by Sakshi Awasthy and Subrat Kumar Seet^{^}

While the broader shift to digital payments is well-established, regional adoption of the Unified Payments Interface (UPI) and its impact on cash demand remain underexplored. Using a dual empirical strategy - an autoregressive distributed lag model and panel quantile regression - this study finds that higher UPI adoption is associated with lower cash demand at both national and subnational levels, with state-level patterns suggesting non-linearity. Among other state-wise factors, income and ATM density are positively associated with cash demand, whereas workforce formalisation and educational attainment are linked to lower cash reliance.

Introduction

Payments underpin all economic activity. In a frictionless environment, the choice of payment mode may have less bearing on real outcomes; however, in practice, transaction costs and information asymmetries render certain payment methods more efficient than others in shaping economic growth (Dubey and Purnanandam, 2023). The shift from cash to digital payments, particularly fast payment systems, has been associated with increased welfare, financial inclusion, credit access, economic formalisation and financial resilience (Bachas et al., 2018; Aguilar et al., 2024; Aurazo and Franco, 2024; Cantú et al., 2024). At the same time, existing literature is also strewn with instances of simultaneous rise in cash and digital payments (Bech et al., 2018; Chen et al., 2020; Caswell et al., 2020), even as the transactional use of cash ebbs, or what is often described as the “paradox of banknotes” (Bailey, 2009). This trend has reinvigorated the debate on the impact of digital payments on cash, with significant implications for currency and liquidity management, underlying economic frictions, and broader macroeconomic policy.

India’s fast payment system, Unified Payments Interface (UPI), launched in 2016, offers a unique empirical setting to study the evolving relationship between cash and digital payments for three key reasons. First, the scale of adoption has been unprecedented. UPI users have surged from around 30 million in 2017 to over 420 million by 2024 (RBI, 2024; Reddy et al., 2024). Transaction volumes are nearing 200 billion a year, accounting for over 80 per cent of total digital payments (RBI, 2025). Second, the launch of UPI closely followed a large-scale financial inclusion drive i.e., Pradhan Mantri Jan Dhan Yojana, creating enabling conditions for widespread digital uptake across socio-economic groups. Finally, notwithstanding the growth in digital payments (especially UPI), currency in circulation has continued to rise, albeit at a slower pace in recent years, reflecting a dynamic interplay between cash and digital modes.

While the broader shift to digital payments is well-established (Nachane et al., 2013; Chaudhari et al., 2019; Raj et al., 2020; Awasthy et al., 2022; RBI, 2023), regional adoption of the UPI and its impact on cash demand at the state-level remain underexplored. Given India’s geographical and income diversity, national aggregates may obscure regional disparities, as digital uptake may be concentrated in select economic clusters, with cash being persistent in other regions. As per estimates, individuals in the top 20 per cent income group are twice as likely to use digital payments as those in the bottom 40 per cent (NPCI, 2020). More recent data show a steeper gradient, with the top 10 per cent by consumption expenditure twice as likely to report the ability to use UPI as the bottom 25 per cent, though the overall ability stands close to 50 per cent (NSO, 2025). As digital payments become central to economic activity, identifying regions that are excluded or lagging behind is crucial - not only to promote inclusive access but also to address infrastructure gaps and risks to consumer protection.

Against this backdrop, the paper examines the impact of UPI on cash usage by modelling cash demand at both national and subnational levels. Specifically, the study addresses four key research questions: (a) What is the impact of UPI on cash demand at the all-India aggregate level? (b) What regional patterns emerge in the adoption of UPI and cash? (c) How does UPI influence cash demand across states? and (d) Does this impact vary by state’s income levels? Given the limited empirical focus on regional trends, this study provides one of the first state-level assessments of cash to UPI substitution in India.

The remainder of the paper is structured as follows: Section II reviews the literature, followed by descriptive analysis in Section III. Section IV outlines the data and methodology, while Section V presents the empirical results. Section VI concludes. Technical details and additional estimation outputs are presented in Annexures I–III.

II. Related Literature

There exists a substantial body of theoretical and empirical literature on the determinants of money demand (Friedman, 1999; Alvarez and Lippi, 2009). The demand for cash is traditionally attributed to three primary motives: the transaction motive linked to economic activity (Fisher, 1911); the precautionary motive, reflecting the need for liquidity in uncertain situations; and the speculative motive, driven by expectations about interest rate movements (Keynes, 1954). Building on this, money demand is reconceptualised as a stable function of wealth, incorporating expected returns on alternative assets such as bonds, equities, and durable goods (Friedman, 1956). The seminal inventory (Baumol, 1952) and portfolio (Tobin, 1956) theoretical models extend the money demand function by incorporating interest rates and transaction costs. More recent studies emphasise the negative impact of payment innovations on physical currency (Columba, 2009; Oyelami and Yinusa, 2013; Huynh et al., 2014). Concurrently, a growing body of literature highlights the coexistence of cash and digital payments, attributing sustained cash usage to precautionary motives and economic uncertainties (Bech et al., 2018; Caswell et al., 2020; Chen et al., 2020; Ardizzi et al., 2020).

In the Indian context, studies have found a significant negative association between digital payments and currency demand, reflecting a growing substitution effect (Nachane et al., 2013; Bhattacharya and Singh, 2016; Chaudhari et al., 2019; Raj et al., 2020; and Awasthy et al., 2022; Udupa et al., 2025). At the regional level, however, empirical research has largely focussed on digital payment adoption, instead of substitution dynamics. Using transaction level data from PhonePe, Dubey and Purnanandam (2023) find that districts with higher post-UPI cashless payment intensity experienced significantly greater household income growth. Drawing on the same dataset, a report by ICRIER finds that COVID-19 accelerated digital adoption and narrowed disparities in UPI’s user penetration across states and districts (Reddy et al. 2024). The report also identifies key drivers of digital adoption such as income levels, internet access, digital literacy, and financial infrastructure.

III. How does India Pay?

III.1. Aggregate-Level Insights into Payment Choice

India has a diverse payment ecosystem, encompassing both cash and a broad suite of digital options. Currency in circulation (CIC)¹ has normalised from a peak of 14.4 per cent of Gross Domestic Product (GDP) in 2020–21 to 11.7 per cent in 2023–24 and further to 11.2 per cent in 2024–25. CIC growth slowed to 4–6 per cent in recent years, driven by structural shift towards digital payments, post-pandemic normalisation, phased withdrawal of ₹2000 notes, and greater formalisation (Chart 1). A marginal rise (y-o-y) in 2024-25 reflects higher rural demand and election-related spending. Real CIC growth turned negative in 2023-24 and remained modest in 2024-25, suggesting decline in inflation-adjusted cash demand.

In contrast, digital payments (value) as a share of GDP has risen sharply to over 800 per cent, with the pandemic acting as a catalyst for increased adoption in both volume and value terms (Chart 2a). Overall, total digital payments have exhibited robust growth over the last decade (2015-2025), recording a compound annual growth rate of 48 per cent by volume and 12.5 per cent by value. Monthly trends show a broadly sustained digital momentum amid tapering CIC growth (Chart 2b).

The shift away from cash is also evident in the decline in currency-to-demand deposits ratio to 1.31 in 2024-25 from 1.68 in 2015-16² and a steady fall in ATM cash withdrawals (as a share of GDP) since 2018-19 (Charts 3 a and b).

A possible driver behind the decline in cash demand has been the rise of UPI. Transaction volumes logged under the fast payment mode surged to 18,586 crore in 2024-25 from 1,252 crore in 2019-20, with a marked acceleration post COVID-19. In less than a decade, UPI has become a leading payment system, processing more than 17 billion transactions a month and overall, accounting for 84 per cent and 9 per cent of total digital payment volumes and values, respectively, in 2024-25 (Table 1).

The strong UPI rally is underpinned by its open, technology-agnostic architecture that eases development of applications, user-friendly design, and increasing digital awareness (Aurazo et al. 2024). Growing use of UPI for daily low-value transactions is evident from the rising share of peer-to-merchant (P2M) payments, narrowing ticket size of UPI payments (Chart 4a), and the bulk of the P2M volumes falling within the sub-₹500 value band (Chart 4b).

III.2. State-level Insights into Payment Choice

State-level analysis reveals regional variations shaped by income and structural factors. Due to unavailability of granular data on ATM withdrawals, cash usage is proxied by withdrawals from currency chests, which are regional repositories managed by commercial banks on behalf of the Reserve Bank of India. As all freshly issued notes pass through these chests, their withdrawal patterns are assumed to reflect public cash demand. On average, the share of annual cash withdrawals from ATMs (through debit and credit cards) to cash withdrawals at currency chests stands at 80 per cent in 2024-25.

In the absence of disaggregated UPI data, this study employs data from PhonePe (Pulse), a payment service provider accounting for 58 per cent of total UPI transaction volume and 53 per cent of value (Charts 5 a and b). This open-source dataset has been widely used in studies examining UPI diffusion across states and districts (Dubey and Purnanandam, 2023; Reddy et al., 2024).

Two factors support the generalisability of this dataset as a proxy for overall UPI activity: First, PhonePe’s growth trajectory has closely mirrored overall UPI trends in recent years, with correlations between their growths being 0.99 for both volume and value. Second, PhonePe-based state-wise rankings exhibit a strong correlation with total state-wise UPI rankings in 2024, for which data was available (r = 0.97). To ensure comparability, both cash and UPI indicators are normalised by state population, yielding measures of cash and UPI intensities.

Cash intensity varies widely across states and Union Territories (UTs), with Goa, Delhi, Chandigarh, Arunachal Pradesh, Nagaland, Kerala, and Sikkim recording the highest per capita cash withdrawals (Chart 6), reflecting factors such as tourism and service-led cash usage, remittance inflows, rural areas’ cash dependence, limited digital infrastructure, older demography, and security constraints. Recent trends indicate a broad-based and sustained decline in cash usage across most states over the past few years, suggesting a structural rather than transitory shift.

On the digital front, UPI intensity, proxied by PhonePe transactions, remains high in Telangana, Karnataka, Andhra Pradesh, Delhi and Maharashtra in per capita volume terms, aligning closely with the presence of urban centres, economic hubs and regions with high employment-driven migration (Chart 7a). In contrast, UPI uptake remains modest in several cash-dependent regions such as the North-Eastern states (Tripura, Manipur, Meghalaya, Nagaland). Data from a nationwide survey suggest relatively lower inter-state variation in the ability to use UPI for online banking transactions, with a modest skew towards the southern and northern states (NSO, 2025).³ Notably, Chandigarh, Himachal Pradesh, Kerala, Manipur, and Mizoram exhibit high reported ability to use UPI (Chart 7b).

In terms of growth, most states have witnessed a surge in UPI adoption post pandemic (FY: 2022). Although the overall trajectory of UPI payments remains positive across states, the pace of growth has moderated due to high base effect from the pandemic year and a transition towards a more stable, self-propelling adoption curve.

UPI usage, however, continues to be concentrated, with the top 10 states accounting for nearly 80 per cent of total transaction volumes - a pattern that has remained relatively stable over time. Nevertheless, the trend decline in dispersion of UPI adoption across states is evident from the strengthening of sigma (σ) convergence since 2020, albeit at a gradual pace (Chart 8). This slower convergence may reflect heterogeneity in digital infrastructure, extent of formalisation, financial inclusion and literacy, and merchant acceptance across states.

IV. Data and Methodology

At the national level, an auto-regressive distributed lag (ARDL) model is estimated using quarterly data from Q2:2009 to Q4:2024 to assess UPI’s impact on cash demand in nominal and real terms.⁴ Key determinants include GDP, deposit rates (proxied by major banks’ one year lower bounds), the share of high-denomination notes in circulation⁵ (store-of-value proxy), and UPI transaction volumes (substitutive effect)⁶, thereby accounting for transaction, precautionary, and speculative motives. Controlling for the high denomination notes’ share also helps isolate UPI’s impact on CIC, as high-value transactions may distort trends driven by predominantly small-value UPI payments. The sample period chosen reflects the structural shift following the Payment and Settlement Systems Act (2007) and minimises the global financial crisis’s impact. Except for interest rates, all variables are seasonally adjusted and log-transformed. Stationarity checks using the Augmented Dickey-Fuller (ADF) test confirm that all series are I(0) or I(1), validating the ARDL framework. Key shocks, including withdrawal of specified bank notes in 2016 and COVID-19 lockdowns are captured through quarterly dummies.⁷

Building on the macro-level insights, cash determinants at the state level are analysed using fixed-effects⁸ panel quantile regression for 31 Indian states and UTs over the period Q2:2019 to Q1:2025, at the 25th, 50th, and 75th percentiles of the cash distribution. The model accounts for unobserved state-specific heterogeneity and time effects. The sample period, beginning in 2019, captures the phase during which UPI gained traction. To examine heterogeneity across income groups, separate panel regressions are estimated for low, middle, and high-income states, stratified on the 25th, 50th, and 75th percentiles of net state domestic product (current prices).

As mentioned above, cash demand is measured by quarterly currency chest withdrawals and UPI adoption by PhonePe transaction data. In the absence of quarterly subnational GDP, economic activity is proxied using VIIRS VNP46A2 nighttime lights, which provides daily measurements of artificial (human-generated) illumination at \~500-meter spatial resolution. Quarterly state-level aggregates are computed as the sum of the “Gap Filled DNB BRDF Corrected Nighttime Lights” band, using zonal statistics over state boundaries, thereby eliminating any high-frequency volatility. This data has been widely used to estimate output and growth, especially in data-scarce granular geographical levels, and to better capture informal sector activity (Lahiri, 2020; Beyer et al., 2022; Mathen et al., 2024). Other control variables include ATM density (financial infrastructure), employee provident fund organisation (EPFO) net payroll additions (formalisation), Periodic Labour Force Survey (PLFS)’s educational attainment below higher-secondary level (literacy), and Telecom Regulatory Authority of India’s internet subscriptions (digital infrastructure). All variables, except internet subscribers and education attainment levels, are normalised by state population and log-transformed. Year fixed effects control for broad macroeconomic trends, while intra-year shocks like festivals, state elections, and COVID-19 are captured through quarterly dummies. While these regression estimates do not necessarily imply causality, they provide insights on the magnitude of these factors. Cross-state summary statistics and correlation heatmap are provided in Annex I.

V. Impact of UPI on Cash Demand: Empirical Evidence

V.1. National Level Insights

The UPI volumes are negatively associated with cash demand across models both in nominal and real terms, underscoring its role as a substitute for cash (Table 2). Income (GDP) emerges as the primary determinant of cash demand with elasticities ranging from 0.79 to 0.86, indicating a positive association between economic activity and cash usage. Deposit interest rates exhibit a negative and statistically significant effect, reflecting the opportunity cost of holding cash. Conversely, the higher denomination banknotes share shows a small but positive effect, consistent with its store-of-value role (Model 2). The post-estimation diagnostics confirm the absence of serial autocorrelation and heteroscedasticity at 5 per cent level. The error correction coefficient, which captures the speed at which short-run deviations adjust to the long-run equilibrium, shows that 24-30 per cent of deviations are corrected within a single quarter. Moreover, the Bounds test F-statistic exceeds the upper bound of the critical values, confirming the existence of a long-run relationship between these variables.

Owing to the specified bank note withdrawal, the dummy coefficient for Q4:2016 and Q1:2017 is negative and statistically significant (Annex II). Further, dummy variables for both the first and second waves of the pandemic are positive and statistically significant, suggesting that the increase in currency demand during the lockdown was driven by precautionary and store-of-value motives, consistent with previous findings (Caswell et al., 2020; Chen et al., 2020; Awasthy et al., 2022; RBI, 2023).

V.2. State Level Insights

V.2.1. By Cash Quantiles

Consistent with the aggregate regression, economic activity as proxied by nighttime lights exhibits a strong and statistically significant association with cash usage across all states (Table 3, Model 1). While its influence remains consistently positive across the conditional distribution of cash demand, it marginally attenuates from lower to upper quantiles of cash usage (Models 2 – 4).

UPI volumes per capita display a negative and non-linear association, given the negative linear term coupled with a positive squared term. This indicates that increases in UPI usage substitute for cash, however, beyond an estimated threshold (log UPI per capita = 2.18) and as digital adoption matures, the substitution effect moderates, possibly reflecting saturation or behavioural inertia. Plotting the UPI coefficient across different cash quantiles indicates a stronger substitution effect in upper quantiles, implying that in cash-intensive states, digital adoption exerts a stronger dampening impact on cash usage (Chart 9). This pattern may reflect a combination of higher initial cash dependence, policy and market efforts, and steeper early-stage learning curves in digital adoption. Similar non-linear dynamics are observed for UPI value per capita (Table 1:Annex III).

Internet subscriber base, as a proxy for digital infrastructure, exerts only a weak influence, with borderline significance at the median quantile. The degree of formalisation displays a concave relationship with cash demand. Initial formalisation is associated with lower cash reliance, possibly due to improved access to banking and digital wage payments, which wears off later (post log of degree of formalisation = 5.8). This pattern suggests that informal sector remains more cash-intensive, with lower willingness to adopt digital payments (Ligon et al., 2019), possibly owing to limited integration with formal financial networks (Lahiri, 2020). Further, states with higher proportions of population with at least higher secondary education show lower cash demand at median and upper quantiles, reflecting the positive relationship between education and digital alternatives. Structural shocks, along with policy and seasonal dummies such as COVID-19, state elections, festivals and the marriage season are all positively and significantly associated with spikes in cash demand across the distribution, reaffirming its episodic and precautionary nature in line with Raj et al., (2020).

V.2.2. By Income Groups

Although UPI adoption exhibits a non-linear relationship across income groups, mid-income states display the strongest substitution elasticity, indicating that they are at a critical inflection point in the ongoing digital transition (Table 4). Economic activity is positively associated with cash demand in all income groups, but its magnitude is higher in high-income states. ATM density is associated with higher cash usage only in low-income states than in more affluent ones, underscoring their continued dependence on traditional access points. Formalisation of the workforce is negatively associated with cash usage, though only in mid-income states and that too up to a threshold. Additionally, higher education levels are linked with lower cash demand in low and high income states. Similar results prevail for UPI values per capita (Annex III, Table 2).

VI. Conclusion

The study examines the impact of UPI on cash demand in India. Using a dual empirical strategy of autoregressive distributed lag model and panel quantile regression, the article finds that higher UPI adoption is associated with lower cash demand at both national and subnational levels. At the aggregate level, descriptive trends indicate a structural shift in India’s payment landscape, evident from currency growth moderating from pandemic levels and sustained UPI expansion with narrowing ticket sizes. Empirically, income, proxied by GDP, is positively associated with cash demand, while UPI and interest rates exhibit a negative effect.

At the state-level, preferences between cash and UPI, as proxied by PhonePe transactions, display regional variation. Early UPI adopting states continue to retain a dominant share of total UPI payments, however, a broad-based decline in cash demand across states and narrowing inter-state disparities in UPI adoption since the pandemic point to early signs of convergence. Empirical analysis reveals a negative and non-linear association between UPI adoption and cash demand across cash quantiles. While UPI largely substitutes cash, the effect moderates as digital adoption matures, possibly due to saturation or behavioural inertia. Income, proxied by nighttime lights, and ATM density are positively associated with cash demand, whereas workforce formalisation and higher educational attainment are linked to lower cash reliance. Income-group-wise segregation shows that mid-income states exhibit the strongest substitution elasticity, while lower-income states may unlock untapped substitution potential through improved literacy and greater workforce formalisation.

These findings suggest that a one-size-fits-all approach may not be sufficient for adoption and sustained usage of UPI. Region-specific targeted interventions aligned with each state’s demographic, infrastructural, and behavioural context are likely to be effective. Expanding digital infrastructure and financial literacy interventions, incentivising digital wage transfers, and building trust in digital modes may accelerate cash-to-UPI transition across the spectrum.

References:

Aguilar, A., Frost, J., Guerra, R., Kamin, S., and Tombini, A. (2024). Digital Payments, Informality and Economic Growth. BIS Working Papers No. 1196.

Alvarez, F., and Lippi, F. (2009). Financial Innovation and the Transactions Demand For Cash. Econometrica, 77(2), 363-402.

Ardizzi, G., Nobili, A., and Rocco, G. (2020). A Game Changer in Payment Habits: Evidence from Daily Data during a Pandemic. Bank of Italy Occasional Paper, 591.

Aurazo, J., and Franco, C. (2024). Fast Payment Systems and Financial Inclusion. BIS Quarterly Review, p 41.

Awasthy, S., Misra, R., and 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.

Bachas, P., Gertler, P., Higgins, S., and Seira, E. (2018). Digital Financial Services Go a Long Way: Transaction Costs and Financial Inclusion. AEA Papers and Proceedings Vol. 108, pp. 444-448.

Bailey, A. (2009). Banknotes in Circulation–Still Rising. What Does this Mean for the Future of Cash?. In Speech at the Banknote 2009 Conference, Washington DC (Vol. 6).

Baumol, W. J. (1952). The Transactions Demand for Cash: An Inventory Theoretic Approach. The Quarterly Journal of Economics. 66 (4), 545–556.

Bech, M. L., Faruqui, U., Ougaard, F., and Picillo, C. (2018). Payments are a-changin’ But Cash Still Rules. BIS Quarterly Review, March, 67–80.

Beyer, R. C., Hu, Y., and Yao, J. (2022). Measuring Quarterly Economic Growth from Outer Space. IMF Working Paper 22/109.

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

Cantú, C., Frost, J., Goel, T., and Prenio J. (2024). From Financial Inclusion to Financial Health. BIS Bulletin no. 85.

Caswell, E., Smith, H., Learmonth, D., and Pearce, G. (2020). Cash in the Time of Covid. Bank of England Quarterly Bulletin, Q4.

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.

Chen, H., Engert, W., Huynh, K., Nicholls, G., Nicholson, M., and Zhu, J. (2020). Cash and COVID-19: The Impact of the Pandemic on the Demand for and Use of Cash. Bank of Canada Staff Discussion Paper 2020-6.

Columba, F. (2009). Narrow Money and Transaction Technology: New Disaggregated Evidence. Journal of Economics and Business, 61(4), 312-325.

Dubey, T. S., and Purnanandam, A. (2023). Can Cashless Payments spur Economic Growth?. Available at SSRN, 4373602.

Fisher, I. (1911). The Purchasing Power of Money, its Determination and Relation to Credit, Interest and the Crises. Macmillan.

Friedman, M. (1956). The Quantity Theory of Money: A Restatement. Ch. 1 in Studies in the Quantity Theory of Money, ed. by Milton Friedman (Chicago University Press, 1956), 3-21.

Friedman, B. M. (1999). The Future of Monetary Policy: The Central Bank as an Army with only a Signal Corps?. International Finance, 2(3), 321-338.

Huynh, K. P., Schmidt-Dengler, P., and Stix, H. (2014). The Role of Card Acceptance in the Transaction Demand For Money. Bank of Canada Working Paper No. 14-44.

Keynes, J. M. (1954). The General Theory of Employment, Interest, and Money: By John Maynard Keynes. Macmillan.

Lahiri, A. (2020). The Great Indian Demonetization. Journal of Economic Perspectives, 34(1), 55-74.

Ligon, E., Malick, B., Sheth, K., and Trachtman, C. (2019). What Explains Low Adoption Of Digital Payment Technologies? Evidence From Small-Scale Merchants in Jaipur, India. PloS one, 14(7), e0219450.

Mathen, C. K., Chattopadhyay, S., Sahu, S., and Mukherjee, A. (2024). Which Nighttime Lights Data Better Represent India’s Economic Activities and Regional Inequality?. Asian Development Review, 41(02), 193-217.

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.

National Payments Corporation of India (2020). Digital Payments Adoption in India, 2020. NPCI-PRICE Report.

National Statistics Office. (2025). Comprehensive Modular Survey – Telcom, NSS 80th Round. May 29, 2025.

Oyelami, L. O., and Yinusa, D. O. (2013). Alternative Payment Systems Implication for Currency Demand and Monetary Policy in Developing Economy: A Case Study of Nigeria. International Journal of Humanities and Social Science, 3(20), 253–260.

Raj, J., Bhattacharyya, I., Behera S.R., John, J., and Talwar, B.A. (2020). Modelling and Forecasting Currency Demand in India: A Heterodox Approach. Reserve Bank of India Occasional Papers 41 (1), 1–45.

Reddy, A., Kedia, M., and Shukla, S. Diffusion of Digital Payments in India - Insights based on Data from PhonePe Pulse. Indian Council for Research on International Economic Relations (ICRIER) Working Paper. March 2024.

Reserve Bank of India (RBI). (2023). Annual Report 2022-23.

Reserve Bank of India (RBI). (2024). Report on Currency and Finance – India’s Digital Revolution.

Reserve Bank of India (RBI). (2025). Payments Systems Data. Accessed June 2025.

Tobin, James (1956). “The Interest Elasticity of the Transactions Demand for Cash”. Review of Economics and Statistics. 38 (3), 241–247.

Udupa, G., Bhuyan P., Verma D.K., and Kulkarni N. (2025). Economic Activity and Banknotes: New Approaches. RBI May 2025 Bulletin.

Annex I

Annex II

Annex III

^{^} Authors are from the Department of Economic and Policy Research. Valuable insights provided by Dr. Rajiv Ranjan, former Executive Director, Shri M.M. Ramaiah, and Dr. Rakhe Balachandran are gratefully acknowledged. Authors are grateful to the team from Department of Currency Management, including Shri Sanjeev Prakash, CGM-in-Charge; Pradip Bhuyan, and Baswaraj Patil for making available the currency chest data. Authors are thankful to Shri Gunveer Singh, CGM-in-Charge, Department of Payment and Settlement Systems for facilitating access to region-wise UPI data. The views expressed in this paper are those of authors and do not represent the views of the Reserve Bank of India.

¹ Given anonymity associated with cash-based economic transactions, CIC is taken as a proxy for cash demand, in line with previous RBI studies (Nachane et al., 2013; Chaudhari et al., 2019; Raj et al., 2020)

² Since digital payments are backed by bank deposits, mainly demand deposits, a decline in the CIC-to-demand deposits ratio—holding other factors constant—indicates a shift towards digital modes of transaction, whereas an increase in the ratio reflects a rising preference for cash.

³ These estimates are based on unit level data from National Statistical Survey’s Comprehensive Modular Survey – Telcom, 80th Round released on May 29, 2025. The survey questionnaire includes a specific question posed to individual respondents: “Whether able to perform online banking transactions via devices like computers, or mobile?” The response options are: (i) yes, through UPI only; (ii) yes, through net banking or other means (except UPI) only; (iii) yes, both UPI and other means; and (iv) no.

⁴ The following long-run equation is estimated: ln(CiC_t) = ψ₀ + ψ₁ ln (GDP_t) + ψ₂INT_t + ψ₃ HDN_t + ψ₄ ln(1 + UPI_t) + μ_t; where ψ_k are long-run coefficients.

⁵ High denomination notes include ₹500, ₹1000 (before their withdrawal) and ₹2000 notes.

⁶ Since UPI data is unavailable for the period before 2016, log (1 + actual UPI transactions) is used as the variable to ensure continuity. This variable remains constant for pre-2016 quarters, thereby not affecting the estimation.

⁷ A dummy for the ₹2000 note withdrawal in May 2023 was initially included but found insignificant and thus, excluded from the final model. The effect may have been subsumed by the share of high-denomination notes variable, which likely accounts for its explanatory power in the main regression.

⁸ Hausman Test validates the use of fixed effects model over random effects.