Abstract
The concentration of digital payment transactions in just two UPI apps like PhonePe and Google Pay has raised concerns of duopoly in India's digital financial ecosystem. To address this, the National Payments Corporation of India (NPCI) has mandated that no single UPI app should exceed 30 percent of total transaction volume. Enforcing this cap, however, poses a significant computational challenge: how to redistribute user transactions across apps without causing widespread user inconvenience while maintaining capacity limits? In this paper, we formalize this problem as the Minimum Edge Activation Flow (MEAF) problem on a bipartite network of users and apps, where activating an edge corresponds to a new app installation. The objective is to ensure a feasible flow respecting app capacities while minimizing additional activations. We further prove that Minimum Edge Activation Flow is NP-Complete. To address the computational challenge, we propose scalable heuristics, named Decoupled Two-Stage Allocation Strategy (DTAS), that exploit flow structure and capacity reuse. Experiments on large semi-synthetic transaction network data show that DTAS finds solutions close to the optimal ILP within seconds, offering a fast and practical way to enforce transaction caps fairly and efficiently.
Introduction: The UPI Duopoly Challenge
India’s Unified Payments Interface (UPI) has revolutionized digital finance, processing billions of transactions monthly. However, this success is concentrated: over 80% of volume flows through just two applications, PhonePe and Google Pay. This duopoly introduces systemic risks; a technical failure in either platform could paralyze the national economy. To mitigate this, the National Payments Corporation of India (NPCI) proposed a 30% market share cap per app.
Implementing this cap is non-trivial. A simple "tail-drop" policy—blocking transactions once an app hits its limit—would frustrate users and disrupt commerce. Instead, transactions should be rerouted to alternative apps. Since users prefer not to install new software, the challenge is to route all transactions while minimizing the total number of new app installations across the population.
Formalizing the Problem: Minimum Edge Activation Flow (MEAF)
The authors formalize this optimization challenge as the Minimum Edge Activation Flow (MEAF) problem. The system is modeled as a bipartite graph where one set of nodes represents users ($U$) and the other represents apps ($A$).
In this model:
- Solid edges represent apps a user already has installed.
- Dashed edges represent potential new installations (activations).
- Capacities ($c$) on the edges to the sink ($t$) enforce the 30% regulatory limit.
The goal is to find a flow that satisfies all user transaction demands while minimizing the number of activated dashed edges. The paper proves that MEAF is NP-Complete, necessitating efficient heuristic solutions for real-world scale.
The DTAS Algorithm: Prioritizing the "Little Guy"
To solve MEAF at scale, the authors developed the Decoupled Two-Stage Allocation Strategy (DTAS). A key insight driving DTAS is the heavy-tailed nature of user behavior: a small number of "heavy" users generate the majority of transactions. Naive algorithms that process heavy users first often consume the capacity of popular apps, forcing millions of "lightweight" users to install new apps for just one or two transactions.
DTAS reverses this logic by sorting users in ascending order of transaction volume. It operates in two global phases:
- Stage 1: Allocate as many transactions as possible using only pre-installed apps across the entire network.
- Stage 2: Route remaining demand by reusing "extra" apps already activated by other users, installing new apps only as a final resort.
By processing lightweight users first, DTAS ensures they can use their existing apps. Heavy users, who must route large volumes anyway, "absorb" the cost of a new installation, which is more efficient than forcing thousands of small users to do the same.
Experimental Results and Scalability
The authors evaluated DTAS using a semi-synthetic dataset based on Rabobank financial records, mapped to mirror NPCI market share statistics.
The results demonstrate that DTAS is both highly effective and exceptionally fast:
- Near-Optimality: On datasets with up to 100 million transactions, DTAS achieved installation counts nearly identical to the optimal Integer Linear Program (ILP) solutions.
- Linear Scalability: While ILP and Linear Programming (LP) baselines timed out on large datasets, DTAS processed 100 million transactions in seconds.
- Fairness: DTAS improved the Inverse Gini Score (IGS) to 0.53 (vs. 0.34 for ILP), indicating a more balanced distribution of transactions across the app ecosystem.
Conclusion
The research proves that enforcing regulatory transaction caps is computationally feasible without significant user friction. By using intelligent routing heuristics like DTAS, regulators can foster a more competitive and resilient digital payment landscape. Future work will look to incorporate transaction values into the model to further refine systemic risk management and fairness.