Eurozone pent-up demand: big and decisive or over-estimated and uncertain?

It surely must be one, right? With the virus retreating, a window is opening up for a gradual unwinding of lockdown restrictions, which, until very recently, were close to their strictest since the first wave hit the European continent. Meanwhile even the services part of the economy is already showing signs of life (figure 1). Of course, the shape of the recovery path this year still largely depends on the vaccination pace, but on this front things clearly seem to have turned a corner (figure 2). As such the European Commission’s target to have 70% of the adult population (or just under 60% of the total population) vaccinated by September no longer seems remote. Indeed, according to EC President Von der Leyen, this point has recently shifted to end-July.

Such an opening up of the economy obviously raises the question of how strong the recovery in consumer spending will be. The optimists argue that the built-up in (forced) savings will unleash a wave of pent-up demand once people are allowed to make good on postponed consumption, particularly in the hospitality sector (hotels, restaurants, holidays etc.).

A recovery in GDP, as the economy opens up, is perhaps a no-brainer, but its speed and strength will partly depend on this ‘pent-up demand’ effect. Whether this will be ‘big and decisive’, or ‘over-estimated and uncertain’ is the question we will try to answer in this research note.

One key fact is that, since the pandemic, we have seen a strong rise in household saving. So let’s first zoom in on this. How much was saved, why did households save, which countries saved most and what can we say about the key drivers behind this household saving?

We estimate that up until 2021Q2 a cumulative c. EUR600bn[1] - in excess of what normally would have been saved - has ended up in Eurozone household’s piggy banks.[2] This is about 8.3% of gross 2019 disposable income. And, at 5% of GDP it’s almost equal to the current ‘pandemic’ GDP gap. A comparison with the EUR750bn EU Recovery Fund announced last year (whose disbursements will be spread over several years) only further puts the size of these savings in perspective. Should these be unleashed in the form of additional consumption, this would provide a significant boost to economic activity, even when spread out over several quarters.

[1] All figures mentioned here are in current prices.

[2] Of which EUR 517bn in 2020 (particularly in Q2 and Q4); a broad-brush estimate for 2021Q1 is EUR 83bn, but this is subject to quite some uncertainty given a lack of data for that quarter.

So let’s look at the drivers behind the rise in saving. Clearly, 2020 was anything but normal, even when we compare it to previous episodes of economic distress. In table 1 above, we show the breakdown and/or drivers of the additional saving in 2020. We also compare it with two historical episodes (we chose a period spanning one year): i) the Global Financial Crisis (GFC) and ii) the ‘second leg’ of the Sovereign Debt Crisis (SDC). Three things stand out:

Although there is ample reason to assume that a large part of these savings were involuntary due to the specific nature of the pandemic crisis (the lockdown measures constrained people’s ability to spend in sectors like hospitality, transport and recreation), the historical evidence indicates that economic uncertainty and/or financial distress often leads to precautionary saving.

In figure 3 we show those times of economic stress in the form of our recession probability indicator that we developed in 2019. We also constructed a long historical series of the household aggregate savings ratio in the Eurozone.[3] We then selected several of those episodes (as shown in figure 4). The starting point here is the last quarter before the recession probability exceeds 10%. Although there is quite a broad range, most of these episodes show a 0-2 %-points increase in the saving ratio after 4-6 quarters, before it gradually returns to its pre-crisis level in the quarters following.[4] Although many people affected by the Covid crisis received some form of government support, they may still have treated such income support differently. Precautionary saving probably mostly took place in the first wave of the crisis, a view that is also backed by surveys which have been showing a steady rise in confidence in the economy since 2020Q3.

[3] Due to a lack of reliable data, notably before the 1980s, this series is subject to a considerable error margin.

[4] Aside from the 1981/82 recession, on the heels of the 1980 recession, the sovereign debt crisis is the odd one out here. One explanation is that households in the South of Europe were forced to dis-save in order to make ends meet, as their saving ratios at the time were low to start with and government support was basically lacking.

Indeed, to get a better grip on where savings ended up (and why) during the pandemic, we zoom in on the European Commission’s Consumer Sentiment Survey that is held on a monthly basis throughout the bloc.

In particular the breakdowns by age group and income quartile provide some insight into the motivations behind the recent surge in household saving. That, in turn, should give us some pointers as for the future direction of saving.

The figures below show the changes between February 2020 (last month before the pandemic) and April 2021 (last month available) in the net balance of answers in the survey, broken down by age (figure 5) and income quartile (figure 6). A number of observations stand out:

[5] Technically, this particular question asks whether people can put money away in future rather than whether they intend to save more. Still, this may also tell us something about their plans.

A more detailed technical paper by DG ECFIN, which also looks at this issue, concludes that: “[…] the analysis of the consumer survey results by income and age groups does not point to a significant impulse from pent-up consumer spending going forwards. In particular low income earners report low or no accumulated savings, but high savings intentions and low intention to spend money on major purchases. At the same time, the high savings accumulated by the higher income group go hand in hand with increased expectations to save, relative to the pre-pandemic readings, and no increase in the intention to make major purchases. Similarly the relatively large increase in the saving position of older people is not likely to be quickly reversed, given the low propensity to consume for this group.”

The report then also points out that the “collapse in the consumption of durable goods was to a large extent recouped over the summer of last year, when restrictions were eased but also enabled by consumers’ increased recourse to on-line shopping”, but “[…] given that foregone consumption of services can hardly be recouped, the prospects for pent-up demand for services to strongly boost future consumption may thus be limited.”

All in all, then, the distribution of past (excess) savings suggests that only some part of those savings may eventually turn into additional consumer demand in future.

In the appendix we also look at how excess savings have been distributed among the member states and we ponder whether the (disproportional) additional savings racked up in Spain, Italy and Ireland were more of the ‘precautionary’ kind than in member states such as France and Germany. We would answer this question with a “yes”, but immediately add that it is somewhat of a speculative assertion.

Although a significant amount of excess saving during the pandemic was of a forced nature, even those savings were probably largely attributable to households that have a high natural propensity to save, as we argued in the previous section of this note.

Of course, household behavior is not static, as has been demonstrated by the remarkable resilience of consumption in recent quarters, in the face of continuing lockdowns. Quantifying the impact on future consumption is also more complex due to other factors that will likely influence the behavior of households during the recovery phase. We shall discuss two of them in particular:

1. The wealth effect (which, arguably, is most important for the high income brackets)
2. The extent of government support in the recovery phase; i.e. what will happen to disposable income and how will this affect confidence?

These factors will form the building blocks of our future consumption recovery scenario’s that we will discuss in the next section. But first let’s have a closer look at these specific factors.

Whose wealth?

When you become more wealthy, you may spend some of that extra wealth, right? Household net-financial wealth reached a record-high by end 2020 and recent developments in financial markets suggest that Q1 2021 even smashed that record. And note that we are abstaining from sizeable increases in housing values over the past couple of years! As figure 7 below points out, in the past this would have been a clear signal for households to save less.[6] But since 2013 (pretty much coinciding with the tail of the sovereign debt crisis and the start of even more unorthodox central bank policies such as negative interest rates and QE), this no longer holds. The negative correlation between net financial wealth and the saving ratio became less consistent after the GFC in 2008 and has arguably reversed since QE (since 2013/’14 the saving ratio has been stable or rising despite wealth gains).

In other words, although the wealth effect would argue that there could be a positive impact on future consumption, notably among the higher income/wealth groups, recent history has been less kind to the believers in such a wealth effect.

Again, distribution effects (the rich getting richer, basically) could be the explanation for this reversal in correlation. Another explanation is that people have felt the need to save more ‘just because interest rates are so low’.

[6] Basically, this chart would suggest that a 10% increase in wealth would lead to 0.25%-points fall in the saving ratio.

Given that ECB policy only became more unorthodox with the introduction of PEPP and more and more banks have started to charge for saving deposits now, one could make the case that the rise in wealth during the pandemic of 2020/’21 does not necessarily point to a lower ‘equilibrium’ savings rate in future. And to the extent that the excess savings were seen as purely transitory income, we would not bank strongly on the wealth effect when it comes to making any forecasts. But if such a wealth-effect does materialize, a fall in the savings rate towards 10% would be consistent with pre-2013 behavior, as suggested by the chart.

Extraordinary government support, so…

Having said that, the pandemic has also proved to be a major break with the past decades when it comes to fiscal policy. Indeed, whereas fiscal policy during the GFC and sovereign debt crisis more often acted as an attenuator rather than an amplifier, this time it has been quite different. As my colleagues Lize Nauta and Wouter Eijkelenburg point out in this special, the size of government stimulus has been unprecedented in peace times.

In the EU this is encapsulated in a whole different approach with the SURE fund and Resilience and Recovery Facility. This also provided a backdrop which enabled member states to pursue their own pandemic support policies. As we already pointed out earlier, the amount of government support is a key explanation for the upholding of disposable incomes and thus – given the constraints on consumption – the rise in excess savings. Indeed, when we think of the sectoral balances of income-spending flows, the following identity must always hold:

Budget balance of government + Savings balance of private sector + External balance = 0

In other words, assuming no change in external balance, more government support (reflected in a higher budget deficit) should be reflected in a higher private savings balance (which comprises both households and businesses).

In the aggregate this was clearly the case when we look at households only: higher saving by households went hand in hand with rising government support. The variation between member states only further confirms this. The deterioration in the overall budget balance (which includes both discretionary measures related to the pandemic as well as automatic stabilizers) actually correlates quite well with the rise in household saving ratios. This is illustrated by the blue dots in figure 8. The dots are all below the 45 degree line which means that either the external balance deteriorated or (more likely) businesses also saved in aggregate.

This confirms that government support measures have indeed been the key driver behind the rise in savings. Or, put differently, they kept up disposable incomes and thus allowed households to save in the face of consumption constraints.

However, this also raises other important questions. For example, did people really save because they could not consume certain services? Or did they also save because they were concerned that some of that income support will ultimately have to be repaid in the future? (i.e., in the form of higher taxes or lower future government spending).

For excess savings to be spent in future, a key factor is whether households expect governments to continue to provide support during the recovery phase. If support measures are wound down too quickly this would obviously cement the idea among households that saving was ‘necessary’ after all. Or it would lead to concerns about future tax hikes for the higher incomes – something we saw last month in the US with the launch of President Biden’s ‘American Families Plan’.

Moreover, if support is wound down too quickly, excess savings may be used to keep consumption steady but it wouldn’t lead to an additional boost to GDP growth. Ongoing support and stability, on the other hand, could actually mobilise those excess savings and channel them back into the economy through additional consumption.

In that sense, a steady hand fiscal policy – avoiding negative surprises for households – could prove to be an important element in getting at least some of those savings back into the economy once we leave the pandemic behind. With its Resilience and Recovery Facility and the Stability and Growth Pact ‘holiday’ until 2023 the EU seems to have understood this point. But, when supply comes back on, will national governments still react in a Pavlovian way by tightening the purse strings?

At this stage, we would argue things could go both ways, and this only underscores the need for scenarios. Hence, this is what we will explore in the final section of this note.

Given the uncertainties and many factors that will influence the future behavior of households we have constructed three scenario’s: a base case scenario (which is the most likely outcome in our view), an optimistic and a pessimistic version.

A key factor in these scenarios is whether and to what extent past excess savings will return in the form of additional consumption. Because the excess savings during the pandemic likely ended up mostly in the higher income brackets, we first need to make an educated guess of the distribution of that EUR600bn number.

This is obviously a daunting issue, given that this crisis has been far from normal. What we do know is that the aggregate saving rate differs significantly between the income groups during normal times. Experimental data from Eurostat provides an estimate of the 2015 saving ratio by income quintile.

We have transformed this to a quartile distribution – to allow comparison with the Consumer Confidence Survey by the European Commission. We zoom in on question 12 in the survey, which asks whether people are currently dissaving (negative) or saving (positive). The indicator is derived by taking the balance of positive and negative answers.

This confrontation with the Eurostat data in figure 9 (the dotted lines) is actually pretty astonishing in its own right. Moreover, the first quantile was the only income group that reported more ‘dissaving’ between February 2020 and April 2021 as compared to the other income groups (where the net balance of answers also increases with income).

In short, this suggests that this distribution does serve as a rough ‘guestimate’ of where the EUR600bn of additional savings ended up.[7]

For the sake of simplicity we will assume in the following scenarios that the lowest quartile dis-saved EUR50bn, the second neither dis-saved nor saved and that the highest two quartiles saved EUR200bn and EUR450bn respectively.

Base case scenario

As we have demonstrated, a significant amount of the excess savings may not return in the form of additional consumption, although some of it may still do.

(i) Marginal propensity to consume extra savings

In this scenario we therefore assume that the marginal propensity to consume out of a transitory income shock is around 0.2 on average during an expansion. For the first income quartile we put this at 0.30, for second at 0.25 for the third at 0.2 and for the fourth and highest quartile at 0.15.[8] Basically we treat those savings as a ‘windfall’ income gain. Applying the marginal propensities to consume to the additional savings by income quartile gives us an additional impulse of around EUR110bn, which we evenly spread out over four quarters starting 2021Q3 (we assume no additional impulse in 2021Q2, but the halt to excess saving is in itself a positive impulse for consumption.

(ii) Developments in disposable income

Furthermore, whilst compensation of employees should be fully ‘normalised’ by the middle of 2022, mixed income and the gross operating surplus will only have closed half of their gaps with end-2019 levels. The thinking here is that (small) businesses will need to deal with the fall-out of the pandemic crisis. We assume that government support measures are wound down gradually over the course of one year and roughly in tandem with the recovery in employee compensation.

(iii) Inflation

The recovery leads to an additional rise in prices (due to the shift from goods to services), with the contribution to overall inflation being 1%-points annually until end-2022.[9] In the box below we discuss this issue in slightly greater detail.

This scenario results in a 4.1% boost to consumption volume growth this year and +2.8% next year. Compared to our current projections for consumer spending growth (+3.1% and 4.3% respectively), which assume NO additional pent-up demand effect nor an additional impact on prices, this implies somewhat higher consumption growth for 2021 (+1%-point) and lower growth for 2022 (-1.5%-points).

Note that a considerable part of the impetus simply results from excess saving coming to a halt in this scenario, but also that most of those gains are then being ‘eaten away’ by higher prices.

[7] In a recent study our Dutch economists basically arrive at the same conclusion: only a subset of the Dutch population saw their savings rise (c. 35%) and the highest additional savings were reported by the group with an income above the median.

[8] Some support for this assumption is provided in an ECB working paper from 2014, although we should add that it is based on US data.

[9] Although we have pencilled in this effect in our official inflation projections, this is somewhat of a less conservative assumption than in those projections.

One for the optimists…

We do feel that our base scenario is a realistic one, although we do acknowledge the huge uncertainties surrounding these projections. It would certainly appear that financial markets are taking a more optimistic view, seeing the recent upward pressures on inflation expectations and long-term interest rates.

(i) Marginal propensity to consume extra savings

So, in a more optimistic scenario, we now assume that a much bigger chunk of past excess savings returns to the economy in the form of additional spending. The highest estimates of the marginal propensity to consume out of transitory income in the literature range from 0.5 to 0.75. We assume 0.5 in the highest income group and 0.7 in the 3rd quartile. That would mean that some EUR 340bn returns to consumption in future (again, evenly spread over four quarters). This is pretty high to our liking, but – as said – this is one for the optimists.

Given the limitations in services in terms of repeat spending as pointed out in the box on the previous page (going on holiday twice, visiting restaurants at a higher rate than in 2019), this would most likely be in the form of additional spending on goods as well.

(ii) Developments in disposable income

Similar to the base scenario we assume support measures are wound down in tandem with the recovery in employee compensation but we also assume a full recovery in other primary income sources. Moreover, we assume that due to the positive demand-shock in this scenario, those primary incomes (and hence disposable income) revert more quickly to their pre-Covid levels (i.e. by 2022Q1) and continue to rise thereafter at a similar rate as in 2019.

(iii) Inflation

Due to the inflationary nature of this scenario, we assume a bigger impact on consumer prices of 2%-points annually over the next six quarters.

This results in a 6.2% boost to consumption volume growth this year and +2% next year. So in this more optimistic scenario the consumption gap is closed more quickly, but consumption fails to rise much in 2022 projection as inflation dents the impact of pent-up demand. In that sense this is more of a (mild) ‘boom-bust’ overheating scenario.

… and one for the pessimists

If only for the sake of balance, we believe it is also worth looking at more bearish scenario. In this scenario we assume that no past excess savings return in the form of future consumption. Indeed, it is not unthinkable that the side-effects of the pandemic only become visible when governments withdraw their support measures and if we assume this is done too quickly or forcefully, this will not only contribute to ongoing uncertainty but it would also cement the idea that those savings in the past were needed after all. As such, a further rise in excess savings is clearly possible (although we did not take that into account).

More importantly, in this scenario we assume that disposable incomes fall initially because a decline in social contributions plus tax rises cannot be fully offset by a rise in household primary incomes, which themselves are hurt by the weak recovery. Meanwhile, businesses still hike prices in order to restore their profit margins, albeit at the same pace as in our base scenario.

This scenario results in a meagre 0.8% rise in consumption volumes this year and a relatively sharp fall of -3.3% in 2022. In other words, another ‘post-pandemic’ recession.

For comparison purposes, the volume of consumer spending fell around 8% in 2020.

By construction, the base case scenario is the most realistic one in our view. It suggests a decent but far from spectacular return of consumption this year and next.

The 2021 projections are, in a way, less interesting because they enjoy an almost unavoidable ‘base effect’ from the fact that even when additional saving stops, the economy is likely to receive some boost from that. This also features in our current consumption projections for the Eurozone.

However, by pondering a positive and negative scenario we also get some idea of what else is possible. These exercises underscore that both ‘over-heating’ and a (triple dip) ‘recession’ are among the range of possibilities over the next six quarters. That itself, we believe, is an important outcome. Can Europe really afford to have another recession? Or, will too quick a recovery lead to demand outstripping supply and, as a result, will we see further upward pressures on prices on top of the inflation bump that is currently playing out?

A considerable amount of excess savings are likely to stick and that is also reflected in the fact that even in the optimistic scenario the saving ratio does not fall below 8%. In an environment of low or negative interest rates, these savings will continue to exert upward pressures on financial assets.

A key message to Eurozone policy makers would be to provide sufficient policy certainty during the post-pandemic recovery phase so as to mobilize as much as possible any of the ‘footloose’ savings that have been piled up by households.

Given the planned ‘grand reopening’ in many countries over the next few months, it does seems that we will soon find out whether pent-up demand is ‘big and decisive’ or ‘over-estimated and uncertain’.

Zooming in on some of the bigger Eurozone member states, as a general pattern we see saving increased everywhere but that the rise in saving was smaller than one would expect on the basis of the size of their economy in Finland, Germany, Portugal, Austria and France. This is shown by the dark blue line in figure 14.

In Ireland, Belgium, Spain and Italy, on the other hand, the increase in the saving ratio was disproportionally high. Broadly speaking the rise in the saving ratio correlates positively with the change in bank deposits by households. Although –here too- there is considerable variation, which is to be expected given that we are comparing flows with changes in a specific balance sheet item.[10]

The pattern for household borrowing is much more divergent. On balance there is actually a slight positive correlation (i.e. higher borrowing coincides with higher deposit), which is somewhat counter-intuitive, as one would expect higher borrowing to lead to higher deposits, all other things equal.[11]Note that in the chart net positive borrowing is shown as a negative number. Germany, France and Austria are the prime examples, where saving was below-average but net borrowing above average. But Belgium, Greece and Finland distort the overall picture.

So one could ponder that in those countries where saving was above average but borrowing below-average (i.e. Spain, Italy and Ireland), i) households faced more constraints on bank lending during the pandemic and/or ii) stronger motives or incentives among households to reduce/repay their existing debt, iii) fewer possibilities to spend money due to more stringent lockdowns, iv) less need for borrowing.

Combining these observations, one could stipulate that the additional saving in Spain, Italy and Ireland was more driven by pre-cautionary motives than in Germany and France. The implication of that would be – ceteris paribus – that the additional savings racked up in the South of Europe may prove more difficult to ‘mobilise’ in the recovery phase than those (relatively smaller) additional savings in the North. We admit, though, that this is quite speculative. After all, if only part of those savings were involuntary, there could still be a sizeable boost to consumption once these savings are ploughed back into the economy.

[10] The two are correlated (the cross-country correlation between the rise in saving and bank deposits is approximately 0.66) but not 1-for-1 related, as i) savings may end up in other financial assets than bank deposits, such as equities or cash (although buying equity not necessarily changes this if one household buys and the other sells), ii) banks are not the only entity where these excess savings end up, and iii) balance sheet items may change due to valuation changes and/or borrowing.

[11] Simple example: if you borrow for a house (and assume you buy this house from someone), then the aggregate amount of deposits in the system has risen, but so has aggregate borrowing by households.