An Alternative to Sovereign Credit Ratings: PSCF
The downgrading of the UK by credit rating agency (CRA) Moody’s and of Italy by Fitch reminds us that these agencies remain influential, at least in the media. Whether sovereign ratings actually drive government bonds markets or whether they convey useful information is less clear.
The fair yield on a government bond should be a function of the time value of money, inflationary expectations and expected losses arising from default. In theory, ratings inform us about this last driver. But the A,B, C letter grades assigned by the CRAs are not clearly associated with default probabilities or expected loss rates. Instead, it is left to market participants to make this association.
Further, the assignment of a single symbol to the entire term structure of a country’s debt – or at least that which matures one year or more hence – ignores the fact that different maturities embed different levels of risk. In the US, for example, there is widespread agreement that the long-term fiscal situation is unsustainable and there is no political consensus on how to rectify the situation (the so-called ‘fiscal cliff’ debate). On the other hand, the country currently enjoys very low interest rates. This implies that its near term bonds are risk free but its longer maturities carry some risk.
Over the life of a newly-issued 30-year treasury bond, the US dollar (USD) could lose its status as the world’s reserve currency and the government could choose to reschedule debt payments as an alternative to servicing its bonds with freshly-printed money. Inflation affects the voting public, while the brunt of a debt rescheduling would fall heavily on foreigners not eligible to advance their interests through the ballot box.
Finally, sovereign rating upgrades and downgrades are not necessarily timely. They may occur whenever an analyst has a chance to review the country’s situation and organise a ratings committee meeting to build support for a proposed action. Meanwhile governments frequently release new fiscal reports and economic developments routinely alter each nation’s budget trajectory. This constant flow of data can render ratings assessments stale in short order.
Addressing the CRA Weaknesses
The sovereign ratings status quo has at least three inherent disadvantages – a lack of numerical precision; inattention to term structure; and insufficient timeliness. All of these can be addressed by turning to quantitative default probability models. One other disadvantage of current CRA practice, is its lack of transparency. This can also be resolved by making sovereign risk models more open source.
There are advantages conferred by using the Public Sector Credit Framework (PSCF), which is a free, open source government default probability modelling tool. The system can be used by rating agencies, public policy analysts or corporate treasurers investors. PSCF allows the user to set up and perform a multi-year budget simulation with a default point stated in terms of a fiscal ratio. Default probabilities are assumed to be the proportion of trials that exceed the threshold. The budget simulation and the thresholds are all user supplied parameters entered into an Excel workbook. Setting up a simulation in PSCF is not a trivial matter, and hopefully the tool will evolve to provide greater ease-of-use. But the user’s effort is rewarded when the system generates a full term structure of annual default probabilities based on the model created. In contrast to ratings, this term structure of default probabilities provides a readily usable input to estimating fair yields on a government’s bonds.
One source of PSCF’s complexity is its flexibility. Because there is no universally agreed standard for measuring government solvency, it provides a blank slate onto which analysts can etch their views. The only rule in PSCF is that all solvency drivers have to be quantified. While we often hear bromides about political considerations being resistant to quantification, political markets have been effective in doing so. Further, unless each element affecting sovereign credit risk is quantified, it cannot be weighed against other factors.
Today, the most commonly used statistical measure of government solvency is the debt-to-gross domestic product (GDP) ratio. While this metric can be used in PSCF, I would not recommend doing so because debt-to-GDP misses two important effects:
In regard to debt burdens, the UK is protected both by the current low interest rate regime and the fact that the term structure of gilts is relatively back loaded. According to Bloomberg, the average maturity of UK government debt is 14.52 years compared with 5.38 years for the US. This means that a spike in interest rates will take much longer to impact UK debt servicing ability than that of the US.
The two limitations outlined above of the debt-to-GDP ratio are resolved by using the ratio of interest to revenue. Research into historical defaults by major Anglophone and other western governments, suggests that ratios of up to about 30% are sustainable. At this point continued debt service may become so politically unpalatable – due to the crowding out of other spending priorities – that default becomes a feasible option.
Consequently, a reasonable default threshold for a sovereign would be 30%. It may also be appropriate to use a lower threshold for a sub-sovereign or any other government that does not have direct access to a printing press.
With the threshold determined, the next step is to set up a fiscal simulation that produces an annual distribution of future interest to revenue ratios for the target country. Revenues and expenditures can be calculated under a range of interest rate, economic growth rate, inflation rate scenarios generated by the model. Policy choices can also be simulated. Annual deficits can be captured and added to the previous year’s stock of debt, servicing as a basis for the interest expense calculations.
As new national budgets are filed, as new economic forecasts become available, or as new political developments occur, the PSCF model can be updated and rerun to reflect these changing realities.
Since all the major Western European countries are substantially below the 30% threshold none would have measurable near-term default probability if modelled in PSCF with this level. Eventually, scenarios that contain low economic growth and high interest rates produce some level of default probability in the intermediate to long term. Because the UK is running lower deficits than the US (building its stock of debt at a slower rate) and is better insulated against interest rate shocks, it is likely to look better in a PSCF model than the US, despite its paltry economic growth.
While risk analysts cannot accurately predict the future, they now possess the computing power needed to simulate it. Rather than rely on letter ratings that have unclear definitions from CRAs and that are often uninformed by the most recent developments, PSCF provides a way for technically sophisticated analysts to make their own assessments of sovereign credit risk. Treasurers may wish to consider using it.