Keep calm and carry cash: lessons on the unique role of physical currency across four crises
Published as part of the ECB Economic Bulletin, Issue 6/2025.
1 How crises strongly affect cash circulation
Demand for euro banknotes has exhibited robust growth despite ongoing payment digitisation. While the share of cash in daily transactions has declined in the euro area, the value of euro banknote circulation has significantly increased over the past two decades (Chart 1, panel a). This variable serves as a reliable indicator of overall demand – domestic and foreign – as the Eurosystem accommodates requests for banknotes. In fact, the value of outstanding banknotes has consistently maintained a share of over 10% of euro area GDP over the last ten years, with a temporary increase during the COVID-19 pandemic years and a moderation since the second half of 2022 due to higher interest rates. It also represents a consistent portion of around 10% of M3 (broad money) – a measure encompassing other liquid, euro-denominated assets. The sustained demand for cash, despite the proliferation of digital payment alternatives, suggests its distinct utility and imperfect substitutability. This stable overall demand contrasts with the diminishing share of cash in everyday payments, a phenomenon often termed “the paradox of banknotes” (Zamora-Pérez, 2021).
Chart 1
Euro banknotes in circulation
a) Value of euro banknotes in circulation
(left-hand scale: EUR billions; right-hand scale: percentages)
b) Monthly banknote net issuance from the public and banks
(EUR billions)
Sources: ECB Statistical Data Warehouse (SDW) and ECB staff calculations.
Note: In panel a), to ensure consistent comparison across periods of euro area enlargement, the ratios to GDP and M3 are presented on a “changing composition” basis, incorporating new member countries from their respective dates of entry.
The sustained demand for banknotes has been amplified by sharp increases in public demand during major crises, which highlights the unique role and attributes of physical currency.[1] As illustrated in Chart 1, panel b, monthly net issuance data can be decomposed into public demand, domestic bank demand (“vault cash”), and foreign bank demand (“net shipments”). The public’s additional demand for cash is usually moderately positive. However, the onset of sudden crises – such as the 2008 financial turmoil, the 2014-15 sovereign debt crisis in Greece, the outbreak of the COVID-19 pandemic or Russia’s unjustified full-scale invasion of Ukraine in 2022 – triggered immediate and extreme surges in cash acquisition by the public. Foreign bank demand, reflecting net shipments from wholesaler banks to clients outside the euro area, also spiked during globally significant events like the 2008 crisis and showed a response to the war in Ukraine.[2] In contrast, domestic bank demand for “vault cash”, which represents a smaller component of total circulation (4%-6%), shows less sensitivity to crisis events, with the notable exception of a spike when the COVID-19 pandemic intensified in Europe in March 2020. Conversely, both domestic and foreign banks show strong responsiveness to monetary policy, as seen by the sharp negative net flows during and after July 2022 when an ECB interest rate hike significantly increased the opportunity cost of holding cash.[3] While public demand also saw a temporary dip during this period, net flows were never negative and it has since resumed its trend, highlighting its less interest-rate-sensitive drivers. A disaggregated view, presented in Chart 2, further illustrates how the scope of these surges varies with the nature of the crisis. Some crises trigger widespread demand increases across most euro area countries, while others have a more concentrated impact that is primarily regional or national.
Chart 2
Statistical outliers in aggregate banknote circulation and national net issuance
(x-axis: year; y-axis: country, grouped by regional cluster)
Source: ECB staff calculations using the Currency Information System 2 database (2005-24).
Notes: Grey shaded areas denote periods of major crisis events in the affected countries. Outliers for the net issuance of banknotes are computed using the monthly data series of euro area national central banks. The chart displays only positive outliers (vertical blue lines), which correspond to increases in net issuance. For each series, an ARIMA model is first identified and estimated; regression dummies for additive outliers (AO), permanent level shifts (LS) and transitory changes (TC) were introduced and their t‐statistics computed. Dummies exceeding a critical threshold (sample size-based) are selected using a stepwise forward selection/backward elimination routine (with LS as step functions, TC as exponentially decaying pulses with a preset decay rate and AO as isolated shocks). The model is re-estimated and the procedure repeats until no further significant outliers remain, optionally within a user-defined date range. The analysis excludes Croatia, as its accession to the euro area on 1 January 2023 means a sufficiently long and comparable historical data series was not available. Malta is also excluded because its net issuance data exhibit significant intermittency, which poses challenges for the reliable convergence and interpretation of the outlier detection algorithm.
Building on the evidence above, this article examines the role of euro cash during major crises – highlighting the unique value of cash as a safe haven asset and essential contingency payment instrument for emergencies. The normal, day-to-day use of cash in transactions is only part of the story behind cash demand: crises, when they occur, are a major factor shaping cash issuance. The heightened demand for cash during stress periods has already been documented globally (e.g. Bartzsch et al., 2024), however this study provides a focused analysis of selected episodes affecting the euro area. Daily data are examined to more precisely attribute demand surges to specific events. By quantifying these effects and exploring potential causal links, this study contributes to a deeper understanding of the enduring function of cash as a contingency tool. These insights contribute to the operational effectiveness of the Eurosystem, the accuracy of banknote forecasting, and crisis preparedness strategies. The results suggest that the unique attributes of cash – the fact that it is tangible, resilient, offline and widely accepted – become paramount during crises, and can also be leveraged for crisis preparedness. Accordingly, several European and national authorities have issued recommendations to the general public to keep some cash reserves in case of sudden and unexpected contingencies.[4]
2 The role of cash during crises
This section analyses cash demand over time across distinct classes of shocks (public health, military, financial and infrastructure) and varying geographical scope (euro area-wide, regional and national). It analyses four major crisis episodes: the COVID-19 pandemic, Russia’s invasion of Ukraine, the April 2025 Iberian blackout and the sovereign debt crisis in Greece.[5] This selection of diverse scenarios – with different triggers, geographical scope and developments over time – allows us to test the robustness of the role of cash as a critical contingency instrument, moving beyond observations tied to a single type of disruption. While not all types of crisis consistently elicit a strong cash demand response (evidence for major shifts due to trade tensions alone, for instance, is limited), the selected cases provide a stress test of the function of cash when the economy, critical infrastructure or public confidence are significantly challenged.
Our approach to analysing the impact of these crises combines descriptive insights with causal analysis. To describe the crises, we use monthly net banknote issuance data from central banks and credit institutions. These data provide broader trend analysis – which is contextualised using other indicators like uncertainty, sovereign stress indices or survey microdata (ECB, 2024) – to understand behavioural drivers. For causal attribution of the increases in cash demand to the events, we employ a causal inference methodology developed by Brodersen et al. (2015).[6] By narrowing the “event windows” around the onset of crises we can more confidently attribute changes in cash demand to the specific shocks. For geographically localised crises, we typically use issuance in unaffected areas as “synthetic controls” to make counterfactual scenarios more robust. This capacity to infer causality from daily observations, distinguishing crisis effects from pre-existing trends or confounding factors, represents an advance compared with traditional analyses. These are often hampered by the use of lower frequency data, e.g. monthly or quarterly data, on banknote demand.[7]
2.1 The COVID-19 pandemic
The onset of the pandemic in early 2020 triggered an extraordinary and sustained increase in the demand for euro banknotes, illustrating the critical role of cash during prolonged uncertainty. Chart 3, panel a, shows that, by the end of 2020, cumulative net banknote issuance in the euro area had surged by over €140 billion. This represented an increase of more than €85 billion (over 130%) compared with the average annual increase of approximately €55 billion seen in the pre-pandemic years (2015-19). Even in early 2021, the “excess” circulation (actual annual issuance minus the pre-pandemic average annual issuance) remained substantial, totalling around €55 billion by the end of the year. This prolonged increase in banknotes in circulation occurred despite a concurrent, well-documented decline in its use for everyday transactions – driven by health concerns, lockdowns and the accelerated shift to online and contactless payments (Tamele et al., 2021).
Chart 3
Long-term effects of the pandemic on banknote circulation
a) Value of banknotes in circulation (stocks)
(EUR billions)
b) Central banks’ outflows and inflows (flows)
(percentages: 100 = average of banknote flows in previous years)
Source: ECB staff calculations using data from the Currency Information System 2 dataset.
In other words, the pandemic starkly accentuated the so-called paradox of banknotes, due to a sharp rise in cash holdings coupled with weakened banknote flows. Evidence from central bank flows (Chart 3, panel b) shows that banknote outflows from central banks (i.e. withdrawals by commercial banks) in 2020 were initially high in March, but then fluctuated below pre-pandemic averages for much of the year. However, banknote inflows to central banks (i.e. deposits by commercial banks) fell even more strongly. This significant reduction in the return flow of cash, reflecting reduced retail turnover and a public inclination to hold onto banknotes, was the primary driver of the net increase in circulation (Tamele et al., 2021).
Some pandemic-related factors had divergent effects on the different functions of cash, simultaneously decreasing its transactional use while increasing the likelihood of people holding cash reserves at home. For instance, an econometric analysis of survey data shows that both reduced access to cash (e.g. due to temporary closures of bank branches or lockdowns) and fear of contagion were linked to lower transactional use, yet they also statistically increased the propensity to hold cash reserves.[8] In contrast, other factors had convergent effects, influencing both functions in the same direction. A perceived increase in the convenience of cashless alternatives, for example, not only drove down transactional cash demand but was also associated with a lower likelihood of keeping cash holdings at home.
While the pandemic saw an estimated €140 billion increase in cash holdings over two years, reflecting a sustained shift towards its store-of-value function, the initial outbreak precipitated a distinct, acute surge in cash acquisition for immediate liquidity needs. To isolate and quantify this immediate shock, we focus on the period following the first widespread euro area lockdowns. While the first major European lockdown was implemented around 9 March 2020, we take 24 January 2020 as the start of the intervention so as to capture potential anticipatory effects.[9] Given the rapid spread of the pandemic throughout Europe after that date, the use of any clearly unaffected euro area countries as external controls is not possible. Therefore, to estimate what issuance would have been had the pandemic not occurred, we constructed a synthetic control for euro banknote issuance in 2019-20 by leveraging its historical stability.[10] Chart 4 reveals a substantial and statistically significant increase: during the 90-day period following 24 January, average daily net issuance, for all Eurosystem central banks together, hit approximately €616 million. This is nearly double the counterfactual prediction of €320 million, implying a daily causal effect of around €260 million – a 94% relative increase. Cumulatively, this initial 90-day surge added a conservative estimate of approximately €19.5 billion to the counterfactual level of currency in circulation, €10 billion of which can be causally attributed to the month after the first lockdowns were declared on 9 March 2020. This points to people immediately turning to cash for liquidity during high uncertainty due to an unprecedented event – a response distinct from the subsequent longer-term precautionary hoarding driven by ongoing anxieties about contagion and income stability.
Chart 4
The short-term impact of the pandemic on the daily net cash issuance of euro area central banks
a) Daily net banknote issuance | b) Cumulative effect |
|---|---|
(EUR millions, months) | (EUR millions, weeks after event) |
 |  |
Source: ECB staff calculations using the Currency Information System 2 database.
Notes: Panel a) shows the observed daily net banknote issuance (solid blue line), aggregated for all euro area countries, which increases sharply several weeks after the intervention date (yellow vertical line, marking the start of the pandemic). This is a significant upward divergence from the model’s counterfactual prediction (dotted blue line with shaded 95% confidence interval), which estimates the expected net issuance had the pandemic not occurred, based on pre-pandemic trends and daily or calendar seasonality. To capture anticipatory effects, the start of the treatment is taken as 24 January 2020, the date on which the first three confirmed COVID-19 cases in Europe were announced. Panel b) displays the cumulative effect over time since the intervention, illustrating a sharply growing total excess currency issuance attributable to the pandemic especially as of 9 March 2020 (after the sixth week), when the lockdowns started in Europe. It slows slightly after the eighth week and plateaus after three months. The model was trained on data from the one-year period before the intervention, using cash flow series from previous years as controls on a matching-day basis. The post-intervention period covers three months following the onset of the pandemic in Europe. The strong statistical significance of the causal effect (Bayesian one-sided tail-area probability p = 0.001) indicates a very high posterior probability that the observed increase was indeed greater than zero, and not due to random chance, given the model and the data.
2.2 Russia’s invasion of Ukraine
Russia’s unjustified full-scale invasion of Ukraine in February 2022 triggered a significant surge in cash demand, concentrated in several neighbouring countries. This is a common response to the pervasive uncertainty that armed conflicts and geopolitical tensions introduce regarding institutional stability, state capacity and the resilience of critical infrastructures.[11] Among the specific concerns fuelling this demand were fears of potential cyberattacks by Russia on critical digital infrastructure (Rösl and Seitz, 2023). The intensity of this uncertainty immediately following the Russian invasion was reflected in broader metrics like the Geopolitical Risk Index, which recorded its third-highest value of the 21st century in March 2022 (Caldara and Iacoviello, 2022), amplifying the perceived need for a tangible and reliable store of value like physical cash.
Geographical proximity was the key determinant in boosting the demand for euro cash. Chart 5 illustrates this by plotting the monthly deviation of cash issuance from the historical average for each euro area country against its capital’s distance from Kyiv. In the wake of the invasion, countries bordering either Ukraine or Russia (such as Estonia, Latvia, Lithuania, Slovakia and Finland) exhibited markedly higher demand, with issuance levels reaching six to ten standard deviations above their respective historical norms. A deviation of this magnitude is highly unusual. Even countries where credit institutions are significantly engaged in international currency trade, such as Germany and Austria (depicted in green in Chart 5), also experienced unusual excess demand. They recorded issuance up to five standard deviations above the typical patterns. Conversely, as geographical distance from the conflict increases, issuance levels are closer to their historical patterns. This clear spatial gradient strongly supports a precautionary motive, suggesting that people responded to heightened proximity to potential disruptions by accumulating portable liquidity (Beckmann and Zamora-Pérez, 2023).
Chart 5
Exceptionally high cash demand in proximate euro area countries in early 2022
(x-axis: distance in kilometres from the country’s capital to Kyiv; y-axis: standard deviations from historical average issuance)
Source: ECB staff calculations.
Notes: On the y-axis, the time series data on banknote issuance are seasonally adjusted for each country and are standardised. On the x-axis, the physical distance in kilometres from each country’s capital to Kyiv is measured using straight lines.
In the countries bordering the conflict, during its first month, the war led to an estimated 36% causal increase in average daily net banknote issuance. Immediately after the war began, daily net issuance in the affected countries significantly exceeded counterfactual predictions, reaching a peak of €80 million recorded in one day at the end of February 2022 (Chart 6, panel a).[12] In this period, average daily net issuance in the treatment group reached approximately €38 million, compared with a counterfactual estimate of €28 million in the absence of the war. The cumulative impact on net banknote issuance exceeded €211 million (Chart 6, panel b). Starting from zero at the onset of the invasion, cumulative cash demand rose sharply during the initial weeks. The pace of accumulation then slowed, with the cumulative curve flattening and plateauing around the third week. This sustained surge in demand for physical banknotes is particularly striking given that the Baltic States and northern European countries are typically highly digitalised and rely heavily on cashless payment systems.
Chart 6
The effect of Russia’s war in Ukraine on daily net cash issuance in neighbouring euro area countries
a) Daily net banknote issuance | b) Cumulative effect as from outbreak of war |
|---|---|
(EUR millions, months) | (EUR millions, days after event) |
 |  |
Source: ECB staff calculations using the Currency Information System 2 database.
Notes: Panel a) shows the observed daily net banknote issuance (solid blue line), aggregated for Estonia, Latvia, Lithuania, Slovakia and Finland, which increases sharply immediately after the intervention (yellow vertical line, marking the start of the war). This is a significant upward divergence from the model’s counterfactual prediction (dotted blue line with shaded 95% confidence interval), which represents the expected circulation had the war not occurred. Panel b) displays the cumulative effect over time since the intervention, illustrating a steadily growing total excess currency in circulation in the treatment countries, attributable to the war, which plateaus after the third week. The start of the treatment is marked as 24 February 2022, the date of Russia’s full-scale invasion of Ukraine. The model was trained on data from the one-year period prior to the intervention and the post-intervention period covers one month following the onset of the war. The strong statistical significance of the causal effect (Bayesian one-sided tail-area probability p = 0.001) indicates a very high posterior probability that the observed increase was indeed greater than zero, and not due to random chance, given the model and the data.
2.3 The April 2025 Iberian blackout
The critical role of physical cash when digital infrastructures fail was demonstrated during the recent Iberian blackout on 28 April 2025. Shortly after noon Central European Time, the Iberian power grid lost synchronism and separated from the main European network, causing a near-total blackout across the peninsula affecting over 50 million people (ENTSO-E, 2025). While power was restored to half the peninsula by late evening, some areas were only re-energised approximately 22 hours after the blackout started, with widespread consequences for transport and digital infrastructure.
With power and telecommunications down, digital payment systems across the peninsula failed. Physical card spending in affected areas plummeted by an estimated 41-42% compared with unaffected regions or normal levels, while national e-commerce spending dropped by around 54%, contributing to an estimated 34% fall in overall Spanish consumption on that day (CaixaBank Research, 2025; BBVA Research, 2025)[13]. Many point-of-sale terminals, automated teller machines (ATMs) and mobile wallets – including card networks and person-to-person (P2P) services like Bizum – were widely inoperable for hours and only fully restored the following morning. Estimates of direct GDP losses range from €400 million to €1,600 million (CaixaBank Research, 2025; Reuters, 2025). This event transformed cash from one payment option among many into the only means of payment for many of those who held it or could access it, as existing banknotes remained perfectly functional even when digital systems and many ATMs were inoperable.
ATM withdrawals – even though locally affected by the blackout – serve as the best indicator of cash demand, while there were no significant spikes in wholesale (central bank) flows. Examining daily ATM withdrawal patterns from a national sample of approximately 4,500 Spanish ATMs provided by BBVA Research, we observe a dramatic divergence in cash demand. Chart 7 shows an index where 100 represents normal daily demand, with the grey shaded area indicating the typical interquartile range. In the days preceding the blackout (D-7 to D-1) cash withdrawals in both the areas subsequently affected (in mainland Spain) and elsewhere (extra-peninsular areas) fluctuated normally. On the day of the blackout (Day D), ATM withdrawals in the affected zones (blue line) plummeted, reflecting constrained access as ATMs went offline. Conversely, in unaffected zones (yellow line), ATM withdrawals surged significantly above normal levels, indicating strong precautionary demand as citizens sought the security of physical cash during uncertainty.[14] In the immediate aftermath (from D+1 onwards), once power and ATM services had been restored in the affected areas, there was a sharp increase in withdrawals, far exceeding typical levels.[15]
Chart 7
Daily cash withdrawals from ATMs during and after the blackout in Spain
(y-axis: expected cash demand index; 100 = expected demand)
Sources: BBVA Research and ECB staff calculations.
Notes: The chart displays an index of daily ATM withdrawal values (100 = expected normal volume for that day of the week) from over 4,500 Spanish ATMs. The blue line represents areas affected by the blackout (mainland Iberia) and the yellow line unaffected areas (the Balearic and Canary Islands, Ceuta and Melilla). The grey area is the baseline interquartile range (IQR) of typical daily fluctuation. To isolate the impact of the blackout, the “normal” demand baseline is conservatively constructed: it includes comparable historical weekdays, incorporating typical pre-public holiday demand patterns such as those observed around 1 May 2024. This ensures that any surge in 2025 is benchmarked against historically high demand in the run-up to a public holiday, providing a conservative estimate of the distinct effect of the blackout.
This episode illustrates the dual function of cash – as a resilient, offline payment method and also a tangible store of value – during an acute infrastructural failure, as confirmed by causal analysis. ATM withdrawals plummeted in the blackout areas in mainland Spain owing to operational constraints. However, people were likely relying on cash from their personal holdings. According to ECB survey data, 39% of Spaniards kept cash reserves at home as a precaution (ECB, 2024). The spike on the day after the blackout reflected a combination of a “restocking” effect, as individuals sought to replenish their cash holdings after using them, and a possible increase in precautionary reserves. This interpretation is supported by a causal impact analysis, which finds a statistically significant net positive effect on cumulated cash demand, even after accounting for the prior day’s suppressed withdrawals.[16] By contrast, the significant surge in ATM withdrawals in unaffected areas points to uncertainty in mainland Spain spurring precautionary cash withdrawals in extra-peninsular Spain. This happened despite digital systems in these areas remaining functional and card spending declining less than in affected areas (CaixaBank Research, 2025; BBVA Research, 2025).
2.4 Greece in the sovereign debt crisis
During the sovereign debt crisis, heightened market uncertainty driven by political developments and evolving prospects for the macroeconomic adjustment programme led to a sharp increase in banknote demand in Greece. Chart 8 juxtaposes monthly net banknote issuance in Greece with the Composite Indicator of Systemic Sovereign Stress (SovCISS) for Greece. Reflecting the heightened uncertainty, monthly net issuance of banknotes by the Bank of Greece soared to a historic peak of nearly €5 billion in June 2015. Following several events at the end of June 2015, including the decision by the Greek authorities to hold a referendum and the non-prolongation of the macroeconomic adjustment programme, the Greek Government declared a bank holiday and introduced strict capital controls, including daily ATM withdrawal limits. The intensity of this period is mirrored in the SovCISS, which aggregates metrics such as yield spreads, volatility and bid-ask spreads (Garcia-de-Andoain and Kremer, 2018). By July 2015 this indicator had reached the exceptional level of 0.82 on a scale from 0 to 1, closely tracking the spikes in cash issuance in Chart 8. The strong co-movement between the SovCISS and net banknote issuance indicates that the public’s heightened demand for physical currency was closely correlated with periods of elevated sovereign and financial market stress.
Chart 8
Central bank monthly net cash issuance and sovereign stress in Greece
(left-hand scale: EUR billions; right-hand scale: index scale)
Sources: ECB SDW and ECB staff calculations.
Notes: The blue area represents the monthly net banknote issuance in Greece, with reference values shown on the left-hand scale. The yellow line shows the monthly values of the SovCISS for Greece, with reference values shown on the right-hand scale.
From late 2014 to mid-2015, daily net banknote issuance in Greece was well above the level expected in the absence of the crisis, reflecting elevated public perceptions of risk. Using a causal inference approach with synthetic controls (the Netherlands, Austria and Finland),[17] our analysis shows that the escalation of the crisis had a pronounced and measurable effect on cash demand in Greece. Daily net issuance repeatedly spiked well above the model’s counterfactual estimates (Chart 9, panel a). A particularly sharp peak occurred on 18 June 2015, coinciding with a Eurogroup meeting that did not result in an agreement to release additional funds. On that day alone, net issuance exceeded €300 million. Analysis confirms that all individual spikes are causally attributable to the specific events labelled in Chart 8. For illustrative purposes, we show that over the entire post-intervention period of seven months the average daily net issuance in Greece was approximately €72 million above the expected level.[18] The cumulative effect rose steadily from the moment the crisis started to escalate, reaching an estimated total of €11.2 billion six months later (Chart 9, panel b).
Chart 9
The effect of the sovereign debt crisis on daily net cash issuance in Greece
a) Daily net cash issuance | b) Cumulative effect as from debt crisis escalation |
|---|---|
(EUR millions, months) | (EUR millions, months after event) |
 |  |
Source: ECB staff calculations using the Currency Information System 2 database.
Notes: Panel a) shows the observed daily net banknote issuance in Greece (solid blue line) experiencing several peaks after the intervention (yellow vertical line, marking the moment when the sovereign debt crisis started to escalate). This is a significant divergence from the model's counterfactual prediction (dotted blue line with shaded 95% interval), which represents the expected issuance had the crisis not escalated. Panel b) displays the cumulative effect over time since the intervention, showing a steadily growing total excess net currency issuance in Greece that is attributable to the crisis. The start of the treatment is marked as 17 December 2014, when the Athens Stock Exchange plunged by roughly 20% and ten-year Greek government bond yields spiked above 9% following an inconclusive first-round presidential vote in Greece. The model was trained on data from the one-year period prior to the intervention and the post-intervention period covers the seven months following the escalation of the crisis. The strong statistical significance of the causal effect (Bayesian one-sided tail-area probability p = 0.001) indicates a very high posterior probability that the observed increase was indeed greater than zero, and not due to random chance, given the model and the data.
3 Conclusion and implications for public policy
These diverse crisis episodes illustrate that the utility of cash intensifies markedly when stability is threatened – irrespective of the specific nature or geographical scope of the underlying shock, or the degree of digitalisation. Each case study, however, illuminates a distinct dimension of this resilience across different failure points. The pandemic revealed sustained precautionary cash hoarding driven by prolonged uncertainty during a public health emergency. Russia’s unjustified full-scale invasion of Ukraine highlighted rapid, localised demand surges near conflict zones and irrespective of countries’ degree of digitalisation. The Iberian blackout highlighted cash as an indispensable payment method when digital infrastructures fail and also as an important instrument for public reassurance, extending its influence even to areas not directly affected by the initial shock. Finally, Greece’s sovereign debt crisis saw recurrent demand spikes during protracted financial turmoil and political tensions. These cases collectively reveal a consistent pattern: in moments of acute stress, the public often turns to physical currency as a reliable store of value and a resilient means of payment, underscoring the crucial role it plays above and beyond everyday transactional convenience.
This crisis-driven demand for cash stems from its fundamental attributes: it offers distinct psychological and practical utility, explaining its well-documented global staying power. Safe asset theory partly explains flights to government-backed liabilities during uncertainty (Gorton and Ordoñez, 2022). However, the appeal of cash is amplified by its tactile nature, providing comfort and control, and its offline functionality becoming paramount during digital system failures (Bartzsch et al., 2024). Cash offers certainty concerning its nominal value, immediate access and privacy. Heightened loss aversion during crisis, coupled with varied individual perceptions of stability, make cash a tool for satisfying risk-averse individuals’ demand for direct liquidity insurance (Muñoz and Soons, 2022).[19] This crisis-specific utility contributes to the persistent demand for cash that cannot be fully explained by traditional economic factors like interest rates or income (Jobst and Stix, 2017; Goodhart and Ashworth, 2020).
Beyond these individual drivers, the resilience of cash suggests it has broader system-wide advantages that are difficult to quantify. From a systems theory perspective, while digital payment rails are optimised for efficiency (maximising “mean time between failure”), cash provides essential redundancy – a “spare tire” – for the payment system.[20] This redundancy is vital for any system, as no system is infallible. Relatedly, widespread individual cash holdings generate a societal benefit or “positive externality”: a distributed liquidity network for the euro area community when centralised systems fail – a feature digital-only regimes cannot replicate. This makes cash a kind of societal insurance, a low-cost safeguard against major systemic instability. Finally, cash can act as a crucial counterweight to concentrated power within payment systems, fostering market competition (Lagos and Zhang, 2022), and empowering users by providing the option to make unrecorded transactions. This suggests that its latent social benefits may be larger than traditionally estimated (Rösl and Seitz, 2022 and 2024).
These findings and reflections support the growing recognition among authorities that cash is a critical component of national crisis preparedness. Central banks, finance ministries and civil protection agencies in several countries now recommend that households maintain a multi-day cash float for essential purchases. For instance, authorities in the Netherlands, Austria and Finland suggest holding amounts ranging from approximately €70 to €100 per household member or enough to cover essential needs for about 72 hours.[21],[22] Some jurisdictions, like Finland, are even exploring “disruption-proof” ATMs to ensure access during digital failures.[23] This aligns with the understanding that physical currency not only serves to meet individual needs but also contributes to broader systemic resilience.[24]
Ultimately, the evidence underscores the continued importance for central banks and the private sector to ensure an efficient and robust cash supply, encompassing adequate stocks and resilient business continuity plans. Understanding the often heavy-tailed nature of cash demand spikes – where extreme, infrequent events drive disproportionate demand – has profound implications. It means that while day-to-day operational forecasting can rely on more normally distributed demand, the infrastructure and strategic reserves must be prepared for these less predictable, high-impact surges.[25] This ensures that cash, as the only central bank liability directly available to all, can fulfil its role – not just in daily transactions but as a fundamental pillar of economic stability and public confidence, particularly when it is needed most. These imperatives are recognised in the Eurosystem cash strategy, which aims to ensure continued availability, access to and acceptance of cash in the euro area.
References
Armelius, H., Claussen, C.A. and Reslow, A. (2022), “Withering cash: Is Sweden ahead of the curve or just special?”, International Journal of Central Banking, Vol. 18, No 5, pp. 199-238.
Bartzsch, N., Rösl, G. and Seitz, F. (2024), “Cash demand in times of crises: A global perspective”, Bank i Kredyt, Vol. 54, No 1, pp. 1-43.
BBVA Research (2025), The blackout in real time: collapse and resilience of consumption after 28A, 9 May.
Beckmann, E. and Zamora-Pérez, A. (2023), “The impact of war: Extreme demand for euro cash in the wake of Russia’s invasion of Ukraine”, The international role of the euro, ECB, pp. 37-40.
Brodersen, K.H., Gallusser, F., Koehler, J., Remy, N. and Scott, S.L. (2015), “Inferring causal impact using Bayesian structural time-series models”, Annals of Applied Statistics, Vol. 9, No 1, pp. 247-274.
Bundesbank (2022), “Demand for euro banknotes issued by the Bundesbank: current developments”, Monthly Report, April, pp. 67-83.
CaixaBank Research (2025), “Valoración del impacto económico del apagón del día 28 de abril,” Nota breve, 5 May.
Caldara, D. and Iacoviello, M. (2022), “Measuring geopolitical risk”, American Economic Review, Vol. 112, No 4, pp. 1194-1225.
Clipal, R. and Zamora-Pérez A. (2025), “Cash is alive… and somewhat young? Decoupling age, period and cohort from cash use”, Economic Bulletin, Issue 5, ECB, Frankfurt am Main.
De Nederlandsche Bank (2025), NFPS advice: Prepare for a three-day disruption of electronic payments, 20 May.
ECB (2024), Study on the payment attitudes of consumers in the euro area (SPACE 2024), Frankfurt am Main.
European Network of Transmission System Operators for Electricity (ENTSO-E) (2025), Iberian Peninsula Blackout, 28 April.
Garcia-de-Andoain, C. and Kremer, M. (2018), “Beyond spreads: Measuring sovereign market stress in the euro area”, Working Paper Series, No 2185, ECB, Frankfurt am Main, October.
Lagos, R. and Zhang, S. (2022), “The limits of monetary economics: On money as a constraint on market power”, Econometrica, Vol. 90, No 3, pp. 1177-1204.
Lalouette, L., Zamora-Pérez, A., Rusu, C., Bartzsch, N., Politronacci, E., Delmas, M., Rua, A., Brandi, M. and Naski, M. (2021), “Foreign demand for euro banknotes”, Occasional Paper Series, No 253, ECB, Frankfurt am Main, January.
Lerner, J.S., Li, Y., Valdesolo, P. and Kassam, K.S. (2015), “Emotion and decision making”, Annual Review of Psychology, Vol. 66, pp. 799-823.
Ministry of Finance Finland (2022), Securing daily payments.
Oesterreichische Nationalbank (2024), Cash management.
Reuters (2025a), EU tells the public to hold 72 hours of emergency supplies, 26 March.
Reuters. (2025b), Power begins to return after huge outage hits Spain and Portugal, 29 April.
Reuters (2025c), Post-blackout in Spain and Portugal, companies count the cost, 29 April.
Rösl, G. and Seitz, F. (2022), “Cash demand in times of crisis”, Journal of Payments Strategy & Systems, Vol. 16, No 2), pp, 107-119.
Rösl, G. and Seitz, F. (2024), “Uncertainty, politics, and crises: The case for cash”, Latin American Journal of Central Banking, Vol. 5, No 3, 100128.
Tamele, B., Zamora-Pérez, A., Litardi, C., Howes, J., Steinmann, E. and Todt, D. (2021), “Catch me (if you can): assessing the risk of SARS-CoV-2 transmission via euro cash”, Occasional Paper Series, No 259, ECB, Frankfurt am Main, July.
Zamora-Pérez, A. (2021), “The paradox of banknotes: Understanding the demand for cash beyond transactional use”, Economic Bulletin, Issue 2, ECB.
Crises are defined as unstable and critical situations that pose a significant threat to individuals, organisations, or societies, often requiring immediate decision-making under conditions of uncertainty.
Net shipments represent only the formal wholesale channel and do not capture informal flows like tourism and remittances, which constitute a significant portion of foreign demand. Overall, foreign holdings of euro banknotes are estimated to account for up to half of the total value in circulation (Lalouette et al., 2022).
While not visible in this chart of monthly flows, the stock of vault cash held by banks had progressively increased from around 2016, when short-term interest rates were at their lowest, reducing the holding cost. The sharp reduction seen in July 2022 largely represented a return to previous average levels.
Examples, such as announcements by the European Commission and the Austrian, Finnish and Dutch national authorities, are discussed below.
The 2008 global financial crisis is excluded from our in-depth analysis. While it was the largest shock to euro banknote circulation, its effects are well documented, and its global nature complicates causal attribution. We therefore focus our analysis on the COVID-19 pandemic as a recent widespread shock, and the Greek debt crisis to specifically illustrate a context of national financial and political turmoil where causal inference is more robust.
The causal impact analysis employs a Bayesian structural time-series model suggested by Brodersen et al. (2015) to estimate a counterfactual scenario – i.e. the level of central bank issuance or ATM withdrawals had the event not occurred, given pre-event trends and, where available, the behaviour of unaffected control regions.
The previous literature has faced the challenge of isolating crisis-driven demand spikes from underlying seasonality in high-frequency cash data, as cash flows and ATM withdrawals exhibit complex patterns, including daily effects, monthly variations and distinct calendar effects around public holidays. To address this, we use infra-monthly seasonal adjustment techniques (Webel and Smyk, 2023).
Where fear of the virus has opposite effects on the two functions of cash demand, this may be explained by the unobserved (confounding) factor of general risk profiles. Some individuals with higher health-risk aversion decreased their physical payments due to fear of contracting the virus and were simultaneously also more inclined to secure cash as a tangible store of value for emergencies, as individuals might proactively build up home cash reserves.
The earlier date of 24 January 2020 coincides with the first documented COVID-19 case in Europe (and outside Asia). This event was confirmed by national health authorities and received significant media coverage, likely amplifying public risk perception and precautionary behaviour prior to official government measures. However, given the visible, gradual anticipatory effects before the widespread lockdowns (as of 9 March 2020), selecting a date is to some extent arbitrary.
Our approach uses cash flow data from previous years (2015-16, 2016-17, 2017-18 and 2018-19) as explanatory variables. These historical periods were aligned with the period under review, 2019-20, on a matching-day basis, considering factors like the day of the week and its specific occurrence within the year. This method is appropriate because euro banknote issuance has historically been very stable, as evidenced by the relatively narrow maximum-minimum range from 2015 to 2019 (illustrated by the grey area in Chart 3, panel a). While these are not contemporaneous controls, this method improves our estimates in the absence of better alternatives, providing a conservative interpretation (i.e. with wide confidence intervals) of the observed impact.
Given the timing of the observed effects, this phenomenon was likely primarily driven by not only euro cash demand from refugees coming from Ukraine but also a precautionary motive.
The intervention date is 24 February 2022, when Russian military forces entered Ukraine from Belarus, Russia and Crimea. The analysis is based on aggregated daily net banknote issuance for Estonia, Latvia, Lithuania, Slovakia and Finland, which comprise the treatment group, while more distant countries – Spain, France, Italy and Portugal – serve as the control series. The post-intervention period is defined as the first month after the onset of the war.
A more detailed analysis of the Spanish case is provided here, as the data allow for a comparison between affected and unaffected regions. This distinction is not possible with the Portuguese data.
While this surge is probably indicative of a widespread precautionary response to the crisis, it was likely amplified by individuals in these unaffected areas withdrawing cash in advance of planned travel to the blacked-out peninsula.
Complementary analysis from Banco de España and Banco de Portugal confirms a slight post-blackout increase in central bank outflows, driven mostly by ATM denominations (€20, €50) but also in some cases higher demand for large denominations (€100). National weekly ATM data for both countries also support demand in excess of simple restocking, pointing to additional precautionary holdings.
Following Brodersen et al. (2015) and controlling for daily seasonality and calendar effects, we find a statistically significant net positive effect on cumulated cash demand in mainland Spain, even after accounting for the prior day’s suppressed withdrawals. The observed ATM withdrawals in the days after the blackout (average index 379.16) far outstrip what would have been expected had behaviour simply returned to normal after the constraint. (The counterfactual average index was 268.57 and a strong statistical significance of the causal effect was found – a Bayesian one-sided tail-area probability p = 0.009 – indicating a very high posterior probability that the observed increase was indeed greater than zero, and not due to random chance.) Similar models applied to the extra-peninsular demand also confirm a causal increase attributable to the blackout, most likely due to fears the blackout would spread to Spain’s islands as well as Ceuta and Melilla.
Germany, despite its distance from Greece and fiscal situation at the time, is not a suitable control country as the restrictions on withdrawals from ATMs in Greece impacted German-issued banknotes in circulation (Deutsche Bundesbank, 2022).
Daily net issuance stood at €57 million, in stark contrast to the -€15 million predicted by the model had the crisis not intensified. This negative predicted issuance is consistent with typical cash flow seasonality, where certain times of the year historically exhibit net inflows of banknotes back to the central bank. In Greece, this is further enhanced by tourism-driven banknote inflows.
Under conditions of stress and uncertainty, individuals often exhibit heightened loss aversion, meaning the psychological impact of a potential loss becomes disproportionately larger than that of an equivalent gain.
The analogy draws on systems safety engineering, where critical systems incorporate simpler, often manual, backup mechanisms to ensure functionality if primary automated systems fail. A well-known example is the emergency staircase in a skyscraper; while elevators are more efficient for everyday use, the staircase provides an essential, redundant path for egress during a power failure. While the analogy is not perfect – as cash, unlike an emergency staircase, is widely used for daily transactions –the underlying principle holds.
See, for example, Ministry of Finance Finland (2022), Oesterreichische Nationalbank (2024) and De Nederlandsche Bank (2025).
Relatedly, the European Commission’s 72-hour emergency preparedness guidance included cash alongside essentials like water and medicine (Reuters, 2025b). This announcement made by video on the European Commission’s official YouTube Channel was widely covered in the media. However, there is no outline of population preparedness guidance as yet. This will be part of the “EU Preparedness Union Strategy to prevent and react to emerging threats and crises”.
This is available at Suomen Pankki’s website under “'Home emergency kit' for payments”.
Yet, this enduring utility of cash, particularly its store-of-value function and crisis demand, often seems at odds with policies that, intentionally or not, increase friction in its use or aim to reduce its circulation. Measures such as stringent payment limits or the removal of large-denomination banknotes – as seen, for example, in the motivations behind India’s demonetisation – may not adequately account for the positive externalities of a readily available cash stock or the public’s legitimate need for convenient, high-value physical storage, particularly during periods of uncertainty. For this reason, Regulation (EU) 2024/1624 of the European Parliament and of the Council of 31 May 2024 on the prevention of the use of the financial system for the purposes of money laundering or terrorist financing (the new Anti-Money Laundering Regulation) provides that in the event of contingencies where electronic payments are not available, cash payment limits can temporarily be suspended.
This is analogous to the design of Dutch dykes, which are engineered not for average tides but for rare, catastrophic floods, using principles from extreme value theory to model such high-impact surges. A resilient cash supply must similarly be sized for its critical role during infrequent crises, not just for daily transactional flows.