Country Risk Service Rating model description and service methodology Summary

Transcription

1 Country Risk Service Rating model description and service methodology Summary As part of the upgrade to the Economist Intelligence Unit's Country Risk Service, we have developed a new rating model. This is an enhancement of the previous model which has been in operation since in The new model's primary focus is on the three risk categories to which clients can have direct financial exposure: sovereign risk, currency risk and banking sector risk. It also provides ratings for two types of generic risk: political risk and economic structure risk. An overall country risk rating is derived by taking a simple average of sovereign risk, currency risk and banking sector risk. Methodology Model characteristics and structure The model is a signalling model and provides an early warning system for sovereign default, currency crisis and banking sector crisis. The rating for each risk category is determined by a weighted combination of the scores for each of the questions in the model. The model works on a rolling 12-month time horizon, assessing the risk of sovereign default, currency crisis or banking sector crisis during the following 12 months. The model comprises 60 questions and is divided into five sections (politics; economic policy; economic structure; macro economy; financing and liquidity). The number of questions in each section varies but is at least ten for each section. The questions are weighted in accordance with their relevance for the particular type of risk. For example, questions relating to the government's commitment to pay, the public debt to GDP ratio, the transparency of public finances and the government's payment record all have heavy weightings for sovereign risk. Of the 60 questions, 30 are quantitative and 30 are qualitative. Each question has five possible scores, ranging from 0 to 4, with 0 indicating least risk and 4 most risk. For quantitative questions, the scores are determined on the basis of thresholds. For example, there is a question relating to the fiscal balance in the most recent 12 month period; the score for this question ranges from 0 for countries running surpluses or a balanced budget to 4 for countries where the deficit exceeds 5% of GDP. The scores for the quantitative questions are derived from historical data or estimates of a series of variables relevant to country risk, such as the public debt to GDP ratio, the fiscal balance to GDP ratio, the current account balance to GDP ratio, the external debt to GDP ratio, the evolution of foreign exchange reserves, the inflation rate and the GDP growth rate. The data is refreshed on a monthly basis. High-frequency data is used where available, for variables such as inflation, the exchange rate and banks' net foreign asset position. The scores for the qualitative questions are determined by the Economist Intelligence Unit country analyst, who runs the model. For example, analysts are asked to assess how strongly committed the government is to meeting its debt obligations. The answers to the qualitative questions are audited by regional risk managers to ensure consistent assessment across countries.

2 Model specification The questions and variables included in the model were selected on the basis of the Economist Intelligence Unit's collective experience and expertise in assessing country risk since the early 1980s. Many of the questions and variables formed part of the previous model. Some variables in the old model were omitted from the new one, and some new questions were introduced. These decisions were taken on the basis of the experience of the past decade, particularly during the emerging market crises of the late 1990s. Individual variables were assessed for relevance as part of the back-testing procedure detailed below. The results of the tests informed the decision whether variables were included in the model and if so, what weighting to assign to them. While it is convenient to refer to a single country risk model, in fact there is a main model and several different sub-models. These sub-models are applied to different types of countries. All have essentially the same structure and have been carefully calibrated to ensure that the results can be compared across all countries covered by the risk service. These sub-models allow us to assess risk in countries which are exposed to differing market pressures. The main model is designed for emerging markets with substantial amounts of foreigncurrency denominated debt, and is used to assess risk in 100 countries covered by the Country Risk Service. For some OECD countries with virtually no foreign-currency denominated debt, we have developed a variant of the main model. This replaces certain indicators relating to a country's external accounts (foreign exchange reserves, the external financing requirement, external debt/exports) with indicators relating to the public finances or the public debt market. This is applied to countries such as the US and Japan. We have also developed a variant of the OECD model for rating members of European Monetary Union (EMU). There are only minor differences between the OECD and EMU model, designed to identify strains which EMU members may be experiencing as a result of their membership of the single currency and to gauge the risk that such strains might lead the country eventually to exit from EMU. Finally, there is a model for the Euro zone as a whole, which is used when assessing currency risk relating to the Euro. Testing The model has been back tested to the start of 1997 to gauge its predictive power. The backtesting was performed on a reduced form of the model, which scored all the quantitative questions. We also back-tested qualitative questions which had featured in the previous version of the model. Hence, we have an unbiased assessment of the analyst s attitude to risk factors going back in time. Qualitative questions for which we have no equivalent in the old model were eliminated. The testing method entailed recording the scores of all countries on a monthly basis since the start of The model was tested against a database listing countries that experienced a crisis (sovereign, currency, banking sector) and those which did not. The scores of countries experiencing a crisis were recorded at set periods (36, 24 and 12 months) prior to the onset of the crisis. The scores of crisis-free countries were recorded over the same period. The scores for two sets of countries (crisis and non-crisis) were compared with a view to determining whether there was a statistically significant difference between the two. The model performed well during the test period. The Economist Intelligence Unit plans to extend the back testing to the 1980s, in order to gauge its performance during the 1980s debt crisis.

3 Probability of crisis The probability of default (in the case of sovereign risk) and of crisis (in the case of currency and banking sector risk) was calculated on the basis of the back testing. The number of crises associated with each score was summed and expressed as a percentage of the total number of occurrences of that score. For example if there were ten cases of default associated with a score of 40 and the score of 40 occurred a total of 200 times across the entire sovereign database, this would generate a probability of default of 10/200*100=5%. Testing for sovereign default is difficult because of the low number of cases (in absolute terms and by comparison with corporate defaults). Testing is all the more difficult for OECD countries with unblemished records of servicing their debts. In these cases, we believe that the scores generated by the model have a value in ranking OECD sovereigns in terms of creditworthiness even if it is impossible to validate estimate probabilities of default empirically.

4 Appendices Rating bands The Country Risk model calculates a score for each type of risk which can run from This score is then translated into a rating letter, using a system of rating bands. To make it easier for clients to compare our ratings with those of the three main international ratings agencies, we have adjusted the rating band system on the new model. The scale still runs from 0 to 100, but it is divided into ten bands, rather than five bands. The letters assigned to these bands have been changed in order to facilitate comparison between our ratings with those used by Standard and Poor s, Moody's and Fitch. However, clients should remember that our ratings system is the same as that used by the international agencies. We have a very structured approach to calculating a rating, using a model explicitly designed to facilitate cross country comparisons. We may from time to time also take a different view from other ratings agencies on what factors of a countries risk profile are particularly important. As a result, clients should not expect Economist Intelligence Unit risk ratings to neatly track those published by other agencies. For example, fewer countries are likely to be assigned AAA ratings under the Economist Intelligence Unit's system than by the international rating agencies. To reduce the risk of frequent band changes for countries whose score is close to the cusp of two bands, there is a buffer zone encompassing scores ending with the digits 9, 0, 1 and 2. Within this zone, the country analyst has discretion over whether the lower or higher letter grade is assigned. For example, a country with a score of 19 could be rated either AA or A. Clients are encouraged to focus on the letter rating rather than the underlying score, which can be volatile due to noise in the economic data used to evaluate the model. Score Old EIU model band A A B B C C D D E E New EIU model band AAA AA A BBB BB B CCC CC C D S & P's and Fitch AAA AA A BBB BB B CCC CC C D Moody's Aaa Aa A Baa Ba B Caa Ca C D

5 Rating definitions Sovereign risk This risk category measures the risk of a build-up in arrears of principal and/or interest on foreign- and/or local-currency debt that is the direct obligation of the sovereign or guaranteed by the sovereign. The sovereign risk rating is informed by scores for a combination of political, policy, cyclical and structural variables. Currency risk This risk category measures the risk of devaluation against the reference currency (usually the US dollar, occasionally the Euro) of 25% or more in nominal terms over the next 12-month period. The currency risk rating is informed by scores for a combination of political, policy, cyclical and structural variables. Banking sector risk This risk category gauges the risk of a systemic crisis whereby bank(s) holding 10% or more of total bank assets become insolvent and unable to discharge their obligations to depositors and/or creditors. A banking crisis is deemed to occur even if governments restore solvency through large bail-outs and/or nationalisation. A run on banks facing a temporary lack of liquidity rather than underlying solvency problems is not deemed to constitute a crisis, provided that public confidence in the banking system is quickly restored. Banking crises are typically associated with payment difficulties in the corporate or household sectors; bursting of asset price bubbles; currency and/or maturity mismatches. The rating can therefore serve as a proxy for the risk of a systemic crisis in the private sector. The banking sector risk rating is informed by scores for a combination of political, policy, cyclical and structural variables. Political risk This risk category evaluates a range of political factors relating to political stability and effectiveness that could affect a country s ability and/or commitment to service its debt obligations and/or cause turbulence in the foreign-exchange market. The political risk rating informs the ratings for sovereign risk, currency risk and banking sector risk. Economic structure risk This risk category is derived from a series of macroeconomic variables of a structural rather than a cyclical nature. Consequently, the rating for economic structure risk will tend to be relatively stable, evolving in line with structural changes in the economy. The economic structure risk rating informs the ratings for sovereign risk, currency risk and banking sector risk. Overall country risk This risk rating is derived by taking a simple average of the scores for sovereign risk, currency risk and banking sector risk.

6 List of factors taken into account in the Economist Intelligence Unit Country Risk model 1: risks to economic assumptions/estimates 31: default history POLITICS/INSTITUTIONS 32: financial regulation and supervision 2: external conflict MACROECONOMY 3: governability/social unrest 33: real OECD GDP growth 4: electoral cycle 34: credit as % of GDP, growth 5: orderly transfers 35: real GDP growth, 48 months 6: event risk 36: real GDP growth, 12 months 7: sovereignty risk 37: inflation, 48 months 8: institutional effectiveness 38: inflation, direction 9: corruption 39: trade-weighted real exchange rate 10: corruption/state intervention in banks 40: exchange-rate misalignment 11: commitment to pay 41: exchange-rate volatility ECONOMIC POLICY 42: export receipts growth, 12 months 12: quality of policymaking/policy mix 43: current-account balance, 12 months 13: monetary stability 44: asset price bubble 14: use of indirect instruments FINANCING AND LIQUIDITY 15: real interest rates 45: transfer and convertibility risk 16: fiscal balance/gdp 46: IMF programme 17: fiscal policy flexibility 47: international financial support 18: transparency of public finances 48: access to financing 19: domestic debt 49: gross external financing requirement 20: unfunded pension and healthcare liabilities 50: debt-service ratio 21: exchange-rate regime 51: interest due/exports 22: black-market/dual exchange rate 52: external short-term debt/fx reserves ECONOMIC STRUCTURE 53: % change, fx reserves, actual 23: income level 54: net external debt/exports 24: official data (quality/timeliness) 55: FDI/gross financing requirement 25: current-account balance, 48 months 56: import cover 26: volatility of GDP growth 57: OECD short-term interest rates 27: reliance on a single goods export 58: non performing loans 28: external shock/contagion 59: banks' credit management 29: public debt/gdp 60: banks' foreign asset position 30: gross external debt/gdp