The zero-day close, an idea also for transfer pricing?

There are two industries we watch particularly closely. One is legal technology, where the work is judgment applied to documents, much like our own field. The other is accounting technology, the software industry around the finance function, where whatever changes tends to reach the tax department shortly after. The idea moving accounting technology at the moment is the zero-day close. Half of it transfers to transfer pricing almost unchanged. The other half runs into something particular about the discipline, and it is the more interesting half.

Books that can close on any day

The zero-day close aims at the month-end close, the reconstruction of a period after it ends. Transactions are booked all month, and once the month is over the accounting team reconciles accounts, books accruals, chases exceptions, and reports. In a zero-day close, transactions are recorded, reconciled, and traceable as they occur, so the period end requires little additional work and the books can close on any day. The ambition is older than the term. Cisco described closing its books on an hour’s notice twenty-five years ago, and vendors have sold close software on the promise of a continuous close for years.

The idea moved to the center of the conversation in August, when Sarah Friar, OpenAI’s chief financial officer, named a zero-day close as one of her team’s two ambitions, next to continuously updated forecasting. In the version she describes, approved spending plans, ledger actuals, purchase orders, accruals, and transaction details connect into one continuously reconciled view. AI prepares a first explanation for every variance and flags the exceptions that need attention, and the finance team validates the numbers and owns the sign-off. OpenAI says plainly that it is still building toward this. The goal is a finance function that knows the company’s position while decisions can still change the outcome, instead of one that reconstructs the quarter after it is over.

The transfer pricing equivalent

Transfer pricing has its own period-end reconstruction, and its own version of the idea. The equivalent of a zero-day close would be determining the arm’s length price on the day the transaction occurs, with the entire analysis concluded on the spot. The field already works in two steps. Prices are set prospectively from a policy when transactions happen, and whether those prices were right is tested afterwards against actual outcomes, usually at year end. Same-day final pricing would fold the second step into the first.

Tax administrations have thought about it too. In 2020 the OECD published Tax Administration 3.0, a discussion paper on where digital tax administration could go. Its storyline for multinationals imagines governments publishing transfer pricing algorithms and rules, certifying the pricing software a company runs on, and engaging in real time only when a flagged transaction needs review. In that vision, compliance shifts from auditing transactions years later to assuring the systems that price them. The paper calls itself a vision rather than a plan, and the direction is clear. Tax administrations can imagine transfer prices that are approved the day they are booked.

What can close on day zero

For part of the workload, our answer is yes, and the sooner the better. Some transactions are already priced by rule. The OECD’s own guidelines let groups charge low value-adding services within the group at a 5 percent markup, several jurisdictions run safe harbours with fixed margins for routine services, and the United States prices qualifying support services at cost. Where the regulator has pre-agreed the answer, nothing is gained by waiting. An agent can apply the markup as the transaction is booked, and the number is final the same day.

The second yes reaches every intercompany transaction, however it is priced. Around every pricing conclusion sits a set of numbers that should agree with each other and often do not. The price booked in the ledger should match the policy. The policy should match the intercompany agreement. The documentation should quote the figures the ledger actually shows, and segmented entity results should be current rather than a year old. In most groups these drift apart during the year and are reconciled backwards, at documentation time or under audit, which is the same scramble the zero-day close removes from accounting. This part of the idea transfers whole. Records, policies, agreements, and documentation can be tied together the day a transaction is booked, and agents that watch the ledger all year make that tying continuous. It is the part of operational transfer pricing we build for.

The part that deserves the year

The end-to-end analysis is where we think transfer pricing sits in a different category of domain. An analysis behaves like writing or like product development, work that moves through stages and changes direction as things are discovered, more than it behaves like bookkeeping. The OECD Transfer Pricing Guidelines resolve comparability into five factors, the contractual terms, the functions performed with the assets used and risks assumed, the characteristics of the property or service, the economic circumstances of the parties and their market, and the business strategies they pursue. Several of these show themselves only over time. Which entity actually controlled a risk is evidenced by conduct across the year rather than by the booking entry. Markets move mid-year. A restructuring that looked routine in the spring can be the year’s controversy by autumn.

The discipline’s own machinery says the same. Prices set during the year are tested against outcomes and trued up at the close. Taxpayers in some jurisdictions correct the booked price in the return through compensating adjustments. For intangibles that are hard to value, the guidelines expressly allow a tax administration to treat how things actually turned out as evidence about the original price. Even an advance pricing agreement, the most prospective instrument the field has, is negotiated over years, runs for a fixed term, and rests on critical assumptions that reopen it when they break. Bookkeeping has revision machinery of its own, estimates trued up as actuals arrive and errors corrected, and there each revision is the exception. In transfer pricing the revision of the price itself is scheduled from the start. The machinery exists because the right answer often cannot be known on the day.

That openness is the point of the arm’s length principle, not a flaw in it. The principle settles the tax base between countries on the economics of each case. The alternative, splitting group profit by a fixed formula, has been proposed for as long as the principle has existed, and OECD members have consistently rejected it, in substance because a formula cannot see the circumstances of the individual case. The profession is comfortable closing the routine cases by formula. For everything else, the fairness of the principle lives in the door it keeps open for factors no formula in the world could have anticipated. A price concluded the day of the transaction and never revisited gives up exactly that.

None of this argues for slowness. The standing complaint about the arm’s length principle is the opposite one, that analyses drag, certainty arrives years late, and disputes run past the facts that caused them. Agents change that. The work inside each stage compresses from weeks to hours or days, waiting time falls out of the cycle, and monitoring runs all year instead of once. Our position is that compressing the stages is different from deleting them. The analysis keeps its stages and its window for judgment, and agents make each pass through that window cheap enough to repeat whenever the facts move. A conclusion reached that way is stronger, and it still is not a same-day conclusion.