VoxEU Column COVID-19 Macroeconomic policy Europe's nations and regions

Fears for the future: Saving dynamics after the Covid-19 outbreak

The Covid-19 pandemic has seen a surge in household savings across the world. This column exploits information from the Bank of Italy’s Special Survey of Italian Households to break down recent saving patterns in Italy. It finds an increase in precautionary savings associated with factors such as higher job uncertainty and fears of a protracted pandemic. Additionally, it suggests that a lasting precautionary attitude could slow the decumulation of piled up savings, even as the pandemic slows down.

Soon after the COVID-19 outbreak, households’ savings surged worldwide, especially in the advanced economies (Figure 1). In Italy, after the very high levels attained during the first wave of the pandemic (above 20% in the spring of 2020), the saving rate remained significantly above pre-pandemic figures at the end of last year despite the partial relaxation of mobility restrictions. Consistently, 39% of households interviewed in the Bank of Italy’s Special Survey of Italian Households (SSIH) between the end of February and the beginning of March 2021 accumulated savings in 2020; before the pandemic shock this share was usually much lower (roughly 30% in 2016)  (Bank of Italy 2018). Several factors might have contributed to this phenomenon, including: a heightened precautionary attitude; the fear of contagion, which discouraged households from engaging in consumption of certain services; reduced consumption possibilities caused by the government mandated restrictions (see, among others, Immordino et al. 2021 and Guglielminetti and Rondinelli 2021 for the case of Italy, Ercolani 2020a for the US, Christelis et al. 2021 for the euro area).

As the pandemic is progressively brought under control, also with the crucial help of vaccines, an important and highly debated issue regards the prospects of savings and consumption in the nearer and more distant future (see, among others, Bilbie et al. 2021 and Romei 2021). In Italy, for example, the information contained in the SSIH may suggest that some households could maintain a precautionary attitude. First, between the end of February and the start of March, roughly two thirds of the households thought that the current health crisis would have been resolved by the end of 2022 and the rest saw it continuing at least until 2023. Secondly, on average, households assigned a probability of roughly 50% to the chance of experiencing another pandemic in the next ten years,1 quite a large figure if compared with the evidence available so far.2 Finally, about half of the households foresaw a worsening of the general economic situation and of the labour market in the next 12 months. In addition to macroeconomic conditions, individual circumstances may affect saving intentions and possibilities as well: more than 25% of households claimed that they had difficulty in making ends meet;3 roughly 25% saw their financial situation worsening after the pandemic;4 almost one fourth of household heads judged their own employment situation or that of a family member as uncertain.5 These last factors are partly interconnected: half of those facing an uncertain employment situation were also experiencing financial difficulties.

Figure 1 Household saving rates across some advanced countries


Source: Bureau of Economic Analysis and Eurostat.
Notes: Gross household saving rates from Quarterly Sectoral Accounts.

Some of the abovementioned variables actually correlate with the desired saving that households would have accumulated to protect themselves against contingencies, such as uncertainty about future income or unexpected health shocks.6 Figure 2 shows that, on average, desired precautionary saving is significantly higher for those who believed that the health emergency would have lasted beyond 2022, as compared to those expecting an earlier end. Further, those attaching a probability higher than 75% to a new pandemic event occurring in the next ten years reported a significantly higher level of precautionary saving. In contrast, differing perceptions on the macroeconomic outlook were not clearly associated to desired precautionary saving levels. 

Figure 2 Italian households’ desired precautionary saving: breakdown by expectations and financial situation



Source: our calculation from Wave 4 of the SSIH (February-March 2021). The desired level of precautionary savings is divided by the OECD equivalence scale, that attributes a coefficient of 1 to the first adult in the family, 0.5 to additional family members who are at least 14 years old and 0.3 to those younger than 14.
Notes: *p<0.1, **p<0.05, ***p<0.01. Stars denote the significance of the test of equal means between the two groups within each survey question listed on the x-axis. 

Regarding individual conditions, both households reporting their own employment situation as uncertain and those claiming to make ends meet with difficulty declared a desired precautionary saving significantly higher than the rest of the households (Figure 2). Indeed households declaring financial difficulties may be liquidity- or credit-constrained, thus being more inclined to accumulate savings to face unexpected income losses that would otherwise force them to cut spending (for more details, see Ercolani 2016 and references therein).

To investigate the extent to which actual savings in 2020 are associated to the desired precautionary saving, we recall a standard saving regression as in Mody et al. (2012), where the saving rate is explained by measures of income, wealth, and proxies for precautionary savings, and adapt it to our data and context (for details, see Ercolani et al. 2021). In sum, our results show that the probability of having saved in 2020 was sustained by several factors, such as the desired precautionary saving and the fear of contagion, whilst it was negatively affected by households’ financial difficulties.

Hence, a tension emerges. On the one hand, there is a group of households – those with financial difficulties – who desired to accumulate (precautionary) saving, but had no resources to do so; this likely contributed to curb the actual saving dynamics. On the other hand, precautionary motives may have exerted an upward pressure on savings for those not in financial difficulties. While the behaviours of these two groups of households suggest that saving dynamics are shaped by contrasting forces, they both point to reduced consumption growth and moderate pressures from pent-up demand (see Guglielminetti and Rondinelli, 2021).  

Heightened precautionary attitudes could signal the emergence of ‘scarring effects’: consumers who live through periods of high unemployment or marked financial turbulence may remain pessimistic about the economic situation and hence accumulate more savings (see, among others, Malmendier and Sheng Shen 2021). The COVID-19 pandemic could indeed leave persistent scars on the public’s perceptions; for example, 85% of the households in the SSIH declare that the outbreak of COVID-19 has raised their perception about the risk of a new epidemic occurring in the next ten years.7 

All in all, our analysis suggests that the pandemic and its scars have reinforced households’ precautionary attitudes; this could slow the decumulation of the savings piled up in 2020, at least for households without financial difficulties. Indeed, this seems consistent with some information taken directly from the SSIH; the share of households who will expect to save in 2021 is roughly 45%, 6%above the share of those that actually saved in 2020.

Policies aimed at reducing actual and perceived health and economic risk would be particularly welcome in the current context (Scarpetta et al. 2020). Making national health system more resilient to pandemics, and more generally strengthening their quality, in Italy as well as in other countries, would not only save future lives but could also mitigate the households’ precautionary attitudes, favouring the post-crisis recovery (see Ercolani 2020b and 2021 for more details). Moreover, in order to attenuate the uncertainty stemming from employment conditions, it would be appropriate to enhance policies aimed at preventing discontinuities in working histories and income, such as hiring incentives for permanent employment (Sestito and Viviano 2016, and Ciani et al. 2019). 

Authors’ note: The views here expressed represent those of the authors and not necessarily reflect those of the Bank of Italy. We would like to thank for useful comments and suggestions A. Brandolini, P. Catte, P. Del Giovane, A. Finicelli, S. Neri, L. F. Signorini, I. Visco, G. Zevi, R. Zizza and F. Zollino.


Bank of Italy (2018), Survey on Italian Household Income and Wealth 2016, Statistics.

Bilbie F, G Eggertson and G Primiceri (2021), “US ‘excess savings’ are not excessive”,, March.

Christelis, D, D Georgarakos, T Jappelli and G Kenny (2021), “Heterogeneous effects of Covid-19 on households' financial situation and consumption: Cross-country evidence from a new survey”,, June.

Ciani, E, Grompone, A and E Olivieri (2019). “Long-term unemployment and subsidies for permanent employment”, Temi di discussione (Economic working papers) 1249, Bank of Italy.

Ercolani, V (2016), “The precautionary saving: theories, measurements and policies”, Economic Bulletin and Financial Stability Report Articles, Bank of Portugal.

Ercolani, V (2020a), “Covid-Induced precautionary saving in the US: the role of the unemployment rate”, Covid-19 Note, July, Bank of Italy.

Ercolani, V (2020b), “Strengthening national health systems and dampening precautionary attitudes”, Covid-19 Note, November, Bank of Italy.

Ercolani, V (2021), “The macroeconomic impact of infrastructure investment: a review of channels”, Questioni di Economia e Finanza (Occasional Papers), 613, Bank of Italy.

Ercolani, V, E Guglielminetti and C Rondinelli (2021), “Fears for the future: saving dynamics after the Covid-19 outbreak”, Covid-19 Note, June, Bank of Italy.

Gambacorta, R, A Rosolia and F Zanichelli (2020), “All in it together, but with differences: The finances of European households through the pandemic”,, April.

Guglielminetti, E and C Rondinelli (2021), “Consumption and saving patterns in Italy during Covid-19”, Questioni di Economia e Finanza (Occasional Papers), Bank of Italy, forthcoming.

Huggett, M (1993), “The risk-free rate in heterogeneous-agent incomplete-insurance economies”, Journal of Economic Dynamics and Control 17(5-6).

Kaplan, G, G Violante, and J Weidner (2014), “The wealthy hand-to-mouth”, Brookings Papers on Economic Activity 1

Immordino, G, T Jappelli, T Oliviero and A Zazzaro (2021), “Fear of Covid-19 contagion and consumption: evidence from a survey of Italian households,” CSEF Working Papers 601, Centre for Studies in Economics and Finance (CSEF), University of Naples, Italy.

Malmendier, U and L Sheng Shen (2021), “Scarred consumption”,, March.

Martin, I and R Pindyck (2021), “Welfare costs of catastrophes: lost consumption and lost lives”, The Economic Journal 131(634).

Romei, V (2021), “Global savers’ $5.4tn stockpile offers hope for post-Covid spending”, Financial Times, 18 April.

Sandman, P (2007), “A severe pandemic is not overdue-it’s not when but if”, Center for infectious disease research and policy.

Scarpetta, S, M Queisser, A Garnero, and S Konigs (2020), “Supporting people and companies to deal with Covid-19: Options for an immediate employment and social policy response”,, April.

Sestito, P and E Viviano (2016), “Hiring incentives and/or firing cost reduction? Evaluating the impact of the 2015 policies on the Italian labour market”, Questioni di Economia e Finanza (Occasional Papers) 325, Bank of Italy.


1 Households have been asked the following question: “In your opinion, what is the probability that a new pandemic event will occur in the next 10 years?”.

2 By rough calculations, this figure would imply a 5% probability per year for a new pandemic event. This frequency looks high if compared to those extrapolated in other studies. For example, as discussed in Sandman (2007), since the 18th century there have been 9-10 pandemics (excluding Covid-19), implying a probability of a pandemic of roughly 3% per year. Further, Martin and Pindyck (2021) calibrate their epidemic model for two different scenarios: a low- and a high-risk one in which on average the annual probability of a pandemic is 2% and 4%, respectively.

3 According to data from the Bank of Italy Survey on Household Income and Wealth (SHIW), the percentage of households at risk of poverty in income and liquid wealth reached 19.5% in 2016 (Bank of Italy 2018, Gambacorta et al. 2020), around the same order of magnitude of the percentage of hand-to-mouth households – i.e. holding little or no liquid wealth – estimated by Kaplan et al. (2014).

4 The question was asked with reference both to a household’s current situation and to that prior to the Covid-19 crisis.

5 The individual measure of uncertainty is a dummy variable equal to one if the self-reported probability of (1) losing a job is higher than 25% for employed household members, and (2) finding a job is lower than 25% for unemployed individuals. As expected, the share of uncertain households is higher among households whose head is self-employed or unemployed; for the employees, this share increases with the instability of the contract. 

6 The question in the survey is: “Approximately, how much should your household put aside for unexpected events, e.g. health problems or other emergencies?”.

7 After the question reported in footnote 1, households have been asked: “In your opinion, has your assessment – about the risk of a new pandemic in the next 10 years – increased following the outbreak of the Covid-19 pandemic?”.

3,779 Reads