Cashless Finance: How Universal Payments Interface (UPI) will turbo-charge consumer & small business finance in India

Starting about now (late August 2016), 19 banks in India are rolling out mobile apps for UPI, or Universal Payments Interface. Starting with an idea from Nandan Nilekeni, the RBI governor Raghuram Rajan, and the National Payments Corporation of India (NPCI) led by A.P. Hota have really put their weight behind UPI as a backbone for future payments in India. UPI along w/ Aadhar, automated KYC etc have been dubbed by Nandan as a "Whatsapp" moment for finance in India. Whatsapp took over from SMS because it moved away from the old generation (voice channel) to the data / internet channel, and it was seamlessly integrated into your phone book; and had a real-time nifty user experience. That's what's coming in payments with UPI!

Universal Payments Interface (UPI): A Quick Primer

UPI is a brilliantly simple arrangement for payments between bank accounts. It initially builds off a very popular money transfer method between banks, Immediate Payment Service (IMPS). Consider a payor (Ananya or A) banking with ICICI bank making a payment of Rs. 1000 to payee (Bhavesh or B) with HDFC bank. In IMPS, the payee (Bhavesh) shares his account credentials (Name, Account Number, HDFC branch IFSC code) to payor (Ananya). Ananya registers Bhavesh's credentials as a payee on her HDFC net banking or mobile banking app (which involves a one-time pin on mobile for verification). Once registered, Ananya can initiate a payment for Rs. 1000 to Bhavesh, and the money is transferred instantlyto his HDFC account. There is a aggregate limit of Rs. 2 lakh / day (i.e. Rs 200,000 or ~USD 3075/day) for all her IMPS transfers. This is a non-trivial number especially in India. IMPS has zoomed in popularity: launched in 2010, it has (as of July 2016) 171 member banks, 3.59 crore (i.e. 35.9 million) monthly transactions, Rs. 26662 crores/month amounts ( > USD$4B/month) and growing strongly.

UPI: "Push" Model (Payor to Payee)

UPI builds initially off IMPS and makes it even simpler: Ananya and Bhavesh don't need to exchange hard-to-remember credentials: she just needs his phone number or aadhar number or virtual payment address or VPA (eg: bhavesh@hdfc).

Behind the scenes, NPCI resolves the phone number or aadhar or virtual payment address (w/ the payee bank if needed) to get the IMPS credentials, and invokes the already proven IMPS system to make the instant payment. Thats it! This is called "push" model of payment. The initial limit on UPI transactions is Rs. 1 lakh/day - but is likely to get relaxed to the IMPS Rs. 2L/day over time as the banks get comfortable with the system.

UPI: "Pull" Model (Payee claims from Payor)

Another cool feature of UPI is its support for the "pull" model. Instead of Ananya "push"ing the payments and asking for Bhavesh's information, Bhavesh can send a "COLLECT" request to Ananya (i.e. a "pull" model). The request will come to Ananya's mobile via her bank's app, and she can click accept, and voila, its done. This can be used to settle bills after a restaurant meal, or from a kirana store app (which can separately forward the bill), or mobile phone or utility bill pay. The payor's response to collect can also be to defer (eg: pay the bill 2 days before it is due etc).

One of the nifty applications of "COLLECT" is that the channel for payment is the same as the channel for billing. Earlier you would get a letter (eg: LIC premium or utility pay or IRCTC ticket payment) or SMS or email asking for payment. You would then have to go somewhere else and go through a multiple set of steps to pay (eg: utility bill is a classic example). Now on mobile, it is integrated, and potentially a single click (subject to 2-factor authentication - see below). You could also schedule the payment to be 1-2 days before due date and dispense with it when you see the bill / COLLECT request.

UPI: Performance and Security

Today the success ratios on some payment modes (eg: credit card POS machines, or online gateways) is between 70-80%. UPI can raise this success ratio to 90-95%, and importantly shorten the end-to-end payment time (i.e. it is real time).

A test of UPI's real time & success ratio capability could be in toll payments on tolled highways where a string of QR codes could be placed along a queue and by the time you hit the toll counter, you have paid, and you can move on! Similarly ANY physical queue for payments (eg: in MacDonald's or PVR cinemas) you could probably use your app or the merchant's app to pull in the "menu" (eg: movies or food), make selections while in the physical queue, and pay via your mobile. You can then get a waiting number/receipt/e-ticket and go directly to pick up your food or go to your movie. This is an example of mobile/digital experience intersecting w/ a physical queuing experience.

RBI requires 2-factor authentication for both the push / pull models. The fact that you have your phone in your possession (with its registered number associated with the bank app) is the first factor. The second factor is usually a mobile PIN (MPIN) or biometric authentication (eg: fingerprint check) is required. Newer phones (eg: Lyf) and bank apps (eg: IndusInd) are integrating fingerprint checking as in Apple's iPhone so that to the user it appears as a single-click to pay experience. Note that even though formally it is a 2-factor authentication, the user perceives it as a single click!

NPCI provides an Secure SDK for the UPI apps, certifies apps,  translates global IDs (eg: phone numbers, aadhar) to credentials, passes virtual payment addresses between banks to resolve to credentials, and invokes the core transaction layer (IMPS) between banks. The mobile app, and customer relationships are done by the banks. An ICICI app can be used to invoke other verified accounts (eg: Citibank etc) for payments.

UPI: Paying the Bottom of the Pyramid

UPI payments are not just for high end urban consumers. Payments can be also made to Rupay debit cards or aadhar numbers - and the money will be transferred to the bank account associated with them. You can pay your Dhobhi or servant maid or driver or plumber by paying to their Aadhar card number, and they will get an SMS notification of the payment (as required by IMPS to payee) even if they don't have a cell phone. Another security feature is that the VPA can also be a one-time VPA generated and used just for the transaction on a site. This is another way in which trust is shifted back to the consumer (payor) and her/his banking application rather than being forced to enter what we consider sensitive credentials (CC numbers, CVV) onto a merchant website we dont know if it will be hacked or not.

UPI does require at least one party to have a mobile smartphone, but can enable the smartphone to replace a point-of-sale (POS) machine.Innovators are extending the reach of UPI to ordinary mobilephones and feature phones using SMS, USSD, and offline innovations (eg: offline UPI & sync later, virtual address displays, QR codes - see below). It is also one way transfer (IMPS based) and does not support advanced concepts like refunds, escrow accounts, multi-party transactions etc. These concepts can be potentially overlaid by bank and payment service provider (PSP) innovation.The penetration of smartphones is around 20% (usual inflection point), and growing rapidly: UPI is clearly a bet on the future adoption of smartphones and penetration of data services (GPRS, 3G, 4G). But at the same time, unlike 3G / 4G, it is available instantly across the country. The broader flow is depicted below:

UPI: Innovation by Startups and an Apple-Pay Like Experience

Startups like Ultracash and Jeb (from Vsoft) promise to make the payment experience even simpler.  Ultracash promises Apple Pay-like proximity payments by bringing a payor smartphone close to the payee smartphone (and transfers the secure credentials via audio signals): no data connectivity, NFC chip etc is required. Jeb promises payments to folks like auto-wallahs or vegetable vendors / kirana stores who may not have smartphones themselves, but can display a virtual address  (eg: ramesh-auto@hdfc), or visually via a QR Code (see picture below). No need for POS swipe machines or cards! { On a lighter note: Who knows, maybe even beggars on streets may carry UPI QR codes to direct their receipts to bank accounts! }

UPI payments at a vegetable vendor via a QR code scan (vendor does not need a smart phone, only payer does). Image Credit: Vsoft Inc (Jeb app)

Similarly the Flipkart, Amazon, Snapdeal like e-commerce sites which do cash-on-demand (COD) can switch over to UPI. The smartphone has effectively replaced the POS machine. In India, you can soon pay your driver or maid's monthly salary via UPI in a similar way or to their Aadhar/Rupay accounts. Banks need to accelerate production of such QR codes in addition to RuPay cards to allow peer-to-peer transfers.

How is UPI different from Mobile Wallets, Payment Banks ?

In a mobile wallet you transfer money from you bank into a mobile wallet (eg: PayTM, MobiQwik, Oxigen, Citrus Pay, mRupee, Freecharge, OlaMoney). You can then spend it on other online mobile apps (eg: Uber, Meru taxis, movie tickets, e-commerce sites), or increasingly at offline locations (eg: petrol bunks). For a lay user, once you move money into the wallet, you cannot take it back, and you dont get any interest. To get you to spend / use the balance and develop usage habits, PayTM run incentive programs such as cashback etc. They excel in integration with other mobile / online apps, deep partnerships, and increasingly blur the line between payments and commerce.

Payment gateways like Razorpay are working to create a merchant network by giving them a suite of value-added UPI services (COLLECT, refunds etc). Similarly mobile wallets are starting to decouple the float management (i.e. cash in the wallet) from the customer experience & deep app integration they provide and link those aspect to UPI (via COLLECT). In other words, you could use the same mobile wallet experience and run the transaction through UPI, and the wallet will view UPI as another payment channel or funding account for the transaction that they enable.

UPI is direct inter-bank payment - you can still get interest on your bank balance, and allows banks to potentially catch up with more innovation on their app platforms. Wallets can also see simplified inputs of funds into wallets. Currently UPI does not support mobile wallets (it uses IMPS which is an inter-bank transfer mechanism). But companies like PayTM have acquired a Payment bank license which will blur the boundary from a user perspective. The float in payment bank accounts will receive interest for users, and can be integrated with wallets, and interbank transfers via UPI and IMPS are possible via payment banks. Other payment bank licenses include Airtel (M Commerce), Aditya Birla Nuvo, Vodafone M Pesa, Reliance Industries, FINO, Dept of Posts etc. Payment banks are not allowed to lend (like classical banks): so they will have to drive up e-commerce, and partner with other banks to drive finance (eg: EMI) and other offers to their (presumably) delighted customers. Also once they have a payment bank business they can have their flagship mobile app (eg: PayTM app) integrate other bank accounts for UPI purposes, and focus on the differentiated customer experience and monetizing those eyeballs via referral fees.

UPI and Cashless Finance:

Payments itself is huge, and the brilliant minimalistic design of UPI with scope for remarkable innovation will be a revolution in itself over the next few years. But I believe that payments will be a highly visible tip of the UPI iceberg. A huge revolution awaits when you can embed UPI as the payment or transaction modality to create new financial products, and reduce transaction costs in existing products. Lets consider a few examples:

UPI & Micro-finance / SME Loans / Consumer Finance:

Today micro-finance involves a loan to a self help group and a high transaction cost model of the MFI (micro finance institution) folks going personally and collecting payments. With UPI, this "pull" transaction can be done via a COLLECT request. Alternatively, if the MFI also has a payment bank into which the loan recipients deposit their cash (or receive their inbound payments via UPI), a rule can be written to drive a part of the income to pay back the loan, i.e. the loan repayments can be customized to dynamic cash flows. Interest rates can be dynamic as well, i.e. repay sooner or regularly and have lower interest rates (or have higher interest rates for delays, upto a limit which would trigger MFI personnel visits). Importantly, this transaction history builds up credit history and trustworthiness. Today the spread between cost of funds and market interest rates for MFI recipients is large, and in part driven by high transaction costs of servicing, recovery of loans. Reduction of this transaction costs, and building of credit histories will drive lower average interest rates, and more finance availability. 

More interestingly, the microfinance or SME loan (eg: for a kirana shop) itself can be transformed into a "UPI-only" or "cashless-only" loan, i.e. it is a line-of-credit (LoC)given by the bank or MFI, which can be used only via UPI transactions. This will enable tracking of the transactions (and in the future sharing of the network of transactions, and actors) which can be analyzed for credit histories in a similar way that Google analyzes web graphs for importance of web pages. Again the security of such cashless-only money, and the ability to track a network of transactions could lead to more loan volumes from the bank perspective. A huge number of small businesses can benefit from working capital loans.

The same concept can be used for on-the-spot EMIs for consumer transactions.Lets say you are visiting an Electronics store and bargaining on the price of an air-conditioner worth about Rs. 35,000. As a dimension of negotiation, you or the store manager can use a bank aggregator service like BankBazaar and multiple banks can compete for your EMI business for the high-value transaction on the spot in real-time. This is like an "instant" credit card but only for the transaction before the transaction is done.The payment transaction is processed via the bank aggregator app, and the EMI is then scheduled (as a recurring payment) via UPI. Currently UPI does not natively support recurring payments, but this has to be overlaid atop the basic function via bank aggregator app. 

Escrow accounts, Multi-party transactions, Asset Financing:

Escrow payments have been pioneered by Alipay in China. Money from the payor is held in a trusted escrow account till there is proof of delivery at the payee / delivery location. Then the money is released to the payee.

Lets say we take a more interesting scenario of a residential solar PV escrow account. I have covered rooftop solar in a three part series (Part 1, Part 2, and Part 3 covering policy issues ), and related solar financing issues as well in three parts (Part 1, Part 2, Part 3). A third party installs a solar PV system at a residential home, and it is a zero-down, Solar-as-a-service offering or a hybrid finance (i.e. long term loan). There is a contract with the local utility, say BESCOM, who makes monthly payments based upon the actual production of solar PV. The monthly payment from BESCOM is now sent via UPI to a escrow account, and there is a business rule to split the payments appropriately between asset owner and the residential party.

Going back to the air-conditioner retail purchase. We can now imagine a multi-party transaction in real time involving the a/c manufacturer (eg: Samsung), retailer (eg: Tata Chroma) and the customer and a NBFC financier (eg: Bajaj Finance) or bank aggregator (eg: Bank Bazaar). A custom deal can be put together on the spot for the customer based upon the interests and offers from each party, and the execution of the payments, and subsequent EMI flows to multiple parties done via a set of UPI transactions. The value of UPI again is to minimize credential transfer, and be programmatically composable by applications. The meta data or shared state of such multi-party transactions can be stored on a blockchain (more on that in future articles).

In another article I have covered how companies like NestAway, Coho and others are working with intermediaries like RentSher etc to drive asset-leasing (eg: procuring the air conditioner as-a-service) and asset financing to drive Asset-as-a-service offerings (easy home leases via mobile). Similarly Uber, Ola! can work w/ car companies and financiers (eg: Mahindra & Mahindra and their Mahindra Finance arm) to offer car leasing to qualified drivers. The actualinterest rates charged to the drivers could be a function of driving behaviors, km-driven, and monetized km, and real-time visibility given to drivers. Similarly the insurance premiums could be a dynamic function of safe driving behaviors, number of continuous hours driven, and stress levels of roads in which the vehicle operated. There could also be a revenue sharing embedded with the leasing company if it is a pure lease. At high volumes, the financing / payment streams in these cases again will be cleared via UPI going forward. The application itself that embeds and composes UPI is done via software, and involves multiple parties.

Another multi-party transaction is 3-way receivable financing.  Lets say a small vegetable aggregator startup supplies Reliance Fresh or Metro. Metro may pay the supplier after a week. A financier could come in between to finance the small business and get paid by the credit worthy Metro. This 3-way transaction can be organized via a combination of blockchain and cashless finance/UPI.

In summary, we have at the very beginning and not yet seen the tip of the iceberg of cashless payments. There is an entire iceberg below where UPI and cashless payment transactions are embedded and composed into sophisticated financial products that touch consumers and small businesses.

Twitter: @shivkuma_k

ps: One of the questions below was, since there are bank-to-bank transfers in iMobile apps already, what is the value add of UPI?

A key novelty is "convenience" and "simplicity". Let's illustrate in a few ways: -Today we cannot pay a grocer or kirana store from our phones, since it is not simple enough. Similarly we cannot pay our maids or service providers easily (I dont want to have an ultra long list of payees, and go through a long registration process for each payee). UPI promises to fix this.

- Today with all the bank iMobile apps, we cannot pay Uber or Ola from our phones (only a hand ful of wallets have deals w/ them). With UPI, they will be forced to accept it from any bank.

- SMS existed before Whatsapp. What was the key / radical value prop of Whatsapp? It integrated w/ our address books, so that we could send SMS-equivalents (instant messages) to a name. The entire history of messages was easy to discover. New users got automatically added. We could use it wherever 3G or Wifi existed and overcome exorbitant roaming charges just to send a simple message. Later it expanded to images, videos, and now conversational platforms are all a rage in the community as the future of the "search" box.

- Another way to understand it is to ask: "Whats in a name?" . Complex credentials complicate process. Simple names, and flexibility in names (eg: aadhar numbers, VPAs) will simplify process, and hopefully spur adoption with deep mobile-app integration of UPI payments.

ps: Part of a set posts: "Payment Banks: Reliance Jio vs Airtel & More" , "Financial Inclusion post demonetisation", "Cashless Finance & Universal Payment Interface (UPI)" , "Cashless boosted by demonetisation ...", "Barter & Cash to Cashless & Blockchain... " , "Service vs As-a-Service ... "

pps: Some of this content is based upon a nice tutorial by Razorpay on YouTube

ppps: List of all my LinkedIn Articles