Can private debtors in the Eurozone weather interest rate shocks?

High and rapidly increasing private debt was one of the main causes of the Eurozone crisis. It certainly choked spending in some Member States when the downturn in 2008 set in. Since its peak in early 2015, aggregate Eurozone private debt has finally come down, but it is still higher than in the years prior to the financial crisis: 161% of GDP end 2017 vs. 152% of GDP early 2008 (figure 1).

With monetary easing slowly coming to an end, a frequently heard question is whether private debtors could weather higher interest rates. In this Special we explore whether private debt still poses a risk to economic stability, especially if and when interest rates start to rise. We look at a number of indicators that are associated with interest rate sensitivity such as i) affordability (debt service), ii) maturity distribution, iii) headroom in savings and iv) the availability of (liquid) assets.

We conclude that private debt is less of a risk to economic stability in the Eurozone than prior to the crisis. Yet the former bailout countries plus Finland and Luxembourg still require attention.

History teaches us that rapid increases in private debt are a harbinger of (financial) crises. Figure 2 shows that many Eurozone countries had accumulated significant amounts of private debt[1] in the pre-crisis years. On the upside, that development has stalled in the post-crisis period. In fact, most of the private debt-challenged countries have managed to bring down private debt. On the downside, however, the adjustment following the crisis has been relatively slow and uneven, with the most significant declines seen in Spain, Portugal and Ireland.

[1] With private debt we mean the sum of household debt and debt of non-financial corporations (NFCs). Household debt consists of loans to households and non-profit institutions serving households. NFC debt consists of loans and debt securities. We use non-consolidated data. This may overestimate the debt challenge, especially for the business sector (the ratio of non-consolidated to consolidated debt ranges from 1.04 in Italy to 1.97 in Malta), but consolidated data could likely underestimate the problem. Also, non-consolidated data is more up-to-date.

Indeed, in all but four Member States private debt is still above thresholds often cited by the academic literature. This is illustrated in figures 3 and 4. For example, Cecchetti et al. (2011) argue that total private debt above 100%, household debt above 85% or debt of non-financial corporations above 90% has a negative effect on economic growth. Only in Slovakia, Latvia, Slovenia and Lithuania do we find a private debt ratio below 100%. When looked at separately, in eleven out of nineteen countries, non-financial corporate debt is still above the 90% threshold. Household debt, on the other hand, only surpasses the 85% threshold in the Netherlands and Cyprus.[2]

[2] In especially Ireland, Luxembourg, Malta and Belgium (Netherlands, Latvia and Slovenia to a lesser extent) data on corporate debt likely overestimates the debt burden on the economy. A relatively high amount of loans supplied by non-domestic counterparties is likely to be caused by either the international banking sector setup and/or debt reporting for tax reasons.

How debt can hurt economic growth

Before we move on to indicators of financial instability associated with high levels of debt, we should acknowledge that credit can bring significant benefits to the economy. It helps bring forward productive and profitable investment or it allows households to smooth consumption expenditure over time. However, alongside with debt come debt payments, which eat into (future) consumption and investment expenditure, and hence economic growth. Generally the higher the debt, the higher the debt payments. At some point, the negatives for growth may outweigh the positives, a point also made by Cecchetti et al. (2011).

Moreover, historical experience shows that in case of a(n) (expected) slowdown in economic growth, highly indebted households and businesses try to reduce debt, at the expense of consumption and investment. Slower economic activity means less employment and profit, hence less income for households and businesses to serve their debt. This in turn lowers aggregate expenditure, and so on. Deteriorating credit worthiness could eventually lead to arrears or even defaults on debt payments, and trigger a financial crisis.

In short, high and/or rapidly increasing private debt could bring about economic and financial sector stress, either by inducing an economic slowdown itself or by aggravating an economic downturn brought about by other (external) factors, such as rising interest rates.

In the Eurozone, the era of massive increases in private debt has ended (see also figure 5), but private debt remains at levels deemed unproductive or even risky for the economy. For now, interest rates are at historically low levels, not just in nominal terms but even more so in real terms (figure 6). However, assuming that the ECB will halt its asset purchases by the end of this year, (modest) rate increases seem to be on the cards from the second half of 2019 onwards. Moreover, low sovereign and private credit spreads are not set in stone – a risk highlighted by the recent political turbulence in Italy. The question whether this could pose any risks to economic growth and financial stability in the future should be seen in the context of (still) high debt-GDP ratios.

The level of debt is a rather narrow angle of looking at things. To gauge the affordability of debt and the strain it can put on economic growth, we should also look at debt payments in relation to income. The higher this ratio, the less money households and businesses have left to spend on consumption and investment and the more vulnerable they are to shocks in interest rates and income. We first look at households and then turn to non-financial corporations.

The household debt burden has fallen

For households we look at interest rate payments and debt amortization as a share of income, i.e. the debt service ratio[3]. In all Eurozone countries but Belgium, household debt service ratios have declined when compared to the year just before the crisis (figure 7). In Belgium’s defence, pre-crisis, Belgian households had about the lowest ratio in the Eurozone. In some of the most indebted countries pre-crisis (Ireland, the Netherlands, Portugal and Spain), household debt service ratios have fallen the most, on the back of deleveraging and/or lower interest rates. In two other countries with highly indebted households, Greece and Cyprus, nominal debt payments have fallen as well, but as income is also lower than before the crisis, debt service ratios have remained very high.

Household debt seems quite affordable

Currently, household debt seems quite affordable in most Member States. In neither country the household debt service ratio nears 40%, the vulnerability threshold according to institutions like the BIS (figure 8). But differences between countries are large, with households in Cyprus and Greece the most spending-constrained and in Lithuania and Estonia the least. Average figures could, however, mask the fact that debt and income are distributed differently among households. We use the Household Finance and Consumption Survey of the ECB to dive a bit deeper. We find that indebted households in the bottom 20% income category in Cyprus, Greece, Spain, Italy, Portugal, Slovenia and Estonia have debt service ratios above the vulnerability threshold (145%! in Cyprus). Yet as only 10% to 20% of the low-income households in those countries holds debt, the impact on the total economy from the high debt service ratios for these households could be rather limited. Cyprus stands out, as almost half of its low-income households owns debt.

[3] Due to lack of data, we have calculated a proxy. Debt service is interest payments (incl. FISIM - Financial Intermediation Services Indirectly Measured) + 4-quarter average of short-term loans (loans with initial maturity of <1 year). Income is gross disposable income + interest paid (excl. FISIM) (BIS calculation). We note that this figure tends to underestimate total debt service costs as it does not take into account periodical repayments of the principal of loans with an initial maturity of more than one year.

Interest coverage ratios of non-financials have improved

For non-financial corporations it is customary to look at the interest coverage ratio, i.e. the ratio of earnings to interest payments. When we look at this measure[4], non-financial corporations in the Eurozone now seem less vulnerable to income and interest rate shocks (on average) than prior to the crisis (see figure 9 and 10 below). In all Member States, the interest coverage ratio was higher in 2016 than in 2008. But the magnitude of the improvement differs widely among countries. The largest increase was seen in Lithuania, where the coverage ratio tripled from an already high level.

But also in crisis-hit countries like Spain, Italy and Ireland non-financial corporations managed to substantially lower their debt burden. Lower interest rates have helped corporations in all three countries, but whereas earnings have increased in Spain and Ireland, they have decreased in Italy. At the same time, in Spain and Italy firms have deleveraged, while in Ireland corporate debt has increased. In the other crisis countries, Cyprus and Portugal, the ratio has only improved very little.

[4] For interest payments we use ‘interest paid including FISIM’ and for income ‘operating surplus and mixed income’, which is a very broad measure of earnings, as it, among other things, also includes self-employed incomes.

The Eurozone hosts very solid, but also very vulnerable firms

According to the IMF, non-financial corporations with a coverage ratio below 2 are ‘at risk borrowers’. Cyprus and Luxembourg barely pass that threshold. But also in Portugal, France and Estonia, non-financial corporations appear way more vulnerable than in Lithuania, Slovakia and Germany. The latter seem to have nothing to worry about.

Yet here too, it should be acknowledged that there could be more vulnerable groups within the sector. The recent experience in Italy serves as a case in point. In the run-up to the crisis there was little concern about the (aggregate) non-financial sector debt. Still, non-performing loans rose sharply in the crisis years as certain sectors (construction for example) and small businesses appeared way more vulnerable than average figures had shown, causing a significant headache for regulators.

In short, the average debt burden of households and non-financial corporations has declined, but there is significant variety among countries. Cyprus hosts the most vulnerable households and firms. Households in Greece and the Netherlands also face a relatively large debt burden, and the same holds for firms in Luxembourg, Portugal and France.

In order to determine the vulnerability of households and firms to interest rate shocks it also matters how fast such shocks feed through into their interest rate payments. This first and foremost depends on the maturity distribution of the debt and/or whether households or companies have fixed their interest rates (and at what levels). Indeed, one could argue that it is the interaction between the size of the debt and the maturity that really matters. To assess this vulnerability we look at the share of loans with variable interest rates (or short reset periods)[5] (see figures 11 and 12), as these loans are obviously most prone to the risk of sudden jumps in interest rates. For non-financial corporations we also take into account loans with a remaining maturity of less than one year. It is common that corporate loans are rolled over when they mature, though not necessarily all of the maturing debt in figure 12 will be rolled over.

For households we find that while the share of variable rate loans has fallen in most Member States over the past few years, it is still (very) high in Finland, the peripheral Member States, and several East-European countries. Differences between Member States are large. In Finland and Malta about 90% of total household loans have variable interest rates, whereas in Germany, Belgium and France this is less than 10%.

[5] Our database only includes loan data from monetary financial institutions and not from other counterparts that lend to households and businesses.

Also for non-financial corporations Finland is at the top and Germany at the end of the list. But the difference in the share of variable rate loans is less marked, with around 75% of total loans at a variable interest rate in Finland versus 20% in Germany. Theoretically, to get a clearer picture of vulnerability to interest rate shocks, we should also take into account the extent to which interest rate risk on variable rate loans has been hedged with interest rate swaps. Unfortunately there is no consistent data available and so we have refrained from this[6].

Next to loans, corporations have debt in the form of debt securities. In the Eurozone, the share of debt securities in total corporate debt is (still) rather limited, though, especially the share that is vulnerable to interest rate shocks. Floating rate and short-term debt securities which could be rolled over within a year against a different interest rate make up for less than 20% of total debt securities (figure 13 &14) and less than 5% of total non-financial corporate debt (and GDP) in most Member States. Only in Portugal, the share is rather large at around 70% and 20%, respectively.

[6] Among large corporates, notably those that are not very cyclical, it is quite common to hedge a significant portion or all of the variable interest rate risk with swaps. Hedge ratios of between 60%-90% are common for non-cyclical businesses. So our measure of ‘interest rate reset within a year’ is likely to over-estimate the impact.

Broadly speaking, Finland, several Southern and several East-European Member States seem more vulnerable because higher market rates are likely to feed through quicker than in Germany, France and the Netherlands for example.

Size and starting point matter

Vulnerability to interest rate shocks not only depends on how fast interest rate shocks feed through to debt service costs. It also matters to what extent higher interest rates raise costs when they feed through. If households or businesses have locked in (past) interest rates for a longer period of time, whilst interest rates have declined (as they have in recent years), an interest rate reset not necessarily implies a negative shock for them.

Ideally, a complete maturity breakdown would be required to gauge whether an interest rate reset now would actually raise or depress interest costs. Lacking that information, we rely on a simpler metric (see figures 15 and 16): the difference between the average cost households and businesses are currently paying on existing loans (end-2017) and the rate they would pay, if they would refinance those loans now. This shows that households in Slovakia, Belgium, Germany the Netherlands and Italy are most likely to in fact ‘benefit’ from a reset in the short term; for businesses this holds for those based in Estonia, Portugal, France, Italy and the Netherlands.

Of course, also the size of debt matters. The higher the debt to income ratio, the larger the increase in the debt service ratio if interest costs rise. In this light, especially the combination of a large share of variable interest rate loans and high debt is toxic. Based on these two measures, we expect interest rate shocks to have the largest impact on household debt service ratios in Finland and the former crisis countries Spain, Portugal, Ireland and Cyprus (figure 17), despite years of deleveraging in Spain, Portugal and Ireland.

A simple ‘on the back of an envelope’ calculation shows that if interest costs were to increase by 100 basis points and income is held constant, the debt service ratio of households in Ireland is likely to increase by 1.1%-points in one year, the most of all Eurozone households. Second and third come Cypriot and Finnish households with a 1%-point change.

Turning to non-financial corporations, we expect debt service ratios of firms in Cyprus and Luxembourg to increase the most after an interest rate shock, mainly due to their very sizeable debt levels. This is followed by debt service ratios of firms in Portugal and France.

Real versus nominal

When it comes to future interest rate increases, we should also take into account whether such increases are merely a compensation for higher inflation (or inflation expectations) or whether they also represent higher real interest rates. After all, higher (demand-pull) inflation may also imply higher revenues. As figure 6 (on page 3) underscores, real rates have actually continued to fall in recent years due to gradually rising inflation expectations.

Looking to the future, one could argue that interest rates will only rise (or the ECB will hike rates) if the inflation target is met on the back of an improved economic environment. In such a scenario, higher income and profit may, partly or fully, compensate debtors for higher interest rate payments, i.e. the affordability of debt would not worsen. But, monetary policy in the Eurozone is based on the average performance of its Member States. Important here is that currently Member States are at different stages of the recovery phase, with Ireland and Germany leading the bunch and Greece and Italy dangling at the back. Due to its average-based (i.e. one-size-fits-nobody) policy, the ECB would likely tighten monetary policy too late for those countries that are ahead of the curve and too soon for those still in need of cheap money.

Moreover, an adverse external shock (think of an escalation of geopolitical and trade tensions, faster than expected policy tightening in the US, and lower than expected economic growth) could also increase risk premiums, especially in high debt countries. This, in turn, could also raise interest rates for households and businesses.

Specifically in the case of households it is also relevant to take into account their ability to eat into their savings. After all, when things take a turn for the worse, a high saving rate (especially when these savings are freely adjustable) could help households overcome the first shock of higher interest rates and adapt more gradually rather than being forced to cut back on consumption. Figure 18 below shows both the general saving rate and the freely adjustable saving rate (active saving rate).

The general saving rate is simply disposable income minus consumption divided by disposable income[7]. Part of these so-called savings is actually income that households do not use at the time. Yet another part is money used to fulfill debt obligations. As debt payments do not count as consumption, they are included in saving. In other words, the higher the debt service ratio, the higher the saving rate. Obviously, the money that is used to pay for debt is not freely adjustable and cannot help to cushion interest rate shocks.

While general saving rates in Eurozone Member States have on average declined since the crisis (Germany being the odd one out), the active saving rate (i.e. the general rate minus debt obligations) has in fact increased in all countries except Greece, Cyprus, Belgium and Italy. As such, in most Member States, households are now better able to gradually adapt to income and interest rate shocks than prior to the crisis. In some Member States, however, the rate is still (or again) negative. Based on the active saving rates we expect that especially households in the former bailout countries, Finland, and some East-European countries would have to cut consumption to service debt if debt costs increased.

As an aside, we note that the average household saving rate may overestimate the actual ability of the average household in a Member State to absorb higher debt payments. There can (indeed, is likely to) be a mismatch in the sense that those households that would see their debt servicing costs rise, are not necessarily the ones who currently save.

[7] Augmented with interest paid excluding FISIM

Last but not least, the availability of (liquid) assets could also mitigate the impact to the economy if debt costs rise. Private agents can fulfil debt obligations or maintain consumption by divesting themselves of (financial) assets. We chose to focus on liquid financial assets[8] here, instead of total (financial) assets. As we are more interested in spending constraints and liquidity risk than in solvency risks. In all countries except Cyprus, Greece and Slovenia, household liquid assets as a percentage of income have increased since the crisis. But even in those particular countries household liquid financial assets are substantially larger than debt service costs. In fact, in most Member States household cash and deposits (see figure 19) are even larger than total household debt. The Netherlands, Finland, Estonia and Ireland are the exception. The relatively large amount of household debt compared to liquid financial assets in the Netherlands and Finland (177% and 155% respectively) can be explained by large pension savings in both countries. So the need for precautionary savings is relatively low.

When we turn to non-financial corporations we find that in all Member States deposits are at least five times larger than annual interest payments on debt, and even as much as eighteen times in Slovakia (figure 20)[9]. Deposits are on average about 4.5 times smaller than total debt of non-financial corporations in all countries. Yet total liquid assets (see footnote 8) are larger than debt in all countries but Portugal, Spain and Cyprus.

[8] Usually, liquid assets consist of cash and deposits, bonds, mutual funds, listed shares and products like derivatives. We only look at cash and deposits, as the overall value of financial assets except cash and deposits could fall in downturns when households and business are most likely to need them to repay debt or maintain consumption levels. Moreover, only high income households tend to have other financial assets than cash deposits.

[9] Due to its international banking systems, Luxembourg stands out in figure 20. Luxembourg has a high amount of non-domestic deposits.

The same caveat as to savings applies: debt and assets are unlikely to be equally distributed among households or firms. For households we know for example that on average in Eurozone countries, the 90th income and net wealth percentile holds around half of total household financial assets, while the 20th income percentile holds less than 2.5% and the 20th net wealth percentile less than 1.5%. The upshot is that if we don’t take (liquid) assets into account, we tend to overestimate the vulnerability of households, firms and the economy to interest rate (and income) shocks. Yet if we do, we are likely to underestimate the vulnerability, possibly quite substantially, for certain groups.

In this Special we looked at a number of indicators of economic and financial stability risk stemming from this relatively high level of debt, particularly in the context of potential future increases in market–based interest rates. In particular we looked at debt affordability, maturity profiles, headroom in savings and liquid asset positions. We have combined a number of indicators into the heat map below (table 1). Although a heat map does not always allow for the necessary nuances in comparisons (take for example the high household debt ratio in the Netherlands, which has its counterpart in high pension wealth), it still helps in spotting patterns and getting an overall picture of where risks might be most prominent.

To sort the countries in the heat map, we have first calculated z-scores for each separate variable. We skipped the overall private debt/GDP ratio and replaced the debt/income ratios (for both households and corporates) by the interaction of debt/income multiplied by the share of variable rate loans.

We then added the z-scores by giving each variable a weight of 1 (or -1 in case the variable is considered negatively correlated with interest rate vulnerability). We then ranked the countries from the highest to the lowest overall z-score.

Despite recent deleveraging, private sector debt remains high in many Eurozone Member States. In most countries, debt (either for households and non-financial combined, or separate) remains above levels deemed damaging for long-term growth. Yet, broadly speaking, debt at non-financial corporations seems more of an issue than household debt.

More specifically, we conclude that average vulnerability of households and non-financial corporations to higher interest rates has decreased in recent years. Moreover, the risk of a near-term hike in European interest rates seems rather slim and if it were to happen, it would likely take place against the backdrop of rising (demand pushed) inflation.

Still, several countries, in particular the former bailout countries, Cyprus, Greece, Ireland, Portugal, and Spain, plus Finland and Luxembourg require ongoing attention (although the non-financial corporate debt data especially for Luxembourg, but also Ireland is likely to be skewed due to a relatively high amount of ‘non-domestic loans'). Notably, in Italy private sector debt seems to be (much) less vulnerable.

Household debt challenge has decreased, but vulnerabilities remain

In most Eurozone countries household debt service ratios have declined when compared to the year just before the crisis and, hence, affordability has improved. In neither country the average household debt service ratio nears the 40% vulnerability threshold advocated by institutions such as the BIS. Although, the ratio for low-income households in Southern Member States, Slovakia and Estonia does.

One specific vulnerability to interest rate shocks is the share of short-term loans and/or loans with an interest rate reset over the next 12 months. This share is highest among households in Finland, Malta, the Baltics, Portugal and Spain. Based on their (relatively) low active saving rates and (relatively) high debt service ratios we expect that especially households in the former bailout countries and Finland would have to cut consumption to service debt if debt costs were to increase (although debt ratios themselves are actually low in the Baltics). The availability of sufficient liquid assets (currency and deposits) is one factor that could mitigate the initial impact on consumption, although it is difficult to draw conclusions in this respect based on average data figures.

Corporate resilience has significantly improved but certain countries require attention

Since the crisis, interest coverage of non-financial corporations in the Eurozone has also improved significantly, making them less vulnerable to income and interest rate shocks. But there are huge differences. Non-financial corporations in Cyprus and Luxembourg as well as in Portugal, France and Estonia, still seem more vulnerable than their European counterparts (although the data especially for Luxembourg is likely to be skewed due to a relatively high amount of ‘non-domestic loans'). Vulnerability to interest rate shocks (when only focusing on interest rate resets) is relatively high in Finland and the Baltic states, but when we include short-term debt, most former ‘problem’ countries are also more vulnerable than the average. However, the relatively low debt-GDP and debt-income ratios in the Baltics should limit the overall impact on the economy.

Stephen G Cecchetti, M S Mohanty and Fabirzio Zampolli (2011), “The Real Effects of Debt”, BIS September 2011.