A Fund Working Paper Finds Sovereign Ratings Diverge From What a Simple Debt Model Implies
3 economists writing for the International Monetary Fund have compared the sovereign credit ratings of S&P, Moody’s and Fitch against what a simple model of debt and forecast primary balances would imply, and found that the ratings diverge from it in 3 specific ways. The paper was published on 18 September and runs to 48 pages.
The finding, in the authors’ words, is that ratings “give much more weight to debt relative to forecast primary balances”, that they “understate the effects of the difference between the interest rate and the growth rate”, and that they “give a very large role to country effects”.
| Divergence | What the paper says ratings do |
|---|---|
| Debt against forecast balances | Overweight the stock of debt |
| Interest rate minus growth rate | Understate its effect |
| Country effects | Give them a very large role |
The 3 divergences named in the paper.
What was actually tested
The exercise covers 35 advanced economies and, separately, 97 emerging market and developing economies, annually from 1995 to 2025. Debt ratios come from the Fund’s own forecasting database and the fiscal variable is the mean forecast of the primary balance over the current year and the following 5. Because ratings come in ordered categories rather than as a continuous number, the authors estimate them with ordered probit regressions. The 3 agencies’ ratings move closely together, with correlations above 0.94 in advanced economies and above 0.90 in the emerging and developing group.
The headline results follow from that. Debt is a highly significant determinant of ratings while the effect of forecast primary balances is positive but less significant, and the ratio between the 2 is much smaller than the model implies. Year effects are almost flat, which the authors call surprising given how far the gap between interest rates and growth fell over the period in advanced economies. And country effects are, in their word, surprisingly large.
The country illustration is the part with teeth
The paper makes the size of those country effects concrete. Holding debt and forecast primary balances at the sample average for the emerging and developing group, the expected rating differs enormously by country: Qatar sits at 8, China at 6.7, India at 4.1 and Brazil at 2.6, while Pakistan and Turkiye come out at 1, below investment grade. The debt ratio at which a country would still hold at least a rating of 2 with even odds runs from 161 percent in Qatar and 134 percent in China down to 79 percent in Mexico, and to negative values for Pakistan and others.
| Economy | Expected rating at the same debt and balances |
|---|---|
| Qatar | 8 |
| China | 6.7 |
| India | 4.1 |
| Brazil | 2.6 |
On the 11 point scale the paper uses, where 11 is the highest rating. Same debt ratio, same forecast primary balances, different country.
Among advanced economies the spread is the same shape. At identical debt and forecast balances Germany comes out at 11 and Ireland at 9, while the Baltic countries come out at 2. On the authors’ debt tolerance measure, the level of debt above which a country is predicted to lose a rating of 9 on even odds, Singapore reaches 272 percent and Germany 248, the United States 161, France 135 and Italy 89, while the Baltic states are negative, meaning the estimates imply they would not reach that rating even with no debt at all.
Why it matters: For Gulf and Arab sovereigns this is a paper worth reading with a rating letter in hand, because the one emerging market the authors put at the top of their illustration is a Gulf state. Qatar’s expected rating of 8 against Brazil’s 2.6 on identical debt and identical fiscal forecasts is the cleanest statement anywhere of how much of a rating is not explained by the numbers the agencies say they are looking at. On our reading the practical implication for the region is the debt tolerance figure rather than the rating itself: 161 percent of output for Qatar against 79 percent for Mexico is a different borrowing capacity for the same credit letter.
Outlook: The authors put the conclusion as a choice. Either the model is missing something essential about what determines default, or, in their words, “rating agencies should revisit their methods”. They say preliminary work on market spreads rather than ratings puts relatively more weight on forecast primary balances, closer to the model, though a gap remains. This is a working paper, and it carries the Fund’s standard notice that the views are the authors’ rather than those of the Fund, its executive board or its management. No sovereign rating action was taken on any Gulf state, Egypt or Jordan by any of the 3 agencies on 18 or 19 September.
Sources: International Monetary Fund.
