Does Digital Payment Adoption Reshape the Tax Mix? An Instrumental-Variables Analysis of Tax Composition in 38 OECD Countries, 2000–2022

Abstract

Abstract

Digital payment adoption is widely believed to broaden the visible tax base, and prior work links cashless payments to smaller value-added tax (VAT) compliance gaps. Whether this effect is large enough to shift a country’s tax mix—the share of revenue from personal income tax (PIT), corporate income tax (CIT), and VAT—rather than raising each tax proportionately, remains untested. We examine this using a panel of 38 OECD countries in 2000–2022, instrumenting digital payment adoption with two time-invariant infrastructure-legacy proxies—broadband rollout timing and submarine cable distance—via two-stage least squares, benchmarked against naive OLS and two-way fixed effects. Naive OLS shows digital payment adoption significantly lowering the CIT and VAT shares of revenue, opposite to the base-broadening intuition; these associations vanish under instrumentation and fixed effects for every outcome. Weak-instrument-robust tests corroborate the null for three of the four outcomes; for VAT, where our overidentification and placebo evidence raise the most doubt about instrument validity, the weak-instrument-robust confidence set excludes zero, a result we treat as inconclusive given those same validity concerns rather than as either confirming an effect or corroborating the null. A cluster bootstrap and lagged and extended-control specifications corroborate the null throughout, while a placebo test shows that our instruments correlate with pre-sample tax composition, a genuine limitation we report in full. The two-way fixed-effects results, which do not rely on these instruments and are therefore not exposed to that specific confounder (though, like any fixed-effects design, it cannot rule out time-varying confounding), corroborate the same null and are weighted as the most credible evidence. We interpret the results as evidence against an economically meaningful, robust association between digital payment adoption and OECD tax composition, rather than as a causally identified null effect—a pattern at least consistent with the possibility that mature tax administrations had already captured most realizable enforcement gains from digitalization before 2000.