For over a decade, Pakistan’s banking system has been sold a vision of digital transformation. From billboards promising banking in your pocket to policy speeches extolling a cashless future, technology was cast as the shortcut to inclusion, efficiency, and growth. The future, we were told, would be smaller branches and bigger apps, fewer tellers and more taps. Yet nearly twenty years into this experiment, the data tell a more complicated story. Not every digital channel pays its way. Some, for now, remain more aspiration than asset.
Between 2006 and 2022, commercial banks poured capital into electronic delivery—mobile apps, internet portals, e-commerce payment rails, point-of-sale terminals, interactive voice systems, and real-time online transfers—while also expanding the automated teller machine footprint. The working assumption behind these investments was deceptively simple: adoption would reduce cost-to-serve, widen the customer base, and lift profitability. What happened instead was asymmetry. The channel that consistently improved returns was not the app or the portal, but the machine on the street corner.
This is not conjecture; it’s what the numbers say when they are given a proper time series interrogation. Using quarterly industry data across sixteen years and an autoregressive distributed lag framework to separate short-run dynamics from long-run relationships, the study finds that ATM usage is positively and significantly linked to bank performance, measured by return on assets. The logic is practical. ATMs are ubiquitous, their interface is familiar, they are open around the clock, and they support the most trusted transaction in the economy: turning a balance into cash.
By contrast, mobile banking drags on profitability. The technology spend is large, the security bill is constant, and the revenue side is thin because adoption is still hesitant. The story is similar for e-commerce. Fraud risk, chargeback frictions, and patchy consumer protection conspire to keep usage concentrated in a relatively small segment. Banks pay to connect gateways, to monitor risk, to staff dispute desks, and to plug compliance gaps. Too often those costs exceed the fees they can credibly charge customers or merchants.
Other channels—internet portals, POS networks, and legacy IVR—are present but statistically insignificant in their contribution to sector profitability in this period. Real-time transfers have grown quickly, but on the financial performance metric in question their impact is neutral rather than transformative.
It would be convenient to blame all this on technology, but that would be the wrong lesson. The results also reflect the macroeconomic and behavioral context in which these channels operate. When growth improves, bank performance improves; when inflation bites, margins erode. Performance is path-dependent—the previous quarter’s results materially shape the next—which means shocks and gains are sticky. Digital adoption is not immune to those cycles. It accelerates when incomes are stable and optimism rises; it falters when uncertainty dominates.
Trust is the real currency here. It is not enough to build and announce a feature; customers must believe it will work as expected, and that if something goes wrong they will be made whole swiftly and predictably. That belief is fragile in a market where phishing, social engineering, and SIM swap fraud are widely discussed and too often felt. The user experience still asks too much of too many: complex passwords, one-time codes that arrive late, interfaces that change too often, and jargon that assumes more fluency than most users possess.
The irony is that Pakistan is not a stranger to digital finance. Telco-led wallets have habituated millions to electronic bill payments and domestic remittances. Instant rails have made account-to-account transfers quick and cheap. But the study’s result is a reminder that banking profitability and national digitization are not synonyms. Wallet activity can be lively while bank returns are muted if banks carry the cost of infrastructure, security, and compliance without capturing enough of the transaction flow to cover it.
There is also a distributional story behind the headline result. ATMs thrive because they meet people where they are. They require no new vocabulary, no new mental model, no fear that a mis-tap will cost a month’s wages. They convert a digital balance into physical cash—still the medium of trust and settlement for daily life—and they do it reliably. Mobile apps invert that. They ask people to move savings into a glass rectangle, to trust that electricity and networks will be available when needed, and to believe that invisible systems will protect their money.
If the conclusion were simply “build more ATMs,” we would have learned nothing. The more useful reading is about sequencing and incentives. First, make the digital experience as reliable, legible, and forgiving as the ATM. That means defaulting to strong but humane authentication; investing in proactive fraud controls that stop dubious transactions before they occur; simplifying interfaces to match the tasks customers actually perform; and, most important of all, guaranteeing fast, fair redress. Second, align incentives so that customers and banks share the benefits of digital adoption rather than treating it as a one-sided cost.
Cost matters as much as design. Banks can shrink the technology bill by collaborating more deliberately with fintechs that build lighter, modular systems—front-ends that ride the existing rails, fraud controls that are risk-based rather than one-size-fits-all, analytics that can be embedded rather than bolted on. The point is not to outsource the core, but to reduce the weight of the non-core. The study’s long horizon makes clear that capitalized software projects do not vanish from the P&L; they amortize slowly, dragging on earnings for years.
None of this diminishes the role of the ATM, which deserves a second act rather than a sunset. The “phygital” bridge can be strengthened by upgrading cash machines into smart kiosks: cash recyclers that reduce replenishment costs, biometric modules that make onboarding safer, and mini-branches that extend services without the fixed cost of a full counter. Most important of all, ATMs and agent points can become the human face of digital—places where staff help people install, enroll, and practice on mobile channels until comfort builds.
If there is a single thread that runs through the results, it is that Pakistan’s digital banking story is not a failure so much as an unfinished transition. The structural headwinds are real—volatile inflation, uneven growth, a large cash economy, and the constant pressure on banks to hold more capital against more risks. In that environment, the temptation is to treat technology as a talisman. Launch the app and assume usage will follow; install the terminal and assume swipes will come. The study’s message is a corrective: technology delivers returns only when it is trusted, understood, and integrated into daily life.
There is a broader geoeconomic angle to consider. Financial infrastructure is increasingly inseparable from trade and investment. Export financing, supply-chain payments, cross-border settlements, and remittances are all migrating onto digital rails. Countries that make those rails reliable become easier to trade with and to invest in. Countries that cannot guarantee reliability raise the cost of doing business with them. Pakistan does not have to match the most advanced systems overnight, but it does have to convince counterparties that its rails are dependable.
The study’s econometrics carry one final, underappreciated implication: the adjustment mechanism back to long-run equilibrium is active. In plain English, when performance deviates from the level supported by fundamentals, it tends to move back over subsequent quarters. That is not a license for complacency; it is a case for urgency. Interventions that lower the cost of digital service, reduce fraud loss, and lift usage will not produce miracles in a month—but they will compound. The same is true in the negative direction: neglect compounds, too.
So what would a sensible next chapter look like? Start with a realism that resists both fatalism and hype. Accept that cash will remain prominent for longer than campaigns would like, and use the ATM and agent network as the scaffolding for digital habit rather than as the enemy of it. Clean up the hard edges in mobile and online journeys: minimize failure points, design for low-bandwidth environments, and set defaults that prevent the most common user errors. Turn redress from a grudging back-office function into a competitive differentiator.
Above all, adopt a sequencing mindset. The history of technology adoption in finance is a history of steps, not leaps. ATMs worked because they met a need with competence. The next step is to make the phone feel like the same promise in a smaller box: a tool that is boringly reliable, fair when it fails, and worth the trade between convenience and risk. When that is achieved, the profitability arithmetic will start to look like the strategy decks imagined. Until then, digital will remain an uneven ledger, promising more than it delivers.
If Pakistan’s banks can take that lesson to heart, they will discover that the dividends of digital come not from the gloss of the interface but from the depth of the trust underneath. In a sector where confidence is capital, that is the only balance sheet that truly compounds.
Read more here: https://ijmres.pk/index.php/IJMRES/article/view/756/304
Follow the PakBanker Whatsapp Channel for updated across Pakistan’s banking ecosystem.