A New Model for UPI

For the longest time, questions have been raised about the business model for UPI and how the industry should charge for its services. The government has prohibited MDR on UPI transactions, but it should still be possible for us to develop a new revenue model.

This article was first published in The Mint. You can read the original at this link, or, if you prefer, listen to me read the article by clicking play below.


It is hard to argue with the success of India’s Unified Payments Interface (UPI). With over 12 billion transactions processed each month, it is already, by an order of magnitude, the largest digital payment system by volume in the world. Given its consistently high rate of growth, National Payments Corporation of India’s (NPCI) stated ambition of crossing 1 billion daily transactions is looking more achievable with each passing month. Despite these truly impressive statistics, however, concern continues to be raised about the long-term viability of its business model.

The MDR Expectation

When it was first launched, most people thought that Third Party Application Providers (TPAPs), the entities we use to access the digital payments ecosystem (Google Pay, PhonePe and the like), would simply levy a fee on UPI transactions akin to the merchant discount rate (MDR) that credit card companies charge. However, in 2019, the government prohibited banks and other service providers from imposing charges of any sort on UPI payments—effectively preventing TPAPs from using UPI-transaction fees as a revenue model.

MDR is charged as a percentage of the value of the transaction. For large-value transactions, this is a cost that merchants find relatively easy to bear, particularly considering that customers who aren’t carrying that much cash on them may not make the purchase otherwise.

For smaller transactions, where margins tend to be thin, merchants often refuse credit card payments because the MDR eats into their profits. It is to address this problem—to ensure that the benefits of the digital payments ecosystem extend to small-ticket transactions—that the government kept UPI free of such a charge.

While none of this has deterred players from continuing to invest in UPI, it has placed a strain on their profitability. Today, the annual losses posted by some of the largest companies in this space are in excess of ₹2,500 crore a year. Unless we can find them a better way to meet their costs, I worry that there will come a time when the system becomes unsustainable.

Proportionate Fees

Credit card companies charge an MDR to help offset the costs they have to incur in guaranteeing payments. Merchants only accept credit card payments because of the assurance that, regardless of whether or not the purchaser pays his credit card bill, the card network will always ensure they get paid. This shifts the risk of customer fraud onto the card network, which, partly in order to offset it, charges a percentage of the transaction’s value as its fee.

While this might make sense in the context of credit cards, it doesn’t translate to UPI in quite the same way in our current context. UPI has been designed differently from credit cards, and, as a result, the same rationale doesn’t apply. UPI transactions are processed in real time, with the bank account of the transferor being debited at the same time as—and by the same amount that—the account of the receiver is credited. As a result, TPAPs do not worry about payment risk in the same way that card companies do.

That being the case, charging fees as a proportion of the value of a UPI transaction is not warranted, since their risk does not vary proportionately with the value of the purchase. The UPI ecosystem just needs to earn enough revenue to meet the cost of running the UPI system, plus a reasonable profit on top of that. Surely, we can find a way to do this without placing a disproportionate burden on the customer.

A Flat Fee

So, what might a good solution look like? Let’s say we charge all customers a flat annual fee of, say, ₹120 for the convenience of using UPI. This is a relatively small amount (₹10 per month), that should be affordable by even those of modest means. For context, the average monthly revenue per user of telecom companies is in the range of ₹100, which suggests that everyone who uses UPI today can afford to pay at least that much a month.

If we collect this amount from the entire UPI user base—roughly 500 million customers today—we should be able to generate a revenue pool of about ₹6,000 crore every year. Shared among four participants in a UPI transaction, this would meet the cost of running the UPI system, and allow them to make a reasonable profit.

Let’s test this hypothesis. At its current scale, each API call in the UPI ecosystem costs roughly 10 paise. Across the four parties in each transaction, this means the cost of a transaction is roughly 40 paise. At 12 billion transactions a month, this would place the expense of participating in the UPI ecosystem at ₹5,760 crore per annum. Which is well within the ₹6,000 crore revenue pool that we can raise if every UPI user were to pay a flat usage fee of ₹120 per year.

Legislative Amendment

One challenge still remains. The prohibition on charging for UPI payments has been framed in such sweeping terms that even an annual fee like this may not be permissible under the law. However, given that the primary reason for imposing this prohibition was to ensure that low-value transactions remain viable, it would be consistent with that objective to make an exception and permit a modest annual fee, as described above. Not only will this put the entire UPI ecosystem back on a path to profitability, it would do so in a way that does not harm the societal objective of enabling deeper penetration of digital payments.

For the sake of all that we have achieved on this front so far, I hope the government considers them both to be equally important regulatory objectives.