The product with no budget line¶

A plain-language companion to the draft working paper "The Unpriced Bundle: Decision Suppression in Software Procurement" (v0.2, September 2026). The paper sets out the model, the three propositions with their proofs, the numeric cross-checks, and the Monte-Carlo report; this text explains the ideas.

The paper this condenses Read the full draft working paper (PDF)

The short version. Ask why a European administration still runs Teams and the usual answers are network effects, habit, or the cost of moving. There is a simpler mechanism underneath all of them, and it operates earlier. Teams arrives inside the Microsoft 365 licence at no incremental price, so it has no budget line. Nothing is authorised for it, no contract renews for it, and no procurement officer is ever asked to compare it with anything. A competitor has to cause a decision to exist before its price and its quality can enter one. The paper models that step and draws out the consequence that matters for policy: inside the region where the decision never convenes, cutting the challenger's price changes nothing at all, while an instrument that forces the comparison to happen costs almost nothing and works on every buyer at once.

Decisions have to be convened¶

Standard models of switching costs and network effects act on the terms of a decision: they make the incumbent's option better or the challenger's option worse. They assume the comparison happens. This paper models whether it happens.

For a separately priced product, the comparison is free and recurrent. The contract renews and someone has to sign. Signing means asking whether this is still the right product. For a component folded into a suite, the licence renews as a block. Dropping the component saves nothing, because the suite price does not fall. Convening a comparison means finding an internal champion willing to spend effort defending a new budget line, taking visible career risk for a benefit that mostly accrues to other people.

The model prices exactly that. The champion bears an initiation cost and captures only a share of the gain. The two together define a wedge: the improvement a challenger must offer before anyone inside the organisation bothers to look.

Worth its price, never weighed¶

The first result separates two ways a bundle loses the challenger its buyers. Some buyers would compare the two and still keep the bundled component, because dropping something that has no price saves nothing; economists who study tying already count those. The others would switch if anyone asked the question, and no one does, because the improvement falls short of the wedge. For them the challenger's merits are not weighed and found wanting. They are not weighed.

On the paper's calibrated illustration, 95% of buyers adopt the challenger when the component has its own price, 80% would if the bundle were evaluated, and 40% do when it sits inside a bundle that nobody reviews. Of the 55 points in between, 15 are the price effect tying models already contain and 40 are decisions that never happen.

None of this requires the incumbent to price aggressively, to bundle strategically, or to do anything at all beyond leaving the component unpriced. In the model the incumbent has no strategic response, which is the conservative assumption: letting it fight back could only make the suppressed region larger.

Why subsidies bounce off¶

The second result is the one to hand a policy maker, because it separates this mechanism from the ones people usually assume.

A suppressed buyer is invariant to every subsidy below its own gap, and then flips at it. Price competition below the gap is inert. The challenger cannot buy the occurrence of a decision with its price, and the subsidy that finally does move a buyer pays that buyer's full residual friction in cash, on top of any genuine price difference.

Price alone does not tell the two mechanisms apart, since across many buyers with different gaps take-up rises smoothly with a subsidy under either one. Timing does. A buyer held by a switching cost responds to price and not to events; a suppressed buyer adopts when a reorganisation or an audit or a mandate happens to convene a comparison, whatever the price. That gives a test procurement data can run: adoption that jumps at such events, and moves little with discounts between them, marks a suppressed estate, where migration funding is money spent against the wrong friction.

The cheap instrument¶

The third result is what dissolves the region. Anything that re-creates the decision event does it.

Itemised pricing is the strongest: give the component a price and every renewal becomes an occasion to compare, for every buyer simultaneously, at close to zero public cost. A periodic duty to evaluate does the same on a clock, with an expected delay set by how often it fires. Disclosure of what the component would cost on its own works on the champion's justification cost, shrinking the wedge without firing any event.

The ranking follows from what each one restores. Itemisation deletes the wedge by regulation. A subsidy pays it, buyer by buyer, in cash. The evaluation duty trades delay for not having to legislate about prices.

The Commission's September 2025 Teams commitments are these instruments entering the record. Microsoft offers the suites without Teams at an appreciably lower price, and has to display the offer without Teams wherever it advertises the offer with it. Read through the model, the pricing arm creates the component's price, hence its budget line, hence the possibility of a decision at all. The display arm is the disclosure instrument. The remedy cannot force the comparison to happen off-cycle, where it still needs a champion.

Where it fits¶

The companion paper Cheaper and Losing explains why European cloud providers keep losing share while being several times cheaper: buyers sort by how differentiated their needs are, and the differentiated layer runs on network effects. This paper compounds that from the demand side. Network effects price the ecosystem's pull once a comparison occurs. The bundle makes sure the comparison does not occur.

The two un-bundlings are complements. Supply-side un-bundling, which is what interoperability rules buy, lets a stack be split across providers. Decision-side un-bundling, which is what itemisation buys, lets a component be questioned at all.

The small print¶

This is a short draft working paper and it claims no identification. The Teams record motivates the model; it does not test it. The champion is reduced-form, with effort, career risk, and attention compressed into one pair of parameters that a fuller principal-agent treatment would price separately. One component is analysed on its own, so suite-level strategy is out of scope, as is the incumbent's counter-pricing. The welfare statements count buyer surplus only. The paper supplies a testable signature, the flat-then-jumping response to subsidy, and the European procurement dataset that would support the test is named in the companion taxonomy paper.