IMF Concludes Cash Usage Falling Globally

While data about digital payments is usually good, measuring cash is problematic. Understanding cash usage is important for regulatory authorities. IMF Working Paper WP/23/62 1 explores the measurement of cash based on ATM cash withdrawals value as an alternative to the traditional cash in circulation (CIC) as a proportion of GDP method. The paper analyses data from 14 countries representing over half the world’s population and concludes that cash usage is falling around the world.

Good cash data would allow central banks to understand and manage banknotes and their fitness, the criminal use of cash, access to cash for individuals without bank accounts, bank costs for providing cash to depositors, and business costs from accepting cash.

This is particularly important where access to cash and cash acceptance has become an issue. Falling demand for cash for purchases can lead to reduced access to cash from banks. When cash use is very low, banks have responded by reducing their ATM networks and, in some cases, even consolidating a few of their branch offices to save costs. Good data would allow central banks to respond in good time.

Cash Share

Khiaonarong and Humphrey have used a ‘Cash Share’ measure to understand cash usage and have compared their findings with CIC/GDP to draw conclusions. In this approach ATM cash withdrawals are expressed as a ratio to the total value of all payments in a country. The authors believe this more closely reflects the use of cash for payments than does CIC, which includes high value notes that are regarded by many as being used largely for hoarding and illegal activities. They see focusing on cash for payments as more important than measuring its other potential uses.

In addition, the value of cash withdrawn from ATMs is a flow measure, not a stock, and thus its value is already adjusted for changes in velocity due to the substitution of cards and other payment instruments for cash over time.

Why use a new method?

If user needs are fully met with the current mix of payment instruments on offer, payment shares should be stable and indicate that, given the benefits from using different payment instruments, cost efficiency is met as well. A declining share of cash, however, suggests that neither goal is currently achieved.

The study looked at data from Advanced Economies (AEs) and Emerging Market Economies (EMEs). In order to assess differences in cash use for payments among economies at different levels of development, a common metric was needed to allow a comparison of local currency payment values. The data for all countries was transformed into purchasing power parity (PPP) US dollars.

The demand side of CIC concerns how the population chooses to spread currency and coin across three general activities:

1. ATM cash withdrawals that are typically used for payments that are known and legal

2. Cash used for hoarding, which are not payments and has 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.

There is good data for ATM cash withdrawals but the value attributable to activities (2) and (3) include some unknown amount. This represents an inevitable bias when (1) is used to represent the value of cash for payments. This bias is regarded as being less than the bias of using all of CIC for payments, hence using ATM cash rather than CIC when choosing to measure the use of cash for payments.

What has happened to cash usage?

Measured as a level, or relative to GDP or total payments, the standard measure and the cash share approach give conflicting information on the use of cash for payments. The data shows different magnitudes of the two measures at the same point in time, different slopes for the rising (or downward) trends they both may have at the same time, and also on some occasions even a different direction in the trend – one rising, the other falling. This occurs in different countries, at different time periods, but is important enough to be obvious when the data for all countries are expressed in US dollars and aggregated across countries.

The main reason for this divergence lies in the fact that CIC includes cash used for payments, hoarding, and illegal use while ATM cash is focused much more on the use of cash for payments alone. In addition, the stock of CIC is not adjusted for the turnover of cash for payments while ATM cash already includes it.

The cash share measure of cash use per adult rose from 2005 and reached a peak in 2017, but fell thereafter. This occurred in absolute terms and as a payment share. The results for the standard measure of cash use (CIC) have not reached a peak and did not fall over this period studied.

Cash is still strongly used for payments in some countries, but in most it has reached a peak and, overall, it is starting to fall using the ATM cash measure – either in levels or as a share of payments.

Using the value of cash withdrawn from ATMs, 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 percentage points) over the 16-year period. Half of the countries already have had low cash shares of 5.5% or less.

Considering the largest and smallest users of cash together, when cash shares are high, share reductions are large. And 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. In this regard, 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.

Which method to use?

Neither measure of cash use is perfect. CIC includes hoarding and illegal use and excludes timely changes in velocity. ATM cash includes velocity and is focused on payments as ATMs are usually stocked with currency reflecting medium denominations – not the highest denominations.

For the authors, the balance favours the ATM cash measure of cash use for payments. This measure may be preferred when policy makers are faced with decisions concerning CBDC and, where cash use is already very low, establishing regulations to ensure access to cash and assuring it can continue to be used for transactions with merchants.


1 - ‘Measurement and Use of Cash by Half the World’s Population.’ Tanai Khiaonarong, David Humphrey.