From Friction to Formality: Why Making Invoicing Easy Is a Tax Collection Strategy

Tax administrations have spent two decades chasing the same insight from different directions: the fastest way to raise revenue sustainably is not to squeeze harder on the taxpayers you already have, but to make it effortless for everyone to comply. Latin America proved this with electronic invoicing. Rwanda proved it with electronic billing machines. Brazil and India proved a related version of the same idea in payments, with systems that let people transact using a name instead of a bank account number. This post connects those threads and proposes a simple, low-cost step that any modernizing tax administration can take next: pairing every Taxpayer Identification Number (TIN) with a Taxpayer Unique Alias (TUA), combined with automatic data completion at the point of invoicing. We also estimate what that costs, review what has worked elsewhere, and share that Antares — a new integrated financial and tax administration platform — will ship with TUA-based online invoicing built in from day one.

1. Easing taxpayer operations is the collection strategy, not a side benefit

For much of the twentieth century, tax administrations treated compliance friction as someone else’s problem: the taxpayer’s, the accountant’s, the software vendor’s. That framing has flipped. The World Bank’s governance team frames it plainly: streamlining tax compliance through digitalization and simplified procedures reduces administrative burdens and frees businesses to invest in growth instead of paperwork, and effective tax administration is now understood as a tool for job creation as much as for revenue collection.¹ The same analysis notes that internal efficiency gains — in taxpayer registration, voluntary compliance mechanisms, online filing, electronic payments, and data analytics — are what actually unlock additional revenue, more than rate increases or audit intensity do.¹

The OECD’s Forum on Tax Administration has been making a parallel argument under the banner of “Tax Administration 3.0”: administrations that embed themselves into taxpayers’ natural business systems, rather than asking taxpayers to step outside their workflow to comply, get better data and better compliance simultaneously.² Electronic invoicing is the clearest expression of this. Because VAT already represents around 20% of total tax revenue in the average country worldwide, and between 30% and 50% of total revenue in many developing economies, even a modest improvement in VAT compliance driven by easier invoicing moves the needle on the whole fiscal accounts.²

The empirical record backs the theory. When Mexico made e-invoicing mandatory, declared revenues rose 14% within three years.³ Uruguay collected an estimated 3.7% more in VAT and corporate income tax in the year e-invoicing became broadly mandatory.³ Peru saw taxable sales reported to the tax authority rise 7% and purchases rise 5% in the first year after the mandate took effect, a result that a separate quasi-experimental study using SUNAT’s own administrative data later confirmed using a staggered difference-in-differences design across adoption waves.^{3 4} El Salvador’s Ministry of Finance reports VAT collection climbing from 3.5% of GDP in 2017 to 8.7% of GDP in 2023, a period that coincides with its e-invoicing rollout.³ In several early-adopting Brazilian states, e-invoicing was associated with a 12% increase in tax revenue and a 2% reduction in informality.³ Italy’s Sistema di Interscambio (SdI), operating since 2019, detected roughly €1.1 billion in VAT fraud in a single year and pushed reported VAT payments up 3.6% even before accounting for the deterrent effect on future misreporting.² Between 2019 and 2022, Italian VAT revenue is estimated to have grown by €1.7–2 billion while fraud detection rose from about €1 billion to €9 billion.⁵

None of these gains came from raising rates or hiring more auditors. They came from lowering the cost, in time and hassle, of doing the right thing — and from the data exhaust that structured, real-time invoicing produces almost as a byproduct. Rwanda’s experience with Electronic Billing Machines illustrates the taxpayer side of that trade just as clearly: VAT declaration and filing time fell from roughly 45 hours a year to about 5, while the tax authority gained real-time visibility into sales that previously existed only on paper receipts vulnerable to forgery and parallel bookkeeping.⁶ That is the whole thesis in one statistic: less taxpayer burden, more government visibility, at the same time, from the same reform.

2. The simple case: a Taxpayer Unique Alias for the TIN

Here is where most e-invoicing systems still leave money on the table. Once a country mandates structured electronic invoices, the technical burden of issuing one shifts from “write it by hand” to “know the counterparty’s TIN, legal name, fiscal address, and tax regime, and enter them correctly into a form or an API call.” That is a real barrier, especially for the smallest taxpayers and for one-off transactions between parties who have never invoiced each other before. A market vendor who has to look up a customer’s 10- or 13-digit TIN, copy it without a typo, and separately supply the customer’s registered name and address exactly as the tax authority has them on file, is being asked to do something structurally harder than what informal commerce required. Every extra required field is a place where a transaction quietly reverts to cash, or to a paper receipt that never reaches the tax administration.

The fix already exists, just not yet in tax administration. It exists in payments. Brazil’s Pix, the instant payment system built and operated by the Central Bank of Brazil, replaced bank account numbers and branch codes with a “Pix key” (chave Pix) — a short, memorable alias based on a phone number, an email address, the person’s own tax ID, or a random string, resolved behind the scenes to the actual account through a directory called DICT.⁷ The result was not a marginal improvement; it was the fastest payment-system adoption on record. Within about a year of its November 2020 launch, Pix had reached 114 million users, or roughly 67% of Brazil’s adult population.⁸ By September 2024, Pix had 538 million registered accounts including businesses, and by 2024 it was processing tens of billions of transactions a year.^{9 10} A parallel mechanism from India tells the same story from a different angle: the Unified Payments Interface’s Virtual Payment Address (VPA) lets any user or merchant transact through something as memorable as name@bank, without ever exposing an account number or routing code, and UPI now clears tens of billions of transactions a month.^{11 12} In both cases, the alias is not a cosmetic convenience. It is the mechanism that let systems designed for financial-sector plumbing become genuinely usable by street vendors and casual counterparties.

There is no reason the same principle cannot apply to the taxpayer identifier itself. A Taxpayer Unique Alias (TUA) is exactly that: a short, memorable, business-chosen or system-suggested string — a trade name, a handle, a phone-number-like token — mapped one-to-one to a taxpayer’s TIN inside the tax administration’s own directory, the same way DICT maps a Pix key to a bank account. A buyer issuing or receiving an invoice would only need to know or type the counterparty’s TUA. The system resolves it instantly and auto-completes everything the invoice legally requires: the TIN itself, the registered legal name, the fiscal address, the applicable tax regime, and any other master-data fields the format demands — the same fields that today have to be looked up, copied, and re-typed by hand, and that are consequently the most common source of invoice rejection at the validation stage. This is a small technical addition on top of infrastructure most modern e-invoicing platforms already have (a taxpayer registry, a validation API, real-time clearance), but it removes exactly the friction that keeps casual, low-volume, and small-business transactions out of the formal invoicing stream. Given that Pix moved a large share of a national economy’s person-to-business payments away from cash within about three years largely on the strength of a memorable alias, and UPI did something similar for India, there is a credible basis for expecting an TUA-plus-autocomplete layer to produce comparable, and possibly exponential, growth in electronic invoice issuance among micro and small taxpayers — precisely the segment every e-invoicing mandate struggles hardest to bring on board.

3. Estimating the cost of adoption

The honest answer is that the cost is small relative to the e-invoicing infrastructure it sits on top of, because an TUA layer reuses rather than replaces what a modern tax administration platform already has: a taxpayer registry, a validation/clearance API, and an invoice data model. Three cost components are worth separating.

Government-side infrastructure. Directory-and-resolution services of this kind are not exotic. Brazil’s central bank built the entire real-time payment engine (SPI) and the DICT alias directory that resolves Pix keys to accounts for roughly USD 4 million, a figure that becomes almost trivial once set against the transaction volume it now clears.¹³ A tax administration adding an alias-resolution table and an autocomplete endpoint to an existing taxpayer registry and e-invoicing clearance system is a considerably smaller undertaking than building a national instant-payment rail from scratch, since the identity, KYC, and registry work is already done for tax purposes. For a country building this as an incremental module on an existing e-invoicing platform, a realistic order of magnitude sits in the same range independent studies give for a full custom e-invoicing solution built from nothing — commonly cited at USD 140,000–400,000 in development cost and six to ten months of implementation time for a complete platform¹⁴ — with the TUA/autocomplete layer representing a fraction of that, since it is additive to, not a replacement for, existing registry and clearance infrastructure.

Taxpayer-side cost. This is the more important number, because it determines adoption. Processing a paper invoice is commonly estimated at €14–20; a structured electronic invoice costs a few euros to process, and under high automation that can fall below €1.¹⁵ Separately, surveys of firms already operating under e-invoicing mandates find that 55% cite upfront cost as a real barrier to adoption, alongside integration complexity.¹⁶ An TUA layer directly attacks the integration and data-entry side of that cost, since auto-completing counterparty data removes one of the most error-prone and time-consuming manual steps in invoice preparation — for the smallest taxpayers, often the single largest recurring cost of compliance, since it is paid in staff time on every invoice rather than once at implementation. For point-of-sale and mobile issuance methods aimed specifically at small taxpayers, the incremental cost to the business is close to zero: Rwanda’s mobile e-invoicing channel, aimed at taxpayers with annual turnover under roughly FRW 20 million (about EUR 16,400), works over ordinary SMS with no dedicated hardware required.¹⁷

Net assessment. For the administration, an TUA/autocomplete module is best estimated as a moderate incremental cost on an existing e-invoicing or taxpayer-registry system — well under the cost of a new platform build, and recoverable quickly given that every percentage point of formalization the analogous Pix and UPI alias systems achieved corresponds, in the tax-collection studies above, to measurable multi-percentage-point gains in reported VAT and income. For the taxpayer, the marginal cost per invoice should approach zero, which is precisely the condition under which the Pix and UPI case studies show adoption becomes exponential rather than linear.

4. Best practices worldwide

A few patterns recur across the countries and systems that have gotten this right, and they are worth stating explicitly because they generalize well beyond invoicing.

Start with the highest-friction, highest-informality segment, not the largest taxpayers. Chile pioneered e-invoicing in the early 2000s and by 2014 had passed a law extending the obligation to all companies within five years, but the design that ultimately produced near-universal adoption phased in company size deliberately.¹⁸ Peru’s rollout likewise moved in graduated waves, starting with larger firms and priority sectors and giving smaller firms more time and lighter-weight tools, which is exactly the design that later let researchers isolate the reform’s causal effect using the staggered timing itself.⁴ Rwanda went a step further and built a specific, nearly cost-free mobile channel exclusively for its smallest VAT-registered taxpayers, rather than asking a street-level vendor to adopt the same software as a large exporter.¹⁷

Make the clearance model real-time, and make rejection immediate and fixable. Italy’s SdI validates every invoice against more than 200 rules before it legally exists for tax purposes; a rejected invoice can be corrected and resubmitted within days, which keeps the compliance loop tight instead of letting errors surface only at audit time months later.¹⁹ The result is not just fraud detection — it is a live picture of the economy that the tax administration did not have before, which is now explicitly the model the European Union’s VAT in the Digital Age (ViDA) initiative is generalizing across all member states by 2030.²⁰

Treat invoicing data as an input to pre-filled returns, not just a compliance record. Italy now offers pre-filled periodic VAT returns to roughly 2.4 million VAT-registered operators, generated directly from the invoice data SdI already has.⁵ This closes the loop: the same data captured to ease invoicing is reused to ease filing, compounding the reduction in taxpayer burden rather than creating a second, separate burden downstream.

Reduce identifiers to something memorable wherever a human has to type or say one. This is the lesson from outside tax administration that this post argues should be imported into it. Pix’s alias-based design, resolved through a central directory, is explicitly credited with making adoption “straightforward” for individuals and, over time, for merchants, even though merchant adoption requires back-end integration and is stickier by nature.⁸ UPI’s VPA does the same for India, explicitly marketed and understood as “an email address for your money.”²¹ Both examples support a general principle: whenever a national digital-infrastructure project requires two unfamiliar parties to identify each other precisely, replacing a long numeric identifier with a short, human-chosen alias resolved by a trusted central directory measurably accelerates adoption.

Build broad coverage before narrowing exemptions, then remove exemptions gradually. Italy first mandated B2G invoicing in 2014, extended to full B2B/B2C in 2019, and only brought micro-enterprises fully into scope in 2024 — a decade-long glide path that let the platform, the taxpayer base, and third-party software ecosystem mature together rather than forcing simultaneous adoption by taxpayers with wildly different capacities.^{22 23}

Taken together, these practices describe a consistent playbook: reduce the marginal cost and cognitive load of the first invoice a small or informal taxpayer ever issues electronically, make the system’s feedback loop fast enough that mistakes are corrected in days rather than discovered at audit, and feed the resulting data back to taxpayers as a benefit (pre-filled returns, faster refunds, fewer inspections) rather than only using it as a stick.

5. Antares: TUAs arrive with online invoicing

We are building this idea directly into Antares, a new integrated financial and tax administration platform designed around the principle that easing taxpayer operations is the most direct route to sustainably higher collection. Antares’ online invoicing module will ship with TUA support from its initial release: every registered taxpayer will be able to claim a short, memorable alias tied one-to-one to their TIN, and any counterparty — from a large exporter’s ERP integration down to a market stall issuing invoices from a phone — will be able to generate a fully compliant invoice by entering that alias alone. The system resolves the alias against the taxpayer registry in real time and auto-completes the TIN, legal name, fiscal address, and applicable tax regime, the same fields that today are the most common source of manual error and invoice rejection.

The design deliberately follows the best practices reviewed above: real-time resolution and validation modeled on the clearance approach that has worked in Italy and across Latin America, a lightweight issuance path aimed at the smallest taxpayers in the spirit of Rwanda’s mobile channel, and an alias layer inspired directly by the adoption curves that Pix and UPI achieved in payments. The goal is not simply to digitize invoicing — every mandate reviewed in this post has already done that — but to remove the last piece of friction that keeps the smallest, most informal, and most occasional transactions outside the formal invoicing stream, which is exactly where the largest untapped compliance gains remain.

Written with the support of claude.com.

References