Prima facie, large withdrawals aren’t because of a change in the denomination mix. Until this behaviour is better understood, a policy response will be incomplete
A reported cash shortage at ATMs has got media attention, but might only be a symptom of underlying behavioural hysteresis in using cash for transactions. The ATM cash problem will quickly be sorted out, but the behavioural trends might need further policy intervention. Cash in the economy consists of two parts: (i) currency with the public (i.e., the cash which households hold) and (ii) currency with banks (which is a transactional balance used for dispensing money through ATMs and bank branches). As of March 31, total system currency was Rs 18.3 trillion (lakh crore), of which households held Rs 17.6 trillion.
Analysts had spotted a sharp rise in cash drawn from banks, particularly in December-January 2018 (see graphic), a puzzling phenomenon, largely unexplained beyond the standard suspects—elections, kharif crop marketing, weddings, etc. While there has also been a fairly large cash withdrawal in the first two weeks of FY19, this was not out of the ordinary, and pretty much in line with previous years. The accompanying graphic shows the growth of currency in circulation (CIC) since 1992.
The line in the second chart is the actual amount of cash, with a clear exponential trend. This growth was arrested in November 2016, but has now crossed the pre-demonetisation. What if demonetisation (DeMo) hadn’t happened? Extrapolation of the exponential trend indicates that the CIC might now have been over Rs 22 trillion, instead of the Rs 18.42 trillion as on April 6, 2018.
The optimal level of cash holdings is theoretically estimated by studying what economists refer to as “the velocity of money”. In common sense terms, this is the following question: Has cash utilisation in creating national income (GDP) become more efficient over time? The graphic shows outstanding cash per unit of GDP over the years. This suggests falling cash utilisation efficiency till FY10 over a long cycle, before a reversion to stable and then rising efficiency till FY15, and then a deterioration again in FY16.
All this was total cash in the system (i.e., CIC). What about the portion of cash held by banks for transactions, the portion pertinent for the current concerns about cash shortages at ATMs? After a sharp rise in cash balances held at bank (tellers and ATMs) in H1FY18 (the re-monetisation effect), these balances dropped sharply in H2, finishing at levels close to the end of FY16. It is difficult to infer either supply or demand behavior from this, but read together with FY18 cash behaviour depicted in accompanying graphic, that seems to suggest that the trend might have been more because of the constricted supply.
Of course, an alternative inference might be that banks have become more efficient in managing their cash holdings. Although purely hypothetical, since it is difficult to construct a counter-factual, the basic problem seems to be that the cash portion of the payments system is operating at the envelope of the economy’s requirement.
As noted above, if we are to base an inference on past trends, the system might be notionally short of about Rs 3 trillion of cash. Adjusting for estimated transactions multipliers and the growth in economic activity, the rise in digital transactions (via NEFT, wallets, cards, UPI and other channels) might have compensated for a large part of this notional deficit, since transaction amounts using these instruments and channels have indeed risen.
A back-of-the-envelope calculations suggest that this gap in cash requirements might be about Rs 300–500 billion (Rs 50,000 crore). So, the probability is quite high that a buffer, which might earlier have existed to absorb a transient demand– supply mismatch, may now have been depleted. Any localised surge in demand might then stress the cash supply logistics. Can any of this be corroborated with patterns of cash withdrawal at ATMs? These are largely in line with increases in cash outstanding till late FY18, when a divergence appears.
However, the nature of the withdrawals at ATMs was distinctly different in FY18. There was a deep slide in transaction volumes in FY18, implying a steep rise in the average size of the transactions. In other words, larger amounts of cash were being drawn in FY18 at every ATM transaction.
This is a change, which might be transient, but has persisted for too many months, and is likely to be behavioural; and it is difficult to explain with the available data. This behaviour is sharply different from the year prior to DeMo, FY16, but is similar to the months in FY17 before November.
To the best of our knowledge, banks had not imposed additional restrictions or charges on withdrawals after the normalisation in April 2017, which might explain this sustained change. And prima facie, this probably is not the result of a change in the denomination mix of notes. Until this behaviour is better understood, a policy response will be incomplete.
With contributions from Tanay Dalal and Vikram Chhabra