Some Thoughts on UPI ecosystem
Recently I was asked for some comments on current state of UPI and interoperability. Putting them below with some minimal editing. Comments are personal.
> What is the status of non-bank merchant acquirers accessing the UPI directly (without going through banks)?
It has started, but I would assume its being taken in controlled activity. BHIM was first non-bank PSP to start acquiring as early as last year, but given commercial agreements, ownership of NPCI, it is practically not a non bank for all practical purpose.
1. PhonePe is the lone non-bank acquirer doing both online & offline merchant acquiring. On the offline retail front, they are now acquiring using traditional sales force (no self-boarding), hence pace is restricted.
2. On the online front, at least one payment gateway I know has built PSP infra and is waiting for green signal to operate as PSP. Non bank gateways have been acquiring since early days using a partner bank, but bigger ones are tentatively getting direct access to rails sooner or later now.
3. PhonePe has also released its own SDK which app developers started using and ~30 apps now have direct UPI integration. Previously tight integrations was possible only through bank partnerships(Uber, Ola). Although some 100 odd apps support UPI through collect requests, native bundling of PSP SDK gives an in-app experience reducing friction as well as failures. I did an analysis of these tightly integrated apps and the pattern I find is non-bank PSPs are allowed until they enter finance sector. PhonePe while having given access to its SDK to 20 odd companies across retail, FMCG, entertainment sector is stopped by YesBank, that has given its SDK when it came to fund houses, MFIs, NBFCs. ICICI and HDFC too have been integrating with financial sector instituitions. I clearly see an unwritten law (or possibly SEBI regulation(?), but technically the need for sharing source account number is possible by non bank PSPs too) here and this will heat up when digital lenders enter in full force. See “UPI Integrations — Android” for data
> Any work on wallet to wallet interoperability at the level of UPI/ IMPS that we know of? What will it take?
Wallets are practically dead. Freecharge acquired by Axis, PayTM being a payments bank, Mobikwik being an aggregator PSP (with no UPI access to their own wallet). That doesn’t leave with many wallet players around with the size that can even ask media to write stories let alone lobby in power corridors. I see increased digital mandate by food coupon majors and they could be new set of PPIs, but its too small a market. The cost of KYC and Aadhaar will only make it impossible to run a wallet business. I think the reliance on UPI for interoperability is adding a high cost for some of these players. Food cards / Tap N Go, can achieve low cost interoperability, cheaper than UPI which requires ongoing investment in maintaining infrastructure, apps and unknown cost (high?) of support etc. Cheaper (neutral) settlement infrastructures that are inclusive might give a push, if at all PPI industry wants interoperability. Beyond a point, the top down push of regulator will only increase cost and unwillingness in interoperability by operators.
> The differences in the way UPI has been implemented as against the original role…
If UPI’s original mandate was to have a mobile first interoperable payments network, freeing up / democratizing payments as enabler for digital economy, there are certain things that block this from technology, business and governance standpoint. Leaving aside the privacy / data debate and role of IndiaStack and Nandan’s / iSpirt vision of data economy, there are other structural issues in UPI preventing further expansion.
1. Purely technically, the APIs haven’t been democratized and that inhibits growth and innovation. I don’t understand when open-banking will become mainstream, but until that, many use cases remain locked.
2. On the privacy / data angle, the centralization choice over decentralized / minimal PII-linked centralization (IMPS) will prove costly over period of time and only way UPI can remain tight is by being a monopoly in instant payments.
3. On access and inclusion, during the India Post Payments Bank launch, I realized the excessive dependence on cards for on-boarding UPI means that debit card is prerequisite. This conflicts the inclusion mandate and positions UPI clearly as a product for carded masses. I heard they have tried some assisted on-boarding for IPPB accounts, but this is hardly good enough. I was able to picture hierarchy of digital payments consumer
4. The original UPI technology architecture design did have wallets and even cards supporting in them. Cards took a wild turn with BharatQR and a post facto UPI-BharatQR integration, while wallets were actively kept out by banks for business reasons and lobbying.
> Other views / comments on UPI
1. I would give app interoperability + enabling large section of banks (Bottom 100 banks in UPI) to transact with ease + PM branding as key factors for success, besides carefully published selective metrics to claim victory and nothing else.
2. Entry of global players is just Samsung and Google if you take out Whatsapp (I found that the limit placed for WA is 5,000 Rs, same as what RBI allows for unencrypted transfers). Amazon has started accepting, while doesn’t seem to invest otherwise. There isn’t anyone and this origin of player is irrelevant, not even significant investments as running infra is doesn’t need much investment.
3. P2M needs numbers publicly put out and selective publishing of metrics isn’t helping anyone.
4. On misses part of UPI 2.0, eMandates will be bigger grudge with industry, but improving reliability at scale is yet to be achieved and should be of higher priority. I did hear NPCI is being more focused on it than adding features. The chat bot RFP for improving customer support is another platform upgrade in the works. (Different matter that NPCI is running a PSP without ticketing system and chat bots are seen as future support tool).