## Measurement and Use of Cash by Half the World’s Population (Chapter summary)

## Source details

**Canonical URL:** [Measurement and Use of Cash by Half the World’s Population (Chapter summary)](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023062-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023062-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023062-print-pdf.pdf.json)

---

### Background and purpose
- Payments instruments have shifted from cash, check, and paper giro-type payments toward electronic instruments including automated clearing house and electronic giro-type transactions, payment cards, and instant (or fast) payments.
- Improved measurement of cash use for payments is needed to assess payment system performance and cost efficiency.
- Paper objectives:
  - Note emerging regulatory developments associated with access to cash and acceptance of cash at the point of sale.
  - Contrast two measures of cash use—Currency in Circulation (CIC) and cash withdrawn from ATMs—across 14 countries.
  - Transform local currency payment values into purchasing power parity (PPP) US dollars to compare and aggregate cash use.
  - Present conclusions and extended country comparisons.

### Two measures of cash use: definitions and conceptual differences
- Currency in Circulation (CIC) and CIC/GDP:
  - Standard stock measure used to monitor cash demand/use.
  - Includes cash used for payments, hoarding, illegal uses, and cash held abroad.
  - Small CIC isolates coins and lower denominations to focus on purchases but remains a stock requiring velocity estimates.
- ATM cash withdrawals (flow measure):
  - Focuses on cash used for payments and excludes much hoarding/illegal use.
  - Conceptually similar to a ratio of M' x V' (stock of payment cash times its velocity).
  - When expressed as a share, ATM cash share = ATM cash as share of domestic payments (excludes large value wire or credit transfers).
- Conceptual advantages of ATM withdrawals:
  - Flow measure captures velocity changes due to substitution toward cards and other instruments.
  - Less distorted by cross-border holdings (example: estimated 30% of Euros and 55% of dollars held abroad).
- Drawbacks and data limitations of ATM measure:
  - Limited data on cash-back at the point of sale (POS) and over-the-counter (OTC) consumer withdrawals.
  - OTC data often commingle consumer withdrawals with retailer withdrawals to make change.
  - European Central Bank OTC data for six countries imply adding OTC consumer withdrawals would raise ATM cash shares by two percentage points for four countries and by six to seven percentage points for two countries.

### Regulatory context and examples
- CIC supply determined by central bank, treasury, and printers; CIC demand splits into ATM withdrawals (payments), hoarding, and circulated cash used for unknown/illegal transactions or used abroad.
- Examples of regulatory actions to preserve access/acceptance of cash:
  - Netherlands: minimum of 3,850 ATMs set by central bank.
  - Norway: amendments clarifying bank obligations to enable deposit/withdrawal of cash.
  - Sweden: rules obliging the six largest banks to provide certain cash services and ensure “adequate” ATM coverage.
  - Switzerland: rules ensuring adequate geographic ATM coverage.
  - United Kingdom: Financial Services Act 2021 enables cash-back at POS without a purchase; regulators given powers to ensure cash withdrawal/deposit facilities and broad coverage of free-to-use ATMs.
  - United States: several large cities and states have prohibited retailers from refusing cash or charging higher prices to cash-paying customers; similar federal legislation is pending.
- Fixed cost elements in providing/accepting cash imply reductions in provision/acceptance do not always lead to proportional expense reductions.

### Empirical comparisons across 14 countries — levels and shares
- Level measures compared per adult:
  - CIC per adult (CICPP) and ATM cash per adult (CASHPP).
  - China: CASHPP rose 98% nominally from 2005 to peak in 2017, then fell 42% over the next three years to 2020 (relative to 2005). CICPP rose 32% up to 2017 and then continued to rise by an additional 47% (relative to 2005).
  - Euro Area and US: CIC rising while ATM cash relatively stable rather than falling.
  - India and Russia: ATM cash rising faster than CIC at different times.
- Inconsistency typology (levels):
  - Problem 1 (one rising, other falling): Australia, China.
  - Problem 2 (one rising/falling, other relatively stable): Euro Area, Norway, Singapore, UK, US.
  - Problem 3 (one rising/falling faster than the other): India, Indonesia, Malaysia, Mexico, Russia, Sweden, Thailand.
- Share measures compared:
  - CIC/GDP and ATM cash share (ATM cash as share of domestic payments).
  - Trends depend on numerator (CIC or ATM cash) and denominator (GDP or value of payments); faster growth in the denominator can cause relative shares to fall even if numerators rise.
  - China: CASHPP level peak in 2017, but shareCASH peaked in 2011 because other payments rose more rapidly.
- Inconsistency typology (shares):
  - Problem 1: Australia, China.
  - Problem 2: Euro Area, India, Indonesia, Malaysia, Norway, Russia, Singapore, UK, US.
  - Problem 3: Mexico, Sweden (Thailand basically parallel, no problem).

### Cross-country cash-share patterns and exact values (Table 3 highlights)
- Exact ATM cash share values in 2005 and 2020, peak year and peak share, and pp change and pp decline since peak:
  - Russia: 2005 = 59% ; 2020 = 21% ; pp change = -38 ; Peak Year = 2010 (75%) ; pp Decline Since Peak = -54
  - Indonesia: 2005 = 72 ; 2020 = 41 ; pp change = -31 ; Peak Year = 2014 (75) ; pp Decline Since Peak = -34
  - Sweden: 2005 = 24 ; 2020 = 4.6 ; pp change = -19 ; Peak Year = 2006 (24) ; pp Decline Since Peak = -20
  - Norway: 2005 = 13 ; 2020 = 2.5 ; pp change = -11 ; Peak Year = 2006 (13) ; pp Decline Since Peak = -11
  - India: 2005 = 1.6 ; 2020 = 17 ; pp change = 15 ; Peak Year = 2015 (22) ; pp Decline Since Peak = -5
  - Thailand: 2005 = 11 ; 2020 = 8.1 ; pp change = -2.8 ; Peak Year = 2005 (11) ; pp Decline Since Peak = -3
  - UK: 2005 = 5.8 ; 2020 = 2.5 ; pp change = -3.3 ; Peak Year = 2009 (7) ; pp Decline Since Peak = -4
  - China: 2005 = 1.1 ; 2020 = 1.3 ; pp change = 0.2 ; Peak Year = 2011 (4) ; pp Decline Since Peak = -2
  - Malaysia: 2005 = 8.1 ; 2020 = 15 ; pp change = 6.7 ; Peak Year = 2018 (17) ; pp Decline Since Peak = -2
  - Australia: 2005 = 2.3 ; 2020 = 1.9 ; pp change = -0.4 ; Peak Year = 2018 (2) ; pp Decline Since Peak = -1
  - Euro Area: 2005 = 6.4 ; 2020 = 9.3 ; pp change = 2.9 ; Peak Year = 2016 (10) ; pp Decline Since Peak = -1
  - Singapore: 2005 = 6.7 ; 2020 = 5.5 ; pp change = -1.2 ; Peak Year = 2014 (7) ; pp Decline Since Peak = -1
  - US: 2005 = 1.0 ; 2020 = 1.3 ; pp change = 0.3 ; Peak Year = 2011 (1) ; pp Decline Since Peak = -0
  - Mexico: 2005 = 10 ; 2020 = 27 ; pp change = 19 ; Peak Year = 2020 (29) ; pp Decline Since Peak = 0
- Notes: pp = percentage points; all data are rounded; data starts in 2009 for Singapore, 2006 for China.

### PPP-adjusted ATM cash per adult in 2020 (exact values, Table 4)
- Top group (Annual; Weekly; Daily):
  - Singapore: Annual $13,790 ; Weekly $265 ; Daily $38
  - Russia: Annual $11,568 ; Weekly $222 ; Daily $32
  - Malaysia: Annual $10,712 ; Weekly $206 ; Daily $29
  - Thailand: Annual $10,603 ; Weekly $204 ; Daily $29
  - Euro Area: Annual $6,300 ; Weekly $121 ; Daily $17
  - Mexico: Annual $4,805 ; Weekly $92 ; Daily $13
  - Australia: Annual $4,299 ; Weekly $83 ; Daily $12
- Bottom group (Annual; Weekly; Daily):
  - China: Annual $3,859 ; Weekly $74 ; Daily $11
  - US: Annual $3,801 ; Weekly $73 ; Daily $10
  - UK: Annual $3,798 ; Weekly $73 ; Daily $10
  - Indonesia: Annual $3,458 ; Weekly $67 ; Daily $9
  - Sweden: Annual $1,459 ; Weekly $28 ; Daily $4
  - India: Annual $1,421 ; Weekly $27 ; Daily $4
  - Norway: Annual $1,042 ; Weekly $20 ; Daily $3
- Total across sample: Annual $80,915 ; Weekly $3,112 ; Daily $222
- Authors remark: annual values may appear high and daily values low given typical withdrawal frequency; weekly or bi-weekly are more representative.

### Aggregate PPP-based findings for the 14-country sample
- Sample coverage: 14 countries accounting for 53% of the world’s population and two-thirds of world GDP.
- Aggregate series:
  - PPPCash = sum of PPP-adjusted annual value of ATM cash per adult across countries.
  - PPPCIC = sum of PPP-adjusted value of currency in circulation per adult across countries.
- Aggregate pattern:
  - PPPCash rises faster than PPPCIC, reaches a peak toward the end of the period, then starts to fall while PPPCIC is still rising.
  - PPPCASHshare first rises, reaches a peak, then falls toward the end of the period while PPPCIC/GDP share is relatively stable.
- Method: PPP-adjusted US dollar values summed across countries prior to forming ratios.

### Conclusions, strengths/weaknesses, and policy implications
- Empirical conclusion:
  - CIC and ATM cash measures often conflict in magnitude and trend direction because CIC is a stock including hoarding, illegal use, and foreign holdings, while ATM cash is a flow focused on payments and captures velocity.
  - Using the ATM cash measure, most countries reached a peak in cash use toward the middle or end of the 2005–2020 sample and are experiencing a downward trend.
  - Four countries with highest cash shares in 2005 (ranging from 13 to 72%) experienced the largest reductions by 2020 (falling between 11 to 54 pp).
  - Half of the countries have low cash shares of 5.5% or less.
  - Cash reductions, when present, follow an inverse logistic curve and fall at a decreasing rate.
- Indicator strengths and weaknesses:
  - CIC: includes hoarding and illegal use; excludes turnover/velocity.
  - ATM cash: includes velocity and better reflects payments; limited by missing POS cash-back and OTC consumer withdrawal data.
  - Overall balance favors ATM cash as the preferred indicator for cash used in payments.
- Policy implications and recommendations:
  - ATM cash measure may be preferred when policymakers consider central bank digital currency (CBDC) decisions.
  - Where cash use is already very low, establish regulations to ensure access to cash and to assure cash can continue to be used for transactions with merchants.

*Source: IMF Working Paper — Measurement and Use of Cash by Half the World’s Population, Chapters 1–5 and Chapter 3 country comparisons.*

### 1. Introduction ........................................................................................................

### 1. Introduction

### Background and motivation
- Payments systems evolve as instruments replace others; physical cash, check, and paper giro-type payments have been and continue to be replaced by electronic instruments including automated clearing house and electronic giro-type transactions, payment cards, and instant (or fast) payments (Bech and Hancock, 2020; Khiaonarong and Humphrey, 2022).
- All payment instruments except cash appear to be reasonably well measured at the national level.
- An improved measure of cash use for payments would provide policy makers with more accurate and timely information to assess payment system performance in meeting user needs and cost efficiency.
- A declining share of cash suggests user needs or cost efficiency are not fully achieved under the current instrument mix.

### Purpose and structure of the paper
- The paper:
  - Notes emerging regulatory developments associated with access to cash and acceptance of cash at the point of sale (Section 2).
  - Contrasts two measures of cash use—Currency in Circulation (CIC) and cash withdrawn from ATMs—across 14 countries (Section 3).
  - Transforms local currency payment values into purchasing power parity (PPP) US dollars to compare cash use across countries and aggregate results for half the world’s population (Section 4).
  - Presents conclusions in Section 5 and additional country comparisons in Annex I.

### High-level empirical result preview
- ATM cash withdrawn per adult (measure of cash used for payments) rose from 2005, peaked in 2017, and fell thereafter both in absolute terms and as a payment share for the aggregated 14-country sample.
- Standard CIC measures do not reach a peak and do not fall over the same period.

### Key organizational elements of the document
- Sections include: regulatory and measurement issues, cross-country comparisons of CIC and ATM measures, PPP conversions and aggregation, and conclusions; Annexes provide extended country comparisons and data sources.

---

### 2. Regulatory and Measurement Issues Concerning Cash Use

### Access to cash and acceptance for payments (regulatory context)
- CIC supply side is determined by government actors: central bank, treasury (seigniorage revenue), and specialized printers.
- CIC demand splits into three activities:
  - (1) ATM cash withdrawals typically used for payments that are known and legal;
  - (2) cash used for hoarding, which are not payments and have little circulation;
  - (3) circulated cash used for unknown transactions involving corruption, tax avoidance, other illegal domestic transactions, and transactions using domestic currency in another country.
- ATM cash withdrawals are known flows but include some amount associated with activities (2) and (3), introducing bias when using ATM withdrawals to represent payment cash. The authors regard this bias as less than the bias of using all of CIC for payments; hence preference for ATM cash as a measure of cash used for payments.
- Falling demand for cash in some countries has led banks to reduce ATM networks and consolidate branches; outsourcing of cash services and merchant efforts to reduce cash acceptance have also occurred.
- Fixed cost elements in providing or accepting cash imply reductions in provision/acceptance do not always lead to proportional expense reductions.

### Examples of regulatory actions to preserve access/acceptance
- Netherlands: central bank set 3,850 ATMs as the minimum number in the country to ensure adequate access to cash services (De Nederlandsche Bank, 2020); a non-regulatory agreement among banks, retailers, and consumers on cash deposits, withdrawals, and acceptance.
- Norway: amendments clarify banks' obligation to enable deposit/withdrawal of cash directly or via arrangements with other cash service providers; government and central bank plan to assess cash's role in natural disasters or merchant non-cash preferences (Norges Bank, 2022).
- Sweden: rules to ensure “adequate” geographic coverage of ATMs; amendment to the Swedish Act on Payment Services obliges the six largest banks to provide certain cash services; enforcement roles for the Swedish Post and Telecom Agency and Swedish Financial Supervisory Authority (Sveriges Riksbank, 2021).
- Switzerland: rules ensure adequate geographic ATM coverage for cash withdrawals.
- United Kingdom: Financial Services Act 2021 enables cash-back at the POS without a purchase (HM Treasury, 2022); Financial Conduct Authority given authority to ensure cash withdrawal/deposit facilities in all communities; Payment Systems Regulator issued a Special Directive to ensure broad coverage of free-to-use ATMs (Payment Systems Regulator, 2021); Bank of England issued a consultation on wholesale cash distribution oversight as of December 2022 (Bank of England, 2022).
- United States: several large cities and several states have made it unlawful for retailers to refuse cash or to charge cash-paying customers higher prices; similar federal legislation is pending (Tarlin, 2021).
- If cash usage falls further in other countries, similar regulations, rules, or voluntary agreements may spread—creating the need for a good indicator of cash use for payments.

---

### 3. Two Measures of Cash Use for Payments

### Standard measure: Currency in Circulation (CIC) and CIC/GDP
- CIC or CIC/GDP has been the standard way to monitor cash demand/use.
- Literature findings:
  - Ashworth and Goodhart (2020): demand for cash fell from its highest use in the late 1940s, flattened during the 1980s, then rose steadily after the early 2000s for a selected group: the Euro Area, Japan, the UK and US.
  - Bech, et al. (2018): CIC/GDP rose for 20 out of 25 countries covered; ratio fell only in 5 countries: China, Norway, Russia, South Africa, and Sweden.
- Small CIC (ratio of coin and lower denominations to GDP) focuses on denominations most used for purchases and attempts to exclude cash used for hoarding, overseas holdings, or illegal activities. Small CIC remains a stock measure and does not capture velocity directly.

### Alternative measure: Cash withdrawn from ATMs (flow)
- ATM cash withdrawals focus on cash used for payments and exclude much hoarding/illegal use.
- When expressed as a ratio to total value of payments (excluding large value wire or credit transfers), ATM cash share is conceptually similar to CIC/GDP.
- Large-value wire transfers in the Euro Area and US are more than 15 times larger than the value of all other payment instruments; including wire transfers would distort cross-country comparisons of cash use. BIS payment data do not separate small and large credit transfers, so both are excluded from the ATM-cash-based ratio.

### Similar approaches and data limitations
- Amromin and Chakravorti (2009) and subsequent work use small CIC to approximate cash used for lower value purchases; ATM-loaded denominations are typically medium value notes commonly used for legal payment transactions, making ATM and small CIC approaches similar in intent.
- When CIC and small CIC were contrasted for 25 countries, more countries showed falling cash use using small CIC than total CIC (Khiaonarong and Humphrey, 2021).
- Payment diary studies are a newer data source, though limited (nine studies found); they suggest younger adults prefer cards over cash relative to older adults (e.g., Kim, et al., 2020 for the US), implying short-term cash use may decline as younger cohorts age into full consumer activity.

### Conceptual and measurement advantages of ATM withdrawals
- ATM cash is a flow measure and thus already adjusts for changes in velocity due to substitution toward cards and other instruments.
- CIC x V = GDP mixes turnover of all payment forms; CIC/GDP gives an implied velocity V that reflects turnover of other payment forms as well. In contrast, value of cash used for payments ≈ M' x V' ≈ ATM cash, where M' is stock of cash used for payments and V' is its velocity.
- ATM cash approximates M' x V' directly (annual flow) without separately estimating M' or V', unlike small CIC which requires an estimate of V' to convert stock to flow.
- For large currencies, significant portions of CIC are held abroad (estimated 30% of Euros and 55% of dollars), distorting CIC/GDP measures for domestic cash use.

### Drawbacks and caveats of ATM withdrawal measure
- Almost no data exist on cash-back at the POS and few observations on over-the-counter (OTC) bank or postal withdrawals by consumers.
- OTC observations comingle consumer OTC withdrawals (presumed payments) with retailer withdrawals to make change—making OTC data unhelpful for isolating consumer payment withdrawals.
- CIC includes cash-back at the POS and OTC cash withdrawn by consumers, as well as hoarding, illegal uses, and cash held outside the country—hence neither CIC nor ATM measures are perfect.
- European Central Bank OTC data for six European countries indicate that, if OTC consumer withdrawals were simply added to ATM cash, ATM cash shares for four of six countries would rise by two percentage points (pp) or less and by six to seven pp for the remaining two—illustrating sensitivity to missing OTC/cash-back data.

### Authors' assessment and preference
- While neither measure is perfect, the authors prefer cash withdrawn from ATMs as a better reflection of the value of cash used for payments than CIC because:
  - It more closely reflects legal payment activity (excluding hoarding and likely illegal uses).
  - It is a flow measure that captures velocity changes over time tied to payment substitution.
  - It avoids cross-border holding distortions inherent in CIC for major currencies.

---

*IMF Working Paper — Measurement and Use of Cash by Half the World’s Population, Chapter: 1. Introduction; Sections 2, 2.1, 2.2*

### 3. Differences in Cash Use Across 14 Countries

### 3. Differences in Cash Use Across 14 Countries

### 3.1 Value of Currency Use Per Adult — Key Findings
- Measures compared: CIC per adult (CICPP) and ATM cash per adult (CASHPP).  
- China:
  - ATM cash per adult (CASHPP) had a 98% rise in nominal terms from 2005 to its peak in 2017.
  - After the peak, and relative to the base period 2005, use of ATM cash fell by 42% over the next three years to 2020.
  - CIC in China (CICPP) rose by 32% up to 2017 and then continued to rise by an additional 47% (relative to 2005).
  - Inconsistency: ATM cash falling while CIC rising (identified as Problem 1).
- Euro Area and US:
  - CIC is rising while ATM cash is relatively stable, rather than falling for more than one year toward the end of the sample (identified as Problem 2).
- India and Russia:
  - ATM cash is rising faster than CIC at different points in time; the reverse occurs for the US and (less so) for the Euro Area (identified as Problem 3).
- Interpretation and preference:
  - The authors consider ATM cash preferable to measure cash used for payments because it includes turnover/velocity; CIC does not and includes non-payment activity (hoarding) and illegal payment activity (extent largely unknown).
- Footnotes and data notes:
  - Domestic payments = sum of ATM cash, all cards, instant payments, checks and direct debits; excludes large value wire or credit transfers.
  - Reversal for the US and Euro Area likely due to significant portion of CIC held and used outside their borders.
  - Australia collects additional case use data including debit card “cash out”, credit card cash advances, cash depot utilization, and consumer survey data.

### 3.1 Value of Currency Use Per Adult — Inconsistency Typology (Table 1)
- Problem 1: One level measure is rising; the other is falling:
  - Australia, China
- Problem 2: One level measure is rising or falling, the other is relatively stable:
  - Euro Area, Norway, Singapore, UK, US
- Problem 3: One level measure rising or falling faster than the other one:
  - India, Indonesia, Malaysia, Mexico, Russia, Sweden, Thailand

### 3.2 Share of Currency Use Per Adult — Key Findings
- Relative measures used: CIC/GDP and ATM cash share (ATM cash as share of domestic payments).
- Two determinants of trend in relative use:
  - Changes in numerator (CIC or ATM cash).
  - Changes in denominator (GDP or value of payments); if denominators rise faster, relative shares fall even if numerator rises.
- China example:
  - CASHPP peak in level occurred in 2017; shareCASH peak occurred in 2011 because other payments rose more rapidly, causing shareCASH to decline starting 2011.
  - CICPP slowly rose 2005–2020 while CIC/GDP constantly fell because GDP consistently rose more rapidly than CIC.
  - Problem 1 (one trend rising while the other falling) persists when using share measures, though attenuated.
- Conclusion: Whether CIC or ATM cash are viewed individually or relatively, they do not provide the same information about the trend in cash use.

### 3.2 Share of Currency Use Per Adult — Inconsistency Typology for Shares (Table 2)
- Problem 1: One relative measure is rising; the other is falling:
  - Australia, China
- Problem 2: One measure is rising or falling, the other is relatively stable:
  - Euro Area, India, Indonesia, Malaysia, Norway, Russia, Singapore, UK, US
- Problem 3: One measure rising or falling faster than other one:
  - Mexico, Sweden
  - Thailand (basically parallel, no problem)

### 4. Comparing Shares of Currency Use Per Adult Across Countries — Key Findings
- Countries with cash share ≥ 12% at some point (Figure 2):
  - Indonesia and Russia reached over 70% at their highest point; by 2020 they fell to 40% and 20%, respectively.
  - Sweden and Norway: start at 24% and 13% in 2005, consistently fall to 4% and 3% in 2020.
  - Mexico: only country with continuous rise for whole period.
  - Malaysia and India: rose initially then leveled off or fell slightly; by 2020 their shares are 15% and 17%, respectively.
- Countries with lower shares (Figure 3) — dispersion:
  - Initial dispersion in 2005 among 6 countries ranges from 11% (Thailand) to 1% (China and US).
  - Dispersion falls slightly to 8 percentage points by 2020 (Euro Area 9%, China and US 1%).
  - Australia, China, and the US: changes in shares between 2005 and 2020 are so small beginning and ending values are almost equal.

### 4. Comparing Shares — Detailed Table Results (Table 3)
- Table 3 columns: (1) Pct. Share in 2005; (2) Pct. Share in 2020; (3) pp change; (4) Peak Year (& Share); (5) pp Decline Since Peak.
- Exact values from Table 3:
  - Russia: in 2005 = 59% ; in 2020 = 21% ; pp change = -38 ; Peak Year = 2010 (75%) ; pp Decline Since Peak = -54
  - Indonesia: in 2005 = 72 ; in 2020 = 41 ; pp change = -31 ; Peak Year = 2014 (75) ; pp Decline Since Peak = -34
  - Sweden: in 2005 = 24 ; in 2020 = 4.6 ; pp change = -19 ; Peak Year = 2006 (24) ; pp Decline Since Peak = -20
  - Norway: in 2005 = 13 ; in 2020 = 2.5 ; pp change = -11 ; Peak Year = 2006 (13) ; pp Decline Since Peak = -11
  - India: in 2005 = 1.6 ; in 2020 = 17 ; pp change = 15 ; Peak Year = 2015 (22) ; pp Decline Since Peak = -5
  - Thailand: in 2005 = 11 ; in 2020 = 8.1 ; pp change = -2.8 ; Peak Year = 2005 (11) ; pp Decline Since Peak = -3
  - UK: in 2005 = 5.8 ; in 2020 = 2.5 ; pp change = -3.3 ; Peak Year = 2009 (7) ; pp Decline Since Peak = -4
  - China: in 2005 = 1.1 ; in 2020 = 1.3 ; pp change = 0.2 ; Peak Year = 2011 (4) ; pp Decline Since Peak = -2
  - Malaysia: in 2005 = 8.1 ; in 2020 = 15 ; pp change = 6.7 ; Peak Year = 2018 (17) ; pp Decline Since Peak = -2
  - Australia: in 2005 = 2.3 ; in 2020 = 1.9 ; pp change = -0.4 ; Peak Year = 2018 (2) ; pp Decline Since Peak = -1
  - Euro Area: in 2005 = 6.4 ; in 2020 = 9.3 ; pp change = 2.9 ; Peak Year = 2016 (10) ; pp Decline Since Peak = -1
  - Singapore: in 2005 = 6.7 ; in 2020 = 5.5 ; pp change = -1.2 ; Peak Year = 2014 (7) ; pp Decline Since Peak = -1
  - US: in 2005 = 1.0 ; in 2020 = 1.3 ; pp change = 0.3 ; Peak Year = 2011 (1) ; pp Decline Since Peak = -0
  - Mexico: in 2005 = 10 ; in 2020 = 27 ; pp change = 19 ; Peak Year = 2020 (29) ; pp Decline Since Peak = 0
- Notes from Table 3:
  - pp = percentage points; all data are rounded; data starts in 2009 for Singapore, 2006 for China.

### 4.1 Cash Use in Advanced and Emerging Economies — Key Finding
- The sample contains both advanced and emerging market economies; high and low cash users appear in both groups, suggesting factors beyond income level and availability of cash substitutes (e.g., relative ATM availability, institutional or cultural differences).
- Table 4: Average ATM Cash Use Per Adult in 2020 (US dollars, PPP) — exact values:
  - Top group:
    - Singapore: Annual $13,790 ; Weekly $265 ; Daily $38
    - Russia: Annual $11,568 ; Weekly $222 ; Daily $32
    - Malaysia: Annual $10,712 ; Weekly $206 ; Daily $29
    - Thailand: Annual $10,603 ; Weekly $204 ; Daily $29
    - Euro Area: Annual $6,300 ; Weekly $121 ; Daily $17
    - Mexico: Annual $4,805 ; Weekly $92 ; Daily $13
    - Australia: Annual $4,299 ; Weekly $83 ; Daily $12
  - Bottom group:
    - China: Annual $3,859 ; Weekly $74 ; Daily $11
    - US: Annual $3,801 ; Weekly $73 ; Daily $10
    - UK: Annual $3,798 ; Weekly $73 ; Daily $10
    - Indonesia: Annual $3,458 ; Weekly $67 ; Daily $9
    - Sweden: Annual $1,459 ; Weekly $28 ; Daily $4
    - India: Annual $1,421 ; Weekly $27 ; Daily $4
    - Norway: Annual $1,042 ; Weekly $20 ; Daily $3
  - Total (sum across sample): Annual $80,915 ; Weekly $3,112 ; Daily $222
- Authors’ remark: annual values may appear high and daily values low given typical withdrawal frequency; weekly or bi-weekly are more representative of individual monitoring.

### 4.2 Aggregate Use of Cash for Half the World’s Population — Key Findings
- Sample coverage: 14 countries accounting for 53% of the world’s population and two-thirds of world GDP.
- Aggregate series shown:
  - PPPCash = sum of PPP-adjusted annual value of ATM cash per adult across countries.
  - PPPCIC = sum of PPP-adjusted value of currency in circulation per adult across countries.
- Aggregate pattern:
  - ATM cash (PPPCash) rises faster than CIC (PPPCIC), reaches a peak toward the end of the period, and then starts to fall while CIC is still rising — mirroring the China-dominated inconsistency observed at the country level.
  - Share series (Figure 5): PPPCASHshare (PPP-adjusted share of ATM cash in payments) first rises, reaches a peak, then falls toward the end of the period while PPPCIC/GDP share is relatively stable.
- Methodological note:
  - Figures 4 and 5 determined by summing US PPP dollar values across countries before forming ratios; sums across 14 countries are obtained prior to division to form graphs.

_Italic: Source: Authors’ calculations (from IMF Working Paper chapter "3. Differences in Cash Use Across 14 Countries")._

### 5. Conclusion

### 5. Conclusion

### Comparison of cash-use indicators
- Standard indicator: value of currency in circulation (CIC), and CIC as a ratio to GDP.
- Alternative indicator proposed: value of cash withdrawn from ATMs, and ATM cash as a ratio to the total value of payments in a country (excluding large value credit transfers).
- All indicators are measured per adult to control for demographic differences across countries.
- Aim: exclude cash uses for hoarding or illegal use, focusing on cash used for payments.

### Key empirical findings
- The CIC measure and the ATM cash measure often give conflicting information on cash use for payments:
  - Different magnitudes at the same point in time.
  - Different slopes for rising or downward trends at the same time.
  - On some occasions, differing trend directions—one rising, the other falling.
- Main reason for divergence:
  - CIC includes cash used for payments, hoarding, and illegal use.
  - ATM cash is focused much more on cash used for payments.
  - CIC is a stock not adjusted for turnover (velocity) while ATM cash includes turnover.
- Aggregate pattern:
  - Cash is still strongly used for payments in some countries, but in most it has reached a peak and, overall, is starting to fall using the ATM cash measure—either in levels or as a share of payments.

### Sample-specific statistics
- The 14 countries in the sample account for half of the world’s population and two-thirds of its GDP.
- Using the ATM cash measure:
  - Four countries with the highest cash shares in 2005 (ranging from 13 to 72%) experienced the largest reductions in cash use by 2020 (falling between 11 to 54 pp) over our 16-year period.
  - Half of the countries already have low cash shares of 5.5% or less.
  - Considering largest and smallest users together: when cash shares are high, share reductions are large; when shares are small, reductions are also small.
  - The pattern of cash reduction, when falling, follows an inverse logistic curve and thus seems to fall at a decreasing rate.
  - All but two of the 14 countries have reached a peak in their cash use toward the middle or end of the sample period (2005-2020) and are currently experiencing a downward trend.

### Strengths and weaknesses of indicators
- CIC:
  - Includes hoarding and illegal use (undesired for payment-focused measurement).
  - Excludes timely changes in velocity.
- ATM cash:
  - Includes velocity.
  - Focused on payments because ATMs are usually stocked with medium denominations and reflect cash used for transactions.
  - Similar in focus to a “small CIC” indicator that excludes high value notes thought to be used mostly for hoarding and illegal activities.
- Overall assessment:
  - Neither measure is perfect.
  - The balance favors the ATM cash measure for assessing cash use for payments.

### Policy implications and recommendations
- The ATM cash measure may be preferred when policymakers face decisions concerning central bank digital currency (CBDC).
- Where cash use is already very low:
  - Establish regulations to ensure access to cash.
  - Assure that cash can continue to be used for transactions with merchants.

*Source: IMF Working Paper — Measurement and Use of Cash by Half the World’s Population, Section 5. Conclusion*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023062-print-pdf.pdf_