Svensson, “The Terms of Trade and the Current Account: The Harberger-Laursen-Meltzer Effects,” Journal of Political Economy, 1983.
Sachs, Jeffrey D. and Daniel Cohen, “LDC Borrowing with Default Risk,” National Bureau of Economic Research Working Paper 925, July 1982.
The authors would like to thank Patrick Conway, Robert Feldman, Mohsin S. Khan, John Montgomery, Peter Montiel, Michael Wattleworth, and Peter Wickham for their comments on an earlier draft of the paper and Ravina Malkani for excellent research assistance. We are indebted to Mr. Piggot of Euromoney, and to Mrs. Toksoz of Economist Intelligence Unit for providing us with their respective country risk ratings. Any errors that remain are, of course, the authors’ responsibility.
In our analysis, we use creditworthiness indicators developed by the Euromoney and the Institutional Investor magazines and the Economist Intelligence Unit. We intend to analyze the indicies developed by the Moody’s and Standard and Poor’s credit rating agencies in forthcoming papers.
See Appendix I for a more detailed discussion of the economic, political and financial variables used in constructing the various creditworthiness indicators.
The EIU rating was not initiated until 1989.
See Gibbons (1993) for a description of the Kendall’ measure of concordance. Basically, the data are assumed to be collected in the form of k ≥ = 3 sets of rankings for η objects by k judges. The sum of the ranks given to the respective objects by the k judges are denoted by R1, R2, …. Rn. The sum of the rank around k(n+1)/2, the expected rank sum under a random assignment, is denoted by S and defined as
If there is complete agreement among the judges on the rankings, the sum of squared deviations around k(n+1)/2 is
A relative measure of agreement is then the ratio of S and J, which is the estimate of the Kendall’ measure of concordance i.e. W = S/J.
For k = 3 ratings and n countries, the statistic Q=k(n-1)W is distributed as a chi-square variate with (n-1) degrees of freedom under the hypothesis that there is no agreement among the three ratings, which we use to formally test for the presence of a relationship among the ratings.
More formally, for each country, j = 1, 2, … n, the point of the analysis is to characterize the extent to which each of the i=1,..,k creditworthiness ratings (or some transformation), yt,i,j, can be represented as an affine function,
of a single, possible unobservable factor, Pj,t, plus a linear least squares projection error, vj,t. Here, pj,t is the first principle component of the Txk matrix of creditworthiness ratings, Yj, for country j. In this analysis, we present results using only the first principle component, since we are analyzing a small number of series (k=2 or k=3). The analysis is carried out for each of the j = 1,…,n countries individually over the time span for which the data is available.
For each country, we calculate
where, λ1 is the largest eigenvalue of the matrix Yj,Yj.
which measures the proportion of the variation of yt,i,j, that can be attributable to the first principle component. We do this both for the raw levels of the creditworthiness ratings, yt,i,j = Ct,i,j and for the log transformation yt,i,j = 100 ln [Ct,i,j/(100-Ct,i,j)], which is the form employed in the regression analysis below.
The complete individual country results are not reported to economize on space, but are available upon request.
Over the longer period from 1982 to 1993, the first principle component, on average, accounts for 92 percent of the variation in the logistic transformed II and EM ratings and 99 percent of the levels of those two ratings. An examination of the individual series again reveals that well over 90 percent of the variation in the levels of these ratings can be attributable to the first principle component.
The literature on country creditworthiness and the possibility of default, not surprisingly, has focused entirely on a country’s external debt. In recent years issues related to a government’s domestic liabilities have also become very important. For a discussion of the relationship between external and domestic debt, see, for instance, Guidotti and Kumar (1991).
In contrast to the cost-benefit approach which we will discuss later, this approach excludes the possibility of a debtor country willingly repudiating debt even when the intertemporal budget constraint holds. While the cost-benefit approach and the associated literature on optimal debt accumulation assumes that the debtor’s intertemporal budget constraint is satisfied, the debt-service capacity approach deals with cases where it is breached. The literature in this area predates the cost-benefit approach with a number of major contributions in the 1970s and early 1980s. The conceptual underpinnings of this approach were provided by the application of the permanent income theory to a nation portrayed as an infinitely lived agent to for a study of balance of payments and debt developments in an intertemporal framework, (see Bazdarich (1978), Dornbusch and Fischer (1980), Sachs (1981), and Razin and Svensson (1983)).
In addition, high export variability could lead to a deterioration in the economy’s ability to adjust to external shocks, by compounding the irregularity in foreign exchange receipts which results from these shocks. Similarly high variability in the terms of trade, GDP growth, as well as in reserves would be expected to adversely affect the country’s ability to meet its external liabilities and hence would have an adverse effect on creditworthiness. However, we were unable to find any empirically significant effect of the volatility variables on the credit-rating indicators.
An earlier study by Freeman (1979) had considered the benefits and costs of debt repudiation by allowing the debtor to consider default as a possible strategy. The analysis of the risk of repudiation was also undertaken by Kharas (1984), Kletzer (1984), Krugman (1985), and Sachs and Cohen (1985). For an early survey of this approach, see Eaton, Gersovitz and Stiglitz (1986).
While the Eaton-Gersovitz approach to debt repudiation (or rescheduling) has been extended in recent years using modern bargaining theories by Eaton (1989), Bulow and Rogoff (1989), and Atkeson (1991), their basic framework still remains valid.
In this context, it has been argued that it may even be in the lender’s interest to write-off part of the debt because a write-off could boost investment in the debtor country and result in better repayment (see Dooley (1989) and Froot (1989)). This issue is complementary to the debt-overhang issue which emphasizes the inability of a debtor country, hampered by illiquidity, to finance desirable investments, as well as the disincentive effects of high debt.
The data set covers eight periods of six months each, between the second half of 1979 and the first half of 1983. The basic methodology is to apply logistic transformation to the creditworthiness rankings and then use regression analysis. Nine economic explanatory variables are considered: debt/GNP; reserves/imports; average export growth rate; GDP growth; terms of trade; concentration of exports; GNP per capita; oil exporter dummy; and lastly, dummy for countries with debt servicing difficulties. An explanatory variable to capture political risk, in the form of a dummy for political turmoil, is also included in some of the regressions.
As discussed in section II, the two indicators are based on different types of sources; Institutional Investor data are based on surveys of bankers, while Euromoney data reflect financial market conditions, based on credit and market indicators.
In the Discriminant Analysis, the starting point is a sample of countries from two or more known groups and the objective is to devise a method of allocating a new country, whose group membership is unknown, to the appropriate group on the basis of that country’s characteristics. In Cluster Analysis, group membership of the sample of countries is unknown and the problem is one of determining the relative position of countries and seeing which groups emerge.
One anomaly in the EM results is that in some regressions the high inflation slope coefficient is significantly positive.
This is reflected in the significant negative coefficients on the low inflation slope dummy variable LO-INF SLP.
Given the large coefficient of the lagged value of the credit rating variable (αL), the long-run effect on a country’s credit rating of the higher interest rate would be αius/(1-αL) where αius is the short-term effect of a higher international interest rate. If we do not include the international interest rate and the lagged dependent variable we find that the rating calculations suggest three distinct regimes: the debt crisis and its immediate aftermath 1981-83, the post debt crisis period 1984-88 and the return of capital flows 1989-1992. However, these regime differences seem to follow the development in the international financial markets and are rendered insignificant with the inclusion of the international interest rate in the regression analysis.
See Table 1 for a summary of the determinants of the available creditworthiness indicators as well as a comparison of these determinants.
For instance, in rating developing countries, European bankers ranked foreign direct investment as fifth in importance, while Asian bankers put it in seventh place, and the Western Hemisphere bankers rank it ninth. In contrast, bankers in the Western Hemisphere ranked fiscal policy fifth, while those in Europe and Asia ranked this policy respectively as the seventh and ninth most important factor.
Forfeiting entails the discounting of medium-term promissory notes or drafts related to an international trade transaction. Repayments are semiannual and discounting is at a fixed rate. “Sell down” is a measure of over-subscription or otherwise of short-term paper.
For each of the above variables, the scores are obtained using the average of the ratios over the preceding two years.