Keep documents that show each project's objective, its novelty or technical risk, the work done and the qualifying expenditure. IRAS can ask for them when it reviews a claim. Because Singapore has no approval step and a light filing procedure, the documentation carries the whole weight when IRAS does look, and IRAS's reviews of R&D claims are detailed. This page sets out what IRAS expects, what to keep, how to link it to the figures in the return and where claims fall short.
What the tax authority expects
Companies must keep records for 5 years from the Year of Assessment. For R&D, IRAS's e-Tax Guide describes the documentation it expects for each project: the objective at the outset, the state of knowledge and why the result was not achievable with existing knowledge, the technical risk or novelty, the systematic plan and the work actually done, the results, and the expenditure by category with the basis for staff cost apportionment. For claims above the threshold the R&D claim form collects a summary of this with the return; for all claims the underlying documents must exist.
Two features shape the records. First, IRAS tests novelty or technical risk against the field, so the project documents should set out what was already known and cite it. Second, the tiers depend on category and location, so the cost records must distinguish staff costs and consumables from other expenditure, and Singapore work from overseas work, project by project.
Records to keep during the year
| Record | Shows | Kept by | When |
|---|---|---|---|
| Project proposal or charter | The objective, the state of the art, the novelty or technical risk | Technical lead | At the start |
| Work plans and hypotheses | The systematic approach | Technical lead | Before each phase |
| Experiment records, test reports, prototypes, version control | The work done and the results, including failures | Engineers and scientists | As the work happens |
| Milestone and closing reports | Outcomes; when the technical risk was resolved | Technical lead | At milestones and at the end |
| Timesheets by project | Staff time on R&D by project | All staff whose costs are claimed | Monthly or better |
| Payroll and CPF records | Remuneration and employer CPF behind staff costs | Payroll | Monthly |
| Purchase records coded to projects | Consumables versus other expenditure; capital items separated | Finance | At purchase |
| Contracts with R&D organisations and overseas providers | Who did the work, where, the 60% portion treated as staff costs and consumables | Finance | On engagement |
| Location records | Where each part of the work was carried out | Project manager | Through the year |
| Grant agreements | Funded expenditure to exclude | Finance | On signing |
None of this needs a special system. A product team's tickets, commits and sprint reviews meet the standard if they are tagged to the project and record what was learned; a laboratory's notebooks and test reports do the same.
Linking records to projects and costs
The return reports the deductions by tier; the R&D claim form reports by project. The company's workpapers connect the two. For each project:
| Item | Content |
|---|---|
| Project | Title, field, dates, objective, the three requirements analysis with pointers to the records |
| Staff costs in Singapore | Employee, time portion, remuneration and CPF, source of the apportionment |
| Consumables in Singapore | Items, amounts, use in the project |
| Outsourced R&D | Provider, location, fee, the 60% portion treated as staff costs and consumables |
| Other R&D expenditure | Overheads, software, utilities, travel, apportionment |
| Overseas R&D | Amounts and the relationship to the trade |
| Exclusions and adjustments | Grants, capital items, expenditure converted to cash payout |
| Reconciliation | Project totals to the tiers in the return and to the accounts |
For the cash payout, add the CPF records showing three full-time local employees above S$1,400 a month for at least six months of the basis period, and a note of which expenditure was converted so it is excluded from the deductions. For multinational groups, add the intercompany agreements that show the Singapore entity bears the cost and undertakes the R&D for its own trade.
Common gaps
- No statement of technical risk. Projects described by the product built rather than by what was uncertain.
- Novelty to the company only. Nothing shows that the result was new in the field.
- Time apportionments without records. Year-end percentages for the whole engineering team.
- Tiers mixed. Overheads and software claimed at the enhanced rate; utilities treated as consumables.
- Location untracked. Overseas team costs claimed at the Singapore rate.
- Cash payout double-counted. Expenditure converted to cash also deducted.
- Grants not excluded.
Illustrative example
Illustrative example. Sentosa Sensors Pte Ltd, a hypothetical company, develops a humidity sensor stable in tropical conditions. Its chief engineer's project charter records that published sensors drift beyond tolerance above 90% relative humidity and sets out the hypotheses about coating chemistry. Test reports record each coating and its drift; timesheets code 60% of four engineers' time to the project; finance codes coatings and substrates as consumables, cloud and rent as other expenditure, and the Malaysian subsidiary's testing as overseas R&D. When IRAS queries the YA 2026 claim, Sentosa produces the project workpaper and the charter, and the claim is allowed as filed.
Related pages
Official sources
Educational content only, not tax advice. See the disclaimer.