Measuring the impact of digital public infrastructure: beyond transaction counts

Analysis · Published 2026-10-20

A green retractable tape measure partly extended across a white background — Photo: Evan-Amos, public domain, via Wikimedia Commons.

Digital public infrastructure is usually described in big numbers: billions of payments a month, more than a billion identities, hundreds of connected services. Those figures show adoption. They do not show whether people's lives got easier, whether public money was saved, or who was left out. Measuring impact requires different questions and different data.

Adoption is not outcome

A discussion note by the Centre for Social and Economic Progress (CSEP), an Indian think tank, makes the distinction central: adoption metrics, such as identity coverage and payment transactions, are widely published, but outcomes, such as reduced costs, improved access to credit or better targeting of benefits, require separate evidence. The two can diverge. A payment system can grow quickly because a few large apps push it, while small merchants see limited benefit; an identity system can reach near-universal enrolment while authentication failures exclude some people from rations.

What the evidence says so far

Fiscal savings: government estimates

An International Monetary Fund working paper published in March 2023, Stacking up the Benefits: Lessons from India's Digital Journey, reviewed India's foundational DPI. It reports the Government of India's estimate that, up to March 2021, savings of about 1.1 percent of GDP in expenditure were achieved through digital infrastructure and related governance reforms, largely from direct benefit transfers that removed duplicate and ineligible beneficiaries. That figure is a government estimate, and attributing savings precisely to technology versus other reforms is difficult. The same paper noted that India still lacked comprehensive data protection legislation at the time, which was later passed in 2023.

Financial inclusion

The World Bank's Global Findex surveys show account ownership among adults in India rising from 35 percent in 2011 to 78 percent in 2021. Accounts opened under the Jan Dhan programme, identity-based simplified onboarding and digital payments all played a part. But Findex also distinguishes owning an account from using it, and inactive accounts were a recurring concern in the surveys.

Field experiments on identity and welfare

Some of the most rigorous evidence comes from randomised evaluations by economists Karthik Muralidharan, Paul Niehaus and Sandip Sukhtankar. A large experiment on biometric smartcards for welfare payments in the former state of Andhra Pradesh, published in the American Economic Review in 2016, found faster, more predictable payments and reduced leakage. A later experiment in Jharkhand on requiring biometric authentication to collect subsidised food found no significant reduction in leakage, but higher transaction costs for beneficiaries and exclusion of some eligible households. Together, they show that the same family of technology can help or harm depending on how and where it is applied.

Payments design

The Bank for International Settlements' 2019 paper The design of digital financial infrastructure: lessons from India (BIS Papers No 106) analysed how identity, payments and data layers fit together, and argued that the public infrastructure model lowered entry barriers for new providers. It is descriptive and analytical rather than a measurement of welfare effects, which remain harder to quantify.

A practical set of indicators

A balanced scorecard for a DPI system should cover at least seven dimensions:

Dimension	Example indicators	Why it matters
Adoption and use	Active users (not just registrations); transactions per active user; share of services integrated.	Separates enrolment from real use.
Reliability	Availability; peak-time latency; authentication and transaction failure rates.	Failures fall hardest on people with no alternative.
Inclusion	Coverage and failure rates by gender, age, region, disability; share of transactions via assisted channels.	Averages hide exclusion.
Cost and time	Cost per transaction for government, providers and users; time to complete key services.	The core promise is cheaper, faster services.
Market structure	Number of active providers; market shares; onboarding time for new participants.	Neutral rails should support competition.
Rights and redress	Grievances received and resolved, and resolution time; data breaches; audit findings.	Trust depends on what happens when things go wrong.
Outcomes	Benefits reaching eligible people; credit access for small firms; tax compliance costs.	The reason the system exists.

How to measure well

Publish disaggregated data. National averages hide the districts, age groups or occupations where systems fail.
Measure failures as carefully as successes. Authentication failure rates and grievance statistics should be routine publications.
Build evaluation in from the start. Phased rollouts create natural comparison groups; randomised or quasi-experimental designs are feasible, as the Indian experiments showed.
Open data access for researchers under privacy-preserving conditions, so findings do not depend only on the operator's own reports.
Be careful with attribution. Savings and inclusion gains usually come from combinations of technology, policy and institutions. Claims that credit technology alone should be read with caution.

Reading the headline numbers

When you see a claim about DPI impact, ask: is it adoption or outcome? Who produced the estimate? Over what period, against what baseline? Does it report who was excluded? A figure that answers these questions is worth far more than a larger one that does not. Our design toolkit includes measurement questions for teams at the planning stage.

Sources

Alonso, C., Bhojwani, T., Hanedar, E., Prihardini, D., Uña, G. and Zhabska, K., Stacking up the Benefits: Lessons from India's Digital Journey, IMF Working Paper WP/23/78 (March 2023).
Centre for Social and Economic Progress, Measuring the economic impact of India's digital public infrastructure: an assessment.
D'Silva, D., Filková, Z., Packer, F. and Tiwari, S., The design of digital financial infrastructure: lessons from India, BIS Papers No 106 (2019).
World Bank, Global Findex Database.
Muralidharan, K., Niehaus, P. and Sukhtankar, S., "Building State Capacity: Evidence from Biometric Smartcards in India", American Economic Review 106(10), 2016, and later work on identity verification in welfare programmes, listed on J-PAL.