by Gautham Udupa^{^}, Pradip Bhuyan^#, Dileep Kumar Verma^#, and Nirupama Kulkarni^{^}

Novel high-frequency measures, in the form of gross tax collections and night-lights are used as proxies for economic activity. There is a positive relationship between night-lights and taxes as well as between night-lights and GDP. Formal economic activity, proxied by tax collections, reduces Notes in Circulation (NiC).

Introduction

Currency plays a central role in the economy as a facilitator of transactions and as a store of value. While the secular growth in cash¹ in India has fallen considerably since the 1990s, Notes in Circulation (NiC) has risen in recent years, possibly in part due to COVID-19 driven surge in demand for precautionary cash holdings. The level of NiC has risen from about ₹2.1 lakh crore in 2001 to about ₹34.8 lakh crore in 2024 (RBI, 2001, 2024).

The Reserve Bank of India (RBI) is responsible for overall management of currency in the country. Currency includes both banknotes and coins. Banknotes are issued and circulated by RBI. The issuing authority of coins is the Government of India (GoI) and the coins are put into circulation by RBI on behalf of GoI. In terms of value, banknotes accounted for around 99 per cent of total currency in circulation at the end of March 2024.

Printing and storage capacities, in particular, cannot be altered at short notice. Forecasting future demand for currency is an important element of currency management. Indent and supply of banknotes vary across denominations (Table 1). It is, therefore, important to develop reliable long-run forecasts for the demand for banknotes that incorporate recent developments in the Indian economy.

The article illustrates how economic activity affects banknotes in circulation and presents an extensive discussion. Formal economic activity is proxied by gross tax collections, whereas night-lights data are taken as a proxy for total economic activity. This article brings in new perspectives to understand the role of economic activity in generating demand for banknotes.

The article is divided into five sections. Section II reviews the literature on the determinants of currency demand, section III provides stylized facts regarding NiC and economic activity, and connects them to night-lights and tax collections; section IV reports and discusses results from econometric tests of the link between these variables and NiC; conclusion is in section V.

II. Literature Review

Theoretical literature on the role of cash as a facilitator of economic transactions goes back several decades and in these models, consumers prefer to transact in cash and therefore hold a part of their liquid wealth in the form of cash (Baumol, 1952; Tobin, 1956). The trade-off is governed by the interest rate on bank deposits – if the interest rate is zero, the consumer is indifferent between holding deposits and cash. The literature since then has identified several frictions, both on the demand side (i.e., consumers) and on the supply side (i.e., banking frictions etc.), that impact the deposit-versus-cash trade-off. Several of these frictions favour using cash over holding bank deposits.

The frictions on the demand side are a result of consumers’ preferences. These include (i) the use of cash as a tool for liquidity monitoring (Kalckreuth et al., 2014), (ii) imperfect substitutability between cash and bank deposits (Drechsler et al., 2017; Wang et al., 2022), (iii) Consumption preferences (Alvarez and Lippi, 2009), (iv) network effects in the adoption of digital payments (Amromin and Chakravorti, 2009; Crouzet et al., 2019), and (v) size of transaction (Wang and Wolman, 2016).

The supply side frictions are an outcome of the depth of the banking sector and that of the financial sector in general. These include (i) ATM and bank branch density (Alvarez and Lippi, 2009; Lippi and Secchi, 2009; Prina, 2015; Dupas et al., 2018; Ragot, 2014), (ii) depth of wholesale funding markets (Lucas and Nicolini, 2015), (iii) the cost of holding cash including interest rate, loss due to theft and neglect, and inflation (Attanasio et al., 2002; Amromin and Chakravorti, 2009; Benati et al., 2021), and (iv) policy uncertainty (RBI, 2023).

Among these frictions, ATM and bank branch density are often used as policy tools for effective cash management which may also reduce cash usage. RBI has in the past initiated measures on opening new bank branches to improve financial inclusion in rural India. Alvarez and Lippi, 2009 and Lippi and Secchi, 2009 show how costlier cash withdrawal technology in the form of a sparser ATM network leads to greater cash holdings. The relative ease with which deposits can be converted into cash and vice versa has a bearing on households’ liquid assets portfolio composition. These costs are governed by the density of bank branches and ATMs, and by whether ATMs accept cash deposits in addition to disbursing cash.

Similarly, randomized control trial experiments have shown that offering free or low-cost bank accounts increases take-up, but the effects are small (Prina, 2015; Dupas et al., 2018). Digital payment methods would impact cash usage by offering alternative methods to hold liquid assets with minimal leakage risk. However, only smaller denomination currencies are sensitive to digital payments such as debit cards, whereas interest rate sensitivity is higher for higher denomination cash (Amromin and Chakravorti, 2009).

In the case of India, the practice of identifying the NiC determinants goes back (Jadhav, 1994; Palanivel and Klein, 1999; Bhattacharya and Joshi, 2001; Bhattacharya and Joshi, 2002; Kumar, 2011; Nachane et al., 2013; Bhattacharya and Singh 2016; Chaudhari et al., 2019; Raj et al., 2020, Awasthy et al., 2022). More recently, there is comprehensive work documenting the determinants using time series methods using weekly, monthly, quarterly, and annual data (Raj et al., 2020). The goal of their weekly model is to support the RBI’s liquidity management objective. In their monthly model, digital payments are incorporated into the methodology. In their quarterly and annual models, other variables such as nominal gross domestic product (GDP) and 1-3-year tenor average deposit rates are included. The main advantage of the quarterly and annual models is that the data goes back in time (up to 1970-71 in the annual model), which allows estimation of time-varying coefficients. Recently, there is work on the development of a cash usage indictor for better currency management (Bhuyan, 2024). In comparison, the goal of this article is to establish the role of economic activity on NiC.

III. Stylized Facts

The demand for physical cash has undergone several changes in the last twenty years. Rapid expansion of bank branches and ATM network, increased penetration of internet-enabled phones, and dramatic changes in payment and settlement systems have possibly contributed to lower CAGR in NiC in the last decade under reference (Table 2).

In each of the two 10-year periods between 2004 and 2024, CAGR of NiC in value was higher than that in volume, indicating a shift towards higher denominations. It is worth noting that the growth rate in NiC (in value terms) in the 10 year period between 2014 and 2024 was significantly lower as compared to that in the previous two decades. Moreover, the growth in NiC was noticeably higher than that in GDP between 1994 and 2004, the gap however significantly reduced in the next two decades.

Between 2005 and 2014, the number of ATMs per lakh adults increased dramatically, at a CAGR of slightly over 25 per cent. Evidence shows that in normal period (i.e. a period not affected by events like Covid etc.) easier access to ATMs reduces households’ cash holdings as they are more comfortable holding lower balances and precautionary holdings are reduced (Alvarez and Lippi, 2009).

As a high-frequency measure of total economic activity, night-lights intensity is analysed. It is a reasonable proxy at high frequency when similar administrative data on GDP is not available at a granular level. Night-lights is a result of total economic activity and its intensity has recently been shown to be correlated with Gross Domestic Product (GDP) used to capture economic activity (Beyer et al., 2022; Beyer et al., 2023).

Night-lights captured by satellites on a daily basis are adjusted for cloud covers to arrive at a very granular measure of monthly average night-light intensity. The resulting image clearly indicates major cities as well as smaller towns and villages (Chart 1).

IV. Results – Economic Activity and NiC

In this section, statistical tests of the relationships between night-lights, taxes, and NiC are conducted. The best dataset available, both in terms of granularity and frequency, is used in each of the tests. For example, monthly state-level data is used to test the relationship between night-lights and taxes, as the data on the latter is only available at the state level. Because data on night-lights capture total economic activity, the relationship between night-lights and gross taxes is expected to be positive. This is tested formally by running fixed effects regressions of logged variables. The estimated coefficients, therefore, reflect the elasticity between night-lights and taxes. Results show a positive and statistically significant relationship between the two (Table 3).

The estimates indicate a more than one-to-one response of night-lights to a given percentage change in tax collections. The elasticity falls considerably to about 0.1 once state-fixed effects (FEs) are included. The most detailed specification with state and state by calendar-month fixed effects leads to an elasticity estimate of just under 0.14.

Similarly, state-level panel data by financial year between 2012 and 2022 is used to test the relationship between gross state domestic product (GSDP) and night-lights. The analysis is similar in spirit to Beyer et al., 2022 and Beyer et al., 2023. In a fixed effects regression with variables represented in logs in model 1, the coefficient is positive (0.632) and statistically significant at 1 per cent level (Table 4). In model 2, 4 lags of both the dependent and independent variables are included. The coefficient magnitude is 0.398 and remains significant at 5 per cent level. The coefficient magnitudes are comparable to estimated values in the existing literature between 0.46 and 0.50 in cross-country panel data (Beyer et al., 2022).

In view of the preceding findings, night-lights and taxes are used to examine their impact on NiC. The augmented model includes night-lights and taxes in addition to the variables used in the previous studies (Raj et al., 2020; RBI, 2023) and covers a time period between August 2014 to November 2022. In particular, an ARDL regression in logs is conducted where the dependent variable is the log of NiC, and independent variables, motivated by RBI, 2023, include (in addition to the two variables of interest) economic policy uncertainty index (Baker et al., 2016), weighted average deposit rates of scheduled commercial banks, log of value of digital transactions, and dummy variables for demonetization and pandemic affected months. Data used in this analysis were monthly time series. All the variables are stationary in log first differences as indicated by Augmented Dickey Fuller (ADF) tests.

ARDL model is chosen as opposed to alternatives as most variables used in the model are integrated of order one, but some variables, such as weighted average deposit rates, are stationary. These factors make ARDL the ideal model to choose (Johansen and Juselius, 1990). The bounds test F-stats are higher than the 1 per cent critical values, indicating that the variables are co-integrated in each of the models (Table 5). The long run coefficients related to the two variables of interest, i.e., night-lights and gross taxes, are reported. In a regression of NiC on all the variables except taxes, the night-lights coefficient is positive but not statistically significant. In a specification that omits night-lights but includes taxes, the coefficient on the latter is negative. Finally, these patterns survive in a regression with all the variables, and the coefficient on taxes is more negative and significant at 1 per cent level. Overall, this implies that formal economic activity, proxied by gross taxes, would lead to a fall in NiC.

V. Conclusion

This article proposes a novel approach to link economic activity to demand for banknotes. It uses night-lights as a proxy for total economic activity, and tax collection as a proxy for formal economic activity. It finds that formal economic activity reduces NiC. The estimated coefficient on night-lights in the full specification is positive but it is not statistically significant. Further research can refine the estimates provided in this paper when more granular and frequent data on measures of formal and total activities are available.

References

Alvarez, Fernando, and Francesco Lippi. (2009). Financial Innovation and the Transactions Demand for Cash. Econometrica, 77 (2), 363–402.

Amromin, Gene, and Chakravorti Sujit. (2009). Whither Loose Change? The Diminishing Demand for Small-denomination Currency. Journal of Money, Credit and Banking, 41 (2-3), 315–35.

Awasthy, S., Misra, R., & Dhal, S. (2022). Cash versus Digital Payment Transactions in India: Decoding the Currency Demand Paradox, Reserve Bank of India Occasional Papers, 43(2), 1-45.

Attanasio, Orazio P, Luigi Guiso, and Tullio Jappelli (2002). The Demand for Money, Financial Innovation, and the Welfare Cost of Inflation: An Analysis with Household Data. Journal of Political Economy 110 (2), 317–51.

Baker, S. R., Bloom, N., and Davis, S. J. (2016), Measuring Economic Policy Uncertainty, The Quarterly Journal of Economics, 131(4), 1593-1636.

Baumol, William J. (1952). The Transactions Demand for Cash: An Inventory Theoretic Approach. The Quarterly Journal of Economics 66 (4), 545–56.

Benati, Luca, Robert E Lucas Jr, Juan Pablo Nicolini, and Warren Weber (2021). International Evidence on Long-Run Money Demand. Journal of Monetary Economics 117, 43–63.

Beyer, Robert CM, Yingyao Hu, and Jiaxiong Yao (2022). Measuring Quarterly Economic Growth from Outer Space. International Monetary Fund.

Beyer, Robert CM, Tarun Jain, and Sonalika Sinha (2023). Lights out? COVID-19 Containment Policies and Economic Activity. Journal of Asian Economics 85, 101589.

Bhattacharya, Kaushik, and Himanshu Joshi (2001). Modelling Currency in Circulation in India. Applied Economics Letters 8 (9), 585–92.

——— (2002). An Almon Approximation of the Day of the Month Effect in Currency in Circulation. Indian Economic Review, 163–74.

Bhattacharya, Kaushik, and Sunny Kumar Singh (2016). Impact of Payment Technology on Seasonality of Currency in Circulation: Evidence from the USA and India. Journal of Quantitative Economics 14, 117–36.

Bhuyan, Pradip (2024). Cash Usage Indicator for India. Reserve Bank of India Bulletin October, 255-269.

Chaudhari, Dipak R, Sarat Dhal, and Sonali M Adki (2019). Payment Systems Innovation and Currency Demand in India: Some Applied Perspectives. Reserve Bank of India Occasional Papers, 40 (2): 33–63.

Crouzet, Nicolas, Apoorv Gupta, and Filippo Mezzanotti (2019). Shocks and Technology Adoption: Evidence from Electronic Payment Systems.

Drechsler, Itamar, Alexi Savov, and Philipp Schnabl (2017). The Deposits Channel of Monetary Policy. The Quarterly Journal of Economics, 132 (4), 1819–76.

Dupas, Pascaline, Dean Karlan, Jonathan Robinson, and Diego Ubfal (2018). Banking the Unbanked? Evidence from Three Countries. American Economic Journal: Applied Economics 10 (2), 257–97.

Jadhav, Narendra (1994). Monetary Economics for India, Macmillan India, New Delhi.

Johansen, Soren, and Katarina Juselius (1990). Maximum Likelihood Estimation and Inference on Cointegration—with Appucations to the Demand for Money. Oxford Bulletin of Economics and Statistics 52 (2), 169–210.

Kalckreuth, Ulf Von, Tobias Schmidt, and Helmut Stix (2014). Using Cash to Monitor Liquidity: Implications for Payments, Currency Demand, and Withdrawal Behavior. Journal of Money, Credit and Banking 46 (8), 1753–86.

Kumar, Saten (2011). Financial Reforms and Money Demand: Evidence from 20 Developing Countries. Economic Systems 35 (3), 323–34.

Lippi, Francesco, and Alessandro Secchi (2009). Technological Change and the Households’ Demand for Currency. Journal of Monetary Economics 56 (2), 222–30.

Lucas Jr, Robert E, and Juan Pablo Nicolini (2015). On the Stability of Money Demand. Journal of Monetary Economics 73, 48–65.

Nachane, DM, AB Chakraborty, AK Mitra, and S Bordoloi (2013). Modelling Currency Demand in India: An Empirical Study. Reserve Bank of India Discussion Paper 39.

Palanivel, Thangavel, and Lawrence R Klein (1999). An Econometric Model for India with Emphasis on the Monetary Sector. The Developing Economies 37 (3), 275–327.

Prina, Silvia (2015). Banking the Poor via Savings Accounts: Evidence from a Field Experiment. Journal of Development Economics 115, 16–31.

Ragot, Xavier (2014). The Case for a Financial Approach to Money Demand. Journal of Monetary Economics 62, 94–107.

Raj, Janak, Indranil Bhattacharyya, Samir Ranjan Behera, Joice John, and Bhimappa Arjun Talwar (2020). Modelling and Forecasting Currency Demand in India: A Heterodox Approach. Reserve Bank of India Occasional Papers 41 (1), 1–45.

RBI (2001). RBI Annual Report 2001.

RBI (2023). RBI Annual Report 2023.

RBI (2024). RBI Annual Report 2024.

Tobin, James (1956). The Interest-Elasticity of Transactions Demand for Cash. The Review of Economics and Statistics 38 (3), 241–47.

Wang, Yifei, Toni M Whited, Yufeng Wu, and Kairong Xiao (2022). Bank Market Power and Monetary Policy Transmission: Evidence from a Structural Estimation. The Journal of Finance, 77 (4), 2093–2141.

Wang, Zhu, and Alexander L Wolman (2016). Payment Choice and Currency Use: Insights from Two Billion Retail Transactions. Journal of Monetary Economics, 84, 94–115.

^{^} : The authors are from the Centre for Advanced Financial Research and Learning (CAFRAL).

^# : The authors are from the Department of Currency Management, Reserve Bank of India (RBI).
Authors are thankful for the valuable comments and suggestions received from Shri B.P. Kanungo and Shri Suman Ray. Views expressed in this article are those of the authors and do not represent the views of the RBI and/or CAFRAL.

¹ The words ‘cash’ and ‘currency’ are used interchangeably in this article.