Public debt: Safe at any speed?

Has government debt reached dangerous levels? Or can it still be raised to bridge the economic crater left by the COVID-19 pandemic? As usual, economists are split on the answers to these critical policy questions. While some point out that current levels signal heightened risk of sovereign debt crises, others take comfort in the historically low interest rates, suggesting that governments could actually sustain much higher debts than in the past. To help clarify the discussion, the article draws on the conventional toolkit used in debt sustainability analyses by institutions like the IMF or the European Commission to estimate debt levels beyond which it would be unsafe to venture.

Here, safety is defined as the ability of the government to keep control of the debt-to-GDP ratio even under persistently adverse economic and financial conditions. That ratio is the usual metric to assess a country’s public debt because a government bond is effectively a claim on future tax revenue. Thus, it makes sense to prevent debt from growing systematically faster than the potential tax base (i.e. to prevent an ever-increasing ratio).

The methods discussed in the article revolve around the interplay between the three determinants of the debt ratio: the deficit to be covered by new debt, the rate of interest (which captures the speed at which debt would grow if interest due was paid only with newly borrowed money), and the rate of economic growth (which inflates the denominator of the ratio). The debt ratio grows faster the higher the deficit, the higher the interest rate and the lower the growth rate.

In normal times, the interest rate tends to exceed GDP growth. Thus, to keep the debt ratio in check, governments must cover at least part of the interest bill with tax revenues. In other words, total public expenditure excluding interest payments must be lower than total revenue, a situation that economists call a primary surplus. To keep the debt ratio constant from one year to the other, the primary surplus must exactly offset the debt impact of a positive interest-growth differential. The greater the differential, the higher the debt-stabilising surplus.

The article describes various methods to estimate the debt level that would be risky to exceed. To do so, a distinction is made between a debt limit and a safe debt boundary. A limit is a threshold beyond which government almost certainly loses control of debt dynamics. This happens when the primary surplus required to stabilise or reduce debt is infeasible. A government cannot raise tax rates indefinitely without severely harming the economy (and ultimately tax revenues) nor compress public expenditure to zero. Thus, beyond the debt limit, an element of luck is needed to avoid default. The safe debt boundary is the largest debt ratio the government could plausibly stabilise or reduce under bad economic and financial conditions, using fiscal policy only. Unlike the debt limit, the boundary can be exceeded without necessarily triggering meaningful fears of default. However, allowing debt into unsafe territory means that there is a not insignificant risk (say above 5 or 10 % probability) that bad shocks could push debt over the limit.

When plausible debt limits can be estimated, the safe debt boundary is found by subtracting a safety buffer from the limit. That buffer reflects normal disturbances affecting growth, interest rates and government budget as well as extreme (i.e. low probability but high impact) events. When no debt limit can be estimated, the safe debt boundary reflects the risk that unfavourable circumstances could force the government to exceed the highest feasible primary surplus in order to stabilise the debt ratio. All the methods discussed in the article are tailored to country-specific circumstances in terms of their historical exposure to shocks to the determinants of debt dynamics.

This article provides a range of illustrative estimates for safe debt boundaries in a sample of advanced economies. These estimates vary substantially across countries and methodologies, ranging from 70 % and 160 % of GDP. Belgium is generally close to the median of the sample. For instance, the most complete and consistent methodology yields a safe debt upper bound of 120 %of GDP

That said, it really must be emphasised that such numbers always hide significant uncertainty about the validity of the underlying statistical models and calibrations, which presume that the past is a good guide for the future. Today, that implicit assumption is even more questionable than usual. The long-term properties of the models, especially regarding the interest-growth differential, warrant close scrutiny before putting them to use. All estimates provided in the article deliberately exclude the last five years of actual data to limit the impact of abnormally low interest rates on the long-term properties of the models. Yet, if much lower or even negative interest-growth differentials were to be part of a new normal, the safe debt boundaries obtained with such tools would be higher, perhaps implausibly so. This confirms that, in unchartered territory, sound judgment remains an even more essential component of any balanced assessment.