- Wealth PMS (50L+)
RBI has released a discussion paper that says: We’ve let you good people live all this time with “free” payment systems, so should we allow banks to start charging now? Specifically for UPI, which has reached volumes of 10 lakh crore rupees per month? And should we charge merchants?
My answer is a very big NO. But I’ll put my case more eloquently because size matters no matter what they tell you in the movies. Also, Context matters.
We give banks our money so we can withdraw it when we like. If they add friction in terms of cost or time to our ability to do that, we will remove our money from the banking system and transact in cash. The government hates money going out of the banking system because it generates money on which tax is not paid, i.e. black money, and the RBI hates it because it takes away money from the banking system that could be lent to people and businesses.
To incentivise people to keep money in banks, allow them to transact in a free manner. Banks earn money from this same money we give them in return for being able to pay: it’s called “float“. Float is the money in savings bank accounts or current accounts, where banks pay us very little interest compared to what they can earn on it. “Spread” is what they earn from our money that sits in a bank minus what they pay us. That spread is huge and at the very least, is ₹ 85,000 cr. per year and probably 100,000 cr.+ in reality for the banking system. This is way more than they need for any cost they incur for maintaining UPI (or indeed any electronic system).
The government has also made it illegal to transact relatively small amounts in cash (even more than 20K). If that’s to work, then payments must be free at least at the retail end, so that people don’t try to go to cash for fear of having to pay a transaction fee when they need to pay someone. Small merchants, tiny shops and the like, will not even come on board given their low margins, and go back to cash (even if it’s illegal). People will refuse to pay if their bank charges a fee for UPI.
Most importantly, UPI’s cost is next to nothing. Banks are paid in two ways: currently, the government subsidises them with 1,300 cr. per year, enough to meet costs and more. But banks get much larger amounts by just keeping a larger amount of float money in the banking system. The difference is disproportionate – the cost of a UPI system to the core company that manages the transaction flow and settlement (NPCI) is 400 cr. a year at the very max. The amounts earned in float are close to 100,000 cr., and then, banks get Rs. 1,300 cr. from the government. Essentially, UPI is paid for.
Oh, wait, won’t we stifle innovation if we make it free? This is laughable because:
If we innovated when it was free by an act of Parliament (meaning, not temporary), then the argument that innovation won’t work unless it’s paid falls flat.
Shouldn’t UPI be like a toll road? Pay for it even if it’s public infra? In India, very few roads are tolled. And the more popular a road gets, the more likely that the courts and government will make it free. This is also why city roads aren’t tolled – it is friction for people to use it, and they need to be used seamlessly if the city has to carry on as a good business center. UPI is the biggest and most used payment mechanism: introducing friction with a fee will just slow it down, and that’s not good for business in India.
What about players like GPay and PhonePe, through which platforms most UPI transactions happen? My answer is: what about them anyhow? They continued to operate when the government mandated it to be free. They have business models in advertising to earn money from the customers. They now charge a “platform fee” for some transactions where the merchant wants them to pay. They have a BBPS model that allows them to make money from bill payments. There’s lending, there’s the great customer base they can monetize. UPI is a gateway like free search is a gateway. The point is: they aren’t dying, and they’re in this knowing that UPI is free. If it remains free and they shut shop, others will quickly take their place. It is not necessary to change the rules to help them.
They got customers by giving cashbacks – paying customers to use their product. Others without deep pockets or who couldn’t lose money didn’t compete. If they leave the space, we’ll see more level-headed competition without cash-back-based artificial market creation.
India has a lot of banks. Banks keep the money of the public: individuals, corporates and institutions. This money is available to them for lending, which in turn creates more deposits in the banking system, and so on. The power of this system depends on the confidence in the banking system, because if everyone withdrew their deposits, banks will go belly up. One part of that confidence is that if I keep money in a bank, I should be able to take it out as cash when I like and that I should be able to use that to pay someone else when I like.
The traditional method of paying someone was:
Now the first part involved going to a bank, giving the teller a “self” cheque and taking cash out. This is expensive, it involves people in the branch, and then you can only draw money when the branch is open. The capitalist in you is thinking: Why can’t we charge a customer for this?
If you charge a customer to take her own money out of the bank, she will refuse to put any money into the bank. She will demand her salary in cash, and pay her expenses in cash and save the rest in her house. The bank will have nothing left to lend. This is a disaster. So you make withdrawals of cash free of charge.
Remember this: Banks fear that people withdraw their money in cash and go outside the banking system, reducing what they have available to lend.
To reduce expenses, banks automated cash withdrawals by installing “ATM” machines. Again, in India, banks said, “can we charge for this?” because that’s a question that’s forever in their heads. This question was initially answered with “Yes”. But RBI, over the years, discovered that banks were charging random fees when customers withdrew their money at other banks’ ATMs, or sometimes even at their own.
They didn’t like it. So, in 2008, the RBI gave an order that made all ATM transactions entirely free, including at other banks, from 2009. In their approach paper, they said:
Since then, RBI has reduced the number of “free” transactions to five (at the bank’s own ATMs) or three (at other bank ATMs) for the most part. This has, however, come about because now ATMs are not the only way to transact – you have other digital modes to transfer money to someone else.
You can draw a cheque on your account at any time and withdraw money or transfer it to someone else’s account. Local cheques have been free for retail customers for nearly as long as I remember, but even this went through an evolution process. If you gave your bank a cheque from an account you held inside the same bank, it was quite easy and free. If it was from another bank in the same location (“local” cheque), you had some delay while they sent the cheque for verification to the other bank, who would then transfer the money to your bank. Then there were outstation cheques, where banks could charge a fee. And they did.
In 2008, RBI was furious that banks were charging between 0.4% to 1% of the cheque value for outstation cheques. They brought it down to between Rs. 50 and Rs. 150 per cheque
Over time, cheque fees were regulated downwards. A “Cheque Truncation System” (CTS) was created so that banks didn’t need to courier cheques to each other. They just uploaded a scanned copy to the CTS, which is again run by NPCI now. CTS would reduce costs substantially, and the three large grids in the country (Delhi, Mumbai, Chennai) cover most of the country.
The RBI has mandated that banks cannot charge outstation or collection charges for the most part, now that banks are on CTS. There is a charge per cheque that banks pay to CTS (roughly Rs. 0.50 to Rs. 1.00 per cheque) but banks absorb this fee.
Why? Because if you charge for cheques, people will withdraw their money in cash and start using cash instead. This fear is why payments are regularly made free for customers. Even merchants pay only a small overall handling charge for very high volume of cheques.
RBI wanted less paper – cash and cheques are all paper and expensive to maintain. Electronic transactions were a pain because banks hated to talk to each other anyhow, and if you ask them to talk to each other electronically, they’d probably bust a blood vessel. So RBI went ahead and developed the system of electronic fund transfer (EFT, later National EFT or NEFT). Another faster system called Real Time Gross Settlement (RTGS) was also created by the RBI.
NEFT and RTGS were both initially chargeable (at Rs. 5 to Rs. 50 per transaction). RBI, in 2019, decided that it will not charge the banks anything to use the RTGS/NEFT frameworks. (Link) And asked, nicely, that banks should therefore pass on the love to their customers also.
In six months, RBI was furious that banks weren’t passing on the “free” charges to customers. So they mandated all NEFT transactions would be free for savings bank customers if made online. (Link)
RTGS is only for transactions above Rs. 200,000. Most banks don’t charge for RTGS, and in any case, charges are limited to Rs. 50 per transaction.
Both RTGS and NEFT operated 24 hours a day, 7 days a week. They’re truly online systems and can be initiated at any time. Banks do not get paid to offer these for free but are mandated to do free or low-cost customer fees for all such transactions.
There’s a question now about whether this should be free at all to customers. RBI does incur a cost to maintain these systems themselves and it is in effect a subsidy. Should they recover this? Depends on what it costs – and honestly, it looks like it’s so tiny it doesn’t even show up on the RBI Balance sheet.
A quick look at assets shows only one item that might have all the technology cost they would have incurred (“Other assets”), and the numbers in there add up to a max of Rs. 2,000 cr. (which would be over years, not just now)
If they expense it out, it’s again too small to even be shown anywhere – at best, it is “Miscellaneous expenses” which is Rs. 1000 cr. a year for all of their tech operations (which is far more than just NEFT/RTGS of course).
Given that RBI spends Rs. 4984 cr. on printing currency notes, it seems like the cost that RBI bears to do electronic transfers is less than half of that. At 1,000 cr. this RBI doesn’t need to “recover” this cost, imposing it on the banking system which then will gleefully charge customers even higher fees and, in all, create a mess that encourages people to just go to cash instead.
For payments, you could use debit and credit cards. They’re free for you, but merchants pay. Merchants will pay up to 2% per transaction, though for debit cards, it might be as low as 0.4%. This means merchants must be capable of taking cards (have a card machine) and be willing to pay this extra fee.
This is expensive and silly for most merchants. If your margins are like 5% to 8% or so, for most small stores, you can’t afford to pay 1/5th of that for a transaction charge. Plus, there are machine costs and then settlement pains. You get money after X days, and the lack of visibility is horrible for smaller merchants. Many merchants simply add the transaction charge to their bills (“For credit card, 2% extra”) and push customers to use cash or UPI.
There are 92 cr. debit cards in the country, and around 8 cr. credit cards. And for now, there are only 65 lakh POS terminals that can accept these cards. The penetration is next to nothing compared to UPI.
UPI is interesting: it is built on the backbone of IMPS, an instant payment service started by NPCI. At Capitalmind, we wrote about this in 2011 (!) where you could transfer money instantly to someone else. This is fantastic, but it wasn’t made popular until UPI came around.
For IMPS, you still needed a web-based browser, the account number and the IFSC code of the person you needed to pay. With UPI, you just needed a QR code, and it’s super convenient. The cost of UPI – for both merchants and for customers – is completely free. There is no charge at all to use UPI, and this is mandated in law. This has resulted in an absolutely skyrocketing change in UPI usage.
The amount of money in “cheap” savings accounts (pays around 3.5% to 4% a year) or current accounts (mostly pay no interest) is the float for a bank. This money earns much more: currently, even overnight parking of money in RBI earns 5.15%, but banks have way better avenues to lend and earn even more on it.
Don’t confuse this with term fixed deposits – this is longer term money, and let’s assume that spread is a “pure banking” spread (earns the bank money on lending at a higher rate anyhow, and compares to an NBFC)
Here’s how much money banks make on float (current plus savings accounts) right now. Compare this to what NPCI spends on all the technology it maintains (UPI+everything else)
Float generates nearly 140,000 crore rupees of income for banks. NPCI which runs the settlement infrastructure for UPI, IMPS, Cheque truncation, ATM financial switch etc. is only spending 680 cr. per year. Banks get back a whopping 1,300 cr. per year from the government. There is adequate return for float.
This sounds huge, and it is. This float income is only available to banks. Even wallets cannot earn float income (they have to keep the wallet amounts in an escrow bank account and can’t earn interest on it). NBFCs don’t have such accounts or float income at all. Given this, the float income is primarily used to pay for the one thing they get the float for: payments and transfers.
If you look at the technology expenditure of banks as a whole, it’s still a tiny expense compared to the float income generated.
And CASA has gone up – as a percentage of total deposits, CASA is now a much higher value:
In banking, cross-subsidy is a feature, not a bug. You get the privilege of float, for the cost of easy payments. The easier you make payments, the less likely it is that we use cash to transact.
Usage of cash in transactions has reduced to a large extent. If you look at ATM transactions using debit cards, the growth in transactions is slowing. In fact, we only see as much cash being transacted in ATMs as two years back, whereas growth was higher, in line with broad money supply growth earlier.
The difference is so large that I had to use a separate scale for UPI (otherwise cash would look tiny).
From this it seems true that UPI’s rise has resulted in lower ATM transactions for cash. A lower cash transaction volume reduces costs for the banking system (cash handling charges, ATM transaction fees between banks which may be ₹ 7-15 per transaction, transportation, soiled notes etc.)
Note: People often say that total cash in the system hasn’t come down. That is okay because cash is used to keep or hoard money. This has illegitimate uses (black money generated by corrupt government officials and private people). But it also has legitimate uses, such as a woman who is harassed by her husband or family, squirrelling away money for potential future freedom. This is not money which is transacted often – and does not pose a large threat to the financial system. The fear is if you use currency to transact regularly, you bypass the banking system, which then struggles to raise money to lend etc. Transactional currency tends to be what people draw from ATMs, and if that’s not growing, it’s a good thing.
If you remove P2P payments, that is the bulk of the system – nearly 8.5 lakh crores per month goes between individuals. Can’t we have merchants pay, either? It’s a commercial activity, so won’t they pay? Well, the answer is still likely to be “no”.
Small merchants – those who take less than Rs. 2000 per transaction – simply don’t have the margins to sustain a charge. For a person making 5% margins in a kirana shop, a cost of 0.5% on a payment mechanism hurts them immensely (it’s 10% of their margin!) They will simply refuse and go back to cash.
Will larger merchants pay? Some will, and perhaps do even now, just to get their settlement data correctly. They might pay as part of a payment management system, which providers can charge for. But to do this for all merchants, even for relatively lower volume merchants, is counter productive, because like they do with credit cards, they might just add a “transaction charge” for UPI, and that will reduce volumes even for the “high end” merchants that do more than ₹ 2000 per transaction.
The downside of charging some merchants versus others based on the size of transactions or volumes is that you disincentivize disincentivises. If I’m a “free UPI” merchant and doing a new 21st transaction in a month, or more than Rs. 2000 in transaction size, causes me to pay a fee, I’ll just start taking cash from the 20th transaction onwards. This is a lousy mechanism to grow payments, and we’ve seen growth rock in the free world with adequate compensation.
It does. But even those costs have been coming down substantially. NPCI is a not-for-profit and earns nearly 40% post-tax margins. (1200 cr. revenue, 400 cr. net profit). This has allowed them to keep reducing the cost that they charge banks. Oh, and even the UPI software is built off an open-source stack, so they’ve incurred a lower cost too.
In 2020, the government mandated that there must be absolutely no charge – to the merchant, a sender or a receiver of money – if they used UPI. This is now part of an act of Parliament, the Payments and Settlements Act. and not something temporary.
The government also stated that UPI is a public digital good and that it will be free. This is comforting, but their basis is also that if we force people to use digital payments, we can’t also impose a large cost on them that hurts them. Especially merchants.
Note: The RBI cannot allow UPI to be charged, regardless of this paper, because it requires an act of Parliament to be changed.
This is utter nonsense. Nearly all the innovation in the technology world has happened on the internet protocols, which are free to use. When data was charged on a kilobyte basis, the internet wasn’t used quite as much as now when you have large bandwidth that you pay a subscription fee for. Google built a search engine for free and sold advertisements. Their payments arm in India will do that, too – even if they can’t use the data, they can sell ads that are shown to customers.
There’s been innovation in UPI through the time it was free. You can invest in IPOs through UPI. Loans can be initiated and repaid through UPI Pay and Collect. UPI can be used to invest in mutual funds, divide restaurant bills between friends, even to buy government bonds through RBI. None of this required UPI to be paid for anyone. Why should payment be a barrier to innovation? It hasn’t been in the tech world and shouldn’t ever be.
Too often, free is considered bad. But in public infrastructure, especially digital financial infrastructure, free is the right way. The idea should be to promote more and more people into digital payments, which allows for far greater things to happen on top of this infra. There’s no scarce resource in financial transactions that we need to tell people to use less of (like electricity, water or such) – in fact, the more people transact, the better it is for the economy.
Keeping UPI free is what will make India a far better place for payments in the world. We’re already the cheapest and fastest mechanism for payments, and our float income for banks is so high that they can invest the tiny amounts needed, in comparison with that float income, to keep UPI free.
We have a weekly newsletter where we talk about investing and finance. It goes out to 35000+ readers already. Sign up here!
Invest with us!
Capitalmind Wealth PMS is India’s lowest-cost Portfolio Management Service for investors who deploy INR 50 Lakhs and more. Already managing 1000+ crore across 700+ clients!