³Ô¹ÏºÚÁÏÍø

Spreads versus German yields for the euro-area countries with the highest debt-to-GDP ratios have increased significantly since September 2021. Even if spreads are not yet at the worrying levels of previous stress episodes (and have decreased since the European Central Bank’s 15 June that it is finally working on a new fragmentation tool), this increase could still represent a risk at a time when growth is slowing quickly because of the energy crisis caused by Russia’s invasion of Ukraine, and when many countries have historically high levels of debt after the highly expansive fiscal policy put in place during the COVID-19 crisis.

Moreover, the ECB plans to tighten its monetary policy to tame inflation in the euro area. As after the June ECB Governing Council meeting, net asset purchases are coming to an end. Pandemic emergency purchase programme (PEPP) net purchases ended in March 2022, and public sector purchase programme (PSPP) net purchases ended on 30 June. Furthermore, the ECB has announced it will hike policy rates by 25 basis points in July and (at least) by an additional 25 bps in September, which will bring its deposit rate to 0% for the first time since June 2014. Overall, at time of writing, market participants expect the ECB to increase its relevant policy rate by around 225 bps by the end of 2023 (Figure 1).

The expectation of higher policy rates and the ending of asset purchases, combined with slower growth and high levels of debt, has worried markets and has resulted in a significant increase in spreads.

Even if these risks are not negligible, fears should also not be exaggerated. Debt-to-GDP ratios are historically high, but where are these ratios headed in the next few years? Will policy rate hikes be translated quickly into higher debt-servicing costs for governments? We investigate these two questions by looking at six main scenarios to assess how different interest rate levels, and in particular different spread levels, would impact the interest paid on the stock of debt (the so called ‘implicit rate’) and the debt-to-GDP ratios in euro-area countries with debt-to-GDP ratios above 100%: Belgium, France, Greece, Italy, Portugal and Spain.

Debt evolution

But, first, how does debt evolve over time? The total government debt in year t+1 is equal to the debt level in t minus the primary balance (ie the balance excluding interest rate payments), plus the interest paid on the debt owed in year t. We use the IMF’s April 2022 World Economic Outlook for the levels of nominal GDP and primary account balances from 2022 to 2027.

Despite rising interest rates, the implicit interest rate paid by countries on their debts (ie total interest payments divided by the total debt at the end of the previous year) should rise more gradually because much of the current debt that will prevail in coming years was issued in the low interest environment of the last decade.

To account for this, we compiled all the debt that had been issued up to 1 January 2022 and the corresponding coupons and maturity. We did this to account for the specific maturity schedules of countries' debts. For any given future year, part of that debt will still be active. Our database enables us to compute the interest that will be paid on this fraction of the debt. The remainder of the debt is what will be issued subsequently (either to roll over previous debt or because of future primary deficits). To make it simple, we assumed that all the debt issued in the future is issued with a maturity equal to the weighted average residual maturity of government debt over the past 10 years (see Table 1 in the annex)

We then built scenarios on the evolution of interest rates in future years, and in particular on how spreads could evolve during the next six years, eg if they mimic noteworthy historical levels:

• Market expectations for each country’s interest rates (ie forwards interest rates),
• Severe scenario: spreads reached in the wake of the COVID-19 pandemic,
• Very severe: spreads reached during the euro crisis,
• Low spreads: levels at the end of 2019,
• Current levels of spreads (as of 23 June 2022).
• For Italy, we also included a ‘political risk’ scenario with a spread equal to the peak spread that followed the formation of the Lega-M5S government in 2018.
• Finally, we also show WEO forecasts computed by the IMF.

In all these scenarios (except for the first based on countries’ forward rates, and the IMF’s forecasts) spreads were added to the market expectations of German rates (eg of 10-year rates, visible in Figure 1). Thus we don’t question what markets expect from the ECB or how the ECB’s hikes will be transmitted to the German ‘safe’ rate, but instead focus on spread levels. For each interest-rate scenario, we computed the evolution of implicit interest rates and the resulting debt-to-GDP ratios up to 2027.

The main and obvious limitation of this exercise is that the forecasts of the primary account balance and levels of nominal GDP are fully exogenous. They are the same in all scenarios and are not impacted by the level of interest rates, which vary significantly across our scenarios. This is an acceptable assumption for small differences in interest rates (compared to those assumed in the IMF forecasts) but not for large differences, meaning that for the high spreads scenarios, our results definitely constitute a floor: debt-to-GDP ratios will be higher because GDP would be smaller than the forecasts we use. However, despite this limitation, we believe that our exercise illustrates broadly how different levels of interest rates would impact implicit rates and debt-to-GDP ratios in the analysed countries.

Main findings

In all the countries studied, whatever the scenario – even with rapidly increasing market rates – implicit rates paid by these euro-area governments will increase much more gradually than policy and market rates (Figure 2). This is because countries have been borrowing at very low – sometimes negative – rates in recent years. In addition, countries have responded to low rates by increasing the average residual maturity of their debt from around six years in 2010 to eight years in 2022 (Figure 4 in the annex). The longer maturity of low-interest debt mechanically slows down the impact of interest rate hikes on debt service.

The only scenario in which the implicit rate increases quickly is that in which countries would face a spread similar to that at the peak of the euro crisis. However, this is a very strong assumption because this means that countries would face for six years a scenario that, even during the euro crisis, they faced for only a few months (see Figure 5 in the annex). This very strong assumption is slightly counterbalanced by the fact that for this scenario we don’t use the market rates at which bonds were exchanged, but the highest coupons that countries had to offer to sell their bonds at auction during the crisis. Given the tough conditions some of these countries faced at the time – sometimes even losing market access – this seems like a better proxy than peak market rates that result from panic sales at the worst point of the crisis.

The special case of Greece is telling in that regard. A large share of Greek debt (around 50%) comprises very long-term loans from the European Financial Stability Facility and the European Stability Mechanism, with a low rate (average currently at 1.37%) based on the funding costs of these institutions. Consequently, Greece’s implicit rate is not very sensitive to the increase in its interest rates in the next few years, whatever the scenario, as the ESM and EFSF rates are not affected by the increase in spreads that Greece faces. Official loans thus represent a significant buffer against interest rate increases for Greece.

Thanks to the combination of these gradual increases in the interest rate paid on the debt and high nominal GDP growth in 2022 and 2023, a favourable interest rate-growth differential (r-g) should lead to falling debt-to-GDP ratios in all countries (Figure 3). But after 2023, the situation varies. The expected reduction in inflation, more moderate real output growth returning to potential after the post-COVID-19 rebound, and the increase in interest rates, would lead to less-favourable conditions for all countries, but the overall evolution of the debt-to-GDP ratio will be very different for different countries.

Unlike what the current debate – mainly focused on the risk surrounding Italy’s situation – could make us believe, the debt-to-GDP trend is expected to be more favourable in Italy, Greece and Portugal, than in France and Spain, and much more favourable than in Belgium. While the first three countries could possibly witness a strong fall in their debt-to-GDP ratios up to 2027 (in all scenarios in Greece and Portugal, and at least in the most favourable cases in Italy, see Figure 3), France and Spain should see their debt-to-GDP ratios stagnate, or increase slightly, after the decrease of the first couple of years.

This difference is mainly due the slower reductions in the primary deficits of France and Spain over the considered period (still expected to be at 2.3% and 1.9% of GDP respectively in 2027) compared to Italy’s (which is expected to already come back quickly to 1% next year and reach only 0.1% in 2027). Much more worrying is the situation of Belgium, which is the only country expected to see its debt-to-GDP ratio increase significantly after 2023, whichever scenario is considered. This is mainly because the IMF foresees the Belgian government sustaining primary deficits at a high level until 2027 (when it is expected to be as high as 4.3%; all forecasts of primary balances can be found in Table 3 of the annex), because of social benefit and aging .

Conclusions

Despite their simplicity and limitations, as discussed above, our simulations allow us to draw two main conclusions. First, with expected high nominal output growth and implicit rates that should only increase very gradually, debt-to-GDP ratios should continue to fall, despite the expected relatively quick increase in ECB policy rates in 2022 and 2023. This is because euro-area countries have locked-in low rates thanks to the combination of low rates for a decade and an increase in the average maturity of their debts.

Second, however, after 2023 the situation might vary across countries. Countries with relatively high primary deficits are not expected to see a further fall in their debt-to-GDP ratios. But the countries that will quickly reduce their primary deficits should see their debt-to-GDP ratios continue to fall, as long as spreads are kept relatively in check. This is particularly the case for Italy, which is expected see its debt fall in all scenarios involving a spread below 200 bps. This shows that current fears about debt sustainability are misguided, but also that it is important for the ECB to design a new anti-fragmentation tool that will keep spreads at a level justified by structural differences between countries, and not by self-fulfilling prophecies.

Annex: more details on how simulations are built

Computing debt levels for coming years

In order to construct our debt-to-GDP simulations, we created a database with the full list of active government securities from Bloomberg. Each active security has an associated date of maturity, coupon and outstanding amount. This enabled us to compute, for the coming years, the implicit rate paid by countries on the debt already issued as of 1 January 2022. The implicit rate is defined, for a given year y, as the average coupon weighted by the outstanding amount of the securities that had not yet reached maturity at the end of year y-1.

New levels of total debt were computed based on the following formula:

Debt _y = Debt _y-1 - bal_y + r * Debt_y-1

where bal is the primary balance, r is the implicit rate paid on the debt. Forecasts for the primary balance are from the April 2022 IMF World Economic Outlook.

The difference between debt in year y and still-active debt that has already been issued gives us the amount of newly issued debt in a given year. Governments issue new debt with a coupon that varies with our different scenarios (see details below), enabling us to compute an updated implicit rate based on the already-issued debt (at the beginning of 2022) and the debt issued in the subsequent periods (2022, 2023, 2024, etc).

For example, the implicit rate paid in 2024 is the weighted average of the rate paid on the remaining debt (in 2024) that was already active at the beginning of 2022, on the additional debt issued in 2022 with coupon x and on the additional debt issued in 2023 with coupon y.

We then divided the resulting debt by nominal GDP forecasts, also from the IMF’s World Economic Outlook, to obtain future debt-to-GDP ratios.

Dealing with non-marketable debt

In addition to marketable debt securities, countries also borrow money in other ways. For each country, these loans represent a different share of their total debt. We computed the total amount of non-marketable debt as the current difference between the total general government debt and the sum of all active debt securities. We assumed that this amount remains unchanged over the course of our period of analysis and that interest paid on these loans is equal to each country’s current implicit rate. This is a strong assumption but given the very small portion of the overall debt that these non-marketable debts represent in euro-area countries, our results are not affected significantly by this simplification.

Greece is different since a very large share of its debt (around 50%) is composed of very long-term loans from official institutions (EFSF and ESM). The rate paid by Greece on these loans is equal to their cost of funding plus margin and fees, which thus vary with the evolution of the interest rate paid by these institutions. The cost of funding for the ESM and EFSF is slightly above Germany’s and the average funding maturity is 6.3 years. Historically, the rate is approximately 50 basis points above the German rate. So, to make it simple, we assumed in our scenarios that the interest paid by Greece on its loans is equal to the German 6YR forward plus 50 basis points.

Dealing with inflation-indexed bonds

Italy and France are the only two countries analysed where the current share of active bonds indexed to inflation is significant (above 1%). Given current levels of inflation, it is important to take them into account to compute the evolution of the nominal level of debt. The outstanding amount of money owed for inflation-indexed bonds that have already been issued therefore increases at the same rate as the consumer price index (forecasts from the IMF’s World Economic Outlook). For computation reasons, we assumed that countries do not issue bonds indexed to inflation in future periods.

Elaborating the different interest rate scenarios

For each country, we constructed a set of scenarios on the evolution of the interest rate in the years to come. To do so, we first made a simplifying assumption about the average maturity of newly issued debt. We assumed that the maturity of all newly issued debt is equal to the mean of the average residual maturity of government debt since 2010. The average residual maturity of debt has quickly evolved over the past years, in a low rate environment (Figure 4). The average residual maturity of Belgian debt, for instance, reached a historic high of nearly 11 years in 2022.

Table 1 reports the average maturity since 2010. These are the numbers we used in our simulations for the maturity of newly issued debt in the years to come. For instance, we assumed that the maturity of all newly issued Belgian debt is equal to 8 years.

Our first interest-rate scenario is based on market expectations. We looked at the forward rates for each country for the issued debt with a maturity equal to that country’s average residual maturity. For instance, we used the values of the forward rate for the Belgian 8-year debt security for the next six years.

Other simulations are based on different spread scenarios. We looked at episodes since the creation of the euro which resulted in different levels of spreads. More specifically, we look at the levels of spreads during the euro crisis, the pre-COVID-19 period, the beginning of COVID-19 and the most recent period. While the euro crisis and the beginning of COVID-19 saw high-spread scenarios for most countries studied, the pre-COVID 19 period provides a low-spread scenario and the current spread is intermediate. For each scenario, we computed the peak spread versus the equivalent German yield during each given period, and add this value to the German forward.

In the specific case of the euro crisis and because certain countries (like Greece and Portugal) lost market access, we did not take the yield of the debt securities (which reached 40% in Greece, without actually reflecting borrowing costs since Greece never issued a debt security with a 40% coupon). Instead, we looked at the maximum coupon at which countries issued debt during that period. Table 2 reports the assumed spread against the German rate for each scenario.

By adding these spreads to the forward of the German yield, we obtain different trajectories of the country-specific interest rates (Figure 5). These interest rates enabled us to compute the implicit rates reported in the main section, and in five debt-to-GDP ratios using the method described above.