Return on Research Capital: Measuring R&D Productivity
Table of Contents
If a company spent $30 million on research and development last year and earned $180 million of gross profit this year, its return on research capital (RORC) under a one-year convention is 6.0×. That ratio can help you decide what to investigate. It cannot tell you whether last year’s projects earned $180 million, whether the next $30 million will pay off, or whether the business earned its cost of capital.
To use RORC well, decide which spending could plausibly relate to the output, check what the company expensed versus capitalized, and pair the ratio with a measure that reflects what the research was meant to accomplish.
Return on Research Capital Formula
RORC can refer to different analytical measures. This article starts with the one-year convention used in some equity research: compare this fiscal year’s gross profit with the prior fiscal year’s reported R&D expense.[1]
Gross profit is revenue less cost of revenue. If a filing calls that dollar amount gross margin, use the dollars rather than the margin percentage. The denominator is an annual expense flow, not a balance of accumulated research assets. A 6.0× reading means the business reported $6 of gross profit this year per $1 of R&D expense last year. Existing products, price changes and other inputs all contribute to gross profit, so the multiple does not isolate what the research produced.
Return on invested capital (ROIC) answers a different question: how much after-tax operating profit a company earns against the capital invested in its operations. ROIC can be compared with a cost of capital expressed on the same basis. The gross-profit-to-R&D multiple is a screening ratio and has no such hurdle.
Choose the Lag Before Ranking Companies
Start with when spending could show up in your chosen outcome. A new software feature can reach users before it makes a measurable contribution to paid adoption. A drug candidate may pass several development milestones before it generates product sales at all. If projects take different lengths of time, an average of spending across relevant years can make the screen less dependent on a single unusual year. Even a well-chosen lag cannot match consolidated gross profit to a specific spending cohort.[2]
Microsoft: One Numerator, Three Spending Windows
In its Form 10-K for the year ended June 30, 2025, Microsoft reported $281,724 million of revenue and $87,831 million of cost of revenue, leaving $193,893 million of gross profit. Holding that amount fixed shows how much of the difference between ratios comes from which year’s expense you use.[3]
| Denominator | R&D expense ($ millions) | Multiple |
|---|---|---|
| FY2024 (one-year lag) | 29,510 | 6.57× |
| FY2023 (two-year lag) | 27,195 | 7.13× |
| Average of FY2023 and FY2024 | 28,352.5 | 6.84× |
The two-year ratio is higher because FY2023 spending was lower. It gives no evidence that two years is the better lag for Microsoft. With the same filing you can also calculate FY2025 R&D intensity: $32,488 million of R&D expense divided by $281,724 million of revenue, or 11.53%. That percentage describes investment effort relative to the size of the business; it is neither the same ratio nor a measure of output.
Read the Accounting Notes Before Comparing Companies
Under U.S. GAAP, R&D costs within the general research and development guidance are usually expensed as incurred. Separate guidance can apply: for example, certain costs of software to be marketed are capitalized after technological feasibility is established and before general release. Under IAS 38, research is expensed and internal development is recognized as an intangible asset once all the recognition criteria are met. So two businesses doing similar work can report different R&D expenses. Look for expense embedded in another line, current additions to development assets, and a separate amount for amortization of past costs.[4][5] See intangible assets and goodwill for the distinction between reporting an asset and valuing it.
Hurco: What the Expense Line Leaves Out
Machine tool maker Hurco reported about $33.0 million of gross profit for the year ended October 31, 2025. Its FY2024 financial statement notes identify $3.9 million of R&D expense within selling, general and administrative expenses and a separate $1.6 million of marketed-software development costs capitalized that year.[6]
- Reported expense basis: $33.0 million ÷ $3.9 million ≈ 8.46×.
- Disclosed expense plus new capitalized costs: $33.0 million ÷ ($3.9 million + $1.6 million) ≈ 6.00×.
The second measure is a denominator sensitivity using the disclosed amounts. It is neither total R&D cash spending nor a second observation on the same reported-expense definition. Hurco also reported $1.6 million of FY2024 software amortization. That identical amount allocates earlier capitalized costs; it is not the $1.6 million of new additions used here. Hurco had an FY2025 operating loss of about $10.3 million despite the positive gross-profit multiple. Hurco also identified machine sales volume, product mix and costs related to tariffs as factors affecting FY2025 gross profit. Those factors help explain why the multiple cannot isolate the results of its engineering work.[6]
Cross-industry rankings add another problem: a licensing company and a manufacturer can have very different gross margins and ways of funding development. Even if you reconcile expense and additions for both, the two multiples still combine different profit structures and development cycles. Read the Microsoft and Hurco examples as separate tests of the measure.
Match the Measure to the Research Goal
A gross-profit screen is most useful when commercial results are already visible and the businesses have similar economics. If the research is meant to cut factory costs, a busy quarter or a higher selling price could lift gross profit without any change in the process. Measure the cost or time change directly if you can hold the relevant operating conditions reasonably constant.[7]
| Question | Measure | What to check |
|---|---|---|
| How much does the company commit? | R&D expense as a share of revenue | Identify capitalized costs and any large changes in revenue. This is input intensity. |
| Are new products selling? | Share of sales from launches within a defined period, such as the past three years | Use the same launch window; check for acquired products, cannibalization and differences in margins. |
| Is a pipeline progressing before sales? | Milestone conversion, such as the share of a cohort that reaches the next trial stage | Define the starting cohort, time allowed and treatment of projects still in progress. Reaching a stage does not guarantee approval or sales. |
| Did a process improve? | Change in cost per unit, defect rate or lead time | Compare like operating conditions and separate volume, mix and input price effects. |
| Is a project worth funding? | Risk-adjusted project net present value | Include incremental outlays, future cash flows, timing and failure scenarios. The result depends on the assumptions. |
Capitalize R&D Separately to Analyze ROIC
If your question is how research affects company-wide ROIC, you can model past expensed R&D as an investment whose benefits last more than one year. An analyst can spread that cost across an assumed useful life and add the unamortized balance to invested capital. This is a change to the analyst’s model, not to the issuer’s financial statements. The useful life asks how many periods the model allocates a cohort’s cost over; the lag in the RORC screen asks which spending period to compare with an observed outcome.[8]
Build the Research Asset One Spending Year at a Time
Suppose a company expensed $30 million, $45 million, $60 million and $75 million of R&D in Years 1 through 4. Assume there are no earlier unamortized cohorts and none of this spending was already capitalized in the reported accounts. Add each year’s spending at year-end and amortize it evenly over three years, starting the following year. For example, the $45 million Year 2 cohort starts amortizing in Year 3 at $15 million a year. Its balances at the ends of Years 2 through 5 are $45 million, $30 million, $15 million and zero.
| Spending year | Opening balance | Additions | Amortization | Closing balance |
|---|---|---|---|---|
| Year 1 | 10 | 0 | 10 | 0 |
| Year 2 | 30 | 0 | 15 | 15 |
| Year 3 | 60 | 0 | 20 | 40 |
| Year 4 | 0 | 75 | 0 | 75 |
| Total | 100 | 75 | 45 | 130 |
The closing balance reconciles to $100 million opening balance + $75 million of spending − $45 million of amortization = $130 million. In a real model, collect enough years to calculate both the opening and closing balances. If some costs are already recorded as development assets, keep the adjustment to the expensed slice and leave the already capitalized slice in the baseline, unless you explicitly model and reconcile both.
Adjust Operating Profit and Capital Together
Suppose Year 4 reported operating profit (EBIT) is $100 million and invested capital is $500 million at the start and $550 million at the end. Add back the $75 million of expense you’re modeling and deduct $45 million of amortization. Add the opening research asset to opening capital and the closing research asset to closing capital. Apply the same illustrative 25% normalized operating tax rate in both cases to calculate net operating profit after tax (NOPAT).
| Measure | Reported basis | R&D adjusted |
|---|---|---|
| Operating profit | 100 | 130 |
| NOPAT at 25% tax | 75 | 97.5 |
| Opening invested capital | 500 | 600 |
| Closing invested capital | 550 | 680 |
| Average invested capital | 525 | 640 |
| ROIC: NOPAT ÷ average capital | 14.29% | 15.23% |
Using a two-year life with the same convention gives $52.5 million of Year 4 amortization and research asset balances of $82.5 million at the start and $105 million at the end. Adjusted ROIC drops to 14.85%. Choose plausible lives from product replacement cycles, customer use and evidence of how fast earlier innovations lose relevance. Shorten the assumed life or write down a cohort if the modeled benefits no longer make sense. Show how the conclusion changes across plausible lives. The adjustment can raise or lower ROIC because it changes both profit and capital; the balance is modeled cost remaining, not an appraisal of what the research assets could sell for.
How to Use RORC in Company Research
- Choose a business and an outcome. Compare businesses that sell similar products into similar markets, or start with one company over time. Segment analysis can help, but use a segment ratio only if spending and gross profit both cover that segment.
- Write down the definitions and lag. Use the same currency, units, fiscal period length, and accounting scope in both amounts. Document the base window and an alternative before seeing which ranks best.
- Reconcile the numbers with the filing. Check expense versus capitalized additions and whether acquisitions or disposals leave the two years covering different businesses. Keep a reported measure and label any broader version. If comparable data are missing, skip the comparison or mark that period as a break in the series.
- Inspect what moved. Higher gross profit from pricing or sales mix, lower prior R&D, or a change in reporting can all lift the ratio. When R&D is zero or unavailable, the ratio is undefined or cannot be computed. Flag a very small denominator instead of treating an outsized multiple as evidence of productivity.
- Check an outcome closer to the work. Look for launch sales, milestone progress, unit economics or project cash flows. If a peer ranking reverses under plausible assumptions, investigate the difference before drawing a conclusion.
Frequently Asked Questions
Sources
- Klett + Hamer, Eli Lilly equity analysis, example of the one-year formula in use.
- Zvi Griliches, Issues in Assessing the Contribution of Research and Development to Productivity Growth, sections 2.1 and 2.4–2.5.
- Microsoft FY2025 Form 10-K, consolidated income statements and Note 1. Ratios calculated from reported data.
- FASB, ASU 2023-05 (reproduced ASC 730 language) and FRIC materials (Appendix B, software capitalization).
- IFRS Foundation, IAS 38 Intangible Assets; IASB staff, January 2026 software development discussion paper.
- Hurco FY2025 Form 10-K, Item 7 and Note 1. Ratios calculated from rounded reported data.
- OECD and Eurostat, Oslo Manual 2018, chapter 8, sections 8.3.1–8.3.2.
- Aswath Damodaran, Research and Development Expenses and Research and Development Expenses: Implications for Profitability Measurement and Valuation. The worked example uses illustrative spending and timing assumptions.
