How Much Does Frontier Growth Depend on Catch-Up Growth?
Developing countries grow faster by absorbing technology from rich countries. Everyone knows this. The reverse question gets almost no attention: how much does rich-country growth depend on poor-country growth?
Call it spillback growth — the share of advanced economy GDP growth driven by catch-up dynamics elsewhere. The theoretical case writes itself. Billions of people getting richer expand the global market for frontier innovations. They buy more exports. They deepen financial integration. In endogenous growth models, a larger market raises returns to R&D, which should accelerate frontier innovation.
The empirical case tells a different story.
What the literature finds
Two recent papers give us the best estimates. The IMF's April 2024 World Economic Outlook (Chapter 4, "Trading Places") finds that growth spillovers from G20 emerging markets to advanced economies have roughly tripled since 2000. But they still explain only about 5% of advanced economy GDP variation. Almost all of that traces back to a single country: China.
A World Bank working paper reaches an even weaker conclusion. In the authors' preferred specification, emerging-to-advanced spillovers come out statistically insignificant. The Brookings Institution documents a similar pattern — the "asymmetry of global spillovers" still runs overwhelmingly from rich to poor, not the other way around.
So the honest headline: frontier growth depends surprisingly little on catch-up growth through measurable channels. What dependence exists runs almost entirely through China.
Will this change? Some BOTECs
China's outsized role reflects a transient feature of the current moment, not a deep structural fact. China happened to industrialise first at sufficient scale and trade intensity to register in the data. The question: what happens as other large economies catch up?
India reaches China's current GDP (~$30-44T) by the mid-2040s on most projections. But India's growth model leans services-heavy and less trade-intensive. India accounts for roughly 2.5% of global goods trade versus China's 15%. A reasonable estimate: India generates one-third to one-half of China's spillback intensity per dollar of GDP.
The non-China emerging market bloc collectively grows from ~$50T to $130T+ over the next two decades. Even at low per-dollar spillback intensity, the aggregate starts to matter through sheer scale.
Putting these together, total spillback effects could roughly double by 2045. That takes us from ~5% to ~10% of advanced economy GDP variation. Not nothing, but still not the dominant story.
The wild card: market size and innovation
These estimates all capture trade and financial channels — what VAR models can measure. The potentially larger effect operates on longer timescales. A bigger global market raises returns to frontier R&D. The global research workforce may double. Global R&D spending could grow from ~$1T to $2T+. The induced innovation effects on frontier economies could prove substantial.
But this channel operates too slowly for business-cycle econometrics to detect. It may represent the most important mechanism the current literature misses entirely.
Why this matters
Frontier growth remains largely self-sustaining through direct, measurable channels. The feedback loop from catch-up growth appears weaker than intuition suggests.
But the policy implications cut the other direction. Even if spillback effects explain only 5-10% of advanced economy growth variation, the benefit-cost ratio of increasing them could run high. The reason: catch-up growth has far more variance than frontier growth. Rich countries grow at 1.5-2.5% per year with stubborn regularity. Developing country growth rates range from -2% to 8%+. A development economist with a given budget can plausibly shift more global GDP growth than a domestic macroeconomist with the same resources. Through spillback channels, some of that growth flows back to the frontier.
In a world where the non-China emerging bloc heads toward $130T+ and the global research force may double, rich-country governments that neglect development policy leave their own long-run growth on the table. The case for development converges with the case for national interest — and that convergence may prove more politically durable than altruism alone.
Spillback growth: How much does frontier growth depend on catch-up growth?
What the latest research says
The most comprehensive recent treatment is Chapter 4 of the IMF's April 2024 World Economic Outlook, 'Trading Places.'[1] Using a Bayesian VAR framework, the IMF finds that G20 emerging market shocks now explain approximately 5% of GDP variation in advanced economies — up from near-zero in the early 2000s. Spillovers from G20 emerging markets have roughly tripled over this period. A plausible growth acceleration in G20 emerging markets (excluding China) could boost world growth by about 0.5 percentage points.
But here's the punchline: it's almost entirely China.
A January 2025 IMF working paper by Abdel-Latif and Popescu[2] uses a global VAR covering 63 countries to decompose these effects. A 1 percentage point Chinese demand shock raises advanced economy GDP by about 0.24% after one year and 0.46% after three years. For other G20 emerging markets — India, Brazil, Indonesia, Mexico, Turkey — the effect on advanced economies is below 0.1%. Essentially zero.
This is consistent with Huidrom, Kose, and Ohnsorge's World Bank paper (WPS 8093, later published in International Finance in 2020), which found that spillovers from the EM7 to G7 economies were statistically insignificant in their preferred specification.
So today's answer to 'how much is frontier growth spillback growth?' is: not much — maybe 5% of variation, driven overwhelmingly by a single country.
But how much might this change by 2045?
The current estimates reflect a world where China is the only emerging economy large and trade-integrated enough to move advanced economy GDP. That world is changing fast. Here are some rough back-of-the-envelope estimates for how spillback effects might evolve.
India as the next China-scale spillback source. China began generating measurable spillovers to advanced economies when its GDP (PPP) reached roughly $15-20 trillion and its share of global trade hit ~12-15%. Standard GDP projections have India reaching $30-44 trillion by 2050, with GDP per capita around $26k — roughly where China was in the early 2020s. If India follows a similar path of trade integration, it could become a second major source of spillback growth by the 2040s.
But there's a crucial caveat: China's spillovers operate heavily through commodity demand and manufacturing supply chains. India's growth model is more services-oriented and less trade-intensive (exports are ~20% of GDP vs. China's ~20%, but India's share of global goods trade is only ~2.5% compared to China's ~15%). A reasonable estimate might be that India at China's current GDP generates perhaps one-third to one-half of China's current spillback effect — call it 0.1-0.2% on advanced economy GDP per 1 percentage point demand shock, rather than China's 0.4-0.5%.
The Indonesia and broader EM question. Indonesia's GDP is projected to reach $11 trillion by 2050, making it the world's fourth largest economy. Brazil reaches $8 trillion, Mexico $7 trillion. Individually, these remain too small to generate statistically significant spillbacks (the current literature finds essentially zero for each). But collectively, the non-China EM bloc grows from roughly $50 trillion to $130+ trillion. Even if the per-dollar spillback intensity is low, the sheer scale shift matters.
Crude BOTEC: total spillback share of AE GDP variation by 2045. Spillovers have tripled since 2000 as the EM share of world GDP (PPP) grew from ~35% to ~50%. By 2045, EMs will likely represent 60%+ of world GDP. If we assume spillback intensity scales sub-linearly with EM GDP share (since diminishing returns to trade integration partially offset the scale gains), a reasonable projection might be:
Period | EM share of world GDP | EM spillbacks as % of AE GDP variation |
Early 2000s | ~35% | ~1-2% |
Today (2024) | ~50% | ~5% |
~2035 | ~55% | ~7-8% (if India integrates) |
~2045 | ~60%+ | ~8-12% (with India + Indonesia) |
This implies spillback effects roughly doubling over the next two decades — a meaningful shift, but still leaving frontier growth overwhelmingly self-generated.
The market size channel remains the wild card. The estimates above are all based on trade and financial linkage channels, which is what the VAR literature captures. The potentially larger effect — that a bigger global market increases returns to frontier R&D — is theoretically compelling but empirically unquantified. If the global research force doubles (as my GDP projections suggest) and global R&D spending doubles from ~$1 trillion to ~$2+ trillion, the induced innovation effects on frontier economies could be substantial. But this operates on longer timescales than business-cycle VAR models can detect, and may be the most important channel that the current literature misses entirely.
Why this matters
The honest summary is: frontier growth currently depends surprisingly little on catch-up growth through measurable channels — perhaps 5% of variation, almost all via China. This could roughly double by 2045 as India and other large EMs grow, but frontier growth will likely remain largely self-sustaining through direct channels.
The deeper question is whether the indirect channels — market size effects on innovation incentives, the doubling of the global research workforce, the expansion of the idea pool — are larger than what shows up in VARs. If they are, then catch-up growth may be far more important for the frontier than the current empirical literature suggests. We just can't measure it well yet.
For anyone thinking about long-run growth trajectories — including in the context of AI scaling, where the size of the global compute and researcher base matters enormously — this gap between the measurable and the theoretical is where the most important action likely is.
This has a policy implication worth stating plainly: as spillback effects grow, rich countries have an increasingly selfish reason to care about development. The traditional case for development aid and trade integration rests on moral arguments — helping the global poor because it's the right thing to do. But if frontier growth is partly downstream of catch-up growth, then stagnation in India or sub-Saharan Africa isn't just their problem — it's a drag on innovation incentives, export demand, and R&D workforce growth in the US and Europe too. We are increasingly in the same boat. The IMF's finding that EM spillovers have tripled since 2000 and are now 'comparable to those from advanced economies' suggests this interdependence will only deepen. In a world where the non-China EM bloc is projected to grow from $50 trillion to $130+ trillion, where the global research force may double, and where AI development costs are pegged to GDP, rich-country governments that neglect development policy are effectively leaving their own long-run growth on the table. The case for development is converging with the case for national interest — and that convergence may prove more politically durable than altruism alone.
Finally, even if the overall magnitude of spillback growth remains modest, the benefit-cost ratio of policy efforts to increase it could be remarkably high. The reason is simple: there's far more variance in catch-up growth than in frontier growth. Rich countries grow at 1.5-2.5% per year with stubborn regularity — the policy levers available to shift that are well-explored and yield marginal gains. But developing country growth rates range from-2% to 8%+, and the determinants are far more sensitive to policy intervention: trade access, institutional reform, infrastructure investment, technology transfer. A development economist armed with a given budget can plausibly shift more GDP growth globally than a domestic macroeconomist with the same resources — and through spillback effects, some of that growth flows back to the frontier. The implication is that even if spillbacks only account for 5-10% of advanced economy GDP variation, the elasticity of that channel to policy effort may be substantially higher than the elasticity of direct domestic growth policy.