Brief
Innovation & Digital Policy · July 2026
New firm-level evidence from the OECD shows two things at once: Canadian firms that build core digital technologies become measurably more productive, and Canada's share of exactly that kind of innovation is shrinking against the very partners it is now diversifying trade toward. With Budget 2025's innovation reforms now law and a patent box on the table, the evidence arrives at the moment policy is actually moving.
Page 1 · Executive summary
Bottom line
The OECD's 2025 study of Fourth Industrial Revolution (4IR) patenting, built on the full population of Canadian firms rather than a sample, finds that when Canadian firms develop core 4IR technologies such as connectivity, software, and IT hardware, their labour and multi-factor productivity improve relative to peers.1 That is the productivity payoff Canada has been chasing for a decade, and it is measurable: a firm that doubles its 4IR patent output is associated with roughly a 0.9 percent gain in multi-factor productivity, 1.4 percent more employment, and 1.7 percent more value added within three years.1 The same study shows Canada's 4IR patenting growing at a third of the global pace and its advantage in core technologies collapsing, while Korea, Sweden, Germany, and the United Kingdom, the economies Canada's trade diversification is aimed at, hold or strengthen theirs.1 Canada is producing less of precisely the innovation that the evidence says pays, at the moment it most needs the growth.
The findings, in plain terms:
What Canada needs to do
The rest of this brief is the detail: what the study found, how Canada compares beyond the United States, why the current programs miss, and what should change.
The detail
Canada's productivity record frames everything that follows. Labour productivity growth has been among the worst in the G7 and declined every year since the pandemic except a short 2020 rebound. Multi-factor productivity took one of its largest drops since 1980 in 2021. Business investment fell away after the 2014 oil-price crash and never recovered its trajectory.12
The innovation inputs tell the same story. Business R&D has been flat near 1 percent of GDP for a decade against an OECD average of 2 percent, even though Canadian government support for R&D is above the OECD average. Canada's total R&D intensity has stagnated near 1.7 percent of GDP while the OECD average sits at 2.7 percent, and the World Intellectual Property Organization lists Canada among the economies with the largest declines in share of global R&D since 2000.13 Generous inputs, weak uptake, poor output: that is the standing puzzle. The OECD's patent study is useful because it moves the question from the economy in aggregate to the firms where the decision to innovate actually happens.
The study itself, by Calvino, Dechezleprêtre, De Lyon, Dernis, and Vítková at the OECD's science and technology directorate, identifies patents across 21 4IR technology fields, grouped into core technologies (connectivity, software, IT hardware), enabling technologies (AI, data management, data security, and others), and application domains (smart vehicles, consumer goods, healthcare, agriculture). Its Canadian econometrics run on the full population of firms through confidential Statistics Canada microdata, every firm rather than a commercial sample, which is what makes its findings unusually credible.1
Canadian innovation debates default to the US comparison, and on that axis the story is familiar: the United States gained massive 4IR specialization in core and enabling technologies over the decade while Canada lost its own.1 But Canada is deliberately reducing its dependence on the American market, deepening CETA with the European Union, trading under the CPTPP with Japan and now the United Kingdom, and holding a bilateral agreement with Korea since 2015. The relevant question has changed: how does Canada's digital innovation base compare with the partners it is pivoting toward?
The answer from the patent data is uncomfortable. The economies Canada is diversifying into are, almost uniformly, 4IR-strengthening economies.1
| Economy | 4IR position, 2017-21 | Decade trend |
|---|---|---|
| Korea | Top four in IP5 patent families and at USPTO; R&D intensity of 5 percent of GDP, second only to Israel.13 | Held near the top. |
| Sweden | Top ten in IP5 4IR families with roughly a quarter of Canada's population; R&D intensity 3.6 percent.13 | Out-patents Canada in 4IR. |
| United Kingdom | Grew its revealed advantage in enabling and application 4IR technologies.1 | Strengthening. |
| Germany | Increased specialization in application domains; industrial 4IR anchor of the EU.1 | Strengthening where it chooses. |
| China | Third-largest 4IR contributor, jumping from about 5 to 16 percent of IP5 filings; dominant in core technologies.1 | Surged. |
| Canada | Seventh at USPTO, tenth in IP5, under 2 percent of world 4IR patents; core-technology advantage fell from 1.4 to 0.7.1 | Declined across all three categories. |
Why this matters for diversification is straightforward. Trade agreements set the terms of exchange; the innovation base determines what a country has to exchange. If Canada enters its deepened European, Indo-Pacific, and Korean relationships as a supplier of resources and application-layer products while its partners own the core digital stack, the diversification succeeds on volume and fails on value, reproducing with new partners the branch-plant pattern Canada has spent decades regretting with the old one. The same OECD data shows Canadian 4IR inventions are protected in only two jurisdictions on average, fewer than almost any other economy, with 37 percent first filed in the United States and 29 percent protected only there.1 Canadian innovators are not even filing where Canadian trade policy is going.
Trade agreements set the terms of exchange. The innovation base determines what you have to exchange. Canada is fixing the first while the second erodes.
Global 4IR patenting more than doubled between 2007-11 and 2017-21 and now accounts for 19 percent of everything filed at the US patent office. Canada participated at roughly a third of the pace, about 3 percent annual growth against 8 percent globally.1 The sharpest measure is revealed technological advantage: Canada's core-technology index fell from 1.4 to 0.7, from clearly specialized to clearly not, while enabling technologies held (1.15 to 1.14) and application domains slipped (1.3 to 1.18). Within the detail, old strengths in connectivity, power supply, and consumer goods disappeared, while new specializations emerged in agriculture (an index of 3.5, among the highest recorded), 3D systems, and healthcare, and core AI jumped from 1.4 percent to 21.9 percent of Canada's enabling-technology patents.1
The consolation is calibre. Canadian 4IR patents score above the world average on originality, radicalness, and citations by later inventions.1 Canada's problem is not the quality of what it invents but the volume, the drift of the mix away from core technologies, and the direction of travel.
One company reshapes how the national numbers should be read. BlackBerry accounts for nearly 30 percent of all 4IR patents filed by Canadian residents between 2000 and 2021, peaking above 1,200 a year in 2010-12 and falling to roughly 30 by 2021. Nortel, the second name on the list, is gone entirely.1 Read one way, the national decline is less broad-based than the curves suggest. Read the other way, it is worse: Canada's 4IR capacity was so concentrated that one firm's fall moved the national statistics, and nothing of comparable scale replaced it. The rest of the leaderboard, ATI, TD Bank, Thales Canada, BCE, Geotab, Mitel, RBC, Shopify, is a long tail of much smaller portfolios.1 Concentration was the risk; the risk materialized; the replacement never arrived. And the pipeline points the wrong way: young Canadian firms patent mostly in application domains, healthcare and agriculture above all, not in the core technologies where the productivity return concentrates.1
The econometric core of the paper asks who files 4IR patents and what happens as they file more. Larger and more productive firms are more likely to hold them, in both the Canadian and international samples, but the entry thresholds are lower for 4IR than for other patenting: firms can get into this game at smaller size and lower productivity than conventional R&D-heavy invention requires.1
The growth payoff, stated plainly
In the full-population Canadian data, a firm that doubles its 4IR patent output is associated with roughly a 0.9 percent gain in multi-factor productivity, a 1.4 percent increase in employment, and a 1.7 percent increase in value added over a three-year window, relative to comparable firms, controlling for its other patenting.1 Modest per firm, compounding across an economy, and consistent with the federal government's own consultation finding that patent-owning businesses grow faster and pay higher wages.8
The gains are not evenly distributed. The productivity improvement is driven entirely by core 4IR technologies: connectivity, software, IT hardware. Application-domain patents are associated with firms scaling up but not becoming more productive. And the evidence suggests smaller firms capture the largest productivity gains, while large firms mainly grow.1 Invention in the plumbing pays differently than invention in the fixtures.
The correlation-not-causation caveat applies and the authors make it themselves. Three other cautions belong on the record: patents are the tip of the innovation iceberg and say nothing about adoption by the firms that never file one; the data ends in 2021, so the entire generative-AI wave is invisible and nobody yet knows the post-2021 shape of these curves; and "Canada losing advantage" partly reflects others surging rather than Canadian activity falling absolutely.1 But the pattern, population-wide, robust to controls, specific to core technologies, and strongest where theory predicts, is the best firm-level evidence yet assembled that the Fourth Industrial Revolution's productivity gains have begun to materialize for the Canadian firms that participate in building it.
Canada does not lack innovation programs. It lacks alignment between what the programs reward and what the evidence says pays. Three files illustrate it.
The Scientific Research and Experimental Development credit delivers over 4.2 billion dollars a year and is the backbone of Canadian R&D support.6 The effectiveness question is not whether it funds research but whether it moves the needle: through a decade of generous SR&ED, business R&D sat flat at 1 percent of GDP while most OECD countries rose.1 Critics have long argued the program's structure favours large and foreign-owned claimants over the domestic firms it was meant to grow, concerns the federal government has itself acknowledged in launching its review, and the 2011 expert panel on federal R&D support reached a version of the same conclusion, recommending a rebalance from indirect tax credits toward direct, targeted support.6
Budget 2025 delivered the largest SR&ED reform in a generation, now law: the enhanced 35 percent credit's expenditure limit doubled to 6 million dollars, phase-out thresholds raised to a 15-to-75-million-dollar range, eligibility extended to Canadian public corporations, and capital expenditures restored. From April 2026 the CRA adds an elective pre-claim approval process, 90-day processing for approved claims, and simplified forms.6 These are real improvements to generosity and predictability. What they do not do is direct the subsidy toward the kind of R&D with the demonstrated productivity return. SR&ED remains technology-neutral by design; the OECD evidence suggests neutrality has a cost when the payoff is concentrated in core digital technologies.
The Global Innovation Clusters, the renamed superclusters, carry roughly 2 billion dollars in federal commitments. The criticism record is substantial and specific. The Auditor General found the program's founding claim of a fifty-fold return had no analytical basis and no plan to measure it. The Parliamentary Budget Officer found spending far behind schedule in the early years. Innovation scholars, including the University of Toronto's Innovation Policy Lab, argue the program's indicators, project counts, co-investment totals, partnership numbers, measure activity rather than what matters: productivity growth, firms' sales and exports, and whether business R&D actually rises. And the Council of Canadian Innovators' critique lands close to this brief's evidence: the program counts IP assets rather than asking whether Canadian firms own strategic IP that lets them operate globally.7 Set against the OECD findings, the design flaw is exact: a program built to generate collaborative projects in application domains, in an economy whose productivity problem lives in core technologies, measured by indicators that cannot detect the difference.
The most consequential open file is the patent box, a reduced tax rate on income from Canadian-held IP, designed to fix the pattern the OECD data documents: Canadian inventions filed primarily in the US and commercialized elsewhere. The Commons science committee recommended it, consultations ran in 2024, Budget 2025 stated the intent to implement one, and details remain pending as of mid-2026; British Columbia is meanwhile running its own patent box consultation through June 2026.89 Alongside it, Budget 2025 renewed the SME-facing IP toolkit, ElevateIP, the Patent Collective, and the NRC's IP Assist, committed to a national Intellectual Property Performance Review, and introduced a Productivity Super-Deduction allowing immediate expensing of patents, data network infrastructure, and computers.6 The direction is right. The test is whether the patent box arrives with substance requirements that reward IP developed and held in Canada, and whether the Performance Review measures ownership and productivity rather than filing counts.
1. Tilt support toward core digital technology. The productivity return concentrates in connectivity, software, and IT hardware. SR&ED's new generosity, the clusters' next funding phase, and the Strategic Innovation Fund should carry an explicit core-4IR lens, weighting support toward the technologies with the demonstrated payoff rather than treating a smart-consumer-goods project and a network-infrastructure project as equivalent.
2. Build on the verified strengths rather than lamenting the lost ones. The data shows where Canada is genuinely strong and getting stronger: smart agriculture, at a revealed advantage of 3.5 among the highest anywhere; health technology; 3D systems; a core AI share that jumped from 1.4 to 21.9 percent of enabling patents; and a quantum ecosystem Ottawa is now backing with 334 million dollars through the Defence Industrial Strategy.16 The play is to build core-technology capacity underneath these strengths, the data systems, connectivity, and software infrastructure of agri-tech and health-tech, so the productivity-bearing layer of these value chains is Canadian too.
3. Fix what programs measure. Adopt the innovation scholars' standard: evaluate the clusters and the major funds on productivity growth, business R&D intensity, firm sales and exports, and Canadian IP ownership, not on project and partnership counts. What gets measured is what program managers optimize.
4. Deliver the patent box with teeth. Announce the design, include substance requirements tying the preferential rate to R&D performed and IP held in Canada, and pair it with the promised IP-backed financing framework so young firms can borrow against the patents the regime encourages them to keep.
5. Point the trade diversification agenda at 4IR demand. CETA, the CPTPP, and bilateral technology partnerships should be worked as market-creation instruments for Canadian core technology: digital trade chapters, mutual recognition, government-procurement access, and standards cooperation that give Canadian connectivity, cyber, and data firms a first-mover route into the partners Canada is courting anyway. Filing strategy should follow: two-jurisdiction protection is not a diversification posture.
6. Solve for scale, not just start-ups. The BlackBerry lesson is that Canada's 4IR base lacked redundancy at the top. The Budget 2025 capital measures, the billion-dollar Venture and Growth Capital Catalyst and the forthcoming 750-million-dollar growth-stage strategy, are aimed at the right gap; their test is whether they keep scaling firms, and their IP, Canadian.6
This evidence lands in an unusually active policy moment. Budget 2025, tabled in November 2025 as the government's first, made innovation, AI, and IP its economic centrepiece and its SR&ED reforms have received Royal Assent. The patent box decision is pending federally while British Columbia consults on its own through June 2026. A national Intellectual Property Performance Review is being launched. A Minister of Artificial Intelligence and Digital Innovation now exists, Statistics Canada is standing up a program to measure AI's economic impact, and the CRA's overhauled SR&ED administration takes effect in April 2026.69 The question this brief raises, whether the newly generous toolkit will point at the technologies where the productivity evidence says the return actually is, will be answered in the design details of these files over the coming year. It is the right year to be asking it.
The shape of the whole argument is simple. The technologies of the Fourth Industrial Revolution are not a productivity promise waiting to be redeemed someday; for the firms building their core, in the only full-population dataset we have, the returns have begun. Canada's problem is that fewer of its firms are building them, in a mix drifting toward the lower-payoff end, on a base that never replaced its fallen champions, while the partners it is diversifying toward pull ahead. That is a more specific problem than "productivity," and specific problems are the ones policy can aim at.