State Aid Assessment of Large R&D Projects
R10 is a decade of case-handling practice written up by the people who handled the cases. It shows which test actually decided the outcome, which one was mostly assumed, and how little of either was ever tested in court.
What R10 is, and the one constraint that governs reading it
R10 is a regulatory practice review published in 2006 in a specialist state-aid law quarterly, synthesising ten years of one authority's case handling from 1996 to 2006. It was written by two of that authority's case handlers in a personal capacity. It is descriptive and doctrinal, not empirical, and carries an explicit caveat: the views are the authors' own and are not an official position.
R10's headline is stability. It describes the framework as one of the most stable sets of rules in its field, with a vast body of decision practice and very few court rulings. That combination — high stakes, wide discretion, demanding tests, almost no litigation — is the unusual feature of the decade.
The two-layer assessment
The structural idea is worth extracting from its legal setting. Aid granted under an approved scheme is examined twice, at different moments and to different depths.
- Scheme designed
- First assessment — scheme approved
- Large project draws on the scheme
- Second assessment — individual notification
- Decision
First layer — at scheme approval
- Assesses the scheme as designed, in the abstract
- Sets the rules all aid under it must obey
- Sees no particular project
- Clears ordinary awards without further examination
Second layer — at individual grant
- Triggered by size, not by suspicion
- Rests on the presumed more distortive impact of large projects
- Looks in particular at the incentive effect
- Applies the ordinary criteria more rigorously, not new ones
R10 is explicit about the logic: individual-notification thresholds rest precisely on the presumed more distortive impact of large projects, hence the additional layer. The second layer is not a different test — it is the same criteria applied with more scrutiny, plus one criterion carrying most of the weight.
The notification triggers
Four triggers operated during the period. Only the first depends on both limbs being exceeded together — the detail most often lost in summary.
NOTIFICATION TRIGGERS — HISTORICAL REGULATORY VALUES, FRAMEWORK OF 1996, IN FORCE AT MOST UNTIL 31 DECEMBER 2006
| Situation | Threshold | Notes |
|---|---|---|
| Individual project under an approved scheme | Prior notification where project costs exceed ECU 25 million and the aid has a gross grant equivalent above 5 million | Both limbs must be exceeded. R10 states the same figures elsewhere in the successor currency unit |
| Project under a multi-country intergovernmental research initiative | 40 million project costs and 10 million aid equivalent | Both measured against total project costs across all participating states. Set by authority letter of 2 May 1997 |
| Ad hoc aid, outside any authorised scheme | Always notifiable | Unless below the de minimis threshold, introduced by a 2001 regulation |
| Steel sector | Stricter notification obligations | Under a decision of 18 December 1996, until the relevant sectoral treaty expired |
Regulatory values of one framework in one jurisdiction, for a period ending no later than 31 December 2006, in the unit of account as printed in the source. Not current law.
Where aid is not a straight grant, the second limb must be converted into a gross grant equivalent before the threshold can be tested — soft loans against a reference rate, repayable advances treated as grants, guarantees valued under a separate notice. That machinery, with the stage definitions and permitted intensities, is on R&D stage classification and aid intensity.
The incentive effect — the decisive test
R10 identifies this as the test that determined outcomes in the second layer. Everything else was the ordinary criteria applied more rigorously; this is what the assessment turned on.
WHAT THE INCENTIVE-EFFECT TEST LOOKED AT
| Category | Factors | Source |
|---|---|---|
| Quantifiable factors | Changes in R&D spending; changes in people assigned to R&D activities; R&D spending as a proportion of total turnover | Framework text — the authority had to take particular account of these |
| Contextual factors | Market failures; additional costs of cross-border cooperation; other relevant factors indicated by the notifying state | Framework text |
| Alternative routes to a finding | Aid may also be permitted where it expands the scope of the research or speeds it up — in both cases the intervention has an incentive effect | Framework text |
| Practice-derived | Technical and economic risks; difficulties in financing the project; competition from outside the framework's territory | Decision practice, not framework text |
| Practice-derived | An explicit counterfactual — figures for the case where the large project does not take place | Decision practice, recorded as a requirement on the notifying state |
Framework provisions and practice as recorded in R10, for the period to 31 December 2006. The framework-text versus decision-practice split is R10's own: the second is what case handlers did, not what the rules said.
Two presumption rules shaped how the test was run. The authority may assume a necessary incentive where the recipient is a small or medium enterprise — but large projects are usually run by large firms, so the presumption was in practice unavailable and the state had to prove the effect. The second rule is heightened scrutiny, attached to two situations: close-to-market projects undertaken by large firms, and cases where a significant proportion of the R&D expenditure was already made before the aid application.
How the incentive effect was decided in practice
The impact-on-competition test, and what usually happened to it
Under the general compatibility limb, the authority had to verify that the aid does not adversely affect trading conditions to an extent contrary to the common interest. R10's account of how that substantive test operated is the most candid passage in the paper.
Where the analysis was performed, R10 preserves the reasoning in one semiconductor case, and it is instructive because the intuitive answer is wrong. The beneficiary was at that moment the only significant producer within the framework's territory in the stand-alone flash-memory segment — an apparently strong position.
The reasoning in the semiconductor case
Establish the true geographic market
Semiconductors were treated as a worldwide market, not a regional one. Local dominance is only meaningful if the market is local.
Rank the beneficiary in that market
On a worldwide view the beneficiary was only the eighth largest — a very different position from the one the regional view suggested.
Ask what drives location decisions here
Site choice was driven by production cost reduction rather than proximity to consumers, so aid to a site confers no customer-facing advantage.
Conclude on distortive effect
No significant distortive effect could be anticipated. The strong-looking regional position was outweighed by global ranking, market geography and the logic of plant siting.
The transferable part is the sequence, not the outcome: market definition first, position within the properly defined market second, and the economics of the decision the aid influences third — an ordering close to how risk and competitive position are handled in risk and uncertainty in R&D financial analysis.
Choice of legal basis
Two treaty limbs were available. R10 records that the general compatibility limb — aid to facilitate the development of certain activities — was the basis for by far the predominant part of large R&D cases.
THE TWO BASES AS R10 DESCRIBES THEM
| Basis | What it requires | How often it was used |
|---|---|---|
| General compatibility limb | Aid must not adversely affect trading conditions to an extent contrary to the common interest | By far the predominant part of large R&D cases |
| Common-interest limb | The state must demonstrate in the notification that an important project of common interest exists. Available for individual projects only — never for schemes | Rare. R10 cites a large microelectronics cooperation programme |
As recorded in R10 for the period to 31 December 2006. The framework's permitted intensities applied under the common-interest limb too, so the more exceptional basis unlocked no higher ceiling — it changed the argument the state had to make, not the money available.
How practice changed across the decade
The substantive rules barely moved. What changed sat around the edges, in instruments, procedure and the authority's own working methods.
- 1996Framework enters force with the individual-notification thresholds. Stricter, separate notification obligations set for the steel sector by a December decision
- May 1997Higher, separate thresholds set by letter for the multi-country intergovernmental research initiative
- 1999 to 2000Published reference rates relied on for the aid element in soft loans; a guarantees notice becomes the reference for valuing guarantees. The relevant world-trade-agreement provisions on R&D support expire from 1 January 2000
- 2001De minimis regulation and the small-and-medium-enterprise block exemption arrive. The latter is expressly not applicable to large individual R&D grants
- 2001 onwardAround 65 large R&D cases individually assessed, with formal investigation in only about ten — proportionally fewer than over the whole period from 1996
- 2004An amending regulation clarifies eligible-cost treatment for equipment, buildings and land; it does not apply to large individual grants but is used as interpretive authority. An obsolescence communication removes the 30-working-day deadline for individual cases under approved schemes, leaving the two-month deadline
- Across the periodPre-notification meetings become standard, joint technical assessment across services routine, and commissioned external expertise grows — driven by stage-classification difficulty and the incentive-effect test
- 2005 to 2006Framework prolonged at most to 31 December 2006 to permit revision. A strengthened economic approach and an explicit balancing test are announced, weighing positive impact towards an objective of common interest against distortion of trade and competition
Taking 1996 as the reference year, R10 records around 25 formal investigations opened, approximately 10 withdrawn after the opening decision, and approximately 10 ending in a negative or partly negative decision, sometimes with recovery of aid already paid. These are findings about one authority's caseload, not rates that generalise. The single annulment turned on procedure — the two-month deadline had expired before the authority decided to open the formal investigation — so a decade of high-stakes, discretionary assessment produced almost no judicial guidance on the substantive tests.
What travels, and what does not
The values do not travel. The structure does, and it is a serviceable model for any two-stage approval regime — including internal capital allocation under the financial frame for R&D management.
The structural ideas worth keeping
- Approve the mechanism once; re-examine the individual case only when size crosses a stated line
- Make the trigger conjunctive where two dimensions of risk matter — project cost and size of support
- Name one test as decisive rather than scoring everything equally
- Require the counterfactual: what happens if this does not proceed
- Drop the presumption of good faith where the applicant could have proceeded anyway
- Anchor timing on when the decision was made, not when the paperwork cleared
- Define the market before assessing position within it
One structural caveat applies as across this stream. R10 reached this library inside a reading set assembled for a literature review, and it is the only regulatory analysis in that set — see eleven R&D management papers compared. It was not designed to describe funding regimes in general.
What to carry forward
- Everything numeric here is a historical regulatory value of one framework, in one jurisdiction, for a period ending no later than 31 December 2006. Never quote it as current.
- Large projects were assessed twice — at scheme approval and again at grant — because size was presumed to bring distortion, not because anything was suspected.
- The first notification trigger was conjunctive: project cost above one limb and aid above the other, together.
- The incentive effect was decisive. Large firms had to prove it, and the proof included a counterfactual for the project not proceeding.
- Starting work after applying nationally but before approval did not by itself defeat the incentive effect. The question was whether the prospect of the grant was sufficiently likely to induce the decision.
- The competition test was usually assumed from compliance with the other conditions. Where it was performed, market definition came first and local dominance did not settle it.
Frequently asked questions
Can I use these thresholds to work out whether our project needs notification?
No. They belong to a framework that applied at most until 31 December 2006 and a replacement was already announced when R10 was written. Use the page to understand how the assessment was structured, then obtain the framework currently in force and take advice on it.
Why does the first trigger need both limbs exceeded?
Because the two limbs measure different things: the scale of the research effort and the scale of the public support. A very large project attracting modest support, and a small project attracting disproportionate support, are different risks. The rule of the period required project cost above one figure and aid gross grant equivalent above the other, together.
What made the incentive effect the decisive test?
Because the other criteria in the second layer were the ordinary R&D rules applied more rigorously, while this one asked whether the aid changed anything. The framework directed the authority to take particular account of changes in R&D spending, in people assigned to R&D and in R&D as a share of turnover, and to treat aid as less favourable where the effect was not evident.
Does starting a project before approval destroy the incentive effect?
Not under this framework. The operative moment was the application to the national authorities, not the authority's approval, and a court held that commencement before notification cannot by itself defeat inducement. The authority had to assess whether the prospect of the grant was sufficiently likely to have induced the decision.
Why was the impact-on-competition test hardly ever performed?
R10 records that for large R&D projects the absence of adverse effect on trading conditions has usually been assumed where all the framework's conditions were met, particularly the incentive effect. Explicit analysis appeared in only a few decisions. The paper does not explain when the authority chose to depart from that practice.
Where do the aid intensities and R&D stage definitions live?
On the sibling page. This page covers the two-layer structure, the notification triggers, the incentive-effect and competition tests, the choice of legal basis and how practice changed. Stage classification, permitted intensities, bonuses and eligible costs are treated separately, and carry the same period-bound warning.
References and source attribution
- R10 — large R&D projects: a decade of regulatory practice under an R&D state-aid framework. Legal and regulatory practice review in a specialist state-aid law quarterly, 2006, covering practice from 1996 to 2006; 8 printed pages; 57 footnotes; no tables and no figures, all content in prose and quoted framework provisions. Written by two case handlers of the authority in a personal capacity, with an explicit statement that the views are not an official position. Sections used here: the double assessment, the notification obligation and its thresholds, the legal basis, the incentive effect, the impact on competition, and the account of how practice changed.
- The framework of 1996 on state aid for research and development, prolonged several times and in force at most until 31 December 2006, together with the authority letter of 2 May 1997 setting separate thresholds, the December 1996 sectoral decision on steel, the 2001 de minimis regulation, the 2000 guarantees notice, and the 2004 amending regulation on eligible costs. All are instruments quoted within R10 and were not supplied to this library. Their values are historical regulatory values of their period.
- The court ruling on inducement and pre-notification commencement, and the single annulment on procedural grounds, are recorded in R10 without being reproduced here.
- Eleven copyrighted journal articles on R&D project management, supplied as a reading set assembled by a student for a literature review and profiled for this library. Front matter, abstracts, framework sections, tables and figures were read; article bodies were not reproduced, and all content here is paraphrase. R10 is the only regulatory analysis in that set, and the set is not a systematic or representative survey of the field.
- Supplied teaching source for this library (research methods and research process materials). Used here for page conventions and voice; it does not treat public funding or regulatory assessment.
Suggested questions for Ask KEVOS
- Explain the two-layer assessment structure and where it could apply inside our own capital approval process.
- Draft the counterfactual section of a funding case — what happens if this project does not proceed.
- What evidence would demonstrate that a research project sits outside our normal day-to-day R&D activity?
- Summarise why a strong regional market position did not defeat the competition test in the semiconductor case.
- List the questions a reviewer should ask before accepting that public support changed a firm's decision.
- What has to be checked before any historical funding threshold is used in a live business case?