Canadian Agriculture: Struggling to Grow, or Struggling to be Recognized? (Part 3)

Table of ContentsTogglePart 3: Returns on Investment, Adoption Gaps, and the Road AheadPicking up the ThreadParts1and2of this series established a productive tension at the heart of Canadian agricultural policy. On one hand, Canada’s total factor productivity (TFP) growth has held up well by international standards, consistently outperforming comparators like Australia and tracking close to the United States over the 2001–2020 period. On the other hand, the investment inputs that sustain long-run productivity growth such as public R&D expenditure, extension infrastructure, and technology adoption capacity show signs of strain. Theriskis not that Canadian agriculture is in crisis today, but that it could be running on borrowed time.Part 2found Canada’s public agricultural R&D spending has declined in real terms and that the private sector has not stepped in to fill the gap to the same extent as it has in Australia, where private R&D has grown at roughly 4.2% annually since 2005. It also flagged the adoption gap: generating research is only half the equation, and Canada’s extension and knowledge-transfer infrastructure is comparatively underexamined.Part 3 takes up three questions that flow from that foundation: Why is TFP growth declining broadly, not just in Canada but across most major agricultural producers? What does the return on public R&D investment actually look like for Canada, Australia, and the United States? And, where are the most tractable leverage points for improving outcomes, in funding levels, regulatory conditions, or the space between research output and farm-level adoption?Why is TFP Declining Globally?The TFP deceleration documented inPart 1is not a Canadian anomaly. Declining or stagnating productivity growth is visible across most major agricultural producers over the 2001–2020 period, including countries that have increased public R&D investment. That pattern immediately complicates the most common policy prescription: simply spend more on agricultural research.Fuglie 2018explores agricultural productivity at a global scale, finding steady growth across the world, though he presents that growth has slowed among industrialized countries, while developing countries represent the highest growth. Fuglie also discusses potential reasons for the lower growth observed in industrialized countries, citing decreases in per acre yield growth, (though this tends to be offset by intensification), the expansion of cropland over other land uses, as well as movement towards no-till and other conservation practices. On the other hand, factors influencing the higher growth rates among developing countries span from investment in research, policy reform, and rising farm family education leading to technology adoption.The Australia PuzzleAustralia offers the sharpest illustration of the disconnect between investment and productivity. As Part 2 established, private R&D in the Australian agricultural sector has grown at approximately4.2% annuallysince 2005, a substantial and sustained increase. Yet over the same period, Australia’s agricultural TFP has turned negative, making it something of an outlier even among countries experiencing slower growth. The Australian Bureau of Agricultural and Resource Economics and Sciences(ABARES) reportingshows that Australia’s public-sourced R&D funding has grown at a comparatively modest 1.4% annually since 2005–06, while funding for extension and advisory services declined at an average annual rate of 5.7% over the same window, and federal- and state-government R&D expenditure declined at average annual rates of 3.0% and 1.9% respectively. While the country has had success in growing privately sourced funding, there is still a prevailing question of to what extent this has offset the slowdown in public funding for the sector’s overall productivity. Would outcomes have been worse if that role was not filled, as seems to be the case in Canada currently? Thispaperreviews and discusses the Australian policy landscape and outcomes within a larger time frame concerning agricultural TFP.Table 1 below puts this in relief: Australia’s mean TFP trend growth rate between 2001 and 2020 was approximately −0.7%, the weakest of the three comparator countries, despite R&D spending in the same period that was proportionally substantial relative to its agricultural output base. Two quick definitions before the numbers: GPV (gross production value) is the total market value of a country’s agricultural output, and PPP (purchasing power parity) is the exchange-rate adjustment used to make spending comparable across countries. Except where noted, all monetary figures in this series — R&D spending and GPV alike — are expressed in standardized, PPP-adjusted international USD, not in domestic currencies (A$, C$, or US$).Table 1. Mean Public R&D Spending and TFP Growth, 2000-2020 (Canada, Australia, United States)CountryMean R&D 2000-2020 (M PPP)Mean TFP Growth 2001-2020 (%)Latest YearLatest GPV (M USD)AUS741.1-0.7202250,982CAN850.42.4202251,558USA5,482.70.72022365,711Sources: GRAPE (Global Research on Agriculture: Personnel & Expenditures dataset) v1.0.0, Zenodo 15507361 (R&D, M 2017 PPP USD); USDA ERS (U.S. Department of Agriculture, Economic Research Service) AgTFPInternational2022 (TFP, OLS log-linear trend growth rate); FAOSTAT (Food and Agriculture Organization Corporate Statistical Database) QV element 152 (GPV). Relationship is descriptive.Several non-exclusive explanations emerge from the literature for this kind of disconnect between investment and measured productivity:Lag effects –Current TFP reflects the R&D investment of years prior, not of today. Recentestimatessuggest lags as long as 50 years to show up in productivity metrics. A sustained increase in private R&D since 2005 may not yet be visible in TFP outcomes through 2020.Climate variability –TFP is particularly difficult to measure cleanly in agricultural systems with high climate variability. Australia’sexposureto recurring drought conditions introduces substantial noise into productivity estimates, potentially masking real output gains.Innovationcomplexity –More recent agricultural innovations tend to be more incremental, more systems-integrated, and slower to diffuse to farm-level practice than the varietal and mechanization breakthroughs of earlier decades.Input cost effects –Rising input costs inflate the denominator of TFP even as gross output grows, compressing measured productivity ratios.The Broader PatternAustralia is not alone in showing TFP deceleration despite investment. The United States, with by far the largest absolute public agricultural R&D spending of the three countries, also shows modest mean TFP growth of around 0.7% annually between 2001 and 2020. Canada, with the highest mean TFP growth at 2.4% over the same period, invested at roughly the same absolute scale as Australia but in a much larger agricultural economy (see Table 1; R&D from GRAPE v1.0.0, Zenodo 15507361; TFP trend growth from USDA ERS AgTFPInternational2022).The implication is that the TFP deceleration story is not primarily a funding story—at least not in the short run.USDA ERS analysisof the global slowdown points to structural factors that likely apply across all three focal countries to varying degrees: a slowing diffusion frontier for the major productivity-enhancing technologies of the 20th century, policy and public-attitude barriers that slow adoption of newer technologies in developed countries specifically, and measurement challenges that are particularly acute in agriculture. This is offered as the most likely explanation given the cross-country pattern, not as a finding established by this analysis on its own.Figure 1. Agricultural TFP trend growth rates by decade (top) and mean public R&D over the preceding 10 years (bottom) for Canada, Australia, and the United States. The TFP bar row covers six decades (1961–2020); the R&D line covers five decades (1980–2020) because GRAPE v1.0.0 coverage for Canada begins in 1960, providing only one year within the 1961–1969 preceding-decade window—that point is omitted for Canada as unreliable. Australia and the United States have sufficient coverage from 1970 onwards. The peak in Canada’s R&D line at 2000 reflects mean spending over 1991–2000, Canada’s highest-spending decade in real terms before the sustained decline documented in Parts 1 and 2.Sources: USDA ERS AgTFPInternational2022; GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. Relationship is descriptive—confounded by terms of trade, private R&D, and farm structureWhat Does the Return on R&D Investment Look Like?This section presents the original analytical contribution of Part 3: an implied-return figure that asks a deliberately simple question. If we take gross agricultural production value (GPV) as the output measure and public R&D expenditure as the input measure, how does the ratio between them compare across countries and over time?This is an illustrative exercise, not a causal return-on-investment estimate. Agricultural output is shaped by weather, global commodity prices, private sector investment, farm structure, and policy—none of which are held constant in this calculation. The framing is best understood as a rough measure of output efficiency relative to public investment: how much agricultural output is associated, descriptively, with each dollar of public R&D.All figures use GPV in constant 2014–2016 international USD (FAOSTAT, element code 152) against public R&D expenditure from GRAPE v1.0.0 (van Dijk et al. 2025, Zenodo 15507361) in million 2017 PPP (purchasing power parity) USD. The two currency bases are not identical, the FAOSTAT Geary-Khamis and GRAPE 2017 ICP bases differ by approximately 3–5%, and the mismatch is flagged as a stated limitation throughout.Implied Return Ratios (2022)Table 2 presents the GPV per public R&D dollar at the most recent available data point (2022), under three lag assumptions: contemporaneous, with R&D lagged 10 years, and with R&D lagged 15 years. The 10- and 15-year lags reflect the consensus in the agricultural R&D literature that the productivity impact of research investment takes a decade or more to materialize at the farm level.Table 2. Implied GPV/Public R&D Ratios by Country and Lag Specification (2022)CountryLagGPV (M USD)Public R&D (M PPP)GPV/R&D RatioAustraliaContemporaneous50,982670.676.0R&D lagged 10 yrs50,982849.160.0R&D lagged 15 yrs50,982654.677.9CanadaContemporaneous51,558870.659.2R&D lagged 10 yrs51,558843.261.1R&D lagged 15 yrs51,558877.458.8United StatesContemporaneous365,7115,11471.5R&D lagged 10 yrs365,7115,25769.6R&D lagged 15 yrs365,7115,94261.6Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. GPV in constant 2014–16 int. USD; R&D in million 2017 PPP USD. Descriptive ratio—not a causal return estimate.Several features of Table 2 are worth noting. Canada’s contemporaneous GPV/R&D ratio (59.2) is the lowest of the three countries, reflecting both its moderate agricultural output scale and its higher public R&D spending relative to Australia. When R&D is lagged 10 years, Canada’s ratio rises modestly to 61.1, the smallest shift of the three countries, suggesting that the relationship between past R&D and current output is relatively stable in the Canadian case. By the 15-year lag, however, the ratio dips slightly below its starting point, to 58.8. Read together, the three Canadian figures (59.2 → 61.1 → 58.8) trace the lag structure of public R&D directly: a decade of spending growth is associated with a modest near-term gain, while spending from 15 years prior—an older, smaller R&D base—corresponds to a slightly lower ratio. This is consistent with the lag effects discussed in Section 2 and with Canada’s current position in the R&D cycle, where recent spending growth has not yet had a full 15 years to show up in output.Australia, by contrast, shows the largest swing across lag specifications: from 76.0 contemporaneously to 60.0 at a 10-year lag, as the elevated R&D spending of the prior decade enters the denominator. This pattern is consistent with the interpretation that the private R&D growth documented in Part 2 has raised Australia’s investment base substantially without yet generating proportionate output gains, reinforcing the TFP puzzle described in Section 2.The United States shows a steady decline across lag specifications (71.5 → 69.6 → 61.6), reflecting the consistently high and growing public R&D base. The large absolute GPV keeps the ratio comparatively elevated even as the denominator grows.Implied Return Over Time: The Indexed FigureFigure 2 makes the cross-country divergence visible in a way the raw ratios in Table 2 understate. Each country’s series is indexed to its own first available year (1990 for the contemporaneous panel; 2000 and 2005 for the 10- and 15-year lag panels, respectively, since lagging the R&D series shortens the usable run). On the contemporaneous panel, Canada’s indexed series climbs steadily above its own baseline, reaching roughly 150–190 by the 2010s, while the United States rises more modestly and Australia oscillates close to its own starting point with no clear trend. The picture shifts once R&D is lagged. At the 10-year lag, Canada’s climb is even more pronounced, its series rises furthest above 100 of the three countries, while Australia’s series sits persistently below 100 for most of the 2000s and 2010s, consistent with the TFP puzzle described in Section 2: a rapidly growing R&D base without a matching rise in output. At the 15-year lag, Canada remains the strongest performer through most of the series, but Australia’s line turns sharply upward in the final years shown, climbing back toward and above 100. Taken together, Canada’s indexed return has been on a more consistently rising trajectory across all three lag specifications than either comparator, while Australia’s pattern is the most volatile, swinging from clearly below baseline to a late recovery depending on which lag is used. This volatility is itself informative: it suggests Australia’s investment-to-output relationship is more sensitive to the choice of lag than Canada’s, which is consistent with Australia’s R&D base having grown unevenly over the period rather than at a steady rate.What this means in plain terms: no matter how the timing of R&D spending relative to output is sliced, Canada has consistently generated more agricultural output per public research dollar than either Australia or the United States — and that edge has grown over time rather than shrinking. Australia’s story is the opposite: its return on R&D has swung unpredictably, falling well behind for most of the 2000s and 2010s before a late recovery, which points to a much less stable link between what it spends and what it gets back.Figure 2. GPV/R&D ratio indexed to own-country baseline (base year = 100), for Canada (blue), Australia (orange), and the United States (green), across contemporaneous, 10-year lag, and 15-year lag specifications. Grey lines in the contemporaneous panel = all other countries in the GRAPE sample, capped at the 95th percentile; the lagged panels omit the background countries since fewer of them have enough years of data to survive the lag.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. Descriptive ratio—not a causal return estimate.The raw ratios in Figure 3 underline the scale gap that the indexed figure deliberately removes. In the most recent year, the United States generates roughly $71 of agricultural output per public R&D dollar contemporaneously, compared with roughly $76 for Australia and $59 for Canada, figures driven as much by the size of each country’s R&D base as by underlying productivity. Canada’s raw ratio is consistently the lowest of the three across all three lag specifications, which mechanically reflects its position between Australia’s smaller agricultural economy and the United States’ much larger one, relative to a public R&D budget closer in absolute size to Australia’s. The trajectories also diverge: Canada’s raw ratio rises steadily from the early 1990s through the most recent data, narrowing a gap that was much wider in the 1990s; the United States holds the highest contemporaneous ratio for most of the series before Australia’s late upturn overtakes it; and Australia’s ratio is comparatively flat through the 2000s before that sharp rise in the final years. These patterns mirror the indexed story in Figure 2, but Figure 3 makes clear that the underlying dollar magnitudes are not directly comparable across countries without that indexing step.Breaking these numbers down: the raw dollar figures aren’t comparable across countries on their own, because a country with a much bigger farm sector will naturally show a bigger number here, regardless of how efficiently it turns research spending into output. That’s why Figure 2’s indexed version, which tracks each country’s change relative to its own starting point rather than the raw dollar amount, is the one to rely on for comparing performance between Canada, Australia, and the United States.Figure 3. Raw GPV/R&D ratio (M USD per M PPP USD) for Canada, Australia, and the United States, across contemporaneous, 10-year lag, and 15-year lag specifications. Note: large values reflect output scale relative to R&D spending and are not directly comparable across countries without indexing. Use Figure 2 for cross-country comparison.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361.R&D Spending and TFP Growth: The Cross-Country PatternTable 1 summarizes the country-level averages underlying Figure 4. Canada’s position, high mean TFP growth (2.4%) with a moderate public R&D base (mean $850M PPP, 2000–2020) places it above the OLS (ordinary least squares) fit line in Figure 4, suggesting that Canadian agriculture has generated comparatively strong productivity growth per dollar of public research investment over this period. Australia sits on the opposite side of the fit line—its mean R&D base is roughly comparable to Canada’s, but its mean TFP growth is negative, placing it well below the line and reinforcing the disconnect discussed in Section 2a. The United States sits almost directly on the fit line: its much larger R&D base corresponds to a TFP growth rate close to what the global cross-country relationship would predict, with no strong over- or under-performance in either direction. Read together, the three focal countries span the range of outcomes visible in the broader scatter—Canada above the line, the United States on it, and Australia below it—despite Canada and Australia having broadly similar public R&D levels, which is itself evidence that funding level alone is a weak predictor of productivity outcome.Figure 4. Country-level scatter of mean public agricultural R&D (log₁₀ scale, M 2017 PPP USD) versus mean TFP trend growth rate (%, 2001–2020). Dashed line = OLS fit (descriptive, not causal). Canada (blue), Australia (orange), and USA (green) are highlighted.Sources: R&D: GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. TFP: USDA ERS AgTFPInternational2022. Relationship is confounded by country size, farm structure, and other factors.Barriers: Supply Side and Demand SideThe standard framing for agricultural innovation policy focuses on the supply side: how much is being invested in R&D, by whom, and through what institutional channels. Parts 1 and 2 of this series largely engaged with that framing. But the data in Sections 2 and 3 above suggest that the relationship between investment and productivity outcome is looser than a straightforward supply-side model would predict.This section examines both sides of the innovation chain, supply and demand, with particular attention to the demand side, which receives comparatively little attention in the Canadian policy literature.Supply-Side Barriers (Recap)Part 2covered the main supply-side barriers in detail; a brief summary is included here for completeness.Intellectual Property (IP) protections and farmers’ privilege –The balance between IP protection for plant breeders and the traditional farmers’ privilege to save seed shapes the private R&D incentive structure. The Canadian Food Inspection Agency (CFIA) finalized updated Plant Breeders’ Rights Regulations in April 2026, narrowing farmers’ privilege to small grain crops; see Section 5 for the policy detail and the resulting debate.SR&ED complexity –The Scientific Research and Experimental Development tax creditprogram favours well-capitalized firms with dedicated compliance resources, creating a barrier for smaller agricultural innovators.Regulatory timelines –Approval timelines for genetically modified varieties represent a material lag between research output and farm-level availability.AAFC research centre consolidation –The scheduled closure and consolidation of Agriculture and Agri-Food Canada (AAFC) research facilities represents a near-term supply-side risk that has not yet been fully reflected in the spending data. AAFC announced in January 2026 that it would close seven research centres and satellite farms, part of a broader federal workforce reduction; the House of Commons Standing Committee on Agriculture and Agri-Food has since called for the closures to be reversed (seeRealAgriculture).Demand-Side Barriers: The Underexamined SideThe gap between research output and changed farm practice is poorly measured in Canada, and it represents the most original contribution of this series. Even if supply-side constraints were fully resolved—if R&D spending were adequate, IP frameworks balanced, and regulatory timelines shortened—there would remain a substantial question about whether innovations reach farms and are adopted at scale. The available evidence suggests the answer is often no, or not quickly.Farm financial constraints –Part 2 documented the margin pressures facing Canadian farm operations. Capital-constrained farms face higher effective hurdle rates for technology adoption, particularly for precision-agriculture tools that require upfront equipment investment and ongoing subscription costs.Extension services –ABARES data for Australia shows that public agricultural advisory and extension funding has declined at an average annual rate of approximately5.7%in real terms since 2005–06, a sustained multi-decade contraction even as private R&D investment has grown over the same period. While no equivalent national time-series exists for Canada, the institutional trend toward commodity organization-led extension and away from publicly funded provincial extension services is broadly similar, and has been reinforced recently by the closure and staffing reductions at several AAFC research facilities discussed in Section 4a. This represents a structural weakening in the knowledge-transfer link between research output and farm decision-making.Technology adoption rates –The 2021 Census of Agriculture provides the most granular available picture of precision-agriculture adoption among Canadian farms.Figure 5 reveals a pronounced geographic concentration of precision-agriculture adoption. GPS/auto-steer adoption is highest in Saskatchewan (about 47% of all farms) and Manitoba (about 41%), with Alberta close behind (about 31%); Ontario shows comparatively stronger uptake of GIS mapping and variable-rate application than the Prairie provinces, though at lower GPS adoption (around 23%). New Brunswick, Nova Scotia, and Newfoundland and Labrador show minimal adoption across all four technology categories, generally in the single digits. Prince Edward Island is a notable exception to the Atlantic pattern, with GPS and variable-rate adoption (around 20% each) closer to Ontario’s than to its immediate regional neighbours—likely reflecting its potato and row-crop heavy production mix, which is more compatible with precision-ag tools than the mixed and horticultural operations more common elsewhere in Atlantic Canada.Figure 5. Share of all farms reporting use of selected precision-agriculture practices (GPS/auto-steer, GIS mapping, variable-rate application, and drones) by province, 2021 Census of Agriculture. Percentages are calculated against the total number of farms in each province (all NAICS — North American Industry Classification System — categories), since oilseed/grain-specific denominators are near-zero in several Atlantic provinces; western provinces (AB, SK, MB), where grain and oilseed production dominates, are the most directly comparable.Source: Statistics Canada, Census of Agriculture 2021, Table 32-10-0379-01. Released 2022-05-11.This pattern likely reflects a combination of farm size effects (Prairie operations are on average much larger and more mechanized), crop type (row crops are better suited to precision-agriculture tools than many Eastern mixed or horticultural operations), and capital availability. The adoption data do not allow clean isolation of these factors, but the regional concentration is itself a policy-relevant finding: precision-agriculture supports that are designed around Prairie-scale operations may not transfer effectively to the rest of the country.Scale and operator age –Both farm size and operator age are well-documented correlates of technology adoption. Part 1 documented the ongoing consolidation of Canadian farm operations and the aging of the operator population, and the2021 Census of Agriculturesharpens that picture. The number of census farms fell to 189,874 in 2021, down 1.9% from 2016, even as average farm size has grown substantially over the longer run—from roughly 676 to 809 acres between 2001 and 2021. The distribution is increasingly bimodal: farms of 2,240 acres or more accounted for only 8.9% of all farms in 2021 but generated 37.3% of total operating revenue, while farms under 240 acres made up 52.3% of the total but contributed only about a quarter of revenue. On the operator side, the average age of Canadian farm operators rose to 56.0 years in 2021 (median 58.0), and the share of operators aged 55 and older climbed from 54.5% in 2016 to 60.5% in 2021, while the share of younger operators continued to decline. Combined with the precision-agriculture adoption gradient in Figure 5, this points toward a structural pattern rather than a regional quirk: adoption is concentrated among a shrinking pool of larger, Prairie-dominant operations, while the median Canadian farm—smaller, and run by an operator closer to retirement age than to a multi-decade payback horizon—sits further from the technology-adoption frontier described in Part 1. Larger operations have stronger financial incentives to adopt labour-saving and yield-improving technologies; older operators face shorter payback horizons and, on average, lower returns to technology learning.Where are the Leverage Points?The preceding sections point toward a more granular set of leverage points than the standard policy discussion about R&D funding levels. The evidence does not support the conclusion that simply increasing public agricultural R&D spending is sufficient to close Canada’s productivity trajectory risk. The Australia case illustrates that investment and outcome can diverge substantially, and the adoption data suggest that the return on existing research is partially constrained by the capacity to put it into practice.This section identifies the areas where the evidence points toward tractable improvements, without prescribing a specific funding level or policy instrument.IP regime modernisation –This is no longer purely prospective: the CFIA finalized amendments to the Plant Breeders’ Rights Regulations in April 2026 (SOR/2026-74), with the changes coming into force shortly after. The most consequential change narrows farmers’ privilege, the right to save and reuse seed from a protected variety, to small grain crops such as cereals and pulses, removing the privilege for horticultural and ornamental varieties where it had previously applied.Producer groups broadly supported the changeas a confidence-building measure for plant breeders investing in Canada, while the National Farmers Union opposed it on the grounds that narrowing farmers’ privilege increases production costs and entrenches reliance on proprietary varieties. A well-calibrated IP framework that maintains private sector R&D incentives while preserving farmer access to saved seed represents a genuine leverage point, one that does not require new public expenditure—though the regulation as finalized has already drawn pushback on exactly that balance, and its effect on adoption and farm-level costs will only be observable with a few years of data.SR&ED simplification –Simplifying the SR&ED tax credit regime for smaller agricultural innovators and agribusiness firms would improve the return on an existing public commitment without necessarily increasing its total cost.Extension and knowledge transfer –The case for investment in knowledge-transfer infrastructure is arguably stronger than the case for additional research investment at the margin, precisely because the adoption gap is the underexamined variable. Extension services operate at the link between research output and farm practice, the point where the evidence suggests the current system is weakest.International R&D spillovers –Canada benefits disproportionately from research conducted in the United States and elsewhere. The size of the US public agricultural R&D enterprise (mean $5.5B PPP, 2000–2020) dwarfs Canada’s, and much of the resulting knowledge is applicable in Canadian conditions. A strategy that emphasizes Canada’s capacity to absorb and adapt internationally generated knowledge, rather than simply increasing domestic R&D outlays, may generate higher returns per public dollar.Industry-led R&D models –Australia’s Rural Research and Development Corporation (RDC) model—in which industry levy funds are matched by government contributions and channelled through commodity-specific bodies—offers a structural comparison worth examining in the Canadian context. There are 15 RDCs covering Australia’s major agricultural sectors, collectively investing roughly AUD$825 million annually; theAustralian government matches eligible R&D levy contributions up to 0.5% of each industry’s gross value of production. The model does not simply increase total investment; it changes the governance of how investment decisions are made and how knowledge transfer is organized.Conclusion: What the Series has EstablishedAcross three posts, this series has worked through the evidence on Canadian agricultural productivity with a specific goal: to replace a crisis narrative with a more precise one. The crisis framing—in which Canada’s agricultural sector is falling behind, underinvested, and in need of urgent intervention—is not entirely wrong, but it lacks specificity about where the problems are, which levers are most tractable, and what the counterfactual is.What the data show is a more textured picture. Canada’s TFP performance is genuinely strong by international standards, and its implied return on public R&D investment, descriptive as that measure is, compares favourably with Australia and the United States on several specifications. The risk is not that Canadian agriculture is already in decline, but that the investment trajectory and structural conditions that have sustained that performance may not hold over the next decade.The most important finding of this series may be the one that is hardest to measure: the adoption gap. The chain from public R&D investment to research output to farm-level practice to productivity outcome has multiple links, and the weakest link is not always the one that gets the most policy attention. Extension capacity, precision-agriculture adoption rates, farm financial constraints, and the scale and age structure of the operator population all shape whether research investment actually reaches the farm gate.The dominant policy narrative focuses on spending levels. The more useful question and the one this series has tried to make clearer is what happens to that spending after it leaves the public accounts.A fourth and final post in this series will take up the question left open here: what do the solutions actually look like? Drawing on the concerns raised across Parts 1, 2, and 3, Part 4 will weigh the leverage points introduced in Section 5—IP modernisation, SR&ED simplification, extension investment, international R&D spillovers, and Australia’s industry-led RDC model—against each other, and assess where the evidence points toward genuine leverage and where it does not. Table of ContentsTogglePart 3: Returns on Investment, Adoption Gaps, and the Road AheadPicking up the ThreadParts1and2of this series established a productive tension at the heart of Canadian agricultural policy. On one hand, Canada’s total factor productivity (TFP) growth has held up well by international standards, consistently outperforming comparators like Australia and tracking close to the United States over the 2001–2020 period. On the other hand, the investment inputs that sustain long-run productivity growth such as public R&D expenditure, extension infrastructure, and technology adoption capacity show signs of strain. Theriskis not that Canadian agriculture is in crisis today, but that it could be running on borrowed time.Part 2found Canada’s public agricultural R&D spending has declined in real terms and that the private sector has not stepped in to fill the gap to the same extent as it has in Australia, where private R&D has grown at roughly 4.2% annually since 2005. It also flagged the adoption gap: generating research is only half the equation, and Canada’s extension and knowledge-transfer infrastructure is comparatively underexamined.Part 3 takes up three questions that flow from that foundation: Why is TFP growth declining broadly, not just in Canada but across most major agricultural producers? What does the return on public R&D investment actually look like for Canada, Australia, and the United States? And, where are the most tractable leverage points for improving outcomes, in funding levels, regulatory conditions, or the space between research output and farm-level adoption?Why is TFP Declining Globally?The TFP deceleration documented inPart 1is not a Canadian anomaly. Declining or stagnating productivity growth is visible across most major agricultural producers over the 2001–2020 period, including countries that have increased public R&D investment. That pattern immediately complicates the most common policy prescription: simply spend more on agricultural research.Fuglie 2018explores agricultural productivity at a global scale, finding steady growth across the world, though he presents that growth has slowed among industrialized countries, while developing countries represent the highest growth. Fuglie also discusses potential reasons for the lower growth observed in industrialized countries, citing decreases in per acre yield growth, (though this tends to be offset by intensification), the expansion of cropland over other land uses, as well as movement towards no-till and other conservation practices. On the other hand, factors influencing the higher growth rates among developing countries span from investment in research, policy reform, and rising farm family education leading to technology adoption.The Australia PuzzleAustralia offers the sharpest illustration of the disconnect between investment and productivity. As Part 2 established, private R&D in the Australian agricultural sector has grown at approximately4.2% annuallysince 2005, a substantial and sustained increase. Yet over the same period, Australia’s agricultural TFP has turned negative, making it something of an outlier even among countries experiencing slower growth. The Australian Bureau of Agricultural and Resource Economics and Sciences(ABARES) reportingshows that Australia’s public-sourced R&D funding has grown at a comparatively modest 1.4% annually since 2005–06, while funding for extension and advisory services declined at an average annual rate of 5.7% over the same window, and federal- and state-government R&D expenditure declined at average annual rates of 3.0% and 1.9% respectively. While the country has had success in growing privately sourced funding, there is still a prevailing question of to what extent this has offset the slowdown in public funding for the sector’s overall productivity. Would outcomes have been worse if that role was not filled, as seems to be the case in Canada currently? Thispaperreviews and discusses the Australian policy landscape and outcomes within a larger time frame concerning agricultural TFP.Table 1 below puts this in relief: Australia’s mean TFP trend growth rate between 2001 and 2020 was approximately −0.7%, the weakest of the three comparator countries, despite R&D spending in the same period that was proportionally substantial relative to its agricultural output base. Two quick definitions before the numbers: GPV (gross production value) is the total market value of a country’s agricultural output, and PPP (purchasing power parity) is the exchange-rate adjustment used to make spending comparable across countries. Except where noted, all monetary figures in this series — R&D spending and GPV alike — are expressed in standardized, PPP-adjusted international USD, not in domestic currencies (A$, C$, or US$).Table 1. Mean Public R&D Spending and TFP Growth, 2000-2020 (Canada, Australia, United States)CountryMean R&D 2000-2020 (M PPP)Mean TFP Growth 2001-2020 (%)Latest YearLatest GPV (M USD)AUS741.1-0.7202250,982CAN850.42.4202251,558USA5,482.70.72022365,711Sources: GRAPE (Global Research on Agriculture: Personnel & Expenditures dataset) v1.0.0, Zenodo 15507361 (R&D, M 2017 PPP USD); USDA ERS (U.S. Department of Agriculture, Economic Research Service) AgTFPInternational2022 (TFP, OLS log-linear trend growth rate); FAOSTAT (Food and Agriculture Organization Corporate Statistical Database) QV element 152 (GPV). Relationship is descriptive.Several non-exclusive explanations emerge from the literature for this kind of disconnect between investment and measured productivity:Lag effects –Current TFP reflects the R&D investment of years prior, not of today. Recentestimatessuggest lags as long as 50 years to show up in productivity metrics. A sustained increase in private R&D since 2005 may not yet be visible in TFP outcomes through 2020.Climate variability –TFP is particularly difficult to measure cleanly in agricultural systems with high climate variability. Australia’sexposureto recurring drought conditions introduces substantial noise into productivity estimates, potentially masking real output gains.Innovationcomplexity –More recent agricultural innovations tend to be more incremental, more systems-integrated, and slower to diffuse to farm-level practice than the varietal and mechanization breakthroughs of earlier decades.Input cost effects –Rising input costs inflate the denominator of TFP even as gross output grows, compressing measured productivity ratios.The Broader PatternAustralia is not alone in showing TFP deceleration despite investment. The United States, with by far the largest absolute public agricultural R&D spending of the three countries, also shows modest mean TFP growth of around 0.7% annually between 2001 and 2020. Canada, with the highest mean TFP growth at 2.4% over the same period, invested at roughly the same absolute scale as Australia but in a much larger agricultural economy (see Table 1; R&D from GRAPE v1.0.0, Zenodo 15507361; TFP trend growth from USDA ERS AgTFPInternational2022).The implication is that the TFP deceleration story is not primarily a funding story—at least not in the short run.USDA ERS analysisof the global slowdown points to structural factors that likely apply across all three focal countries to varying degrees: a slowing diffusion frontier for the major productivity-enhancing technologies of the 20th century, policy and public-attitude barriers that slow adoption of newer technologies in developed countries specifically, and measurement challenges that are particularly acute in agriculture. This is offered as the most likely explanation given the cross-country pattern, not as a finding established by this analysis on its own.Figure 1. Agricultural TFP trend growth rates by decade (top) and mean public R&D over the preceding 10 years (bottom) for Canada, Australia, and the United States. The TFP bar row covers six decades (1961–2020); the R&D line covers five decades (1980–2020) because GRAPE v1.0.0 coverage for Canada begins in 1960, providing only one year within the 1961–1969 preceding-decade window—that point is omitted for Canada as unreliable. Australia and the United States have sufficient coverage from 1970 onwards. The peak in Canada’s R&D line at 2000 reflects mean spending over 1991–2000, Canada’s highest-spending decade in real terms before the sustained decline documented in Parts 1 and 2.Sources: USDA ERS AgTFPInternational2022; GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. Relationship is descriptive—confounded by terms of trade, private R&D, and farm structureWhat Does the Return on R&D Investment Look Like?This section presents the original analytical contribution of Part 3: an implied-return figure that asks a deliberately simple question. If we take gross agricultural production value (GPV) as the output measure and public R&D expenditure as the input measure, how does the ratio between them compare across countries and over time?This is an illustrative exercise, not a causal return-on-investment estimate. Agricultural output is shaped by weather, global commodity prices, private sector investment, farm structure, and policy—none of which are held constant in this calculation. The framing is best understood as a rough measure of output efficiency relative to public investment: how much agricultural output is associated, descriptively, with each dollar of public R&D.All figures use GPV in constant 2014–2016 international USD (FAOSTAT, element code 152) against public R&D expenditure from GRAPE v1.0.0 (van Dijk et al. 2025, Zenodo 15507361) in million 2017 PPP (purchasing power parity) USD. The two currency bases are not identical, the FAOSTAT Geary-Khamis and GRAPE 2017 ICP bases differ by approximately 3–5%, and the mismatch is flagged as a stated limitation throughout.Implied Return Ratios (2022)Table 2 presents the GPV per public R&D dollar at the most recent available data point (2022), under three lag assumptions: contemporaneous, with R&D lagged 10 years, and with R&D lagged 15 years. The 10- and 15-year lags reflect the consensus in the agricultural R&D literature that the productivity impact of research investment takes a decade or more to materialize at the farm level.Table 2. Implied GPV/Public R&D Ratios by Country and Lag Specification (2022)CountryLagGPV (M USD)Public R&D (M PPP)GPV/R&D RatioAustraliaContemporaneous50,982670.676.0R&D lagged 10 yrs50,982849.160.0R&D lagged 15 yrs50,982654.677.9CanadaContemporaneous51,558870.659.2R&D lagged 10 yrs51,558843.261.1R&D lagged 15 yrs51,558877.458.8United StatesContemporaneous365,7115,11471.5R&D lagged 10 yrs365,7115,25769.6R&D lagged 15 yrs365,7115,94261.6Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. GPV in constant 2014–16 int. USD; R&D in million 2017 PPP USD. Descriptive ratio—not a causal return estimate.Several features of Table 2 are worth noting. Canada’s contemporaneous GPV/R&D ratio (59.2) is the lowest of the three countries, reflecting both its moderate agricultural output scale and its higher public R&D spending relative to Australia. When R&D is lagged 10 years, Canada’s ratio rises modestly to 61.1, the smallest shift of the three countries, suggesting that the relationship between past R&D and current output is relatively stable in the Canadian case. By the 15-year lag, however, the ratio dips slightly below its starting point, to 58.8. Read together, the three Canadian figures (59.2 → 61.1 → 58.8) trace the lag structure of public R&D directly: a decade of spending growth is associated with a modest near-term gain, while spending from 15 years prior—an older, smaller R&D base—corresponds to a slightly lower ratio. This is consistent with the lag effects discussed in Section 2 and with Canada’s current position in the R&D cycle, where recent spending growth has not yet had a full 15 years to show up in output.Australia, by contrast, shows the largest swing across lag specifications: from 76.0 contemporaneously to 60.0 at a 10-year lag, as the elevated R&D spending of the prior decade enters the denominator. This pattern is consistent with the interpretation that the private R&D growth documented in Part 2 has raised Australia’s investment base substantially without yet generating proportionate output gains, reinforcing the TFP puzzle described in Section 2.The United States shows a steady decline across lag specifications (71.5 → 69.6 → 61.6), reflecting the consistently high and growing public R&D base. The large absolute GPV keeps the ratio comparatively elevated even as the denominator grows.Implied Return Over Time: The Indexed FigureFigure 2 makes the cross-country divergence visible in a way the raw ratios in Table 2 understate. Each country’s series is indexed to its own first available year (1990 for the contemporaneous panel; 2000 and 2005 for the 10- and 15-year lag panels, respectively, since lagging the R&D series shortens the usable run). On the contemporaneous panel, Canada’s indexed series climbs steadily above its own baseline, reaching roughly 150–190 by the 2010s, while the United States rises more modestly and Australia oscillates close to its own starting point with no clear trend. The picture shifts once R&D is lagged. At the 10-year lag, Canada’s climb is even more pronounced, its series rises furthest above 100 of the three countries, while Australia’s series sits persistently below 100 for most of the 2000s and 2010s, consistent with the TFP puzzle described in Section 2: a rapidly growing R&D base without a matching rise in output. At the 15-year lag, Canada remains the strongest performer through most of the series, but Australia’s line turns sharply upward in the final years shown, climbing back toward and above 100. Taken together, Canada’s indexed return has been on a more consistently rising trajectory across all three lag specifications than either comparator, while Australia’s pattern is the most volatile, swinging from clearly below baseline to a late recovery depending on which lag is used. This volatility is itself informative: it suggests Australia’s investment-to-output relationship is more sensitive to the choice of lag than Canada’s, which is consistent with Australia’s R&D base having grown unevenly over the period rather than at a steady rate.What this means in plain terms: no matter how the timing of R&D spending relative to output is sliced, Canada has consistently generated more agricultural output per public research dollar than either Australia or the United States — and that edge has grown over time rather than shrinking. Australia’s story is the opposite: its return on R&D has swung unpredictably, falling well behind for most of the 2000s and 2010s before a late recovery, which points to a much less stable link between what it spends and what it gets back.Figure 2. GPV/R&D ratio indexed to own-country baseline (base year = 100), for Canada (blue), Australia (orange), and the United States (green), across contemporaneous, 10-year lag, and 15-year lag specifications. Grey lines in the contemporaneous panel = all other countries in the GRAPE sample, capped at the 95th percentile; the lagged panels omit the background countries since fewer of them have enough years of data to survive the lag.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. Descriptive ratio—not a causal return estimate.The raw ratios in Figure 3 underline the scale gap that the indexed figure deliberately removes. In the most recent year, the United States generates roughly $71 of agricultural output per public R&D dollar contemporaneously, compared with roughly $76 for Australia and $59 for Canada, figures driven as much by the size of each country’s R&D base as by underlying productivity. Canada’s raw ratio is consistently the lowest of the three across all three lag specifications, which mechanically reflects its position between Australia’s smaller agricultural economy and the United States’ much larger one, relative to a public R&D budget closer in absolute size to Australia’s. The trajectories also diverge: Canada’s raw ratio rises steadily from the early 1990s through the most recent data, narrowing a gap that was much wider in the 1990s; the United States holds the highest contemporaneous ratio for most of the series before Australia’s late upturn overtakes it; and Australia’s ratio is comparatively flat through the 2000s before that sharp rise in the final years. These patterns mirror the indexed story in Figure 2, but Figure 3 makes clear that the underlying dollar magnitudes are not directly comparable across countries without that indexing step.Breaking these numbers down: the raw dollar figures aren’t comparable across countries on their own, because a country with a much bigger farm sector will naturally show a bigger number here, regardless of how efficiently it turns research spending into output. That’s why Figure 2’s indexed version, which tracks each country’s change relative to its own starting point rather than the raw dollar amount, is the one to rely on for comparing performance between Canada, Australia, and the United States.Figure 3. Raw GPV/R&D ratio (M USD per M PPP USD) for Canada, Australia, and the United States, across contemporaneous, 10-year lag, and 15-year lag specifications. Note: large values reflect output scale relative to R&D spending and are not directly comparable across countries without indexing. Use Figure 2 for cross-country comparison.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361.R&D Spending and TFP Growth: The Cross-Country PatternTable 1 summarizes the country-level averages underlying Figure 4. Canada’s position, high mean TFP growth (2.4%) with a moderate public R&D base (mean $850M PPP, 2000–2020) places it above the OLS (ordinary least squares) fit line in Figure 4, suggesting that Canadian agriculture has generated comparatively strong productivity growth per dollar of public research investment over this period. Australia sits on the opposite side of the fit line—its mean R&D base is roughly comparable to Canada’s, but its mean TFP growth is negative, placing it well below the line and reinforcing the disconnect discussed in Section 2a. The United States sits almost directly on the fit line: its much larger R&D base corresponds to a TFP growth rate close to what the global cross-country relationship would predict, with no strong over- or under-performance in either direction. Read together, the three focal countries span the range of outcomes visible in the broader scatter—Canada above the line, the United States on it, and Australia below it—despite Canada and Australia having broadly similar public R&D levels, which is itself evidence that funding level alone is a weak predictor of productivity outcome.Figure 4. Country-level scatter of mean public agricultural R&D (log₁₀ scale, M 2017 PPP USD) versus mean TFP trend growth rate (%, 2001–2020). Dashed line = OLS fit (descriptive, not causal). Canada (blue), Australia (orange), and USA (green) are highlighted.Sources: R&D: GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. TFP: USDA ERS AgTFPInternational2022. Relationship is confounded by country size, farm structure, and other factors.Barriers: Supply Side and Demand SideThe standard framing for agricultural innovation policy focuses on the supply side: how much is being invested in R&D, by whom, and through what institutional channels. Parts 1 and 2 of this series largely engaged with that framing. But the data in Sections 2 and 3 above suggest that the relationship between investment and productivity outcome is looser than a straightforward supply-side model would predict.This section examines both sides of the innovation chain, supply and demand, with particular attention to the demand side, which receives comparatively little attention in the Canadian policy literature.Supply-Side Barriers (Recap)Part 2covered the main supply-side barriers in detail; a brief summary is included here for completeness.Intellectual Property (IP) protections and farmers’ privilege –The balance between IP protection for plant breeders and the traditional farmers’ privilege to save seed shapes the private R&D incentive structure. The Canadian Food Inspection Agency (CFIA) finalized updated Plant Breeders’ Rights Regulations in April 2026, narrowing farmers’ privilege to small grain crops; see Section 5 for the policy detail and the resulting debate.SR&ED complexity –The Scientific Research and Experimental Development tax creditprogram favours well-capitalized firms with dedicated compliance resources, creating a barrier for smaller agricultural innovators.Regulatory timelines –Approval timelines for genetically modified varieties represent a material lag between research output and farm-level availability.AAFC research centre consolidation –The scheduled closure and consolidation of Agriculture and Agri-Food Canada (AAFC) research facilities represents a near-term supply-side risk that has not yet been fully reflected in the spending data. AAFC announced in January 2026 that it would close seven research centres and satellite farms, part of a broader federal workforce reduction; the House of Commons Standing Committee on Agriculture and Agri-Food has since called for the closures to be reversed (seeRealAgriculture).Demand-Side Barriers: The Underexamined SideThe gap between research output and changed farm practice is poorly measured in Canada, and it represents the most original contribution of this series. Even if supply-side constraints were fully resolved—if R&D spending were adequate, IP frameworks balanced, and regulatory timelines shortened—there would remain a substantial question about whether innovations reach farms and are adopted at scale. The available evidence suggests the answer is often no, or not quickly.Farm financial constraints –Part 2 documented the margin pressures facing Canadian farm operations. Capital-constrained farms face higher effective hurdle rates for technology adoption, particularly for precision-agriculture tools that require upfront equipment investment and ongoing subscription costs.Extension services –ABARES data for Australia shows that public agricultural advisory and extension funding has declined at an average annual rate of approximately5.7%in real terms since 2005–06, a sustained multi-decade contraction even as private R&D investment has grown over the same period. While no equivalent national time-series exists for Canada, the institutional trend toward commodity organization-led extension and away from publicly funded provincial extension services is broadly similar, and has been reinforced recently by the closure and staffing reductions at several AAFC research facilities discussed in Section 4a. This represents a structural weakening in the knowledge-transfer link between research output and farm decision-making.Technology adoption rates –The 2021 Census of Agriculture provides the most granular available picture of precision-agriculture adoption among Canadian farms.Figure 5 reveals a pronounced geographic concentration of precision-agriculture adoption. GPS/auto-steer adoption is highest in Saskatchewan (about 47% of all farms) and Manitoba (about 41%), with Alberta close behind (about 31%); Ontario shows comparatively stronger uptake of GIS mapping and variable-rate application than the Prairie provinces, though at lower GPS adoption (around 23%). New Brunswick, Nova Scotia, and Newfoundland and Labrador show minimal adoption across all four technology categories, generally in the single digits. Prince Edward Island is a notable exception to the Atlantic pattern, with GPS and variable-rate adoption (around 20% each) closer to Ontario’s than to its immediate regional neighbours—likely reflecting its potato and row-crop heavy production mix, which is more compatible with precision-ag tools than the mixed and horticultural operations more common elsewhere in Atlantic Canada.Figure 5. Share of all farms reporting use of selected precision-agriculture practices (GPS/auto-steer, GIS mapping, variable-rate application, and drones) by province, 2021 Census of Agriculture. Percentages are calculated against the total number of farms in each province (all NAICS — North American Industry Classification System — categories), since oilseed/grain-specific denominators are near-zero in several Atlantic provinces; western provinces (AB, SK, MB), where grain and oilseed production dominates, are the most directly comparable.Source: Statistics Canada, Census of Agriculture 2021, Table 32-10-0379-01. Released 2022-05-11.This pattern likely reflects a combination of farm size effects (Prairie operations are on average much larger and more mechanized), crop type (row crops are better suited to precision-agriculture tools than many Eastern mixed or horticultural operations), and capital availability. The adoption data do not allow clean isolation of these factors, but the regional concentration is itself a policy-relevant finding: precision-agriculture supports that are designed around Prairie-scale operations may not transfer effectively to the rest of the country.Scale and operator age –Both farm size and operator age are well-documented correlates of technology adoption. Part 1 documented the ongoing consolidation of Canadian farm operations and the aging of the operator population, and the2021 Census of Agriculturesharpens that picture. The number of census farms fell to 189,874 in 2021, down 1.9% from 2016, even as average farm size has grown substantially over the longer run—from roughly 676 to 809 acres between 2001 and 2021. The distribution is increasingly bimodal: farms of 2,240 acres or more accounted for only 8.9% of all farms in 2021 but generated 37.3% of total operating revenue, while farms under 240 acres made up 52.3% of the total but contributed only about a quarter of revenue. On the operator side, the average age of Canadian farm operators rose to 56.0 years in 2021 (median 58.0), and the share of operators aged 55 and older climbed from 54.5% in 2016 to 60.5% in 2021, while the share of younger operators continued to decline. Combined with the precision-agriculture adoption gradient in Figure 5, this points toward a structural pattern rather than a regional quirk: adoption is concentrated among a shrinking pool of larger, Prairie-dominant operations, while the median Canadian farm—smaller, and run by an operator closer to retirement age than to a multi-decade payback horizon—sits further from the technology-adoption frontier described in Part 1. Larger operations have stronger financial incentives to adopt labour-saving and yield-improving technologies; older operators face shorter payback horizons and, on average, lower returns to technology learning.Where are the Leverage Points?The preceding sections point toward a more granular set of leverage points than the standard policy discussion about R&D funding levels. The evidence does not support the conclusion that simply increasing public agricultural R&D spending is sufficient to close Canada’s productivity trajectory risk. The Australia case illustrates that investment and outcome can diverge substantially, and the adoption data suggest that the return on existing research is partially constrained by the capacity to put it into practice.This section identifies the areas where the evidence points toward tractable improvements, without prescribing a specific funding level or policy instrument.IP regime modernisation –This is no longer purely prospective: the CFIA finalized amendments to the Plant Breeders’ Rights Regulations in April 2026 (SOR/2026-74), with the changes coming into force shortly after. The most consequential change narrows farmers’ privilege, the right to save and reuse seed from a protected variety, to small grain crops such as cereals and pulses, removing the privilege for horticultural and ornamental varieties where it had previously applied.Producer groups broadly supported the changeas a confidence-building measure for plant breeders investing in Canada, while the National Farmers Union opposed it on the grounds that narrowing farmers’ privilege increases production costs and entrenches reliance on proprietary varieties. A well-calibrated IP framework that maintains private sector R&D incentives while preserving farmer access to saved seed represents a genuine leverage point, one that does not require new public expenditure—though the regulation as finalized has already drawn pushback on exactly that balance, and its effect on adoption and farm-level costs will only be observable with a few years of data.SR&ED simplification –Simplifying the SR&ED tax credit regime for smaller agricultural innovators and agribusiness firms would improve the return on an existing public commitment without necessarily increasing its total cost.Extension and knowledge transfer –The case for investment in knowledge-transfer infrastructure is arguably stronger than the case for additional research investment at the margin, precisely because the adoption gap is the underexamined variable. Extension services operate at the link between research output and farm practice, the point where the evidence suggests the current system is weakest.International R&D spillovers –Canada benefits disproportionately from research conducted in the United States and elsewhere. The size of the US public agricultural R&D enterprise (mean $5.5B PPP, 2000–2020) dwarfs Canada’s, and much of the resulting knowledge is applicable in Canadian conditions. A strategy that emphasizes Canada’s capacity to absorb and adapt internationally generated knowledge, rather than simply increasing domestic R&D outlays, may generate higher returns per public dollar.Industry-led R&D models –Australia’s Rural Research and Development Corporation (RDC) model—in which industry levy funds are matched by government contributions and channelled through commodity-specific bodies—offers a structural comparison worth examining in the Canadian context. There are 15 RDCs covering Australia’s major agricultural sectors, collectively investing roughly AUD$825 million annually; theAustralian government matches eligible R&D levy contributions up to 0.5% of each industry’s gross value of production. The model does not simply increase total investment; it changes the governance of how investment decisions are made and how knowledge transfer is organized.Conclusion: What the Series has EstablishedAcross three posts, this series has worked through the evidence on Canadian agricultural productivity with a specific goal: to replace a crisis narrative with a more precise one. The crisis framing—in which Canada’s agricultural sector is falling behind, underinvested, and in need of urgent intervention—is not entirely wrong, but it lacks specificity about where the problems are, which levers are most tractable, and what the counterfactual is.What the data show is a more textured picture. Canada’s TFP performance is genuinely strong by international standards, and its implied return on public R&D investment, descriptive as that measure is, compares favourably with Australia and the United States on several specifications. The risk is not that Canadian agriculture is already in decline, but that the investment trajectory and structural conditions that have sustained that performance may not hold over the next decade.The most important finding of this series may be the one that is hardest to measure: the adoption gap. The chain from public R&D investment to research output to farm-level practice to productivity outcome has multiple links, and the weakest link is not always the one that gets the most policy attention. Extension capacity, precision-agriculture adoption rates, farm financial constraints, and the scale and age structure of the operator population all shape whether research investment actually reaches the farm gate.The dominant policy narrative focuses on spending levels. The more useful question and the one this series has tried to make clearer is what happens to that spending after it leaves the public accounts.A fourth and final post in this series will take up the question left open here: what do the solutions actually look like? Drawing on the concerns raised across Parts 1, 2, and 3, Part 4 will weigh the leverage points introduced in Section 5—IP modernisation, SR&ED simplification, extension investment, international R&D spillovers, and Australia’s industry-led RDC model—against each other, and assess where the evidence points toward genuine leverage and where it does not. Table of ContentsToggle Table of ContentsToggle Part 3: Returns on Investment, Adoption Gaps, and the Road AheadPicking up the ThreadParts1and2of this series established a productive tension at the heart of Canadian agricultural policy. On one hand, Canada’s total factor productivity (TFP) growth has held up well by international standards, consistently outperforming comparators like Australia and tracking close to the United States over the 2001–2020 period. On the other hand, the investment inputs that sustain long-run productivity growth such as public R&D expenditure, extension infrastructure, and technology adoption capacity show signs of strain. Theriskis not that Canadian agriculture is in crisis today, but that it could be running on borrowed time.Part 2found Canada’s public agricultural R&D spending has declined in real terms and that the private sector has not stepped in to fill the gap to the same extent as it has in Australia, where private R&D has grown at roughly 4.2% annually since 2005. It also flagged the adoption gap: generating research is only half the equation, and Canada’s extension and knowledge-transfer infrastructure is comparatively underexamined.Part 3 takes up three questions that flow from that foundation: Why is TFP growth declining broadly, not just in Canada but across most major agricultural producers? What does the return on public R&D investment actually look like for Canada, Australia, and the United States? And, where are the most tractable leverage points for improving outcomes, in funding levels, regulatory conditions, or the space between research output and farm-level adoption?Why is TFP Declining Globally?The TFP deceleration documented inPart 1is not a Canadian anomaly. Declining or stagnating productivity growth is visible across most major agricultural producers over the 2001–2020 period, including countries that have increased public R&D investment. That pattern immediately complicates the most common policy prescription: simply spend more on agricultural research.Fuglie 2018explores agricultural productivity at a global scale, finding steady growth across the world, though he presents that growth has slowed among industrialized countries, while developing countries represent the highest growth. Fuglie also discusses potential reasons for the lower growth observed in industrialized countries, citing decreases in per acre yield growth, (though this tends to be offset by intensification), the expansion of cropland over other land uses, as well as movement towards no-till and other conservation practices. On the other hand, factors influencing the higher growth rates among developing countries span from investment in research, policy reform, and rising farm family education leading to technology adoption.The Australia PuzzleAustralia offers the sharpest illustration of the disconnect between investment and productivity. As Part 2 established, private R&D in the Australian agricultural sector has grown at approximately4.2% annuallysince 2005, a substantial and sustained increase. Yet over the same period, Australia’s agricultural TFP has turned negative, making it something of an outlier even among countries experiencing slower growth. The Australian Bureau of Agricultural and Resource Economics and Sciences(ABARES) reportingshows that Australia’s public-sourced R&D funding has grown at a comparatively modest 1.4% annually since 2005–06, while funding for extension and advisory services declined at an average annual rate of 5.7% over the same window, and federal- and state-government R&D expenditure declined at average annual rates of 3.0% and 1.9% respectively. While the country has had success in growing privately sourced funding, there is still a prevailing question of to what extent this has offset the slowdown in public funding for the sector’s overall productivity. Would outcomes have been worse if that role was not filled, as seems to be the case in Canada currently? Thispaperreviews and discusses the Australian policy landscape and outcomes within a larger time frame concerning agricultural TFP.Table 1 below puts this in relief: Australia’s mean TFP trend growth rate between 2001 and 2020 was approximately −0.7%, the weakest of the three comparator countries, despite R&D spending in the same period that was proportionally substantial relative to its agricultural output base. Two quick definitions before the numbers: GPV (gross production value) is the total market value of a country’s agricultural output, and PPP (purchasing power parity) is the exchange-rate adjustment used to make spending comparable across countries. Except where noted, all monetary figures in this series — R&D spending and GPV alike — are expressed in standardized, PPP-adjusted international USD, not in domestic currencies (A$, C$, or US$).Table 1. Mean Public R&D Spending and TFP Growth, 2000-2020 (Canada, Australia, United States)CountryMean R&D 2000-2020 (M PPP)Mean TFP Growth 2001-2020 (%)Latest YearLatest GPV (M USD)AUS741.1-0.7202250,982CAN850.42.4202251,558USA5,482.70.72022365,711Sources: GRAPE (Global Research on Agriculture: Personnel & Expenditures dataset) v1.0.0, Zenodo 15507361 (R&D, M 2017 PPP USD); USDA ERS (U.S. Department of Agriculture, Economic Research Service) AgTFPInternational2022 (TFP, OLS log-linear trend growth rate); FAOSTAT (Food and Agriculture Organization Corporate Statistical Database) QV element 152 (GPV). Relationship is descriptive.Several non-exclusive explanations emerge from the literature for this kind of disconnect between investment and measured productivity:Lag effects –Current TFP reflects the R&D investment of years prior, not of today. Recentestimatessuggest lags as long as 50 years to show up in productivity metrics. A sustained increase in private R&D since 2005 may not yet be visible in TFP outcomes through 2020.Climate variability –TFP is particularly difficult to measure cleanly in agricultural systems with high climate variability. Australia’sexposureto recurring drought conditions introduces substantial noise into productivity estimates, potentially masking real output gains.Innovationcomplexity –More recent agricultural innovations tend to be more incremental, more systems-integrated, and slower to diffuse to farm-level practice than the varietal and mechanization breakthroughs of earlier decades.Input cost effects –Rising input costs inflate the denominator of TFP even as gross output grows, compressing measured productivity ratios.The Broader PatternAustralia is not alone in showing TFP deceleration despite investment. The United States, with by far the largest absolute public agricultural R&D spending of the three countries, also shows modest mean TFP growth of around 0.7% annually between 2001 and 2020. Canada, with the highest mean TFP growth at 2.4% over the same period, invested at roughly the same absolute scale as Australia but in a much larger agricultural economy (see Table 1; R&D from GRAPE v1.0.0, Zenodo 15507361; TFP trend growth from USDA ERS AgTFPInternational2022).The implication is that the TFP deceleration story is not primarily a funding story—at least not in the short run.USDA ERS analysisof the global slowdown points to structural factors that likely apply across all three focal countries to varying degrees: a slowing diffusion frontier for the major productivity-enhancing technologies of the 20th century, policy and public-attitude barriers that slow adoption of newer technologies in developed countries specifically, and measurement challenges that are particularly acute in agriculture. This is offered as the most likely explanation given the cross-country pattern, not as a finding established by this analysis on its own.Figure 1. Agricultural TFP trend growth rates by decade (top) and mean public R&D over the preceding 10 years (bottom) for Canada, Australia, and the United States. The TFP bar row covers six decades (1961–2020); the R&D line covers five decades (1980–2020) because GRAPE v1.0.0 coverage for Canada begins in 1960, providing only one year within the 1961–1969 preceding-decade window—that point is omitted for Canada as unreliable. Australia and the United States have sufficient coverage from 1970 onwards. The peak in Canada’s R&D line at 2000 reflects mean spending over 1991–2000, Canada’s highest-spending decade in real terms before the sustained decline documented in Parts 1 and 2.Sources: USDA ERS AgTFPInternational2022; GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. Relationship is descriptive—confounded by terms of trade, private R&D, and farm structureWhat Does the Return on R&D Investment Look Like?This section presents the original analytical contribution of Part 3: an implied-return figure that asks a deliberately simple question. If we take gross agricultural production value (GPV) as the output measure and public R&D expenditure as the input measure, how does the ratio between them compare across countries and over time?This is an illustrative exercise, not a causal return-on-investment estimate. Agricultural output is shaped by weather, global commodity prices, private sector investment, farm structure, and policy—none of which are held constant in this calculation. The framing is best understood as a rough measure of output efficiency relative to public investment: how much agricultural output is associated, descriptively, with each dollar of public R&D.All figures use GPV in constant 2014–2016 international USD (FAOSTAT, element code 152) against public R&D expenditure from GRAPE v1.0.0 (van Dijk et al. 2025, Zenodo 15507361) in million 2017 PPP (purchasing power parity) USD. The two currency bases are not identical, the FAOSTAT Geary-Khamis and GRAPE 2017 ICP bases differ by approximately 3–5%, and the mismatch is flagged as a stated limitation throughout.Implied Return Ratios (2022)Table 2 presents the GPV per public R&D dollar at the most recent available data point (2022), under three lag assumptions: contemporaneous, with R&D lagged 10 years, and with R&D lagged 15 years. The 10- and 15-year lags reflect the consensus in the agricultural R&D literature that the productivity impact of research investment takes a decade or more to materialize at the farm level.Table 2. Implied GPV/Public R&D Ratios by Country and Lag Specification (2022)CountryLagGPV (M USD)Public R&D (M PPP)GPV/R&D RatioAustraliaContemporaneous50,982670.676.0R&D lagged 10 yrs50,982849.160.0R&D lagged 15 yrs50,982654.677.9CanadaContemporaneous51,558870.659.2R&D lagged 10 yrs51,558843.261.1R&D lagged 15 yrs51,558877.458.8United StatesContemporaneous365,7115,11471.5R&D lagged 10 yrs365,7115,25769.6R&D lagged 15 yrs365,7115,94261.6Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. GPV in constant 2014–16 int. USD; R&D in million 2017 PPP USD. Descriptive ratio—not a causal return estimate.Several features of Table 2 are worth noting. Canada’s contemporaneous GPV/R&D ratio (59.2) is the lowest of the three countries, reflecting both its moderate agricultural output scale and its higher public R&D spending relative to Australia. When R&D is lagged 10 years, Canada’s ratio rises modestly to 61.1, the smallest shift of the three countries, suggesting that the relationship between past R&D and current output is relatively stable in the Canadian case. By the 15-year lag, however, the ratio dips slightly below its starting point, to 58.8. Read together, the three Canadian figures (59.2 → 61.1 → 58.8) trace the lag structure of public R&D directly: a decade of spending growth is associated with a modest near-term gain, while spending from 15 years prior—an older, smaller R&D base—corresponds to a slightly lower ratio. This is consistent with the lag effects discussed in Section 2 and with Canada’s current position in the R&D cycle, where recent spending growth has not yet had a full 15 years to show up in output.Australia, by contrast, shows the largest swing across lag specifications: from 76.0 contemporaneously to 60.0 at a 10-year lag, as the elevated R&D spending of the prior decade enters the denominator. This pattern is consistent with the interpretation that the private R&D growth documented in Part 2 has raised Australia’s investment base substantially without yet generating proportionate output gains, reinforcing the TFP puzzle described in Section 2.The United States shows a steady decline across lag specifications (71.5 → 69.6 → 61.6), reflecting the consistently high and growing public R&D base. The large absolute GPV keeps the ratio comparatively elevated even as the denominator grows.Implied Return Over Time: The Indexed FigureFigure 2 makes the cross-country divergence visible in a way the raw ratios in Table 2 understate. Each country’s series is indexed to its own first available year (1990 for the contemporaneous panel; 2000 and 2005 for the 10- and 15-year lag panels, respectively, since lagging the R&D series shortens the usable run). On the contemporaneous panel, Canada’s indexed series climbs steadily above its own baseline, reaching roughly 150–190 by the 2010s, while the United States rises more modestly and Australia oscillates close to its own starting point with no clear trend. The picture shifts once R&D is lagged. At the 10-year lag, Canada’s climb is even more pronounced, its series rises furthest above 100 of the three countries, while Australia’s series sits persistently below 100 for most of the 2000s and 2010s, consistent with the TFP puzzle described in Section 2: a rapidly growing R&D base without a matching rise in output. At the 15-year lag, Canada remains the strongest performer through most of the series, but Australia’s line turns sharply upward in the final years shown, climbing back toward and above 100. Taken together, Canada’s indexed return has been on a more consistently rising trajectory across all three lag specifications than either comparator, while Australia’s pattern is the most volatile, swinging from clearly below baseline to a late recovery depending on which lag is used. This volatility is itself informative: it suggests Australia’s investment-to-output relationship is more sensitive to the choice of lag than Canada’s, which is consistent with Australia’s R&D base having grown unevenly over the period rather than at a steady rate.What this means in plain terms: no matter how the timing of R&D spending relative to output is sliced, Canada has consistently generated more agricultural output per public research dollar than either Australia or the United States — and that edge has grown over time rather than shrinking. Australia’s story is the opposite: its return on R&D has swung unpredictably, falling well behind for most of the 2000s and 2010s before a late recovery, which points to a much less stable link between what it spends and what it gets back.Figure 2. GPV/R&D ratio indexed to own-country baseline (base year = 100), for Canada (blue), Australia (orange), and the United States (green), across contemporaneous, 10-year lag, and 15-year lag specifications. Grey lines in the contemporaneous panel = all other countries in the GRAPE sample, capped at the 95th percentile; the lagged panels omit the background countries since fewer of them have enough years of data to survive the lag.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. Descriptive ratio—not a causal return estimate.The raw ratios in Figure 3 underline the scale gap that the indexed figure deliberately removes. In the most recent year, the United States generates roughly $71 of agricultural output per public R&D dollar contemporaneously, compared with roughly $76 for Australia and $59 for Canada, figures driven as much by the size of each country’s R&D base as by underlying productivity. Canada’s raw ratio is consistently the lowest of the three across all three lag specifications, which mechanically reflects its position between Australia’s smaller agricultural economy and the United States’ much larger one, relative to a public R&D budget closer in absolute size to Australia’s. The trajectories also diverge: Canada’s raw ratio rises steadily from the early 1990s through the most recent data, narrowing a gap that was much wider in the 1990s; the United States holds the highest contemporaneous ratio for most of the series before Australia’s late upturn overtakes it; and Australia’s ratio is comparatively flat through the 2000s before that sharp rise in the final years. These patterns mirror the indexed story in Figure 2, but Figure 3 makes clear that the underlying dollar magnitudes are not directly comparable across countries without that indexing step.Breaking these numbers down: the raw dollar figures aren’t comparable across countries on their own, because a country with a much bigger farm sector will naturally show a bigger number here, regardless of how efficiently it turns research spending into output. That’s why Figure 2’s indexed version, which tracks each country’s change relative to its own starting point rather than the raw dollar amount, is the one to rely on for comparing performance between Canada, Australia, and the United States.Figure 3. Raw GPV/R&D ratio (M USD per M PPP USD) for Canada, Australia, and the United States, across contemporaneous, 10-year lag, and 15-year lag specifications. Note: large values reflect output scale relative to R&D spending and are not directly comparable across countries without indexing. Use Figure 2 for cross-country comparison.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361.R&D Spending and TFP Growth: The Cross-Country PatternTable 1 summarizes the country-level averages underlying Figure 4. Canada’s position, high mean TFP growth (2.4%) with a moderate public R&D base (mean $850M PPP, 2000–2020) places it above the OLS (ordinary least squares) fit line in Figure 4, suggesting that Canadian agriculture has generated comparatively strong productivity growth per dollar of public research investment over this period. Australia sits on the opposite side of the fit line—its mean R&D base is roughly comparable to Canada’s, but its mean TFP growth is negative, placing it well below the line and reinforcing the disconnect discussed in Section 2a. The United States sits almost directly on the fit line: its much larger R&D base corresponds to a TFP growth rate close to what the global cross-country relationship would predict, with no strong over- or under-performance in either direction. Read together, the three focal countries span the range of outcomes visible in the broader scatter—Canada above the line, the United States on it, and Australia below it—despite Canada and Australia having broadly similar public R&D levels, which is itself evidence that funding level alone is a weak predictor of productivity outcome.Figure 4. Country-level scatter of mean public agricultural R&D (log₁₀ scale, M 2017 PPP USD) versus mean TFP trend growth rate (%, 2001–2020). Dashed line = OLS fit (descriptive, not causal). Canada (blue), Australia (orange), and USA (green) are highlighted.Sources: R&D: GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. TFP: USDA ERS AgTFPInternational2022. Relationship is confounded by country size, farm structure, and other factors.Barriers: Supply Side and Demand SideThe standard framing for agricultural innovation policy focuses on the supply side: how much is being invested in R&D, by whom, and through what institutional channels. Parts 1 and 2 of this series largely engaged with that framing. But the data in Sections 2 and 3 above suggest that the relationship between investment and productivity outcome is looser than a straightforward supply-side model would predict.This section examines both sides of the innovation chain, supply and demand, with particular attention to the demand side, which receives comparatively little attention in the Canadian policy literature.Supply-Side Barriers (Recap)Part 2covered the main supply-side barriers in detail; a brief summary is included here for completeness.Intellectual Property (IP) protections and farmers’ privilege –The balance between IP protection for plant breeders and the traditional farmers’ privilege to save seed shapes the private R&D incentive structure. The Canadian Food Inspection Agency (CFIA) finalized updated Plant Breeders’ Rights Regulations in April 2026, narrowing farmers’ privilege to small grain crops; see Section 5 for the policy detail and the resulting debate.SR&ED complexity –The Scientific Research and Experimental Development tax creditprogram favours well-capitalized firms with dedicated compliance resources, creating a barrier for smaller agricultural innovators.Regulatory timelines –Approval timelines for genetically modified varieties represent a material lag between research output and farm-level availability.AAFC research centre consolidation –The scheduled closure and consolidation of Agriculture and Agri-Food Canada (AAFC) research facilities represents a near-term supply-side risk that has not yet been fully reflected in the spending data. AAFC announced in January 2026 that it would close seven research centres and satellite farms, part of a broader federal workforce reduction; the House of Commons Standing Committee on Agriculture and Agri-Food has since called for the closures to be reversed (seeRealAgriculture).Demand-Side Barriers: The Underexamined SideThe gap between research output and changed farm practice is poorly measured in Canada, and it represents the most original contribution of this series. Even if supply-side constraints were fully resolved—if R&D spending were adequate, IP frameworks balanced, and regulatory timelines shortened—there would remain a substantial question about whether innovations reach farms and are adopted at scale. The available evidence suggests the answer is often no, or not quickly.Farm financial constraints –Part 2 documented the margin pressures facing Canadian farm operations. Capital-constrained farms face higher effective hurdle rates for technology adoption, particularly for precision-agriculture tools that require upfront equipment investment and ongoing subscription costs.Extension services –ABARES data for Australia shows that public agricultural advisory and extension funding has declined at an average annual rate of approximately5.7%in real terms since 2005–06, a sustained multi-decade contraction even as private R&D investment has grown over the same period. While no equivalent national time-series exists for Canada, the institutional trend toward commodity organization-led extension and away from publicly funded provincial extension services is broadly similar, and has been reinforced recently by the closure and staffing reductions at several AAFC research facilities discussed in Section 4a. This represents a structural weakening in the knowledge-transfer link between research output and farm decision-making.Technology adoption rates –The 2021 Census of Agriculture provides the most granular available picture of precision-agriculture adoption among Canadian farms.Figure 5 reveals a pronounced geographic concentration of precision-agriculture adoption. GPS/auto-steer adoption is highest in Saskatchewan (about 47% of all farms) and Manitoba (about 41%), with Alberta close behind (about 31%); Ontario shows comparatively stronger uptake of GIS mapping and variable-rate application than the Prairie provinces, though at lower GPS adoption (around 23%). New Brunswick, Nova Scotia, and Newfoundland and Labrador show minimal adoption across all four technology categories, generally in the single digits. Prince Edward Island is a notable exception to the Atlantic pattern, with GPS and variable-rate adoption (around 20% each) closer to Ontario’s than to its immediate regional neighbours—likely reflecting its potato and row-crop heavy production mix, which is more compatible with precision-ag tools than the mixed and horticultural operations more common elsewhere in Atlantic Canada.Figure 5. Share of all farms reporting use of selected precision-agriculture practices (GPS/auto-steer, GIS mapping, variable-rate application, and drones) by province, 2021 Census of Agriculture. Percentages are calculated against the total number of farms in each province (all NAICS — North American Industry Classification System — categories), since oilseed/grain-specific denominators are near-zero in several Atlantic provinces; western provinces (AB, SK, MB), where grain and oilseed production dominates, are the most directly comparable.Source: Statistics Canada, Census of Agriculture 2021, Table 32-10-0379-01. Released 2022-05-11.This pattern likely reflects a combination of farm size effects (Prairie operations are on average much larger and more mechanized), crop type (row crops are better suited to precision-agriculture tools than many Eastern mixed or horticultural operations), and capital availability. The adoption data do not allow clean isolation of these factors, but the regional concentration is itself a policy-relevant finding: precision-agriculture supports that are designed around Prairie-scale operations may not transfer effectively to the rest of the country.Scale and operator age –Both farm size and operator age are well-documented correlates of technology adoption. Part 1 documented the ongoing consolidation of Canadian farm operations and the aging of the operator population, and the2021 Census of Agriculturesharpens that picture. The number of census farms fell to 189,874 in 2021, down 1.9% from 2016, even as average farm size has grown substantially over the longer run—from roughly 676 to 809 acres between 2001 and 2021. The distribution is increasingly bimodal: farms of 2,240 acres or more accounted for only 8.9% of all farms in 2021 but generated 37.3% of total operating revenue, while farms under 240 acres made up 52.3% of the total but contributed only about a quarter of revenue. On the operator side, the average age of Canadian farm operators rose to 56.0 years in 2021 (median 58.0), and the share of operators aged 55 and older climbed from 54.5% in 2016 to 60.5% in 2021, while the share of younger operators continued to decline. Combined with the precision-agriculture adoption gradient in Figure 5, this points toward a structural pattern rather than a regional quirk: adoption is concentrated among a shrinking pool of larger, Prairie-dominant operations, while the median Canadian farm—smaller, and run by an operator closer to retirement age than to a multi-decade payback horizon—sits further from the technology-adoption frontier described in Part 1. Larger operations have stronger financial incentives to adopt labour-saving and yield-improving technologies; older operators face shorter payback horizons and, on average, lower returns to technology learning.Where are the Leverage Points?The preceding sections point toward a more granular set of leverage points than the standard policy discussion about R&D funding levels. The evidence does not support the conclusion that simply increasing public agricultural R&D spending is sufficient to close Canada’s productivity trajectory risk. The Australia case illustrates that investment and outcome can diverge substantially, and the adoption data suggest that the return on existing research is partially constrained by the capacity to put it into practice.This section identifies the areas where the evidence points toward tractable improvements, without prescribing a specific funding level or policy instrument.IP regime modernisation –This is no longer purely prospective: the CFIA finalized amendments to the Plant Breeders’ Rights Regulations in April 2026 (SOR/2026-74), with the changes coming into force shortly after. The most consequential change narrows farmers’ privilege, the right to save and reuse seed from a protected variety, to small grain crops such as cereals and pulses, removing the privilege for horticultural and ornamental varieties where it had previously applied.Producer groups broadly supported the changeas a confidence-building measure for plant breeders investing in Canada, while the National Farmers Union opposed it on the grounds that narrowing farmers’ privilege increases production costs and entrenches reliance on proprietary varieties. A well-calibrated IP framework that maintains private sector R&D incentives while preserving farmer access to saved seed represents a genuine leverage point, one that does not require new public expenditure—though the regulation as finalized has already drawn pushback on exactly that balance, and its effect on adoption and farm-level costs will only be observable with a few years of data.SR&ED simplification –Simplifying the SR&ED tax credit regime for smaller agricultural innovators and agribusiness firms would improve the return on an existing public commitment without necessarily increasing its total cost.Extension and knowledge transfer –The case for investment in knowledge-transfer infrastructure is arguably stronger than the case for additional research investment at the margin, precisely because the adoption gap is the underexamined variable. Extension services operate at the link between research output and farm practice, the point where the evidence suggests the current system is weakest.International R&D spillovers –Canada benefits disproportionately from research conducted in the United States and elsewhere. The size of the US public agricultural R&D enterprise (mean $5.5B PPP, 2000–2020) dwarfs Canada’s, and much of the resulting knowledge is applicable in Canadian conditions. A strategy that emphasizes Canada’s capacity to absorb and adapt internationally generated knowledge, rather than simply increasing domestic R&D outlays, may generate higher returns per public dollar.Industry-led R&D models –Australia’s Rural Research and Development Corporation (RDC) model—in which industry levy funds are matched by government contributions and channelled through commodity-specific bodies—offers a structural comparison worth examining in the Canadian context. There are 15 RDCs covering Australia’s major agricultural sectors, collectively investing roughly AUD$825 million annually; theAustralian government matches eligible R&D levy contributions up to 0.5% of each industry’s gross value of production. The model does not simply increase total investment; it changes the governance of how investment decisions are made and how knowledge transfer is organized.Conclusion: What the Series has EstablishedAcross three posts, this series has worked through the evidence on Canadian agricultural productivity with a specific goal: to replace a crisis narrative with a more precise one. The crisis framing—in which Canada’s agricultural sector is falling behind, underinvested, and in need of urgent intervention—is not entirely wrong, but it lacks specificity about where the problems are, which levers are most tractable, and what the counterfactual is.What the data show is a more textured picture. Canada’s TFP performance is genuinely strong by international standards, and its implied return on public R&D investment, descriptive as that measure is, compares favourably with Australia and the United States on several specifications. The risk is not that Canadian agriculture is already in decline, but that the investment trajectory and structural conditions that have sustained that performance may not hold over the next decade.The most important finding of this series may be the one that is hardest to measure: the adoption gap. The chain from public R&D investment to research output to farm-level practice to productivity outcome has multiple links, and the weakest link is not always the one that gets the most policy attention. Extension capacity, precision-agriculture adoption rates, farm financial constraints, and the scale and age structure of the operator population all shape whether research investment actually reaches the farm gate.The dominant policy narrative focuses on spending levels. The more useful question and the one this series has tried to make clearer is what happens to that spending after it leaves the public accounts.A fourth and final post in this series will take up the question left open here: what do the solutions actually look like? Drawing on the concerns raised across Parts 1, 2, and 3, Part 4 will weigh the leverage points introduced in Section 5—IP modernisation, SR&ED simplification, extension investment, international R&D spillovers, and Australia’s industry-led RDC model—against each other, and assess where the evidence points toward genuine leverage and where it does not. Part 3: Returns on Investment, Adoption Gaps, and the Road Ahead Part 3: Returns on Investment, Adoption Gaps, and the Road Ahead Part 3: Returns on Investment, Adoption Gaps, and the Road Ahead Part 3: Returns on Investment, Adoption Gaps, and the Road Ahead Picking up the ThreadParts1and2of this series established a productive tension at the heart of Canadian agricultural policy. On one hand, Canada’s total factor productivity (TFP) growth has held up well by international standards, consistently outperforming comparators like Australia and tracking close to the United States over the 2001–2020 period. On the other hand, the investment inputs that sustain long-run productivity growth such as public R&D expenditure, extension infrastructure, and technology adoption capacity show signs of strain. Theriskis not that Canadian agriculture is in crisis today, but that it could be running on borrowed time.Part 2found Canada’s public agricultural R&D spending has declined in real terms and that the private sector has not stepped in to fill the gap to the same extent as it has in Australia, where private R&D has grown at roughly 4.2% annually since 2005. It also flagged the adoption gap: generating research is only half the equation, and Canada’s extension and knowledge-transfer infrastructure is comparatively underexamined.Part 3 takes up three questions that flow from that foundation: Why is TFP growth declining broadly, not just in Canada but across most major agricultural producers? What does the return on public R&D investment actually look like for Canada, Australia, and the United States? And, where are the most tractable leverage points for improving outcomes, in funding levels, regulatory conditions, or the space between research output and farm-level adoption? Picking up the ThreadParts1and2of this series established a productive tension at the heart of Canadian agricultural policy. On one hand, Canada’s total factor productivity (TFP) growth has held up well by international standards, consistently outperforming comparators like Australia and tracking close to the United States over the 2001–2020 period. On the other hand, the investment inputs that sustain long-run productivity growth such as public R&D expenditure, extension infrastructure, and technology adoption capacity show signs of strain. Theriskis not that Canadian agriculture is in crisis today, but that it could be running on borrowed time.Part 2found Canada’s public agricultural R&D spending has declined in real terms and that the private sector has not stepped in to fill the gap to the same extent as it has in Australia, where private R&D has grown at roughly 4.2% annually since 2005. It also flagged the adoption gap: generating research is only half the equation, and Canada’s extension and knowledge-transfer infrastructure is comparatively underexamined.Part 3 takes up three questions that flow from that foundation: Why is TFP growth declining broadly, not just in Canada but across most major agricultural producers? What does the return on public R&D investment actually look like for Canada, Australia, and the United States? And, where are the most tractable leverage points for improving outcomes, in funding levels, regulatory conditions, or the space between research output and farm-level adoption? Picking up the ThreadParts1and2of this series established a productive tension at the heart of Canadian agricultural policy. On one hand, Canada’s total factor productivity (TFP) growth has held up well by international standards, consistently outperforming comparators like Australia and tracking close to the United States over the 2001–2020 period. On the other hand, the investment inputs that sustain long-run productivity growth such as public R&D expenditure, extension infrastructure, and technology adoption capacity show signs of strain. Theriskis not that Canadian agriculture is in crisis today, but that it could be running on borrowed time.Part 2found Canada’s public agricultural R&D spending has declined in real terms and that the private sector has not stepped in to fill the gap to the same extent as it has in Australia, where private R&D has grown at roughly 4.2% annually since 2005. It also flagged the adoption gap: generating research is only half the equation, and Canada’s extension and knowledge-transfer infrastructure is comparatively underexamined.Part 3 takes up three questions that flow from that foundation: Why is TFP growth declining broadly, not just in Canada but across most major agricultural producers? What does the return on public R&D investment actually look like for Canada, Australia, and the United States? And, where are the most tractable leverage points for improving outcomes, in funding levels, regulatory conditions, or the space between research output and farm-level adoption? Picking up the Thread Parts1and2of this series established a productive tension at the heart of Canadian agricultural policy. On one hand, Canada’s total factor productivity (TFP) growth has held up well by international standards, consistently outperforming comparators like Australia and tracking close to the United States over the 2001–2020 period. On the other hand, the investment inputs that sustain long-run productivity growth such as public R&D expenditure, extension infrastructure, and technology adoption capacity show signs of strain. Theriskis not that Canadian agriculture is in crisis today, but that it could be running on borrowed time.Part 2found Canada’s public agricultural R&D spending has declined in real terms and that the private sector has not stepped in to fill the gap to the same extent as it has in Australia, where private R&D has grown at roughly 4.2% annually since 2005. It also flagged the adoption gap: generating research is only half the equation, and Canada’s extension and knowledge-transfer infrastructure is comparatively underexamined.Part 3 takes up three questions that flow from that foundation: Why is TFP growth declining broadly, not just in Canada but across most major agricultural producers? What does the return on public R&D investment actually look like for Canada, Australia, and the United States? And, where are the most tractable leverage points for improving outcomes, in funding levels, regulatory conditions, or the space between research output and farm-level adoption? Parts1and2of this series established a productive tension at the heart of Canadian agricultural policy. On one hand, Canada’s total factor productivity (TFP) growth has held up well by international standards, consistently outperforming comparators like Australia and tracking close to the United States over the 2001–2020 period. On the other hand, the investment inputs that sustain long-run productivity growth such as public R&D expenditure, extension infrastructure, and technology adoption capacity show signs of strain. Theriskis not that Canadian agriculture is in crisis today, but that it could be running on borrowed time. Part 2found Canada’s public agricultural R&D spending has declined in real terms and that the private sector has not stepped in to fill the gap to the same extent as it has in Australia, where private R&D has grown at roughly 4.2% annually since 2005. It also flagged the adoption gap: generating research is only half the equation, and Canada’s extension and knowledge-transfer infrastructure is comparatively underexamined. Part 3 takes up three questions that flow from that foundation: Why is TFP growth declining broadly, not just in Canada but across most major agricultural producers? What does the return on public R&D investment actually look like for Canada, Australia, and the United States? And, where are the most tractable leverage points for improving outcomes, in funding levels, regulatory conditions, or the space between research output and farm-level adoption? Why is TFP Declining Globally?The TFP deceleration documented inPart 1is not a Canadian anomaly. Declining or stagnating productivity growth is visible across most major agricultural producers over the 2001–2020 period, including countries that have increased public R&D investment. That pattern immediately complicates the most common policy prescription: simply spend more on agricultural research.Fuglie 2018explores agricultural productivity at a global scale, finding steady growth across the world, though he presents that growth has slowed among industrialized countries, while developing countries represent the highest growth. Fuglie also discusses potential reasons for the lower growth observed in industrialized countries, citing decreases in per acre yield growth, (though this tends to be offset by intensification), the expansion of cropland over other land uses, as well as movement towards no-till and other conservation practices. On the other hand, factors influencing the higher growth rates among developing countries span from investment in research, policy reform, and rising farm family education leading to technology adoption.The Australia PuzzleAustralia offers the sharpest illustration of the disconnect between investment and productivity. As Part 2 established, private R&D in the Australian agricultural sector has grown at approximately4.2% annuallysince 2005, a substantial and sustained increase. Yet over the same period, Australia’s agricultural TFP has turned negative, making it something of an outlier even among countries experiencing slower growth. The Australian Bureau of Agricultural and Resource Economics and Sciences(ABARES) reportingshows that Australia’s public-sourced R&D funding has grown at a comparatively modest 1.4% annually since 2005–06, while funding for extension and advisory services declined at an average annual rate of 5.7% over the same window, and federal- and state-government R&D expenditure declined at average annual rates of 3.0% and 1.9% respectively. While the country has had success in growing privately sourced funding, there is still a prevailing question of to what extent this has offset the slowdown in public funding for the sector’s overall productivity. Would outcomes have been worse if that role was not filled, as seems to be the case in Canada currently? Thispaperreviews and discusses the Australian policy landscape and outcomes within a larger time frame concerning agricultural TFP.Table 1 below puts this in relief: Australia’s mean TFP trend growth rate between 2001 and 2020 was approximately −0.7%, the weakest of the three comparator countries, despite R&D spending in the same period that was proportionally substantial relative to its agricultural output base. Two quick definitions before the numbers: GPV (gross production value) is the total market value of a country’s agricultural output, and PPP (purchasing power parity) is the exchange-rate adjustment used to make spending comparable across countries. Except where noted, all monetary figures in this series — R&D spending and GPV alike — are expressed in standardized, PPP-adjusted international USD, not in domestic currencies (A$, C$, or US$).Table 1. Mean Public R&D Spending and TFP Growth, 2000-2020 (Canada, Australia, United States)CountryMean R&D 2000-2020 (M PPP)Mean TFP Growth 2001-2020 (%)Latest YearLatest GPV (M USD)AUS741.1-0.7202250,982CAN850.42.4202251,558USA5,482.70.72022365,711Sources: GRAPE (Global Research on Agriculture: Personnel & Expenditures dataset) v1.0.0, Zenodo 15507361 (R&D, M 2017 PPP USD); USDA ERS (U.S. Department of Agriculture, Economic Research Service) AgTFPInternational2022 (TFP, OLS log-linear trend growth rate); FAOSTAT (Food and Agriculture Organization Corporate Statistical Database) QV element 152 (GPV). Relationship is descriptive.Several non-exclusive explanations emerge from the literature for this kind of disconnect between investment and measured productivity:Lag effects –Current TFP reflects the R&D investment of years prior, not of today. Recentestimatessuggest lags as long as 50 years to show up in productivity metrics. A sustained increase in private R&D since 2005 may not yet be visible in TFP outcomes through 2020.Climate variability –TFP is particularly difficult to measure cleanly in agricultural systems with high climate variability. Australia’sexposureto recurring drought conditions introduces substantial noise into productivity estimates, potentially masking real output gains.Innovationcomplexity –More recent agricultural innovations tend to be more incremental, more systems-integrated, and slower to diffuse to farm-level practice than the varietal and mechanization breakthroughs of earlier decades.Input cost effects –Rising input costs inflate the denominator of TFP even as gross output grows, compressing measured productivity ratios.The Broader PatternAustralia is not alone in showing TFP deceleration despite investment. The United States, with by far the largest absolute public agricultural R&D spending of the three countries, also shows modest mean TFP growth of around 0.7% annually between 2001 and 2020. Canada, with the highest mean TFP growth at 2.4% over the same period, invested at roughly the same absolute scale as Australia but in a much larger agricultural economy (see Table 1; R&D from GRAPE v1.0.0, Zenodo 15507361; TFP trend growth from USDA ERS AgTFPInternational2022).The implication is that the TFP deceleration story is not primarily a funding story—at least not in the short run.USDA ERS analysisof the global slowdown points to structural factors that likely apply across all three focal countries to varying degrees: a slowing diffusion frontier for the major productivity-enhancing technologies of the 20th century, policy and public-attitude barriers that slow adoption of newer technologies in developed countries specifically, and measurement challenges that are particularly acute in agriculture. This is offered as the most likely explanation given the cross-country pattern, not as a finding established by this analysis on its own.Figure 1. Agricultural TFP trend growth rates by decade (top) and mean public R&D over the preceding 10 years (bottom) for Canada, Australia, and the United States. The TFP bar row covers six decades (1961–2020); the R&D line covers five decades (1980–2020) because GRAPE v1.0.0 coverage for Canada begins in 1960, providing only one year within the 1961–1969 preceding-decade window—that point is omitted for Canada as unreliable. Australia and the United States have sufficient coverage from 1970 onwards. The peak in Canada’s R&D line at 2000 reflects mean spending over 1991–2000, Canada’s highest-spending decade in real terms before the sustained decline documented in Parts 1 and 2.Sources: USDA ERS AgTFPInternational2022; GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. Relationship is descriptive—confounded by terms of trade, private R&D, and farm structure Why is TFP Declining Globally?The TFP deceleration documented inPart 1is not a Canadian anomaly. Declining or stagnating productivity growth is visible across most major agricultural producers over the 2001–2020 period, including countries that have increased public R&D investment. That pattern immediately complicates the most common policy prescription: simply spend more on agricultural research.Fuglie 2018explores agricultural productivity at a global scale, finding steady growth across the world, though he presents that growth has slowed among industrialized countries, while developing countries represent the highest growth. Fuglie also discusses potential reasons for the lower growth observed in industrialized countries, citing decreases in per acre yield growth, (though this tends to be offset by intensification), the expansion of cropland over other land uses, as well as movement towards no-till and other conservation practices. On the other hand, factors influencing the higher growth rates among developing countries span from investment in research, policy reform, and rising farm family education leading to technology adoption.The Australia PuzzleAustralia offers the sharpest illustration of the disconnect between investment and productivity. As Part 2 established, private R&D in the Australian agricultural sector has grown at approximately4.2% annuallysince 2005, a substantial and sustained increase. Yet over the same period, Australia’s agricultural TFP has turned negative, making it something of an outlier even among countries experiencing slower growth. The Australian Bureau of Agricultural and Resource Economics and Sciences(ABARES) reportingshows that Australia’s public-sourced R&D funding has grown at a comparatively modest 1.4% annually since 2005–06, while funding for extension and advisory services declined at an average annual rate of 5.7% over the same window, and federal- and state-government R&D expenditure declined at average annual rates of 3.0% and 1.9% respectively. While the country has had success in growing privately sourced funding, there is still a prevailing question of to what extent this has offset the slowdown in public funding for the sector’s overall productivity. Would outcomes have been worse if that role was not filled, as seems to be the case in Canada currently? Thispaperreviews and discusses the Australian policy landscape and outcomes within a larger time frame concerning agricultural TFP.Table 1 below puts this in relief: Australia’s mean TFP trend growth rate between 2001 and 2020 was approximately −0.7%, the weakest of the three comparator countries, despite R&D spending in the same period that was proportionally substantial relative to its agricultural output base. Two quick definitions before the numbers: GPV (gross production value) is the total market value of a country’s agricultural output, and PPP (purchasing power parity) is the exchange-rate adjustment used to make spending comparable across countries. Except where noted, all monetary figures in this series — R&D spending and GPV alike — are expressed in standardized, PPP-adjusted international USD, not in domestic currencies (A$, C$, or US$).Table 1. Mean Public R&D Spending and TFP Growth, 2000-2020 (Canada, Australia, United States)CountryMean R&D 2000-2020 (M PPP)Mean TFP Growth 2001-2020 (%)Latest YearLatest GPV (M USD)AUS741.1-0.7202250,982CAN850.42.4202251,558USA5,482.70.72022365,711Sources: GRAPE (Global Research on Agriculture: Personnel & Expenditures dataset) v1.0.0, Zenodo 15507361 (R&D, M 2017 PPP USD); USDA ERS (U.S. Department of Agriculture, Economic Research Service) AgTFPInternational2022 (TFP, OLS log-linear trend growth rate); FAOSTAT (Food and Agriculture Organization Corporate Statistical Database) QV element 152 (GPV). Relationship is descriptive.Several non-exclusive explanations emerge from the literature for this kind of disconnect between investment and measured productivity:Lag effects –Current TFP reflects the R&D investment of years prior, not of today. Recentestimatessuggest lags as long as 50 years to show up in productivity metrics. A sustained increase in private R&D since 2005 may not yet be visible in TFP outcomes through 2020.Climate variability –TFP is particularly difficult to measure cleanly in agricultural systems with high climate variability. Australia’sexposureto recurring drought conditions introduces substantial noise into productivity estimates, potentially masking real output gains.Innovationcomplexity –More recent agricultural innovations tend to be more incremental, more systems-integrated, and slower to diffuse to farm-level practice than the varietal and mechanization breakthroughs of earlier decades.Input cost effects –Rising input costs inflate the denominator of TFP even as gross output grows, compressing measured productivity ratios.The Broader PatternAustralia is not alone in showing TFP deceleration despite investment. The United States, with by far the largest absolute public agricultural R&D spending of the three countries, also shows modest mean TFP growth of around 0.7% annually between 2001 and 2020. Canada, with the highest mean TFP growth at 2.4% over the same period, invested at roughly the same absolute scale as Australia but in a much larger agricultural economy (see Table 1; R&D from GRAPE v1.0.0, Zenodo 15507361; TFP trend growth from USDA ERS AgTFPInternational2022).The implication is that the TFP deceleration story is not primarily a funding story—at least not in the short run.USDA ERS analysisof the global slowdown points to structural factors that likely apply across all three focal countries to varying degrees: a slowing diffusion frontier for the major productivity-enhancing technologies of the 20th century, policy and public-attitude barriers that slow adoption of newer technologies in developed countries specifically, and measurement challenges that are particularly acute in agriculture. This is offered as the most likely explanation given the cross-country pattern, not as a finding established by this analysis on its own.Figure 1. Agricultural TFP trend growth rates by decade (top) and mean public R&D over the preceding 10 years (bottom) for Canada, Australia, and the United States. The TFP bar row covers six decades (1961–2020); the R&D line covers five decades (1980–2020) because GRAPE v1.0.0 coverage for Canada begins in 1960, providing only one year within the 1961–1969 preceding-decade window—that point is omitted for Canada as unreliable. Australia and the United States have sufficient coverage from 1970 onwards. The peak in Canada’s R&D line at 2000 reflects mean spending over 1991–2000, Canada’s highest-spending decade in real terms before the sustained decline documented in Parts 1 and 2.Sources: USDA ERS AgTFPInternational2022; GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. Relationship is descriptive—confounded by terms of trade, private R&D, and farm structure Why is TFP Declining Globally?The TFP deceleration documented inPart 1is not a Canadian anomaly. Declining or stagnating productivity growth is visible across most major agricultural producers over the 2001–2020 period, including countries that have increased public R&D investment. That pattern immediately complicates the most common policy prescription: simply spend more on agricultural research.Fuglie 2018explores agricultural productivity at a global scale, finding steady growth across the world, though he presents that growth has slowed among industrialized countries, while developing countries represent the highest growth. Fuglie also discusses potential reasons for the lower growth observed in industrialized countries, citing decreases in per acre yield growth, (though this tends to be offset by intensification), the expansion of cropland over other land uses, as well as movement towards no-till and other conservation practices. On the other hand, factors influencing the higher growth rates among developing countries span from investment in research, policy reform, and rising farm family education leading to technology adoption.The Australia PuzzleAustralia offers the sharpest illustration of the disconnect between investment and productivity. As Part 2 established, private R&D in the Australian agricultural sector has grown at approximately4.2% annuallysince 2005, a substantial and sustained increase. Yet over the same period, Australia’s agricultural TFP has turned negative, making it something of an outlier even among countries experiencing slower growth. The Australian Bureau of Agricultural and Resource Economics and Sciences(ABARES) reportingshows that Australia’s public-sourced R&D funding has grown at a comparatively modest 1.4% annually since 2005–06, while funding for extension and advisory services declined at an average annual rate of 5.7% over the same window, and federal- and state-government R&D expenditure declined at average annual rates of 3.0% and 1.9% respectively. While the country has had success in growing privately sourced funding, there is still a prevailing question of to what extent this has offset the slowdown in public funding for the sector’s overall productivity. Would outcomes have been worse if that role was not filled, as seems to be the case in Canada currently? Thispaperreviews and discusses the Australian policy landscape and outcomes within a larger time frame concerning agricultural TFP.Table 1 below puts this in relief: Australia’s mean TFP trend growth rate between 2001 and 2020 was approximately −0.7%, the weakest of the three comparator countries, despite R&D spending in the same period that was proportionally substantial relative to its agricultural output base. Two quick definitions before the numbers: GPV (gross production value) is the total market value of a country’s agricultural output, and PPP (purchasing power parity) is the exchange-rate adjustment used to make spending comparable across countries. Except where noted, all monetary figures in this series — R&D spending and GPV alike — are expressed in standardized, PPP-adjusted international USD, not in domestic currencies (A$, C$, or US$).Table 1. Mean Public R&D Spending and TFP Growth, 2000-2020 (Canada, Australia, United States)CountryMean R&D 2000-2020 (M PPP)Mean TFP Growth 2001-2020 (%)Latest YearLatest GPV (M USD)AUS741.1-0.7202250,982CAN850.42.4202251,558USA5,482.70.72022365,711Sources: GRAPE (Global Research on Agriculture: Personnel & Expenditures dataset) v1.0.0, Zenodo 15507361 (R&D, M 2017 PPP USD); USDA ERS (U.S. Department of Agriculture, Economic Research Service) AgTFPInternational2022 (TFP, OLS log-linear trend growth rate); FAOSTAT (Food and Agriculture Organization Corporate Statistical Database) QV element 152 (GPV). Relationship is descriptive.Several non-exclusive explanations emerge from the literature for this kind of disconnect between investment and measured productivity:Lag effects –Current TFP reflects the R&D investment of years prior, not of today. Recentestimatessuggest lags as long as 50 years to show up in productivity metrics. A sustained increase in private R&D since 2005 may not yet be visible in TFP outcomes through 2020.Climate variability –TFP is particularly difficult to measure cleanly in agricultural systems with high climate variability. Australia’sexposureto recurring drought conditions introduces substantial noise into productivity estimates, potentially masking real output gains.Innovationcomplexity –More recent agricultural innovations tend to be more incremental, more systems-integrated, and slower to diffuse to farm-level practice than the varietal and mechanization breakthroughs of earlier decades.Input cost effects –Rising input costs inflate the denominator of TFP even as gross output grows, compressing measured productivity ratios.The Broader PatternAustralia is not alone in showing TFP deceleration despite investment. The United States, with by far the largest absolute public agricultural R&D spending of the three countries, also shows modest mean TFP growth of around 0.7% annually between 2001 and 2020. Canada, with the highest mean TFP growth at 2.4% over the same period, invested at roughly the same absolute scale as Australia but in a much larger agricultural economy (see Table 1; R&D from GRAPE v1.0.0, Zenodo 15507361; TFP trend growth from USDA ERS AgTFPInternational2022).The implication is that the TFP deceleration story is not primarily a funding story—at least not in the short run.USDA ERS analysisof the global slowdown points to structural factors that likely apply across all three focal countries to varying degrees: a slowing diffusion frontier for the major productivity-enhancing technologies of the 20th century, policy and public-attitude barriers that slow adoption of newer technologies in developed countries specifically, and measurement challenges that are particularly acute in agriculture. This is offered as the most likely explanation given the cross-country pattern, not as a finding established by this analysis on its own.Figure 1. Agricultural TFP trend growth rates by decade (top) and mean public R&D over the preceding 10 years (bottom) for Canada, Australia, and the United States. The TFP bar row covers six decades (1961–2020); the R&D line covers five decades (1980–2020) because GRAPE v1.0.0 coverage for Canada begins in 1960, providing only one year within the 1961–1969 preceding-decade window—that point is omitted for Canada as unreliable. Australia and the United States have sufficient coverage from 1970 onwards. The peak in Canada’s R&D line at 2000 reflects mean spending over 1991–2000, Canada’s highest-spending decade in real terms before the sustained decline documented in Parts 1 and 2.Sources: USDA ERS AgTFPInternational2022; GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. Relationship is descriptive—confounded by terms of trade, private R&D, and farm structure Why is TFP Declining Globally? The TFP deceleration documented inPart 1is not a Canadian anomaly. Declining or stagnating productivity growth is visible across most major agricultural producers over the 2001–2020 period, including countries that have increased public R&D investment. That pattern immediately complicates the most common policy prescription: simply spend more on agricultural research.Fuglie 2018explores agricultural productivity at a global scale, finding steady growth across the world, though he presents that growth has slowed among industrialized countries, while developing countries represent the highest growth. Fuglie also discusses potential reasons for the lower growth observed in industrialized countries, citing decreases in per acre yield growth, (though this tends to be offset by intensification), the expansion of cropland over other land uses, as well as movement towards no-till and other conservation practices. On the other hand, factors influencing the higher growth rates among developing countries span from investment in research, policy reform, and rising farm family education leading to technology adoption. The TFP deceleration documented inPart 1is not a Canadian anomaly. Declining or stagnating productivity growth is visible across most major agricultural producers over the 2001–2020 period, including countries that have increased public R&D investment. That pattern immediately complicates the most common policy prescription: simply spend more on agricultural research. Fuglie 2018explores agricultural productivity at a global scale, finding steady growth across the world, though he presents that growth has slowed among industrialized countries, while developing countries represent the highest growth. Fuglie also discusses potential reasons for the lower growth observed in industrialized countries, citing decreases in per acre yield growth, (though this tends to be offset by intensification), the expansion of cropland over other land uses, as well as movement towards no-till and other conservation practices. On the other hand, factors influencing the higher growth rates among developing countries span from investment in research, policy reform, and rising farm family education leading to technology adoption. Australia offers the sharpest illustration of the disconnect between investment and productivity. As Part 2 established, private R&D in the Australian agricultural sector has grown at approximately4.2% annuallysince 2005, a substantial and sustained increase. Yet over the same period, Australia’s agricultural TFP has turned negative, making it something of an outlier even among countries experiencing slower growth. The Australian Bureau of Agricultural and Resource Economics and Sciences(ABARES) reportingshows that Australia’s public-sourced R&D funding has grown at a comparatively modest 1.4% annually since 2005–06, while funding for extension and advisory services declined at an average annual rate of 5.7% over the same window, and federal- and state-government R&D expenditure declined at average annual rates of 3.0% and 1.9% respectively. While the country has had success in growing privately sourced funding, there is still a prevailing question of to what extent this has offset the slowdown in public funding for the sector’s overall productivity. Would outcomes have been worse if that role was not filled, as seems to be the case in Canada currently? Thispaperreviews and discusses the Australian policy landscape and outcomes within a larger time frame concerning agricultural TFP.Table 1 below puts this in relief: Australia’s mean TFP trend growth rate between 2001 and 2020 was approximately −0.7%, the weakest of the three comparator countries, despite R&D spending in the same period that was proportionally substantial relative to its agricultural output base. Two quick definitions before the numbers: GPV (gross production value) is the total market value of a country’s agricultural output, and PPP (purchasing power parity) is the exchange-rate adjustment used to make spending comparable across countries. Except where noted, all monetary figures in this series — R&D spending and GPV alike — are expressed in standardized, PPP-adjusted international USD, not in domestic currencies (A$, C$, or US$). Australia offers the sharpest illustration of the disconnect between investment and productivity. As Part 2 established, private R&D in the Australian agricultural sector has grown at approximately4.2% annuallysince 2005, a substantial and sustained increase. Yet over the same period, Australia’s agricultural TFP has turned negative, making it something of an outlier even among countries experiencing slower growth. The Australian Bureau of Agricultural and Resource Economics and Sciences(ABARES) reportingshows that Australia’s public-sourced R&D funding has grown at a comparatively modest 1.4% annually since 2005–06, while funding for extension and advisory services declined at an average annual rate of 5.7% over the same window, and federal- and state-government R&D expenditure declined at average annual rates of 3.0% and 1.9% respectively. While the country has had success in growing privately sourced funding, there is still a prevailing question of to what extent this has offset the slowdown in public funding for the sector’s overall productivity. Would outcomes have been worse if that role was not filled, as seems to be the case in Canada currently? Thispaperreviews and discusses the Australian policy landscape and outcomes within a larger time frame concerning agricultural TFP. Table 1 below puts this in relief: Australia’s mean TFP trend growth rate between 2001 and 2020 was approximately −0.7%, the weakest of the three comparator countries, despite R&D spending in the same period that was proportionally substantial relative to its agricultural output base. Two quick definitions before the numbers: GPV (gross production value) is the total market value of a country’s agricultural output, and PPP (purchasing power parity) is the exchange-rate adjustment used to make spending comparable across countries. Except where noted, all monetary figures in this series — R&D spending and GPV alike — are expressed in standardized, PPP-adjusted international USD, not in domestic currencies (A$, C$, or US$). Table 1. Mean Public R&D Spending and TFP Growth, 2000-2020 (Canada, Australia, United States) CountryMean R&D 2000-2020 (M PPP)Mean TFP Growth 2001-2020 (%)Latest YearLatest GPV (M USD)AUS741.1-0.7202250,982CAN850.42.4202251,558USA5,482.70.72022365,711Sources: GRAPE (Global Research on Agriculture: Personnel & Expenditures dataset) v1.0.0, Zenodo 15507361 (R&D, M 2017 PPP USD); USDA ERS (U.S. Department of Agriculture, Economic Research Service) AgTFPInternational2022 (TFP, OLS log-linear trend growth rate); FAOSTAT (Food and Agriculture Organization Corporate Statistical Database) QV element 152 (GPV). Relationship is descriptive. CountryMean R&D 2000-2020 (M PPP)Mean TFP Growth 2001-2020 (%)Latest YearLatest GPV (M USD)AUS741.1-0.7202250,982CAN850.42.4202251,558USA5,482.70.72022365,711Sources: GRAPE (Global Research on Agriculture: Personnel & Expenditures dataset) v1.0.0, Zenodo 15507361 (R&D, M 2017 PPP USD); USDA ERS (U.S. Department of Agriculture, Economic Research Service) AgTFPInternational2022 (TFP, OLS log-linear trend growth rate); FAOSTAT (Food and Agriculture Organization Corporate Statistical Database) QV element 152 (GPV). Relationship is descriptive. CountryMean R&D 2000-2020 (M PPP)Mean TFP Growth 2001-2020 (%)Latest YearLatest GPV (M USD)AUS741.1-0.7202250,982CAN850.42.4202251,558USA5,482.70.72022365,711Sources: GRAPE (Global Research on Agriculture: Personnel & Expenditures dataset) v1.0.0, Zenodo 15507361 (R&D, M 2017 PPP USD); USDA ERS (U.S. Department of Agriculture, Economic Research Service) AgTFPInternational2022 (TFP, OLS log-linear trend growth rate); FAOSTAT (Food and Agriculture Organization Corporate Statistical Database) QV element 152 (GPV). Relationship is descriptive. Mean R&D 2000-2020 (M PPP) Mean TFP Growth 2001-2020 (%) Sources: GRAPE (Global Research on Agriculture: Personnel & Expenditures dataset) v1.0.0, Zenodo 15507361 (R&D, M 2017 PPP USD); USDA ERS (U.S. Department of Agriculture, Economic Research Service) AgTFPInternational2022 (TFP, OLS log-linear trend growth rate); FAOSTAT (Food and Agriculture Organization Corporate Statistical Database) QV element 152 (GPV). Relationship is descriptive. Several non-exclusive explanations emerge from the literature for this kind of disconnect between investment and measured productivity:Lag effects –Current TFP reflects the R&D investment of years prior, not of today. Recentestimatessuggest lags as long as 50 years to show up in productivity metrics. A sustained increase in private R&D since 2005 may not yet be visible in TFP outcomes through 2020.Climate variability –TFP is particularly difficult to measure cleanly in agricultural systems with high climate variability. Australia’sexposureto recurring drought conditions introduces substantial noise into productivity estimates, potentially masking real output gains.Innovationcomplexity –More recent agricultural innovations tend to be more incremental, more systems-integrated, and slower to diffuse to farm-level practice than the varietal and mechanization breakthroughs of earlier decades.Input cost effects –Rising input costs inflate the denominator of TFP even as gross output grows, compressing measured productivity ratios. Several non-exclusive explanations emerge from the literature for this kind of disconnect between investment and measured productivity: Lag effects –Current TFP reflects the R&D investment of years prior, not of today. Recentestimatessuggest lags as long as 50 years to show up in productivity metrics. A sustained increase in private R&D since 2005 may not yet be visible in TFP outcomes through 2020. Climate variability –TFP is particularly difficult to measure cleanly in agricultural systems with high climate variability. Australia’sexposureto recurring drought conditions introduces substantial noise into productivity estimates, potentially masking real output gains. Innovationcomplexity –More recent agricultural innovations tend to be more incremental, more systems-integrated, and slower to diffuse to farm-level practice than the varietal and mechanization breakthroughs of earlier decades. Input cost effects –Rising input costs inflate the denominator of TFP even as gross output grows, compressing measured productivity ratios. Australia is not alone in showing TFP deceleration despite investment. The United States, with by far the largest absolute public agricultural R&D spending of the three countries, also shows modest mean TFP growth of around 0.7% annually between 2001 and 2020. Canada, with the highest mean TFP growth at 2.4% over the same period, invested at roughly the same absolute scale as Australia but in a much larger agricultural economy (see Table 1; R&D from GRAPE v1.0.0, Zenodo 15507361; TFP trend growth from USDA ERS AgTFPInternational2022).The implication is that the TFP deceleration story is not primarily a funding story—at least not in the short run.USDA ERS analysisof the global slowdown points to structural factors that likely apply across all three focal countries to varying degrees: a slowing diffusion frontier for the major productivity-enhancing technologies of the 20th century, policy and public-attitude barriers that slow adoption of newer technologies in developed countries specifically, and measurement challenges that are particularly acute in agriculture. This is offered as the most likely explanation given the cross-country pattern, not as a finding established by this analysis on its own. Australia is not alone in showing TFP deceleration despite investment. The United States, with by far the largest absolute public agricultural R&D spending of the three countries, also shows modest mean TFP growth of around 0.7% annually between 2001 and 2020. Canada, with the highest mean TFP growth at 2.4% over the same period, invested at roughly the same absolute scale as Australia but in a much larger agricultural economy (see Table 1; R&D from GRAPE v1.0.0, Zenodo 15507361; TFP trend growth from USDA ERS AgTFPInternational2022). The implication is that the TFP deceleration story is not primarily a funding story—at least not in the short run.USDA ERS analysisof the global slowdown points to structural factors that likely apply across all three focal countries to varying degrees: a slowing diffusion frontier for the major productivity-enhancing technologies of the 20th century, policy and public-attitude barriers that slow adoption of newer technologies in developed countries specifically, and measurement challenges that are particularly acute in agriculture. This is offered as the most likely explanation given the cross-country pattern, not as a finding established by this analysis on its own. Figure 1. Agricultural TFP trend growth rates by decade (top) and mean public R&D over the preceding 10 years (bottom) for Canada, Australia, and the United States. The TFP bar row covers six decades (1961–2020); the R&D line covers five decades (1980–2020) because GRAPE v1.0.0 coverage for Canada begins in 1960, providing only one year within the 1961–1969 preceding-decade window—that point is omitted for Canada as unreliable. Australia and the United States have sufficient coverage from 1970 onwards. The peak in Canada’s R&D line at 2000 reflects mean spending over 1991–2000, Canada’s highest-spending decade in real terms before the sustained decline documented in Parts 1 and 2. Sources: USDA ERS AgTFPInternational2022; GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. Relationship is descriptive—confounded by terms of trade, private R&D, and farm structure Sources: USDA ERS AgTFPInternational2022; GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. Relationship is descriptive—confounded by terms of trade, private R&D, and farm structure What Does the Return on R&D Investment Look Like?This section presents the original analytical contribution of Part 3: an implied-return figure that asks a deliberately simple question. If we take gross agricultural production value (GPV) as the output measure and public R&D expenditure as the input measure, how does the ratio between them compare across countries and over time?This is an illustrative exercise, not a causal return-on-investment estimate. Agricultural output is shaped by weather, global commodity prices, private sector investment, farm structure, and policy—none of which are held constant in this calculation. The framing is best understood as a rough measure of output efficiency relative to public investment: how much agricultural output is associated, descriptively, with each dollar of public R&D.All figures use GPV in constant 2014–2016 international USD (FAOSTAT, element code 152) against public R&D expenditure from GRAPE v1.0.0 (van Dijk et al. 2025, Zenodo 15507361) in million 2017 PPP (purchasing power parity) USD. The two currency bases are not identical, the FAOSTAT Geary-Khamis and GRAPE 2017 ICP bases differ by approximately 3–5%, and the mismatch is flagged as a stated limitation throughout.Implied Return Ratios (2022)Table 2 presents the GPV per public R&D dollar at the most recent available data point (2022), under three lag assumptions: contemporaneous, with R&D lagged 10 years, and with R&D lagged 15 years. The 10- and 15-year lags reflect the consensus in the agricultural R&D literature that the productivity impact of research investment takes a decade or more to materialize at the farm level.Table 2. Implied GPV/Public R&D Ratios by Country and Lag Specification (2022)CountryLagGPV (M USD)Public R&D (M PPP)GPV/R&D RatioAustraliaContemporaneous50,982670.676.0R&D lagged 10 yrs50,982849.160.0R&D lagged 15 yrs50,982654.677.9CanadaContemporaneous51,558870.659.2R&D lagged 10 yrs51,558843.261.1R&D lagged 15 yrs51,558877.458.8United StatesContemporaneous365,7115,11471.5R&D lagged 10 yrs365,7115,25769.6R&D lagged 15 yrs365,7115,94261.6Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. GPV in constant 2014–16 int. USD; R&D in million 2017 PPP USD. Descriptive ratio—not a causal return estimate.Several features of Table 2 are worth noting. Canada’s contemporaneous GPV/R&D ratio (59.2) is the lowest of the three countries, reflecting both its moderate agricultural output scale and its higher public R&D spending relative to Australia. When R&D is lagged 10 years, Canada’s ratio rises modestly to 61.1, the smallest shift of the three countries, suggesting that the relationship between past R&D and current output is relatively stable in the Canadian case. By the 15-year lag, however, the ratio dips slightly below its starting point, to 58.8. Read together, the three Canadian figures (59.2 → 61.1 → 58.8) trace the lag structure of public R&D directly: a decade of spending growth is associated with a modest near-term gain, while spending from 15 years prior—an older, smaller R&D base—corresponds to a slightly lower ratio. This is consistent with the lag effects discussed in Section 2 and with Canada’s current position in the R&D cycle, where recent spending growth has not yet had a full 15 years to show up in output.Australia, by contrast, shows the largest swing across lag specifications: from 76.0 contemporaneously to 60.0 at a 10-year lag, as the elevated R&D spending of the prior decade enters the denominator. This pattern is consistent with the interpretation that the private R&D growth documented in Part 2 has raised Australia’s investment base substantially without yet generating proportionate output gains, reinforcing the TFP puzzle described in Section 2.The United States shows a steady decline across lag specifications (71.5 → 69.6 → 61.6), reflecting the consistently high and growing public R&D base. The large absolute GPV keeps the ratio comparatively elevated even as the denominator grows.Implied Return Over Time: The Indexed FigureFigure 2 makes the cross-country divergence visible in a way the raw ratios in Table 2 understate. Each country’s series is indexed to its own first available year (1990 for the contemporaneous panel; 2000 and 2005 for the 10- and 15-year lag panels, respectively, since lagging the R&D series shortens the usable run). On the contemporaneous panel, Canada’s indexed series climbs steadily above its own baseline, reaching roughly 150–190 by the 2010s, while the United States rises more modestly and Australia oscillates close to its own starting point with no clear trend. The picture shifts once R&D is lagged. At the 10-year lag, Canada’s climb is even more pronounced, its series rises furthest above 100 of the three countries, while Australia’s series sits persistently below 100 for most of the 2000s and 2010s, consistent with the TFP puzzle described in Section 2: a rapidly growing R&D base without a matching rise in output. At the 15-year lag, Canada remains the strongest performer through most of the series, but Australia’s line turns sharply upward in the final years shown, climbing back toward and above 100. Taken together, Canada’s indexed return has been on a more consistently rising trajectory across all three lag specifications than either comparator, while Australia’s pattern is the most volatile, swinging from clearly below baseline to a late recovery depending on which lag is used. This volatility is itself informative: it suggests Australia’s investment-to-output relationship is more sensitive to the choice of lag than Canada’s, which is consistent with Australia’s R&D base having grown unevenly over the period rather than at a steady rate.What this means in plain terms: no matter how the timing of R&D spending relative to output is sliced, Canada has consistently generated more agricultural output per public research dollar than either Australia or the United States — and that edge has grown over time rather than shrinking. Australia’s story is the opposite: its return on R&D has swung unpredictably, falling well behind for most of the 2000s and 2010s before a late recovery, which points to a much less stable link between what it spends and what it gets back.Figure 2. GPV/R&D ratio indexed to own-country baseline (base year = 100), for Canada (blue), Australia (orange), and the United States (green), across contemporaneous, 10-year lag, and 15-year lag specifications. Grey lines in the contemporaneous panel = all other countries in the GRAPE sample, capped at the 95th percentile; the lagged panels omit the background countries since fewer of them have enough years of data to survive the lag.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. Descriptive ratio—not a causal return estimate.The raw ratios in Figure 3 underline the scale gap that the indexed figure deliberately removes. In the most recent year, the United States generates roughly $71 of agricultural output per public R&D dollar contemporaneously, compared with roughly $76 for Australia and $59 for Canada, figures driven as much by the size of each country’s R&D base as by underlying productivity. Canada’s raw ratio is consistently the lowest of the three across all three lag specifications, which mechanically reflects its position between Australia’s smaller agricultural economy and the United States’ much larger one, relative to a public R&D budget closer in absolute size to Australia’s. The trajectories also diverge: Canada’s raw ratio rises steadily from the early 1990s through the most recent data, narrowing a gap that was much wider in the 1990s; the United States holds the highest contemporaneous ratio for most of the series before Australia’s late upturn overtakes it; and Australia’s ratio is comparatively flat through the 2000s before that sharp rise in the final years. These patterns mirror the indexed story in Figure 2, but Figure 3 makes clear that the underlying dollar magnitudes are not directly comparable across countries without that indexing step.Breaking these numbers down: the raw dollar figures aren’t comparable across countries on their own, because a country with a much bigger farm sector will naturally show a bigger number here, regardless of how efficiently it turns research spending into output. That’s why Figure 2’s indexed version, which tracks each country’s change relative to its own starting point rather than the raw dollar amount, is the one to rely on for comparing performance between Canada, Australia, and the United States.Figure 3. Raw GPV/R&D ratio (M USD per M PPP USD) for Canada, Australia, and the United States, across contemporaneous, 10-year lag, and 15-year lag specifications. Note: large values reflect output scale relative to R&D spending and are not directly comparable across countries without indexing. Use Figure 2 for cross-country comparison.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361.R&D Spending and TFP Growth: The Cross-Country PatternTable 1 summarizes the country-level averages underlying Figure 4. Canada’s position, high mean TFP growth (2.4%) with a moderate public R&D base (mean $850M PPP, 2000–2020) places it above the OLS (ordinary least squares) fit line in Figure 4, suggesting that Canadian agriculture has generated comparatively strong productivity growth per dollar of public research investment over this period. Australia sits on the opposite side of the fit line—its mean R&D base is roughly comparable to Canada’s, but its mean TFP growth is negative, placing it well below the line and reinforcing the disconnect discussed in Section 2a. The United States sits almost directly on the fit line: its much larger R&D base corresponds to a TFP growth rate close to what the global cross-country relationship would predict, with no strong over- or under-performance in either direction. Read together, the three focal countries span the range of outcomes visible in the broader scatter—Canada above the line, the United States on it, and Australia below it—despite Canada and Australia having broadly similar public R&D levels, which is itself evidence that funding level alone is a weak predictor of productivity outcome.Figure 4. Country-level scatter of mean public agricultural R&D (log₁₀ scale, M 2017 PPP USD) versus mean TFP trend growth rate (%, 2001–2020). Dashed line = OLS fit (descriptive, not causal). Canada (blue), Australia (orange), and USA (green) are highlighted.Sources: R&D: GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. TFP: USDA ERS AgTFPInternational2022. Relationship is confounded by country size, farm structure, and other factors. What Does the Return on R&D Investment Look Like?This section presents the original analytical contribution of Part 3: an implied-return figure that asks a deliberately simple question. If we take gross agricultural production value (GPV) as the output measure and public R&D expenditure as the input measure, how does the ratio between them compare across countries and over time?This is an illustrative exercise, not a causal return-on-investment estimate. Agricultural output is shaped by weather, global commodity prices, private sector investment, farm structure, and policy—none of which are held constant in this calculation. The framing is best understood as a rough measure of output efficiency relative to public investment: how much agricultural output is associated, descriptively, with each dollar of public R&D.All figures use GPV in constant 2014–2016 international USD (FAOSTAT, element code 152) against public R&D expenditure from GRAPE v1.0.0 (van Dijk et al. 2025, Zenodo 15507361) in million 2017 PPP (purchasing power parity) USD. The two currency bases are not identical, the FAOSTAT Geary-Khamis and GRAPE 2017 ICP bases differ by approximately 3–5%, and the mismatch is flagged as a stated limitation throughout.Implied Return Ratios (2022)Table 2 presents the GPV per public R&D dollar at the most recent available data point (2022), under three lag assumptions: contemporaneous, with R&D lagged 10 years, and with R&D lagged 15 years. The 10- and 15-year lags reflect the consensus in the agricultural R&D literature that the productivity impact of research investment takes a decade or more to materialize at the farm level.Table 2. Implied GPV/Public R&D Ratios by Country and Lag Specification (2022)CountryLagGPV (M USD)Public R&D (M PPP)GPV/R&D RatioAustraliaContemporaneous50,982670.676.0R&D lagged 10 yrs50,982849.160.0R&D lagged 15 yrs50,982654.677.9CanadaContemporaneous51,558870.659.2R&D lagged 10 yrs51,558843.261.1R&D lagged 15 yrs51,558877.458.8United StatesContemporaneous365,7115,11471.5R&D lagged 10 yrs365,7115,25769.6R&D lagged 15 yrs365,7115,94261.6Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. GPV in constant 2014–16 int. USD; R&D in million 2017 PPP USD. Descriptive ratio—not a causal return estimate.Several features of Table 2 are worth noting. Canada’s contemporaneous GPV/R&D ratio (59.2) is the lowest of the three countries, reflecting both its moderate agricultural output scale and its higher public R&D spending relative to Australia. When R&D is lagged 10 years, Canada’s ratio rises modestly to 61.1, the smallest shift of the three countries, suggesting that the relationship between past R&D and current output is relatively stable in the Canadian case. By the 15-year lag, however, the ratio dips slightly below its starting point, to 58.8. Read together, the three Canadian figures (59.2 → 61.1 → 58.8) trace the lag structure of public R&D directly: a decade of spending growth is associated with a modest near-term gain, while spending from 15 years prior—an older, smaller R&D base—corresponds to a slightly lower ratio. This is consistent with the lag effects discussed in Section 2 and with Canada’s current position in the R&D cycle, where recent spending growth has not yet had a full 15 years to show up in output.Australia, by contrast, shows the largest swing across lag specifications: from 76.0 contemporaneously to 60.0 at a 10-year lag, as the elevated R&D spending of the prior decade enters the denominator. This pattern is consistent with the interpretation that the private R&D growth documented in Part 2 has raised Australia’s investment base substantially without yet generating proportionate output gains, reinforcing the TFP puzzle described in Section 2.The United States shows a steady decline across lag specifications (71.5 → 69.6 → 61.6), reflecting the consistently high and growing public R&D base. The large absolute GPV keeps the ratio comparatively elevated even as the denominator grows.Implied Return Over Time: The Indexed FigureFigure 2 makes the cross-country divergence visible in a way the raw ratios in Table 2 understate. Each country’s series is indexed to its own first available year (1990 for the contemporaneous panel; 2000 and 2005 for the 10- and 15-year lag panels, respectively, since lagging the R&D series shortens the usable run). On the contemporaneous panel, Canada’s indexed series climbs steadily above its own baseline, reaching roughly 150–190 by the 2010s, while the United States rises more modestly and Australia oscillates close to its own starting point with no clear trend. The picture shifts once R&D is lagged. At the 10-year lag, Canada’s climb is even more pronounced, its series rises furthest above 100 of the three countries, while Australia’s series sits persistently below 100 for most of the 2000s and 2010s, consistent with the TFP puzzle described in Section 2: a rapidly growing R&D base without a matching rise in output. At the 15-year lag, Canada remains the strongest performer through most of the series, but Australia’s line turns sharply upward in the final years shown, climbing back toward and above 100. Taken together, Canada’s indexed return has been on a more consistently rising trajectory across all three lag specifications than either comparator, while Australia’s pattern is the most volatile, swinging from clearly below baseline to a late recovery depending on which lag is used. This volatility is itself informative: it suggests Australia’s investment-to-output relationship is more sensitive to the choice of lag than Canada’s, which is consistent with Australia’s R&D base having grown unevenly over the period rather than at a steady rate.What this means in plain terms: no matter how the timing of R&D spending relative to output is sliced, Canada has consistently generated more agricultural output per public research dollar than either Australia or the United States — and that edge has grown over time rather than shrinking. Australia’s story is the opposite: its return on R&D has swung unpredictably, falling well behind for most of the 2000s and 2010s before a late recovery, which points to a much less stable link between what it spends and what it gets back.Figure 2. GPV/R&D ratio indexed to own-country baseline (base year = 100), for Canada (blue), Australia (orange), and the United States (green), across contemporaneous, 10-year lag, and 15-year lag specifications. Grey lines in the contemporaneous panel = all other countries in the GRAPE sample, capped at the 95th percentile; the lagged panels omit the background countries since fewer of them have enough years of data to survive the lag.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. Descriptive ratio—not a causal return estimate.The raw ratios in Figure 3 underline the scale gap that the indexed figure deliberately removes. In the most recent year, the United States generates roughly $71 of agricultural output per public R&D dollar contemporaneously, compared with roughly $76 for Australia and $59 for Canada, figures driven as much by the size of each country’s R&D base as by underlying productivity. Canada’s raw ratio is consistently the lowest of the three across all three lag specifications, which mechanically reflects its position between Australia’s smaller agricultural economy and the United States’ much larger one, relative to a public R&D budget closer in absolute size to Australia’s. The trajectories also diverge: Canada’s raw ratio rises steadily from the early 1990s through the most recent data, narrowing a gap that was much wider in the 1990s; the United States holds the highest contemporaneous ratio for most of the series before Australia’s late upturn overtakes it; and Australia’s ratio is comparatively flat through the 2000s before that sharp rise in the final years. These patterns mirror the indexed story in Figure 2, but Figure 3 makes clear that the underlying dollar magnitudes are not directly comparable across countries without that indexing step.Breaking these numbers down: the raw dollar figures aren’t comparable across countries on their own, because a country with a much bigger farm sector will naturally show a bigger number here, regardless of how efficiently it turns research spending into output. That’s why Figure 2’s indexed version, which tracks each country’s change relative to its own starting point rather than the raw dollar amount, is the one to rely on for comparing performance between Canada, Australia, and the United States.Figure 3. Raw GPV/R&D ratio (M USD per M PPP USD) for Canada, Australia, and the United States, across contemporaneous, 10-year lag, and 15-year lag specifications. Note: large values reflect output scale relative to R&D spending and are not directly comparable across countries without indexing. Use Figure 2 for cross-country comparison.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361.R&D Spending and TFP Growth: The Cross-Country PatternTable 1 summarizes the country-level averages underlying Figure 4. Canada’s position, high mean TFP growth (2.4%) with a moderate public R&D base (mean $850M PPP, 2000–2020) places it above the OLS (ordinary least squares) fit line in Figure 4, suggesting that Canadian agriculture has generated comparatively strong productivity growth per dollar of public research investment over this period. Australia sits on the opposite side of the fit line—its mean R&D base is roughly comparable to Canada’s, but its mean TFP growth is negative, placing it well below the line and reinforcing the disconnect discussed in Section 2a. The United States sits almost directly on the fit line: its much larger R&D base corresponds to a TFP growth rate close to what the global cross-country relationship would predict, with no strong over- or under-performance in either direction. Read together, the three focal countries span the range of outcomes visible in the broader scatter—Canada above the line, the United States on it, and Australia below it—despite Canada and Australia having broadly similar public R&D levels, which is itself evidence that funding level alone is a weak predictor of productivity outcome.Figure 4. Country-level scatter of mean public agricultural R&D (log₁₀ scale, M 2017 PPP USD) versus mean TFP trend growth rate (%, 2001–2020). Dashed line = OLS fit (descriptive, not causal). Canada (blue), Australia (orange), and USA (green) are highlighted.Sources: R&D: GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. TFP: USDA ERS AgTFPInternational2022. Relationship is confounded by country size, farm structure, and other factors. What Does the Return on R&D Investment Look Like?This section presents the original analytical contribution of Part 3: an implied-return figure that asks a deliberately simple question. If we take gross agricultural production value (GPV) as the output measure and public R&D expenditure as the input measure, how does the ratio between them compare across countries and over time?This is an illustrative exercise, not a causal return-on-investment estimate. Agricultural output is shaped by weather, global commodity prices, private sector investment, farm structure, and policy—none of which are held constant in this calculation. The framing is best understood as a rough measure of output efficiency relative to public investment: how much agricultural output is associated, descriptively, with each dollar of public R&D.All figures use GPV in constant 2014–2016 international USD (FAOSTAT, element code 152) against public R&D expenditure from GRAPE v1.0.0 (van Dijk et al. 2025, Zenodo 15507361) in million 2017 PPP (purchasing power parity) USD. The two currency bases are not identical, the FAOSTAT Geary-Khamis and GRAPE 2017 ICP bases differ by approximately 3–5%, and the mismatch is flagged as a stated limitation throughout.Implied Return Ratios (2022)Table 2 presents the GPV per public R&D dollar at the most recent available data point (2022), under three lag assumptions: contemporaneous, with R&D lagged 10 years, and with R&D lagged 15 years. The 10- and 15-year lags reflect the consensus in the agricultural R&D literature that the productivity impact of research investment takes a decade or more to materialize at the farm level.Table 2. Implied GPV/Public R&D Ratios by Country and Lag Specification (2022)CountryLagGPV (M USD)Public R&D (M PPP)GPV/R&D RatioAustraliaContemporaneous50,982670.676.0R&D lagged 10 yrs50,982849.160.0R&D lagged 15 yrs50,982654.677.9CanadaContemporaneous51,558870.659.2R&D lagged 10 yrs51,558843.261.1R&D lagged 15 yrs51,558877.458.8United StatesContemporaneous365,7115,11471.5R&D lagged 10 yrs365,7115,25769.6R&D lagged 15 yrs365,7115,94261.6Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. GPV in constant 2014–16 int. USD; R&D in million 2017 PPP USD. Descriptive ratio—not a causal return estimate.Several features of Table 2 are worth noting. Canada’s contemporaneous GPV/R&D ratio (59.2) is the lowest of the three countries, reflecting both its moderate agricultural output scale and its higher public R&D spending relative to Australia. When R&D is lagged 10 years, Canada’s ratio rises modestly to 61.1, the smallest shift of the three countries, suggesting that the relationship between past R&D and current output is relatively stable in the Canadian case. By the 15-year lag, however, the ratio dips slightly below its starting point, to 58.8. Read together, the three Canadian figures (59.2 → 61.1 → 58.8) trace the lag structure of public R&D directly: a decade of spending growth is associated with a modest near-term gain, while spending from 15 years prior—an older, smaller R&D base—corresponds to a slightly lower ratio. This is consistent with the lag effects discussed in Section 2 and with Canada’s current position in the R&D cycle, where recent spending growth has not yet had a full 15 years to show up in output.Australia, by contrast, shows the largest swing across lag specifications: from 76.0 contemporaneously to 60.0 at a 10-year lag, as the elevated R&D spending of the prior decade enters the denominator. This pattern is consistent with the interpretation that the private R&D growth documented in Part 2 has raised Australia’s investment base substantially without yet generating proportionate output gains, reinforcing the TFP puzzle described in Section 2.The United States shows a steady decline across lag specifications (71.5 → 69.6 → 61.6), reflecting the consistently high and growing public R&D base. The large absolute GPV keeps the ratio comparatively elevated even as the denominator grows.Implied Return Over Time: The Indexed FigureFigure 2 makes the cross-country divergence visible in a way the raw ratios in Table 2 understate. Each country’s series is indexed to its own first available year (1990 for the contemporaneous panel; 2000 and 2005 for the 10- and 15-year lag panels, respectively, since lagging the R&D series shortens the usable run). On the contemporaneous panel, Canada’s indexed series climbs steadily above its own baseline, reaching roughly 150–190 by the 2010s, while the United States rises more modestly and Australia oscillates close to its own starting point with no clear trend. The picture shifts once R&D is lagged. At the 10-year lag, Canada’s climb is even more pronounced, its series rises furthest above 100 of the three countries, while Australia’s series sits persistently below 100 for most of the 2000s and 2010s, consistent with the TFP puzzle described in Section 2: a rapidly growing R&D base without a matching rise in output. At the 15-year lag, Canada remains the strongest performer through most of the series, but Australia’s line turns sharply upward in the final years shown, climbing back toward and above 100. Taken together, Canada’s indexed return has been on a more consistently rising trajectory across all three lag specifications than either comparator, while Australia’s pattern is the most volatile, swinging from clearly below baseline to a late recovery depending on which lag is used. This volatility is itself informative: it suggests Australia’s investment-to-output relationship is more sensitive to the choice of lag than Canada’s, which is consistent with Australia’s R&D base having grown unevenly over the period rather than at a steady rate.What this means in plain terms: no matter how the timing of R&D spending relative to output is sliced, Canada has consistently generated more agricultural output per public research dollar than either Australia or the United States — and that edge has grown over time rather than shrinking. Australia’s story is the opposite: its return on R&D has swung unpredictably, falling well behind for most of the 2000s and 2010s before a late recovery, which points to a much less stable link between what it spends and what it gets back.Figure 2. GPV/R&D ratio indexed to own-country baseline (base year = 100), for Canada (blue), Australia (orange), and the United States (green), across contemporaneous, 10-year lag, and 15-year lag specifications. Grey lines in the contemporaneous panel = all other countries in the GRAPE sample, capped at the 95th percentile; the lagged panels omit the background countries since fewer of them have enough years of data to survive the lag.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. Descriptive ratio—not a causal return estimate.The raw ratios in Figure 3 underline the scale gap that the indexed figure deliberately removes. In the most recent year, the United States generates roughly $71 of agricultural output per public R&D dollar contemporaneously, compared with roughly $76 for Australia and $59 for Canada, figures driven as much by the size of each country’s R&D base as by underlying productivity. Canada’s raw ratio is consistently the lowest of the three across all three lag specifications, which mechanically reflects its position between Australia’s smaller agricultural economy and the United States’ much larger one, relative to a public R&D budget closer in absolute size to Australia’s. The trajectories also diverge: Canada’s raw ratio rises steadily from the early 1990s through the most recent data, narrowing a gap that was much wider in the 1990s; the United States holds the highest contemporaneous ratio for most of the series before Australia’s late upturn overtakes it; and Australia’s ratio is comparatively flat through the 2000s before that sharp rise in the final years. These patterns mirror the indexed story in Figure 2, but Figure 3 makes clear that the underlying dollar magnitudes are not directly comparable across countries without that indexing step.Breaking these numbers down: the raw dollar figures aren’t comparable across countries on their own, because a country with a much bigger farm sector will naturally show a bigger number here, regardless of how efficiently it turns research spending into output. That’s why Figure 2’s indexed version, which tracks each country’s change relative to its own starting point rather than the raw dollar amount, is the one to rely on for comparing performance between Canada, Australia, and the United States.Figure 3. Raw GPV/R&D ratio (M USD per M PPP USD) for Canada, Australia, and the United States, across contemporaneous, 10-year lag, and 15-year lag specifications. Note: large values reflect output scale relative to R&D spending and are not directly comparable across countries without indexing. Use Figure 2 for cross-country comparison.Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361.R&D Spending and TFP Growth: The Cross-Country PatternTable 1 summarizes the country-level averages underlying Figure 4. Canada’s position, high mean TFP growth (2.4%) with a moderate public R&D base (mean $850M PPP, 2000–2020) places it above the OLS (ordinary least squares) fit line in Figure 4, suggesting that Canadian agriculture has generated comparatively strong productivity growth per dollar of public research investment over this period. Australia sits on the opposite side of the fit line—its mean R&D base is roughly comparable to Canada’s, but its mean TFP growth is negative, placing it well below the line and reinforcing the disconnect discussed in Section 2a. The United States sits almost directly on the fit line: its much larger R&D base corresponds to a TFP growth rate close to what the global cross-country relationship would predict, with no strong over- or under-performance in either direction. Read together, the three focal countries span the range of outcomes visible in the broader scatter—Canada above the line, the United States on it, and Australia below it—despite Canada and Australia having broadly similar public R&D levels, which is itself evidence that funding level alone is a weak predictor of productivity outcome.Figure 4. Country-level scatter of mean public agricultural R&D (log₁₀ scale, M 2017 PPP USD) versus mean TFP trend growth rate (%, 2001–2020). Dashed line = OLS fit (descriptive, not causal). Canada (blue), Australia (orange), and USA (green) are highlighted.Sources: R&D: GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. TFP: USDA ERS AgTFPInternational2022. Relationship is confounded by country size, farm structure, and other factors. What Does the Return on R&D Investment Look Like? This section presents the original analytical contribution of Part 3: an implied-return figure that asks a deliberately simple question. If we take gross agricultural production value (GPV) as the output measure and public R&D expenditure as the input measure, how does the ratio between them compare across countries and over time?This is an illustrative exercise, not a causal return-on-investment estimate. Agricultural output is shaped by weather, global commodity prices, private sector investment, farm structure, and policy—none of which are held constant in this calculation. The framing is best understood as a rough measure of output efficiency relative to public investment: how much agricultural output is associated, descriptively, with each dollar of public R&D.All figures use GPV in constant 2014–2016 international USD (FAOSTAT, element code 152) against public R&D expenditure from GRAPE v1.0.0 (van Dijk et al. 2025, Zenodo 15507361) in million 2017 PPP (purchasing power parity) USD. The two currency bases are not identical, the FAOSTAT Geary-Khamis and GRAPE 2017 ICP bases differ by approximately 3–5%, and the mismatch is flagged as a stated limitation throughout. This section presents the original analytical contribution of Part 3: an implied-return figure that asks a deliberately simple question. If we take gross agricultural production value (GPV) as the output measure and public R&D expenditure as the input measure, how does the ratio between them compare across countries and over time? This is an illustrative exercise, not a causal return-on-investment estimate. Agricultural output is shaped by weather, global commodity prices, private sector investment, farm structure, and policy—none of which are held constant in this calculation. The framing is best understood as a rough measure of output efficiency relative to public investment: how much agricultural output is associated, descriptively, with each dollar of public R&D. All figures use GPV in constant 2014–2016 international USD (FAOSTAT, element code 152) against public R&D expenditure from GRAPE v1.0.0 (van Dijk et al. 2025, Zenodo 15507361) in million 2017 PPP (purchasing power parity) USD. The two currency bases are not identical, the FAOSTAT Geary-Khamis and GRAPE 2017 ICP bases differ by approximately 3–5%, and the mismatch is flagged as a stated limitation throughout. Implied Return Ratios (2022) Table 2 presents the GPV per public R&D dollar at the most recent available data point (2022), under three lag assumptions: contemporaneous, with R&D lagged 10 years, and with R&D lagged 15 years. The 10- and 15-year lags reflect the consensus in the agricultural R&D literature that the productivity impact of research investment takes a decade or more to materialize at the farm level. Table 2 presents the GPV per public R&D dollar at the most recent available data point (2022), under three lag assumptions: contemporaneous, with R&D lagged 10 years, and with R&D lagged 15 years. The 10- and 15-year lags reflect the consensus in the agricultural R&D literature that the productivity impact of research investment takes a decade or more to materialize at the farm level. Table 2. Implied GPV/Public R&D Ratios by Country and Lag Specification (2022) CountryLagGPV (M USD)Public R&D (M PPP)GPV/R&D RatioAustraliaContemporaneous50,982670.676.0R&D lagged 10 yrs50,982849.160.0R&D lagged 15 yrs50,982654.677.9CanadaContemporaneous51,558870.659.2R&D lagged 10 yrs51,558843.261.1R&D lagged 15 yrs51,558877.458.8United StatesContemporaneous365,7115,11471.5R&D lagged 10 yrs365,7115,25769.6R&D lagged 15 yrs365,7115,94261.6Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. GPV in constant 2014–16 int. USD; R&D in million 2017 PPP USD. Descriptive ratio—not a causal return estimate. CountryLagGPV (M USD)Public R&D (M PPP)GPV/R&D RatioAustraliaContemporaneous50,982670.676.0R&D lagged 10 yrs50,982849.160.0R&D lagged 15 yrs50,982654.677.9CanadaContemporaneous51,558870.659.2R&D lagged 10 yrs51,558843.261.1R&D lagged 15 yrs51,558877.458.8United StatesContemporaneous365,7115,11471.5R&D lagged 10 yrs365,7115,25769.6R&D lagged 15 yrs365,7115,94261.6Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. GPV in constant 2014–16 int. USD; R&D in million 2017 PPP USD. Descriptive ratio—not a causal return estimate. CountryLagGPV (M USD)Public R&D (M PPP)GPV/R&D RatioAustraliaContemporaneous50,982670.676.0R&D lagged 10 yrs50,982849.160.0R&D lagged 15 yrs50,982654.677.9CanadaContemporaneous51,558870.659.2R&D lagged 10 yrs51,558843.261.1R&D lagged 15 yrs51,558877.458.8United StatesContemporaneous365,7115,11471.5R&D lagged 10 yrs365,7115,25769.6R&D lagged 15 yrs365,7115,94261.6Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. GPV in constant 2014–16 int. USD; R&D in million 2017 PPP USD. Descriptive ratio—not a causal return estimate. Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. GPV in constant 2014–16 int. USD; R&D in million 2017 PPP USD. Descriptive ratio—not a causal return estimate. Several features of Table 2 are worth noting. Canada’s contemporaneous GPV/R&D ratio (59.2) is the lowest of the three countries, reflecting both its moderate agricultural output scale and its higher public R&D spending relative to Australia. When R&D is lagged 10 years, Canada’s ratio rises modestly to 61.1, the smallest shift of the three countries, suggesting that the relationship between past R&D and current output is relatively stable in the Canadian case. By the 15-year lag, however, the ratio dips slightly below its starting point, to 58.8. Read together, the three Canadian figures (59.2 → 61.1 → 58.8) trace the lag structure of public R&D directly: a decade of spending growth is associated with a modest near-term gain, while spending from 15 years prior—an older, smaller R&D base—corresponds to a slightly lower ratio. This is consistent with the lag effects discussed in Section 2 and with Canada’s current position in the R&D cycle, where recent spending growth has not yet had a full 15 years to show up in output.Australia, by contrast, shows the largest swing across lag specifications: from 76.0 contemporaneously to 60.0 at a 10-year lag, as the elevated R&D spending of the prior decade enters the denominator. This pattern is consistent with the interpretation that the private R&D growth documented in Part 2 has raised Australia’s investment base substantially without yet generating proportionate output gains, reinforcing the TFP puzzle described in Section 2.The United States shows a steady decline across lag specifications (71.5 → 69.6 → 61.6), reflecting the consistently high and growing public R&D base. The large absolute GPV keeps the ratio comparatively elevated even as the denominator grows. Several features of Table 2 are worth noting. Canada’s contemporaneous GPV/R&D ratio (59.2) is the lowest of the three countries, reflecting both its moderate agricultural output scale and its higher public R&D spending relative to Australia. When R&D is lagged 10 years, Canada’s ratio rises modestly to 61.1, the smallest shift of the three countries, suggesting that the relationship between past R&D and current output is relatively stable in the Canadian case. By the 15-year lag, however, the ratio dips slightly below its starting point, to 58.8. Read together, the three Canadian figures (59.2 → 61.1 → 58.8) trace the lag structure of public R&D directly: a decade of spending growth is associated with a modest near-term gain, while spending from 15 years prior—an older, smaller R&D base—corresponds to a slightly lower ratio. This is consistent with the lag effects discussed in Section 2 and with Canada’s current position in the R&D cycle, where recent spending growth has not yet had a full 15 years to show up in output. Australia, by contrast, shows the largest swing across lag specifications: from 76.0 contemporaneously to 60.0 at a 10-year lag, as the elevated R&D spending of the prior decade enters the denominator. This pattern is consistent with the interpretation that the private R&D growth documented in Part 2 has raised Australia’s investment base substantially without yet generating proportionate output gains, reinforcing the TFP puzzle described in Section 2. The United States shows a steady decline across lag specifications (71.5 → 69.6 → 61.6), reflecting the consistently high and growing public R&D base. The large absolute GPV keeps the ratio comparatively elevated even as the denominator grows. Implied Return Over Time: The Indexed Figure Figure 2 makes the cross-country divergence visible in a way the raw ratios in Table 2 understate. Each country’s series is indexed to its own first available year (1990 for the contemporaneous panel; 2000 and 2005 for the 10- and 15-year lag panels, respectively, since lagging the R&D series shortens the usable run). On the contemporaneous panel, Canada’s indexed series climbs steadily above its own baseline, reaching roughly 150–190 by the 2010s, while the United States rises more modestly and Australia oscillates close to its own starting point with no clear trend. The picture shifts once R&D is lagged. At the 10-year lag, Canada’s climb is even more pronounced, its series rises furthest above 100 of the three countries, while Australia’s series sits persistently below 100 for most of the 2000s and 2010s, consistent with the TFP puzzle described in Section 2: a rapidly growing R&D base without a matching rise in output. At the 15-year lag, Canada remains the strongest performer through most of the series, but Australia’s line turns sharply upward in the final years shown, climbing back toward and above 100. Taken together, Canada’s indexed return has been on a more consistently rising trajectory across all three lag specifications than either comparator, while Australia’s pattern is the most volatile, swinging from clearly below baseline to a late recovery depending on which lag is used. This volatility is itself informative: it suggests Australia’s investment-to-output relationship is more sensitive to the choice of lag than Canada’s, which is consistent with Australia’s R&D base having grown unevenly over the period rather than at a steady rate.What this means in plain terms: no matter how the timing of R&D spending relative to output is sliced, Canada has consistently generated more agricultural output per public research dollar than either Australia or the United States — and that edge has grown over time rather than shrinking. Australia’s story is the opposite: its return on R&D has swung unpredictably, falling well behind for most of the 2000s and 2010s before a late recovery, which points to a much less stable link between what it spends and what it gets back. Figure 2 makes the cross-country divergence visible in a way the raw ratios in Table 2 understate. Each country’s series is indexed to its own first available year (1990 for the contemporaneous panel; 2000 and 2005 for the 10- and 15-year lag panels, respectively, since lagging the R&D series shortens the usable run). On the contemporaneous panel, Canada’s indexed series climbs steadily above its own baseline, reaching roughly 150–190 by the 2010s, while the United States rises more modestly and Australia oscillates close to its own starting point with no clear trend. The picture shifts once R&D is lagged. At the 10-year lag, Canada’s climb is even more pronounced, its series rises furthest above 100 of the three countries, while Australia’s series sits persistently below 100 for most of the 2000s and 2010s, consistent with the TFP puzzle described in Section 2: a rapidly growing R&D base without a matching rise in output. At the 15-year lag, Canada remains the strongest performer through most of the series, but Australia’s line turns sharply upward in the final years shown, climbing back toward and above 100. Taken together, Canada’s indexed return has been on a more consistently rising trajectory across all three lag specifications than either comparator, while Australia’s pattern is the most volatile, swinging from clearly below baseline to a late recovery depending on which lag is used. This volatility is itself informative: it suggests Australia’s investment-to-output relationship is more sensitive to the choice of lag than Canada’s, which is consistent with Australia’s R&D base having grown unevenly over the period rather than at a steady rate. What this means in plain terms: no matter how the timing of R&D spending relative to output is sliced, Canada has consistently generated more agricultural output per public research dollar than either Australia or the United States — and that edge has grown over time rather than shrinking. Australia’s story is the opposite: its return on R&D has swung unpredictably, falling well behind for most of the 2000s and 2010s before a late recovery, which points to a much less stable link between what it spends and what it gets back. Figure 2. GPV/R&D ratio indexed to own-country baseline (base year = 100), for Canada (blue), Australia (orange), and the United States (green), across contemporaneous, 10-year lag, and 15-year lag specifications. Grey lines in the contemporaneous panel = all other countries in the GRAPE sample, capped at the 95th percentile; the lagged panels omit the background countries since fewer of them have enough years of data to survive the lag. Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. Descriptive ratio—not a causal return estimate. Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. Descriptive ratio—not a causal return estimate. The raw ratios in Figure 3 underline the scale gap that the indexed figure deliberately removes. In the most recent year, the United States generates roughly $71 of agricultural output per public R&D dollar contemporaneously, compared with roughly $76 for Australia and $59 for Canada, figures driven as much by the size of each country’s R&D base as by underlying productivity. Canada’s raw ratio is consistently the lowest of the three across all three lag specifications, which mechanically reflects its position between Australia’s smaller agricultural economy and the United States’ much larger one, relative to a public R&D budget closer in absolute size to Australia’s. The trajectories also diverge: Canada’s raw ratio rises steadily from the early 1990s through the most recent data, narrowing a gap that was much wider in the 1990s; the United States holds the highest contemporaneous ratio for most of the series before Australia’s late upturn overtakes it; and Australia’s ratio is comparatively flat through the 2000s before that sharp rise in the final years. These patterns mirror the indexed story in Figure 2, but Figure 3 makes clear that the underlying dollar magnitudes are not directly comparable across countries without that indexing step.Breaking these numbers down: the raw dollar figures aren’t comparable across countries on their own, because a country with a much bigger farm sector will naturally show a bigger number here, regardless of how efficiently it turns research spending into output. That’s why Figure 2’s indexed version, which tracks each country’s change relative to its own starting point rather than the raw dollar amount, is the one to rely on for comparing performance between Canada, Australia, and the United States. The raw ratios in Figure 3 underline the scale gap that the indexed figure deliberately removes. In the most recent year, the United States generates roughly $71 of agricultural output per public R&D dollar contemporaneously, compared with roughly $76 for Australia and $59 for Canada, figures driven as much by the size of each country’s R&D base as by underlying productivity. Canada’s raw ratio is consistently the lowest of the three across all three lag specifications, which mechanically reflects its position between Australia’s smaller agricultural economy and the United States’ much larger one, relative to a public R&D budget closer in absolute size to Australia’s. The trajectories also diverge: Canada’s raw ratio rises steadily from the early 1990s through the most recent data, narrowing a gap that was much wider in the 1990s; the United States holds the highest contemporaneous ratio for most of the series before Australia’s late upturn overtakes it; and Australia’s ratio is comparatively flat through the 2000s before that sharp rise in the final years. These patterns mirror the indexed story in Figure 2, but Figure 3 makes clear that the underlying dollar magnitudes are not directly comparable across countries without that indexing step. Breaking these numbers down: the raw dollar figures aren’t comparable across countries on their own, because a country with a much bigger farm sector will naturally show a bigger number here, regardless of how efficiently it turns research spending into output. That’s why Figure 2’s indexed version, which tracks each country’s change relative to its own starting point rather than the raw dollar amount, is the one to rely on for comparing performance between Canada, Australia, and the United States. Figure 3. Raw GPV/R&D ratio (M USD per M PPP USD) for Canada, Australia, and the United States, across contemporaneous, 10-year lag, and 15-year lag specifications. Note: large values reflect output scale relative to R&D spending and are not directly comparable across countries without indexing. Use Figure 2 for cross-country comparison. Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. Sources: FAOSTAT QV domain element 152; GRAPE v1.0.0, Zenodo 15507361. R&D Spending and TFP Growth: The Cross-Country Pattern Table 1 summarizes the country-level averages underlying Figure 4. Canada’s position, high mean TFP growth (2.4%) with a moderate public R&D base (mean $850M PPP, 2000–2020) places it above the OLS (ordinary least squares) fit line in Figure 4, suggesting that Canadian agriculture has generated comparatively strong productivity growth per dollar of public research investment over this period. Australia sits on the opposite side of the fit line—its mean R&D base is roughly comparable to Canada’s, but its mean TFP growth is negative, placing it well below the line and reinforcing the disconnect discussed in Section 2a. The United States sits almost directly on the fit line: its much larger R&D base corresponds to a TFP growth rate close to what the global cross-country relationship would predict, with no strong over- or under-performance in either direction. Read together, the three focal countries span the range of outcomes visible in the broader scatter—Canada above the line, the United States on it, and Australia below it—despite Canada and Australia having broadly similar public R&D levels, which is itself evidence that funding level alone is a weak predictor of productivity outcome. Table 1 summarizes the country-level averages underlying Figure 4. Canada’s position, high mean TFP growth (2.4%) with a moderate public R&D base (mean $850M PPP, 2000–2020) places it above the OLS (ordinary least squares) fit line in Figure 4, suggesting that Canadian agriculture has generated comparatively strong productivity growth per dollar of public research investment over this period. Australia sits on the opposite side of the fit line—its mean R&D base is roughly comparable to Canada’s, but its mean TFP growth is negative, placing it well below the line and reinforcing the disconnect discussed in Section 2a. The United States sits almost directly on the fit line: its much larger R&D base corresponds to a TFP growth rate close to what the global cross-country relationship would predict, with no strong over- or under-performance in either direction. Read together, the three focal countries span the range of outcomes visible in the broader scatter—Canada above the line, the United States on it, and Australia below it—despite Canada and Australia having broadly similar public R&D levels, which is itself evidence that funding level alone is a weak predictor of productivity outcome. Figure 4. Country-level scatter of mean public agricultural R&D (log₁₀ scale, M 2017 PPP USD) versus mean TFP trend growth rate (%, 2001–2020). Dashed line = OLS fit (descriptive, not causal). Canada (blue), Australia (orange), and USA (green) are highlighted. Sources: R&D: GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. TFP: USDA ERS AgTFPInternational2022. Relationship is confounded by country size, farm structure, and other factors. Sources: R&D: GRAPE v1.0.0 (van Dijk et al. 2025), Zenodo 15507361. TFP: USDA ERS AgTFPInternational2022. Relationship is confounded by country size, farm structure, and other factors. Barriers: Supply Side and Demand SideThe standard framing for agricultural innovation policy focuses on the supply side: how much is being invested in R&D, by whom, and through what institutional channels. Parts 1 and 2 of this series largely engaged with that framing. But the data in Sections 2 and 3 above suggest that the relationship between investment and productivity outcome is looser than a straightforward supply-side model would predict.This section examines both sides of the innovation chain, supply and demand, with particular attention to the demand side, which receives comparatively little attention in the Canadian policy literature.Supply-Side Barriers (Recap)Part 2covered the main supply-side barriers in detail; a brief summary is included here for completeness.Intellectual Property (IP) protections and farmers’ privilege –The balance between IP protection for plant breeders and the traditional farmers’ privilege to save seed shapes the private R&D incentive structure. The Canadian Food Inspection Agency (CFIA) finalized updated Plant Breeders’ Rights Regulations in April 2026, narrowing farmers’ privilege to small grain crops; see Section 5 for the policy detail and the resulting debate.SR&ED complexity –The Scientific Research and Experimental Development tax creditprogram favours well-capitalized firms with dedicated compliance resources, creating a barrier for smaller agricultural innovators.Regulatory timelines –Approval timelines for genetically modified varieties represent a material lag between research output and farm-level availability.AAFC research centre consolidation –The scheduled closure and consolidation of Agriculture and Agri-Food Canada (AAFC) research facilities represents a near-term supply-side risk that has not yet been fully reflected in the spending data. AAFC announced in January 2026 that it would close seven research centres and satellite farms, part of a broader federal workforce reduction; the House of Commons Standing Committee on Agriculture and Agri-Food has since called for the closures to be reversed (seeRealAgriculture).Demand-Side Barriers: The Underexamined SideThe gap between research output and changed farm practice is poorly measured in Canada, and it represents the most original contribution of this series. Even if supply-side constraints were fully resolved—if R&D spending were adequate, IP frameworks balanced, and regulatory timelines shortened—there would remain a substantial question about whether innovations reach farms and are adopted at scale. The available evidence suggests the answer is often no, or not quickly.Farm financial constraints –Part 2 documented the margin pressures facing Canadian farm operations. Capital-constrained farms face higher effective hurdle rates for technology adoption, particularly for precision-agriculture tools that require upfront equipment investment and ongoing subscription costs.Extension services –ABARES data for Australia shows that public agricultural advisory and extension funding has declined at an average annual rate of approximately5.7%in real terms since 2005–06, a sustained multi-decade contraction even as private R&D investment has grown over the same period. While no equivalent national time-series exists for Canada, the institutional trend toward commodity organization-led extension and away from publicly funded provincial extension services is broadly similar, and has been reinforced recently by the closure and staffing reductions at several AAFC research facilities discussed in Section 4a. This represents a structural weakening in the knowledge-transfer link between research output and farm decision-making.Technology adoption rates –The 2021 Census of Agriculture provides the most granular available picture of precision-agriculture adoption among Canadian farms.Figure 5 reveals a pronounced geographic concentration of precision-agriculture adoption. GPS/auto-steer adoption is highest in Saskatchewan (about 47% of all farms) and Manitoba (about 41%), with Alberta close behind (about 31%); Ontario shows comparatively stronger uptake of GIS mapping and variable-rate application than the Prairie provinces, though at lower GPS adoption (around 23%). New Brunswick, Nova Scotia, and Newfoundland and Labrador show minimal adoption across all four technology categories, generally in the single digits. Prince Edward Island is a notable exception to the Atlantic pattern, with GPS and variable-rate adoption (around 20% each) closer to Ontario’s than to its immediate regional neighbours—likely reflecting its potato and row-crop heavy production mix, which is more compatible with precision-ag tools than the mixed and horticultural operations more common elsewhere in Atlantic Canada.Figure 5. Share of all farms reporting use of selected precision-agriculture practices (GPS/auto-steer, GIS mapping, variable-rate application, and drones) by province, 2021 Census of Agriculture. Percentages are calculated against the total number of farms in each province (all NAICS — North American Industry Classification System — categories), since oilseed/grain-specific denominators are near-zero in several Atlantic provinces; western provinces (AB, SK, MB), where grain and oilseed production dominates, are the most directly comparable.Source: Statistics Canada, Census of Agriculture 2021, Table 32-10-0379-01. Released 2022-05-11.This pattern likely reflects a combination of farm size effects (Prairie operations are on average much larger and more mechanized), crop type (row crops are better suited to precision-agriculture tools than many Eastern mixed or horticultural operations), and capital availability. The adoption data do not allow clean isolation of these factors, but the regional concentration is itself a policy-relevant finding: precision-agriculture supports that are designed around Prairie-scale operations may not transfer effectively to the rest of the country.Scale and operator age –Both farm size and operator age are well-documented correlates of technology adoption. Part 1 documented the ongoing consolidation of Canadian farm operations and the aging of the operator population, and the2021 Census of Agriculturesharpens that picture. The number of census farms fell to 189,874 in 2021, down 1.9% from 2016, even as average farm size has grown substantially over the longer run—from roughly 676 to 809 acres between 2001 and 2021. The distribution is increasingly bimodal: farms of 2,240 acres or more accounted for only 8.9% of all farms in 2021 but generated 37.3% of total operating revenue, while farms under 240 acres made up 52.3% of the total but contributed only about a quarter of revenue. On the operator side, the average age of Canadian farm operators rose to 56.0 years in 2021 (median 58.0), and the share of operators aged 55 and older climbed from 54.5% in 2016 to 60.5% in 2021, while the share of younger operators continued to decline. Combined with the precision-agriculture adoption gradient in Figure 5, this points toward a structural pattern rather than a regional quirk: adoption is concentrated among a shrinking pool of larger, Prairie-dominant operations, while the median Canadian farm—smaller, and run by an operator closer to retirement age than to a multi-decade payback horizon—sits further from the technology-adoption frontier described in Part 1. Larger operations have stronger financial incentives to adopt labour-saving and yield-improving technologies; older operators face shorter payback horizons and, on average, lower returns to technology learning. Barriers: Supply Side and Demand SideThe standard framing for agricultural innovation policy focuses on the supply side: how much is being invested in R&D, by whom, and through what institutional channels. Parts 1 and 2 of this series largely engaged with that framing. But the data in Sections 2 and 3 above suggest that the relationship between investment and productivity outcome is looser than a straightforward supply-side model would predict.This section examines both sides of the innovation chain, supply and demand, with particular attention to the demand side, which receives comparatively little attention in the Canadian policy literature.Supply-Side Barriers (Recap)Part 2covered the main supply-side barriers in detail; a brief summary is included here for completeness.Intellectual Property (IP) protections and farmers’ privilege –The balance between IP protection for plant breeders and the traditional farmers’ privilege to save seed shapes the private R&D incentive structure. The Canadian Food Inspection Agency (CFIA) finalized updated Plant Breeders’ Rights Regulations in April 2026, narrowing farmers’ privilege to small grain crops; see Section 5 for the policy detail and the resulting debate.SR&ED complexity –The Scientific Research and Experimental Development tax creditprogram favours well-capitalized firms with dedicated compliance resources, creating a barrier for smaller agricultural innovators.Regulatory timelines –Approval timelines for genetically modified varieties represent a material lag between research output and farm-level availability.AAFC research centre consolidation –The scheduled closure and consolidation of Agriculture and Agri-Food Canada (AAFC) research facilities represents a near-term supply-side risk that has not yet been fully reflected in the spending data. AAFC announced in January 2026 that it would close seven research centres and satellite farms, part of a broader federal workforce reduction; the House of Commons Standing Committee on Agriculture and Agri-Food has since called for the closures to be reversed (seeRealAgriculture).Demand-Side Barriers: The Underexamined SideThe gap between research output and changed farm practice is poorly measured in Canada, and it represents the most original contribution of this series. Even if supply-side constraints were fully resolved—if R&D spending were adequate, IP frameworks balanced, and regulatory timelines shortened—there would remain a substantial question about whether innovations reach farms and are adopted at scale. The available evidence suggests the answer is often no, or not quickly.Farm financial constraints –Part 2 documented the margin pressures facing Canadian farm operations. Capital-constrained farms face higher effective hurdle rates for technology adoption, particularly for precision-agriculture tools that require upfront equipment investment and ongoing subscription costs.Extension services –ABARES data for Australia shows that public agricultural advisory and extension funding has declined at an average annual rate of approximately5.7%in real terms since 2005–06, a sustained multi-decade contraction even as private R&D investment has grown over the same period. While no equivalent national time-series exists for Canada, the institutional trend toward commodity organization-led extension and away from publicly funded provincial extension services is broadly similar, and has been reinforced recently by the closure and staffing reductions at several AAFC research facilities discussed in Section 4a. This represents a structural weakening in the knowledge-transfer link between research output and farm decision-making.Technology adoption rates –The 2021 Census of Agriculture provides the most granular available picture of precision-agriculture adoption among Canadian farms.Figure 5 reveals a pronounced geographic concentration of precision-agriculture adoption. GPS/auto-steer adoption is highest in Saskatchewan (about 47% of all farms) and Manitoba (about 41%), with Alberta close behind (about 31%); Ontario shows comparatively stronger uptake of GIS mapping and variable-rate application than the Prairie provinces, though at lower GPS adoption (around 23%). New Brunswick, Nova Scotia, and Newfoundland and Labrador show minimal adoption across all four technology categories, generally in the single digits. Prince Edward Island is a notable exception to the Atlantic pattern, with GPS and variable-rate adoption (around 20% each) closer to Ontario’s than to its immediate regional neighbours—likely reflecting its potato and row-crop heavy production mix, which is more compatible with precision-ag tools than the mixed and horticultural operations more common elsewhere in Atlantic Canada.Figure 5. Share of all farms reporting use of selected precision-agriculture practices (GPS/auto-steer, GIS mapping, variable-rate application, and drones) by province, 2021 Census of Agriculture. Percentages are calculated against the total number of farms in each province (all NAICS — North American Industry Classification System — categories), since oilseed/grain-specific denominators are near-zero in several Atlantic provinces; western provinces (AB, SK, MB), where grain and oilseed production dominates, are the most directly comparable.Source: Statistics Canada, Census of Agriculture 2021, Table 32-10-0379-01. Released 2022-05-11.This pattern likely reflects a combination of farm size effects (Prairie operations are on average much larger and more mechanized), crop type (row crops are better suited to precision-agriculture tools than many Eastern mixed or horticultural operations), and capital availability. The adoption data do not allow clean isolation of these factors, but the regional concentration is itself a policy-relevant finding: precision-agriculture supports that are designed around Prairie-scale operations may not transfer effectively to the rest of the country.Scale and operator age –Both farm size and operator age are well-documented correlates of technology adoption. Part 1 documented the ongoing consolidation of Canadian farm operations and the aging of the operator population, and the2021 Census of Agriculturesharpens that picture. The number of census farms fell to 189,874 in 2021, down 1.9% from 2016, even as average farm size has grown substantially over the longer run—from roughly 676 to 809 acres between 2001 and 2021. The distribution is increasingly bimodal: farms of 2,240 acres or more accounted for only 8.9% of all farms in 2021 but generated 37.3% of total operating revenue, while farms under 240 acres made up 52.3% of the total but contributed only about a quarter of revenue. On the operator side, the average age of Canadian farm operators rose to 56.0 years in 2021 (median 58.0), and the share of operators aged 55 and older climbed from 54.5% in 2016 to 60.5% in 2021, while the share of younger operators continued to decline. Combined with the precision-agriculture adoption gradient in Figure 5, this points toward a structural pattern rather than a regional quirk: adoption is concentrated among a shrinking pool of larger, Prairie-dominant operations, while the median Canadian farm—smaller, and run by an operator closer to retirement age than to a multi-decade payback horizon—sits further from the technology-adoption frontier described in Part 1. Larger operations have stronger financial incentives to adopt labour-saving and yield-improving technologies; older operators face shorter payback horizons and, on average, lower returns to technology learning. Barriers: Supply Side and Demand SideThe standard framing for agricultural innovation policy focuses on the supply side: how much is being invested in R&D, by whom, and through what institutional channels. Parts 1 and 2 of this series largely engaged with that framing. But the data in Sections 2 and 3 above suggest that the relationship between investment and productivity outcome is looser than a straightforward supply-side model would predict.This section examines both sides of the innovation chain, supply and demand, with particular attention to the demand side, which receives comparatively little attention in the Canadian policy literature.Supply-Side Barriers (Recap)Part 2covered the main supply-side barriers in detail; a brief summary is included here for completeness.Intellectual Property (IP) protections and farmers’ privilege –The balance between IP protection for plant breeders and the traditional farmers’ privilege to save seed shapes the private R&D incentive structure. The Canadian Food Inspection Agency (CFIA) finalized updated Plant Breeders’ Rights Regulations in April 2026, narrowing farmers’ privilege to small grain crops; see Section 5 for the policy detail and the resulting debate.SR&ED complexity –The Scientific Research and Experimental Development tax creditprogram favours well-capitalized firms with dedicated compliance resources, creating a barrier for smaller agricultural innovators.Regulatory timelines –Approval timelines for genetically modified varieties represent a material lag between research output and farm-level availability.AAFC research centre consolidation –The scheduled closure and consolidation of Agriculture and Agri-Food Canada (AAFC) research facilities represents a near-term supply-side risk that has not yet been fully reflected in the spending data. AAFC announced in January 2026 that it would close seven research centres and satellite farms, part of a broader federal workforce reduction; the House of Commons Standing Committee on Agriculture and Agri-Food has since called for the closures to be reversed (seeRealAgriculture).Demand-Side Barriers: The Underexamined SideThe gap between research output and changed farm practice is poorly measured in Canada, and it represents the most original contribution of this series. Even if supply-side constraints were fully resolved—if R&D spending were adequate, IP frameworks balanced, and regulatory timelines shortened—there would remain a substantial question about whether innovations reach farms and are adopted at scale. The available evidence suggests the answer is often no, or not quickly.Farm financial constraints –Part 2 documented the margin pressures facing Canadian farm operations. Capital-constrained farms face higher effective hurdle rates for technology adoption, particularly for precision-agriculture tools that require upfront equipment investment and ongoing subscription costs.Extension services –ABARES data for Australia shows that public agricultural advisory and extension funding has declined at an average annual rate of approximately5.7%in real terms since 2005–06, a sustained multi-decade contraction even as private R&D investment has grown over the same period. While no equivalent national time-series exists for Canada, the institutional trend toward commodity organization-led extension and away from publicly funded provincial extension services is broadly similar, and has been reinforced recently by the closure and staffing reductions at several AAFC research facilities discussed in Section 4a. This represents a structural weakening in the knowledge-transfer link between research output and farm decision-making.Technology adoption rates –The 2021 Census of Agriculture provides the most granular available picture of precision-agriculture adoption among Canadian farms.Figure 5 reveals a pronounced geographic concentration of precision-agriculture adoption. GPS/auto-steer adoption is highest in Saskatchewan (about 47% of all farms) and Manitoba (about 41%), with Alberta close behind (about 31%); Ontario shows comparatively stronger uptake of GIS mapping and variable-rate application than the Prairie provinces, though at lower GPS adoption (around 23%). New Brunswick, Nova Scotia, and Newfoundland and Labrador show minimal adoption across all four technology categories, generally in the single digits. Prince Edward Island is a notable exception to the Atlantic pattern, with GPS and variable-rate adoption (around 20% each) closer to Ontario’s than to its immediate regional neighbours—likely reflecting its potato and row-crop heavy production mix, which is more compatible with precision-ag tools than the mixed and horticultural operations more common elsewhere in Atlantic Canada.Figure 5. Share of all farms reporting use of selected precision-agriculture practices (GPS/auto-steer, GIS mapping, variable-rate application, and drones) by province, 2021 Census of Agriculture. Percentages are calculated against the total number of farms in each province (all NAICS — North American Industry Classification System — categories), since oilseed/grain-specific denominators are near-zero in several Atlantic provinces; western provinces (AB, SK, MB), where grain and oilseed production dominates, are the most directly comparable.Source: Statistics Canada, Census of Agriculture 2021, Table 32-10-0379-01. Released 2022-05-11.This pattern likely reflects a combination of farm size effects (Prairie operations are on average much larger and more mechanized), crop type (row crops are better suited to precision-agriculture tools than many Eastern mixed or horticultural operations), and capital availability. The adoption data do not allow clean isolation of these factors, but the regional concentration is itself a policy-relevant finding: precision-agriculture supports that are designed around Prairie-scale operations may not transfer effectively to the rest of the country.Scale and operator age –Both farm size and operator age are well-documented correlates of technology adoption. Part 1 documented the ongoing consolidation of Canadian farm operations and the aging of the operator population, and the2021 Census of Agriculturesharpens that picture. The number of census farms fell to 189,874 in 2021, down 1.9% from 2016, even as average farm size has grown substantially over the longer run—from roughly 676 to 809 acres between 2001 and 2021. The distribution is increasingly bimodal: farms of 2,240 acres or more accounted for only 8.9% of all farms in 2021 but generated 37.3% of total operating revenue, while farms under 240 acres made up 52.3% of the total but contributed only about a quarter of revenue. On the operator side, the average age of Canadian farm operators rose to 56.0 years in 2021 (median 58.0), and the share of operators aged 55 and older climbed from 54.5% in 2016 to 60.5% in 2021, while the share of younger operators continued to decline. Combined with the precision-agriculture adoption gradient in Figure 5, this points toward a structural pattern rather than a regional quirk: adoption is concentrated among a shrinking pool of larger, Prairie-dominant operations, while the median Canadian farm—smaller, and run by an operator closer to retirement age than to a multi-decade payback horizon—sits further from the technology-adoption frontier described in Part 1. Larger operations have stronger financial incentives to adopt labour-saving and yield-improving technologies; older operators face shorter payback horizons and, on average, lower returns to technology learning. Barriers: Supply Side and Demand Side The standard framing for agricultural innovation policy focuses on the supply side: how much is being invested in R&D, by whom, and through what institutional channels. Parts 1 and 2 of this series largely engaged with that framing. But the data in Sections 2 and 3 above suggest that the relationship between investment and productivity outcome is looser than a straightforward supply-side model would predict.This section examines both sides of the innovation chain, supply and demand, with particular attention to the demand side, which receives comparatively little attention in the Canadian policy literature. The standard framing for agricultural innovation policy focuses on the supply side: how much is being invested in R&D, by whom, and through what institutional channels. Parts 1 and 2 of this series largely engaged with that framing. But the data in Sections 2 and 3 above suggest that the relationship between investment and productivity outcome is looser than a straightforward supply-side model would predict. This section examines both sides of the innovation chain, supply and demand, with particular attention to the demand side, which receives comparatively little attention in the Canadian policy literature. Supply-Side Barriers (Recap) Part 2covered the main supply-side barriers in detail; a brief summary is included here for completeness.Intellectual Property (IP) protections and farmers’ privilege –The balance between IP protection for plant breeders and the traditional farmers’ privilege to save seed shapes the private R&D incentive structure. The Canadian Food Inspection Agency (CFIA) finalized updated Plant Breeders’ Rights Regulations in April 2026, narrowing farmers’ privilege to small grain crops; see Section 5 for the policy detail and the resulting debate.SR&ED complexity –The Scientific Research and Experimental Development tax creditprogram favours well-capitalized firms with dedicated compliance resources, creating a barrier for smaller agricultural innovators.Regulatory timelines –Approval timelines for genetically modified varieties represent a material lag between research output and farm-level availability.AAFC research centre consolidation –The scheduled closure and consolidation of Agriculture and Agri-Food Canada (AAFC) research facilities represents a near-term supply-side risk that has not yet been fully reflected in the spending data. AAFC announced in January 2026 that it would close seven research centres and satellite farms, part of a broader federal workforce reduction; the House of Commons Standing Committee on Agriculture and Agri-Food has since called for the closures to be reversed (seeRealAgriculture). Part 2covered the main supply-side barriers in detail; a brief summary is included here for completeness. Intellectual Property (IP) protections and farmers’ privilege –The balance between IP protection for plant breeders and the traditional farmers’ privilege to save seed shapes the private R&D incentive structure. The Canadian Food Inspection Agency (CFIA) finalized updated Plant Breeders’ Rights Regulations in April 2026, narrowing farmers’ privilege to small grain crops; see Section 5 for the policy detail and the resulting debate. SR&ED complexity –The Scientific Research and Experimental Development tax creditprogram favours well-capitalized firms with dedicated compliance resources, creating a barrier for smaller agricultural innovators. Regulatory timelines –Approval timelines for genetically modified varieties represent a material lag between research output and farm-level availability. AAFC research centre consolidation –The scheduled closure and consolidation of Agriculture and Agri-Food Canada (AAFC) research facilities represents a near-term supply-side risk that has not yet been fully reflected in the spending data. AAFC announced in January 2026 that it would close seven research centres and satellite farms, part of a broader federal workforce reduction; the House of Commons Standing Committee on Agriculture and Agri-Food has since called for the closures to be reversed (seeRealAgriculture). Demand-Side Barriers: The Underexamined Side The gap between research output and changed farm practice is poorly measured in Canada, and it represents the most original contribution of this series. Even if supply-side constraints were fully resolved—if R&D spending were adequate, IP frameworks balanced, and regulatory timelines shortened—there would remain a substantial question about whether innovations reach farms and are adopted at scale. The available evidence suggests the answer is often no, or not quickly.Farm financial constraints –Part 2 documented the margin pressures facing Canadian farm operations. Capital-constrained farms face higher effective hurdle rates for technology adoption, particularly for precision-agriculture tools that require upfront equipment investment and ongoing subscription costs.Extension services –ABARES data for Australia shows that public agricultural advisory and extension funding has declined at an average annual rate of approximately5.7%in real terms since 2005–06, a sustained multi-decade contraction even as private R&D investment has grown over the same period. While no equivalent national time-series exists for Canada, the institutional trend toward commodity organization-led extension and away from publicly funded provincial extension services is broadly similar, and has been reinforced recently by the closure and staffing reductions at several AAFC research facilities discussed in Section 4a. This represents a structural weakening in the knowledge-transfer link between research output and farm decision-making.Technology adoption rates –The 2021 Census of Agriculture provides the most granular available picture of precision-agriculture adoption among Canadian farms.Figure 5 reveals a pronounced geographic concentration of precision-agriculture adoption. GPS/auto-steer adoption is highest in Saskatchewan (about 47% of all farms) and Manitoba (about 41%), with Alberta close behind (about 31%); Ontario shows comparatively stronger uptake of GIS mapping and variable-rate application than the Prairie provinces, though at lower GPS adoption (around 23%). New Brunswick, Nova Scotia, and Newfoundland and Labrador show minimal adoption across all four technology categories, generally in the single digits. Prince Edward Island is a notable exception to the Atlantic pattern, with GPS and variable-rate adoption (around 20% each) closer to Ontario’s than to its immediate regional neighbours—likely reflecting its potato and row-crop heavy production mix, which is more compatible with precision-ag tools than the mixed and horticultural operations more common elsewhere in Atlantic Canada. The gap between research output and changed farm practice is poorly measured in Canada, and it represents the most original contribution of this series. Even if supply-side constraints were fully resolved—if R&D spending were adequate, IP frameworks balanced, and regulatory timelines shortened—there would remain a substantial question about whether innovations reach farms and are adopted at scale. The available evidence suggests the answer is often no, or not quickly. Farm financial constraints –Part 2 documented the margin pressures facing Canadian farm operations. Capital-constrained farms face higher effective hurdle rates for technology adoption, particularly for precision-agriculture tools that require upfront equipment investment and ongoing subscription costs. Extension services –ABARES data for Australia shows that public agricultural advisory and extension funding has declined at an average annual rate of approximately5.7%in real terms since 2005–06, a sustained multi-decade contraction even as private R&D investment has grown over the same period. While no equivalent national time-series exists for Canada, the institutional trend toward commodity organization-led extension and away from publicly funded provincial extension services is broadly similar, and has been reinforced recently by the closure and staffing reductions at several AAFC research facilities discussed in Section 4a. This represents a structural weakening in the knowledge-transfer link between research output and farm decision-making. Technology adoption rates –The 2021 Census of Agriculture provides the most granular available picture of precision-agriculture adoption among Canadian farms. Figure 5 reveals a pronounced geographic concentration of precision-agriculture adoption. GPS/auto-steer adoption is highest in Saskatchewan (about 47% of all farms) and Manitoba (about 41%), with Alberta close behind (about 31%); Ontario shows comparatively stronger uptake of GIS mapping and variable-rate application than the Prairie provinces, though at lower GPS adoption (around 23%). New Brunswick, Nova Scotia, and Newfoundland and Labrador show minimal adoption across all four technology categories, generally in the single digits. Prince Edward Island is a notable exception to the Atlantic pattern, with GPS and variable-rate adoption (around 20% each) closer to Ontario’s than to its immediate regional neighbours—likely reflecting its potato and row-crop heavy production mix, which is more compatible with precision-ag tools than the mixed and horticultural operations more common elsewhere in Atlantic Canada. Figure 5. Share of all farms reporting use of selected precision-agriculture practices (GPS/auto-steer, GIS mapping, variable-rate application, and drones) by province, 2021 Census of Agriculture. Percentages are calculated against the total number of farms in each province (all NAICS — North American Industry Classification System — categories), since oilseed/grain-specific denominators are near-zero in several Atlantic provinces; western provinces (AB, SK, MB), where grain and oilseed production dominates, are the most directly comparable. Source: Statistics Canada, Census of Agriculture 2021, Table 32-10-0379-01. Released 2022-05-11. Source: Statistics Canada, Census of Agriculture 2021, Table 32-10-0379-01. Released 2022-05-11. This pattern likely reflects a combination of farm size effects (Prairie operations are on average much larger and more mechanized), crop type (row crops are better suited to precision-agriculture tools than many Eastern mixed or horticultural operations), and capital availability. The adoption data do not allow clean isolation of these factors, but the regional concentration is itself a policy-relevant finding: precision-agriculture supports that are designed around Prairie-scale operations may not transfer effectively to the rest of the country.Scale and operator age –Both farm size and operator age are well-documented correlates of technology adoption. Part 1 documented the ongoing consolidation of Canadian farm operations and the aging of the operator population, and the2021 Census of Agriculturesharpens that picture. The number of census farms fell to 189,874 in 2021, down 1.9% from 2016, even as average farm size has grown substantially over the longer run—from roughly 676 to 809 acres between 2001 and 2021. The distribution is increasingly bimodal: farms of 2,240 acres or more accounted for only 8.9% of all farms in 2021 but generated 37.3% of total operating revenue, while farms under 240 acres made up 52.3% of the total but contributed only about a quarter of revenue. On the operator side, the average age of Canadian farm operators rose to 56.0 years in 2021 (median 58.0), and the share of operators aged 55 and older climbed from 54.5% in 2016 to 60.5% in 2021, while the share of younger operators continued to decline. Combined with the precision-agriculture adoption gradient in Figure 5, this points toward a structural pattern rather than a regional quirk: adoption is concentrated among a shrinking pool of larger, Prairie-dominant operations, while the median Canadian farm—smaller, and run by an operator closer to retirement age than to a multi-decade payback horizon—sits further from the technology-adoption frontier described in Part 1. Larger operations have stronger financial incentives to adopt labour-saving and yield-improving technologies; older operators face shorter payback horizons and, on average, lower returns to technology learning. This pattern likely reflects a combination of farm size effects (Prairie operations are on average much larger and more mechanized), crop type (row crops are better suited to precision-agriculture tools than many Eastern mixed or horticultural operations), and capital availability. The adoption data do not allow clean isolation of these factors, but the regional concentration is itself a policy-relevant finding: precision-agriculture supports that are designed around Prairie-scale operations may not transfer effectively to the rest of the country. Scale and operator age –Both farm size and operator age are well-documented correlates of technology adoption. Part 1 documented the ongoing consolidation of Canadian farm operations and the aging of the operator population, and the2021 Census of Agriculturesharpens that picture. The number of census farms fell to 189,874 in 2021, down 1.9% from 2016, even as average farm size has grown substantially over the longer run—from roughly 676 to 809 acres between 2001 and 2021. The distribution is increasingly bimodal: farms of 2,240 acres or more accounted for only 8.9% of all farms in 2021 but generated 37.3% of total operating revenue, while farms under 240 acres made up 52.3% of the total but contributed only about a quarter of revenue. On the operator side, the average age of Canadian farm operators rose to 56.0 years in 2021 (median 58.0), and the share of operators aged 55 and older climbed from 54.5% in 2016 to 60.5% in 2021, while the share of younger operators continued to decline. Combined with the precision-agriculture adoption gradient in Figure 5, this points toward a structural pattern rather than a regional quirk: adoption is concentrated among a shrinking pool of larger, Prairie-dominant operations, while the median Canadian farm—smaller, and run by an operator closer to retirement age than to a multi-decade payback horizon—sits further from the technology-adoption frontier described in Part 1. Larger operations have stronger financial incentives to adopt labour-saving and yield-improving technologies; older operators face shorter payback horizons and, on average, lower returns to technology learning. Where are the Leverage Points?The preceding sections point toward a more granular set of leverage points than the standard policy discussion about R&D funding levels. The evidence does not support the conclusion that simply increasing public agricultural R&D spending is sufficient to close Canada’s productivity trajectory risk. The Australia case illustrates that investment and outcome can diverge substantially, and the adoption data suggest that the return on existing research is partially constrained by the capacity to put it into practice.This section identifies the areas where the evidence points toward tractable improvements, without prescribing a specific funding level or policy instrument.IP regime modernisation –This is no longer purely prospective: the CFIA finalized amendments to the Plant Breeders’ Rights Regulations in April 2026 (SOR/2026-74), with the changes coming into force shortly after. The most consequential change narrows farmers’ privilege, the right to save and reuse seed from a protected variety, to small grain crops such as cereals and pulses, removing the privilege for horticultural and ornamental varieties where it had previously applied.Producer groups broadly supported the changeas a confidence-building measure for plant breeders investing in Canada, while the National Farmers Union opposed it on the grounds that narrowing farmers’ privilege increases production costs and entrenches reliance on proprietary varieties. A well-calibrated IP framework that maintains private sector R&D incentives while preserving farmer access to saved seed represents a genuine leverage point, one that does not require new public expenditure—though the regulation as finalized has already drawn pushback on exactly that balance, and its effect on adoption and farm-level costs will only be observable with a few years of data.SR&ED simplification –Simplifying the SR&ED tax credit regime for smaller agricultural innovators and agribusiness firms would improve the return on an existing public commitment without necessarily increasing its total cost.Extension and knowledge transfer –The case for investment in knowledge-transfer infrastructure is arguably stronger than the case for additional research investment at the margin, precisely because the adoption gap is the underexamined variable. Extension services operate at the link between research output and farm practice, the point where the evidence suggests the current system is weakest.International R&D spillovers –Canada benefits disproportionately from research conducted in the United States and elsewhere. The size of the US public agricultural R&D enterprise (mean $5.5B PPP, 2000–2020) dwarfs Canada’s, and much of the resulting knowledge is applicable in Canadian conditions. A strategy that emphasizes Canada’s capacity to absorb and adapt internationally generated knowledge, rather than simply increasing domestic R&D outlays, may generate higher returns per public dollar.Industry-led R&D models –Australia’s Rural Research and Development Corporation (RDC) model—in which industry levy funds are matched by government contributions and channelled through commodity-specific bodies—offers a structural comparison worth examining in the Canadian context. There are 15 RDCs covering Australia’s major agricultural sectors, collectively investing roughly AUD$825 million annually; theAustralian government matches eligible R&D levy contributions up to 0.5% of each industry’s gross value of production. The model does not simply increase total investment; it changes the governance of how investment decisions are made and how knowledge transfer is organized. Where are the Leverage Points?The preceding sections point toward a more granular set of leverage points than the standard policy discussion about R&D funding levels. The evidence does not support the conclusion that simply increasing public agricultural R&D spending is sufficient to close Canada’s productivity trajectory risk. The Australia case illustrates that investment and outcome can diverge substantially, and the adoption data suggest that the return on existing research is partially constrained by the capacity to put it into practice.This section identifies the areas where the evidence points toward tractable improvements, without prescribing a specific funding level or policy instrument.IP regime modernisation –This is no longer purely prospective: the CFIA finalized amendments to the Plant Breeders’ Rights Regulations in April 2026 (SOR/2026-74), with the changes coming into force shortly after. The most consequential change narrows farmers’ privilege, the right to save and reuse seed from a protected variety, to small grain crops such as cereals and pulses, removing the privilege for horticultural and ornamental varieties where it had previously applied.Producer groups broadly supported the changeas a confidence-building measure for plant breeders investing in Canada, while the National Farmers Union opposed it on the grounds that narrowing farmers’ privilege increases production costs and entrenches reliance on proprietary varieties. A well-calibrated IP framework that maintains private sector R&D incentives while preserving farmer access to saved seed represents a genuine leverage point, one that does not require new public expenditure—though the regulation as finalized has already drawn pushback on exactly that balance, and its effect on adoption and farm-level costs will only be observable with a few years of data.SR&ED simplification –Simplifying the SR&ED tax credit regime for smaller agricultural innovators and agribusiness firms would improve the return on an existing public commitment without necessarily increasing its total cost.Extension and knowledge transfer –The case for investment in knowledge-transfer infrastructure is arguably stronger than the case for additional research investment at the margin, precisely because the adoption gap is the underexamined variable. Extension services operate at the link between research output and farm practice, the point where the evidence suggests the current system is weakest.International R&D spillovers –Canada benefits disproportionately from research conducted in the United States and elsewhere. The size of the US public agricultural R&D enterprise (mean $5.5B PPP, 2000–2020) dwarfs Canada’s, and much of the resulting knowledge is applicable in Canadian conditions. A strategy that emphasizes Canada’s capacity to absorb and adapt internationally generated knowledge, rather than simply increasing domestic R&D outlays, may generate higher returns per public dollar.Industry-led R&D models –Australia’s Rural Research and Development Corporation (RDC) model—in which industry levy funds are matched by government contributions and channelled through commodity-specific bodies—offers a structural comparison worth examining in the Canadian context. There are 15 RDCs covering Australia’s major agricultural sectors, collectively investing roughly AUD$825 million annually; theAustralian government matches eligible R&D levy contributions up to 0.5% of each industry’s gross value of production. The model does not simply increase total investment; it changes the governance of how investment decisions are made and how knowledge transfer is organized. Where are the Leverage Points?The preceding sections point toward a more granular set of leverage points than the standard policy discussion about R&D funding levels. The evidence does not support the conclusion that simply increasing public agricultural R&D spending is sufficient to close Canada’s productivity trajectory risk. The Australia case illustrates that investment and outcome can diverge substantially, and the adoption data suggest that the return on existing research is partially constrained by the capacity to put it into practice.This section identifies the areas where the evidence points toward tractable improvements, without prescribing a specific funding level or policy instrument.IP regime modernisation –This is no longer purely prospective: the CFIA finalized amendments to the Plant Breeders’ Rights Regulations in April 2026 (SOR/2026-74), with the changes coming into force shortly after. The most consequential change narrows farmers’ privilege, the right to save and reuse seed from a protected variety, to small grain crops such as cereals and pulses, removing the privilege for horticultural and ornamental varieties where it had previously applied.Producer groups broadly supported the changeas a confidence-building measure for plant breeders investing in Canada, while the National Farmers Union opposed it on the grounds that narrowing farmers’ privilege increases production costs and entrenches reliance on proprietary varieties. A well-calibrated IP framework that maintains private sector R&D incentives while preserving farmer access to saved seed represents a genuine leverage point, one that does not require new public expenditure—though the regulation as finalized has already drawn pushback on exactly that balance, and its effect on adoption and farm-level costs will only be observable with a few years of data.SR&ED simplification –Simplifying the SR&ED tax credit regime for smaller agricultural innovators and agribusiness firms would improve the return on an existing public commitment without necessarily increasing its total cost.Extension and knowledge transfer –The case for investment in knowledge-transfer infrastructure is arguably stronger than the case for additional research investment at the margin, precisely because the adoption gap is the underexamined variable. Extension services operate at the link between research output and farm practice, the point where the evidence suggests the current system is weakest.International R&D spillovers –Canada benefits disproportionately from research conducted in the United States and elsewhere. The size of the US public agricultural R&D enterprise (mean $5.5B PPP, 2000–2020) dwarfs Canada’s, and much of the resulting knowledge is applicable in Canadian conditions. A strategy that emphasizes Canada’s capacity to absorb and adapt internationally generated knowledge, rather than simply increasing domestic R&D outlays, may generate higher returns per public dollar.Industry-led R&D models –Australia’s Rural Research and Development Corporation (RDC) model—in which industry levy funds are matched by government contributions and channelled through commodity-specific bodies—offers a structural comparison worth examining in the Canadian context. There are 15 RDCs covering Australia’s major agricultural sectors, collectively investing roughly AUD$825 million annually; theAustralian government matches eligible R&D levy contributions up to 0.5% of each industry’s gross value of production. The model does not simply increase total investment; it changes the governance of how investment decisions are made and how knowledge transfer is organized. Where are the Leverage Points? The preceding sections point toward a more granular set of leverage points than the standard policy discussion about R&D funding levels. The evidence does not support the conclusion that simply increasing public agricultural R&D spending is sufficient to close Canada’s productivity trajectory risk. The Australia case illustrates that investment and outcome can diverge substantially, and the adoption data suggest that the return on existing research is partially constrained by the capacity to put it into practice.This section identifies the areas where the evidence points toward tractable improvements, without prescribing a specific funding level or policy instrument.IP regime modernisation –This is no longer purely prospective: the CFIA finalized amendments to the Plant Breeders’ Rights Regulations in April 2026 (SOR/2026-74), with the changes coming into force shortly after. The most consequential change narrows farmers’ privilege, the right to save and reuse seed from a protected variety, to small grain crops such as cereals and pulses, removing the privilege for horticultural and ornamental varieties where it had previously applied.Producer groups broadly supported the changeas a confidence-building measure for plant breeders investing in Canada, while the National Farmers Union opposed it on the grounds that narrowing farmers’ privilege increases production costs and entrenches reliance on proprietary varieties. A well-calibrated IP framework that maintains private sector R&D incentives while preserving farmer access to saved seed represents a genuine leverage point, one that does not require new public expenditure—though the regulation as finalized has already drawn pushback on exactly that balance, and its effect on adoption and farm-level costs will only be observable with a few years of data.SR&ED simplification –Simplifying the SR&ED tax credit regime for smaller agricultural innovators and agribusiness firms would improve the return on an existing public commitment without necessarily increasing its total cost.Extension and knowledge transfer –The case for investment in knowledge-transfer infrastructure is arguably stronger than the case for additional research investment at the margin, precisely because the adoption gap is the underexamined variable. Extension services operate at the link between research output and farm practice, the point where the evidence suggests the current system is weakest.International R&D spillovers –Canada benefits disproportionately from research conducted in the United States and elsewhere. The size of the US public agricultural R&D enterprise (mean $5.5B PPP, 2000–2020) dwarfs Canada’s, and much of the resulting knowledge is applicable in Canadian conditions. A strategy that emphasizes Canada’s capacity to absorb and adapt internationally generated knowledge, rather than simply increasing domestic R&D outlays, may generate higher returns per public dollar.Industry-led R&D models –Australia’s Rural Research and Development Corporation (RDC) model—in which industry levy funds are matched by government contributions and channelled through commodity-specific bodies—offers a structural comparison worth examining in the Canadian context. There are 15 RDCs covering Australia’s major agricultural sectors, collectively investing roughly AUD$825 million annually; theAustralian government matches eligible R&D levy contributions up to 0.5% of each industry’s gross value of production. The model does not simply increase total investment; it changes the governance of how investment decisions are made and how knowledge transfer is organized. The preceding sections point toward a more granular set of leverage points than the standard policy discussion about R&D funding levels. The evidence does not support the conclusion that simply increasing public agricultural R&D spending is sufficient to close Canada’s productivity trajectory risk. The Australia case illustrates that investment and outcome can diverge substantially, and the adoption data suggest that the return on existing research is partially constrained by the capacity to put it into practice. This section identifies the areas where the evidence points toward tractable improvements, without prescribing a specific funding level or policy instrument. IP regime modernisation –This is no longer purely prospective: the CFIA finalized amendments to the Plant Breeders’ Rights Regulations in April 2026 (SOR/2026-74), with the changes coming into force shortly after. The most consequential change narrows farmers’ privilege, the right to save and reuse seed from a protected variety, to small grain crops such as cereals and pulses, removing the privilege for horticultural and ornamental varieties where it had previously applied.Producer groups broadly supported the changeas a confidence-building measure for plant breeders investing in Canada, while the National Farmers Union opposed it on the grounds that narrowing farmers’ privilege increases production costs and entrenches reliance on proprietary varieties. A well-calibrated IP framework that maintains private sector R&D incentives while preserving farmer access to saved seed represents a genuine leverage point, one that does not require new public expenditure—though the regulation as finalized has already drawn pushback on exactly that balance, and its effect on adoption and farm-level costs will only be observable with a few years of data. SR&ED simplification –Simplifying the SR&ED tax credit regime for smaller agricultural innovators and agribusiness firms would improve the return on an existing public commitment without necessarily increasing its total cost. Extension and knowledge transfer –The case for investment in knowledge-transfer infrastructure is arguably stronger than the case for additional research investment at the margin, precisely because the adoption gap is the underexamined variable. Extension services operate at the link between research output and farm practice, the point where the evidence suggests the current system is weakest. International R&D spillovers –Canada benefits disproportionately from research conducted in the United States and elsewhere. The size of the US public agricultural R&D enterprise (mean $5.5B PPP, 2000–2020) dwarfs Canada’s, and much of the resulting knowledge is applicable in Canadian conditions. A strategy that emphasizes Canada’s capacity to absorb and adapt internationally generated knowledge, rather than simply increasing domestic R&D outlays, may generate higher returns per public dollar. Industry-led R&D models –Australia’s Rural Research and Development Corporation (RDC) model—in which industry levy funds are matched by government contributions and channelled through commodity-specific bodies—offers a structural comparison worth examining in the Canadian context. There are 15 RDCs covering Australia’s major agricultural sectors, collectively investing roughly AUD$825 million annually; theAustralian government matches eligible R&D levy contributions up to 0.5% of each industry’s gross value of production. The model does not simply increase total investment; it changes the governance of how investment decisions are made and how knowledge transfer is organized. Conclusion: What the Series has EstablishedAcross three posts, this series has worked through the evidence on Canadian agricultural productivity with a specific goal: to replace a crisis narrative with a more precise one. The crisis framing—in which Canada’s agricultural sector is falling behind, underinvested, and in need of urgent intervention—is not entirely wrong, but it lacks specificity about where the problems are, which levers are most tractable, and what the counterfactual is.What the data show is a more textured picture. Canada’s TFP performance is genuinely strong by international standards, and its implied return on public R&D investment, descriptive as that measure is, compares favourably with Australia and the United States on several specifications. The risk is not that Canadian agriculture is already in decline, but that the investment trajectory and structural conditions that have sustained that performance may not hold over the next decade.The most important finding of this series may be the one that is hardest to measure: the adoption gap. The chain from public R&D investment to research output to farm-level practice to productivity outcome has multiple links, and the weakest link is not always the one that gets the most policy attention. Extension capacity, precision-agriculture adoption rates, farm financial constraints, and the scale and age structure of the operator population all shape whether research investment actually reaches the farm gate.The dominant policy narrative focuses on spending levels. The more useful question and the one this series has tried to make clearer is what happens to that spending after it leaves the public accounts.A fourth and final post in this series will take up the question left open here: what do the solutions actually look like? Drawing on the concerns raised across Parts 1, 2, and 3, Part 4 will weigh the leverage points introduced in Section 5—IP modernisation, SR&ED simplification, extension investment, international R&D spillovers, and Australia’s industry-led RDC model—against each other, and assess where the evidence points toward genuine leverage and where it does not. Conclusion: What the Series has EstablishedAcross three posts, this series has worked through the evidence on Canadian agricultural productivity with a specific goal: to replace a crisis narrative with a more precise one. The crisis framing—in which Canada’s agricultural sector is falling behind, underinvested, and in need of urgent intervention—is not entirely wrong, but it lacks specificity about where the problems are, which levers are most tractable, and what the counterfactual is.What the data show is a more textured picture. Canada’s TFP performance is genuinely strong by international standards, and its implied return on public R&D investment, descriptive as that measure is, compares favourably with Australia and the United States on several specifications. The risk is not that Canadian agriculture is already in decline, but that the investment trajectory and structural conditions that have sustained that performance may not hold over the next decade.The most important finding of this series may be the one that is hardest to measure: the adoption gap. The chain from public R&D investment to research output to farm-level practice to productivity outcome has multiple links, and the weakest link is not always the one that gets the most policy attention. Extension capacity, precision-agriculture adoption rates, farm financial constraints, and the scale and age structure of the operator population all shape whether research investment actually reaches the farm gate.The dominant policy narrative focuses on spending levels. The more useful question and the one this series has tried to make clearer is what happens to that spending after it leaves the public accounts.A fourth and final post in this series will take up the question left open here: what do the solutions actually look like? Drawing on the concerns raised across Parts 1, 2, and 3, Part 4 will weigh the leverage points introduced in Section 5—IP modernisation, SR&ED simplification, extension investment, international R&D spillovers, and Australia’s industry-led RDC model—against each other, and assess where the evidence points toward genuine leverage and where it does not. Conclusion: What the Series has EstablishedAcross three posts, this series has worked through the evidence on Canadian agricultural productivity with a specific goal: to replace a crisis narrative with a more precise one. The crisis framing—in which Canada’s agricultural sector is falling behind, underinvested, and in need of urgent intervention—is not entirely wrong, but it lacks specificity about where the problems are, which levers are most tractable, and what the counterfactual is.What the data show is a more textured picture. Canada’s TFP performance is genuinely strong by international standards, and its implied return on public R&D investment, descriptive as that measure is, compares favourably with Australia and the United States on several specifications. The risk is not that Canadian agriculture is already in decline, but that the investment trajectory and structural conditions that have sustained that performance may not hold over the next decade.The most important finding of this series may be the one that is hardest to measure: the adoption gap. The chain from public R&D investment to research output to farm-level practice to productivity outcome has multiple links, and the weakest link is not always the one that gets the most policy attention. Extension capacity, precision-agriculture adoption rates, farm financial constraints, and the scale and age structure of the operator population all shape whether research investment actually reaches the farm gate.The dominant policy narrative focuses on spending levels. The more useful question and the one this series has tried to make clearer is what happens to that spending after it leaves the public accounts.A fourth and final post in this series will take up the question left open here: what do the solutions actually look like? Drawing on the concerns raised across Parts 1, 2, and 3, Part 4 will weigh the leverage points introduced in Section 5—IP modernisation, SR&ED simplification, extension investment, international R&D spillovers, and Australia’s industry-led RDC model—against each other, and assess where the evidence points toward genuine leverage and where it does not. Conclusion: What the Series has Established Across three posts, this series has worked through the evidence on Canadian agricultural productivity with a specific goal: to replace a crisis narrative with a more precise one. The crisis framing—in which Canada’s agricultural sector is falling behind, underinvested, and in need of urgent intervention—is not entirely wrong, but it lacks specificity about where the problems are, which levers are most tractable, and what the counterfactual is.What the data show is a more textured picture. Canada’s TFP performance is genuinely strong by international standards, and its implied return on public R&D investment, descriptive as that measure is, compares favourably with Australia and the United States on several specifications. The risk is not that Canadian agriculture is already in decline, but that the investment trajectory and structural conditions that have sustained that performance may not hold over the next decade.The most important finding of this series may be the one that is hardest to measure: the adoption gap. The chain from public R&D investment to research output to farm-level practice to productivity outcome has multiple links, and the weakest link is not always the one that gets the most policy attention. Extension capacity, precision-agriculture adoption rates, farm financial constraints, and the scale and age structure of the operator population all shape whether research investment actually reaches the farm gate.The dominant policy narrative focuses on spending levels. The more useful question and the one this series has tried to make clearer is what happens to that spending after it leaves the public accounts.A fourth and final post in this series will take up the question left open here: what do the solutions actually look like? Drawing on the concerns raised across Parts 1, 2, and 3, Part 4 will weigh the leverage points introduced in Section 5—IP modernisation, SR&ED simplification, extension investment, international R&D spillovers, and Australia’s industry-led RDC model—against each other, and assess where the evidence points toward genuine leverage and where it does not. Across three posts, this series has worked through the evidence on Canadian agricultural productivity with a specific goal: to replace a crisis narrative with a more precise one. The crisis framing—in which Canada’s agricultural sector is falling behind, underinvested, and in need of urgent intervention—is not entirely wrong, but it lacks specificity about where the problems are, which levers are most tractable, and what the counterfactual is. What the data show is a more textured picture. Canada’s TFP performance is genuinely strong by international standards, and its implied return on public R&D investment, descriptive as that measure is, compares favourably with Australia and the United States on several specifications. The risk is not that Canadian agriculture is already in decline, but that the investment trajectory and structural conditions that have sustained that performance may not hold over the next decade. The most important finding of this series may be the one that is hardest to measure: the adoption gap. The chain from public R&D investment to research output to farm-level practice to productivity outcome has multiple links, and the weakest link is not always the one that gets the most policy attention. Extension capacity, precision-agriculture adoption rates, farm financial constraints, and the scale and age structure of the operator population all shape whether research investment actually reaches the farm gate. The dominant policy narrative focuses on spending levels. The more useful question and the one this series has tried to make clearer is what happens to that spending after it leaves the public accounts. A fourth and final post in this series will take up the question left open here: what do the solutions actually look like? Drawing on the concerns raised across Parts 1, 2, and 3, Part 4 will weigh the leverage points introduced in Section 5—IP modernisation, SR&ED simplification, extension investment, international R&D spillovers, and Australia’s industry-led RDC model—against each other, and assess where the evidence points toward genuine leverage and where it does not.