I have been using the OECD Composite Leading Indicators as one of my key tools for anticipating turning points in Asian stock markets—particularly in South Korea, whose export-driven economy often responds early to changes in the global manufacturing cycle. Of course this gauges the global economy's direction itself.
The June 2026 data, however, revealed an unusual divergence. The Major G20 indicator edged lower, and all four major European economies lost momentum. Yet South Korea and Japan continued to strengthen, while exports and industrial production remained exceptionally strong in Taiwan and Vietnam.
That raises a more important question: Is Asia beginning to decouple from the traditional global business cycle—or is it simply lagging the downturn?
The evidence suggests that the divergence is real. But it is not yet the kind of broad, self-sustaining regional decoupling that would make Asia independent of the global economy.
Instead, parts of Asia are becoming increasingly connected to a different global cycle—one driven by AI infrastructure investment, semiconductor capacity expansion and supply-chain relocation.
The G20 Has Paused, Not Yet Rolled Over
The OECD Composite Leading Indicator is designed to identify turning points in economic activity relative to its long-term trend.
It is not a forecast of GDP growth, and a reading below 100 does not necessarily indicate a recession. The index is amplitude-adjusted and centered on a long-term average of 100.
The direction of the indicator therefore matters as much as its level.
A CLI above 100 and rising indicates above-trend activity with strengthening momentum. A reading above 100 but falling suggests that activity remains above trend while momentum is weakening. A CLI below 100 but rising can indicate an early recovery.
The OECD describes the CLI as a qualitative signal of the growth cycle rather than a quantitative forecast of economic growth. The data are also revised as component series and trend estimates change.
The June data do not yet confirm a synchronized global rollover.
Changes are measured in index points. OECD data accessed in August 2026.
The Major G20 CLI declined from 100.39 in March and April to 100.38 in May and 100.36 in June. But it remained above its December 2025 level of 100.27 and above its long-term trend.
This is better described as an early loss of momentum than a confirmed downturn.
The G7 is even less consistent with a global rollover. Its CLI remained above 100 and continued to rise marginally through June. The United States also remained firmly in an above-trend, rising phase.
The more important signal is therefore not a synchronized global decline. It is the widening difference between regions and industries inside the global aggregate.
Europe Is Losing Momentum—but Not Collapsing
All four major European economies covered by the OECD moved lower in June.
The decline was most pronounced in Italy and the United Kingdom. France weakened more gradually, while Germany’s CLI also turned lower after reaching a local high early in 2026.
This synchronized directional change is meaningful. But the hard economic data do not yet support a conclusion that Europe has entered a broad and persistent industrial downturn.
Euro-area industrial production was unchanged in June and only 0.1 percent higher than a year earlier. More cyclically sensitive categories were weaker: capital-goods production fell 1.4 percent from May, while intermediate-goods production declined 0.8 percent. Eurostat’s June industrial-production release therefore shows stagnation beneath a stable headline.
Country-level data were mixed.
French manufacturing output fell 1.1 percent in June after declining 1.0 percent in May, with weakness across every major manufacturing branch. INSEE reported especially sharp declines in transport equipment and continued weakness in electrical, electronic and computer equipment.
Italian industrial production fell 1.0 percent from May and 0.6 percent from a year earlier, according to ISTAT.
UK manufacturing output declined 0.5 percent in June following a 0.2 percent fall in May. However, manufacturing still grew 1.0 percent during the second quarter, and computer, electronic and optical production increased 3.0 percent over the quarter. The UK Office for National Statistics therefore shows slowing monthly momentum rather than an unambiguous industrial contraction.
Germany was more resilient. Industrial production increased 0.2 percent in June and 0.7 percent over the second quarter. Exports rose 0.9 percent from May and reached a record seasonally adjusted monthly value.
But manufacturing excluding energy and construction was flat, machinery output fell 3.9 percent and overall industrial production remained 0.1 percent below its year-earlier level. Destatis data point to stabilization rather than a strong recovery.
More recent business surveys suggest that the European slowdown may not be deepening. The euro-area manufacturing PMI rose to 52.8 in the August flash estimate, its highest level in more than four years, while German manufacturing also improved.
That does not invalidate the June CLI signal. It shows why the signal should be interpreted as a warning about momentum rather than proof of a structural decline.
Europe’s underlying problem is a lack of breadth. France and Italy are weak, the UK is uneven, and Germany’s recovery remains shallow. Europe has not displayed an AI-related export and investment engine comparable in scale to the one emerging in the United States and parts of Asia.
Asia Is Not One Cycle
For this analysis, “Asia” refers primarily to four export- and manufacturing-oriented economies: South Korea, Japan, Taiwan and Vietnam.
They are not interchangeable.
South Korea and Taiwan sit near the center of the AI semiconductor, memory and server supply chain. Japan is more exposed to semiconductor equipment, automation, advanced materials and corporate capital expenditure. Vietnam is an increasingly important manufacturing platform shaped by foreign direct investment and supply-chain relocation.
Their common strength therefore cannot be attributed to a single national or industrial model.
South Korea: AI-Led, but No Longer Semiconductor-Only
South Korea has the strongest OECD CLI momentum among the economies examined.
Its CLI rose from 100.81 in December 2025 to 102.87 in June 2026. The 2.06-point six-month increase was much larger than the corresponding gains in the United States, Japan or the G7.
The export data confirm that this was not merely a statistical signal.
Korean exports rose 70.9 percent from a year earlier to a record $102.25 billion in June. Semiconductor exports nearly tripled to $44.82 billion, while computer exports rose 308.8 percent as demand for enterprise SSDs and AI servers increased.
The concentration is substantial.
Semiconductors and SSDs accounted for 83.7 percent of Korea’s ICT exports during the first half, while ICT products accounted for more than half of the country’s total exports. Korea’s Ministry of Trade, Industry and Resources explicitly attributed the increase to expanding AI-server investment, stronger memory demand and higher contract prices.
Yet the recovery was broader than semiconductors alone.
Non-semiconductor exports increased 28 percent in June, and 18 of Korea’s 20 major export categories posted year-on-year growth. Automobiles, ships, steel, general machinery, biohealth products and cosmetics all increased.
The ministry’s complete June trade release shows that the export boom was beginning to spread beyond memory chips.
Domestic production also strengthened. Total industrial production rose 2.3 percent from May, mining and manufacturing output increased 6.4 percent, semiconductor production rose 4.5 percent and facility investment increased 5.8 percent.
The strength continued in July. Exports reached $98.89 billion, the second-highest monthly total on record, while semiconductor exports rose 178.8 percent to $41.01 billion. Non-semiconductor exports increased 26 percent, and Korea’s manufacturing PMI rose from 52.1 in June to 53.1 in July.
The evidence therefore does not support the claim that AI is merely concealing a contracting Korean industrial economy.
But it does show that AI remains the dominant marginal driver. If memory prices or hyperscaler investment weaken, Korea’s exceptional momentum could narrow quickly even if the rest of the export economy remains stable.
Japan: Equipment, Automation and Corporate CAPEX
Japan’s strength is less dramatic than Korea’s or Taiwan’s, but its underlying structure is different.
The OECD CLI continued to rise through June. Final data showed that industrial production increased 1.9 percent from May and 4.9 percent from a year earlier.
Exports rose 19.3 percent year-on-year to ¥10.93 trillion in June, extending their expansion to a tenth consecutive month. In July, exports increased another 23.2 percent to a record ¥11.51 trillion.
Japan’s Ministry of Finance trade releases show continued strength across electrical machinery, semiconductor-related products, transport equipment and industrial machinery.
Corporate investment provides another important signal.
The Bank of Japan’s June Tankan showed large companies planning an 11.5 percent increase in capital expenditure for the fiscal year, up sharply from the 3.3 percent plan recorded in March. Large-manufacturer confidence also improved. The complete survey is available from the Bank of Japan.
Japan’s semiconductor-equipment industry is participating directly in the AI capacity cycle. The Semiconductor Equipment Association of Japan reported record-level billings as investment accelerated in advanced logic, DRAM, HBM and semiconductor packaging. SEAJ’s statistical releases show how AI-related capacity investment is moving upstream into Japanese equipment suppliers.
The strength is not uniform, however.
Core private machinery orders excluding ships and electric utilities fell 12.4 percent in May. That decline warns against describing Japan as a broad investment boom based solely on semiconductor equipment and large-company spending. The Cabinet Office release shows substantial volatility beneath the stronger trend.
Japan is therefore not fully decoupling either. But AI investment, automation demand and rising corporate CAPEX are giving it a stronger buffer than in previous manufacturing slowdowns.
Taiwan: The Cleanest AI Benchmark
Taiwan provides the clearest benchmark for the AI infrastructure cycle.
Exports reached $74.83 billion in June, 40.3 percent higher than a year earlier and the third-highest monthly total on record, according to Taiwan’s Ministry of Finance.
The industrial data were even stronger.
Taiwan’s manufacturing production index increased 24.34 percent from a year earlier in June, while export orders rose 59.4 percent to a record $95.26 billion. Taiwan’s Ministry of Economic Affairs reported particularly strong orders for information and communications products, electronic products, high-performance computing equipment, cloud infrastructure and servers.
The composition of exports shows why Taiwan is an important control case.
Information, communications and audiovisual exports rose 72.3 percent from a year earlier. Electronic-component exports increased 32.8 percent. Together, the two categories accounted for most of the increase.
Traditional industries were not uniformly weak. Their exports rose 6.9 percent in June, while machinery and electrical-equipment exports benefited from semiconductor fabrication, data-center and grid investment.
Taiwan’s momentum also persisted beyond the June CLI reference month. July exports increased another 32.9 percent from a year earlier to $75.30 billion, marking the 33rd consecutive month of annual export growth. Taiwan’s Ministry of Finance reported a monthly trade surplus of $17.17 billion.
Taiwan therefore supports the AI-decoupling hypothesis more strongly than any other economy in the group.
But it also demonstrates the concentration risk. The country’s exceptional growth remains heavily dependent on a narrow set of global technology-investment decisions.
Vietnam Is Driven by a Different Force
Vietnam matters because it helps separate AI investment from supply-chain restructuring.
Vietnam’s first-half GDP increased 8.18 percent from a year earlier. Manufacturing expanded 10.23 percent, while June industrial production rose 12.7 percent year-on-year and 3.5 percent from May.
Industrial production accelerated further in July, rising 14.5 percent from a year earlier as new production capacity came online and companies expanded operations, according to the National Statistics Office of Vietnam.
Exports increased 21 percent during the first half to $266.52 billion. Electronics, computers and components reached $71.16 billion, while machinery and equipment exports totaled $33.24 billion.
But Vietnam’s growth cannot be explained in the same way as Korea’s or Taiwan’s.
Foreign-invested companies generated almost 80 percent of Vietnam’s exports. Registered foreign direct investment rose 61 percent to $34.65 billion in the first half, while disbursed FDI increased 11.2 percent to $13.03 billion. Manufacturing remained the largest recipient.
South Korea was the second-largest source of newly registered FDI, while Japan was also among the major investors. At the same time, the United States remained Vietnam’s largest export market.
The National Statistics Office’s first-half report therefore describes an economy benefiting from both electronics demand and the continuing relocation of manufacturing capacity.
Vietnam is participating in the technology cycle, but its more important structural force is supply-chain reorganization.
Its strength indicates that Asia’s resilience is not an AI story alone.
What History Says About the Lag
The conventional assumption is that export-oriented Asian economies follow the global cycle with a delay.
The historical OECD data only partly support that view.
Using monthly G20, Korean and Japanese CLIs from 2005 through 2023, this analysis compared one-, three- and six-month changes at leads and lags of up to 12 months.
Positive lags indicate that the G20 moved first. Negative lags indicate that the Asian economy moved first.
For Korea, the strongest full-sample relationship appeared when Korea led the G20 by approximately one month using three-month changes and by two to three months using six-month changes.
After excluding the acute COVID disruption, Korea’s lead widened to approximately three months, with correlations above 0.70.
Japan showed the opposite pattern. The G20 generally led Japan by one to two months.
This difference is economically plausible.
Korean export orders, semiconductor inventories and manufacturing expectations often react early to changes in the global goods cycle. Korea can therefore behave as an early indicator of global manufacturing demand rather than a passive follower.
Japan’s broader industrial structure, domestic demand and corporate investment cycle have historically responded with a clearer delay.
The relationship was not stable, however.
During 2021–2023, the G20 led Korea by approximately one to two months. The post-pandemic inflation, inventory and monetary-tightening cycle therefore looked more like the conventional lag relationship.
This regime instability prevents a simple conclusion.
If Korea’s current strength is once again leading a broader global recovery, the G20 may eventually turn higher. If the post-2021 relationship persists, Korea may weaken after the G20. If Korea, Japan, Taiwan and Vietnam remain strong while the G20 and Europe continue to deteriorate, the evidence for a structural break will become much stronger.
China and the G20 Benchmark
The official OECD Major G20 aggregate remains the principal global benchmark in this analysis.
It contains 16 economies, including China, India, Indonesia, South Korea and Japan, as well as the major North American and European economies. China therefore remains inside the official G20 series.
Removing China would create a different, non-official index and require reconstructing the OECD’s chain-linked purchasing-power-parity weighting methodology.
China is not included in the country-level real-economy comparison.
The purpose of this article is to cross-check leading-indicator signals using independently verifiable production, trade, investment and business-survey data. Concerns about the transparency and independent verification of some Chinese economic statistics make an identical country-level comparison difficult.
This does not mean that every Chinese statistic is false. It means that China requires a separate analytical framework using external trade partners, commodity flows, shipping, corporate disclosures and other third-country evidence.
China’s OECD CLI, which was below 100 and falling in June, still affects the official G20 aggregate. This composition effect is one reason why a small decline in the G20 should not automatically be interpreted as a synchronized global downturn.
Two Structural Forces Are Reshaping the Cycle
The evidence points to two different structural forces operating across Asia.
The first is the AI infrastructure cycle:
United States AI investment → advanced processors → HBM and memory → semiconductor equipment and materials → servers and networking → data centers → power infrastructure.
South Korea, Taiwan and Japan are positioned at different points along this chain.
Korea supplies advanced memory, HBM and enterprise SSDs. Taiwan supplies leading-edge logic chips, servers and computing hardware. Japan supplies semiconductor equipment, materials, automation and precision manufacturing technologies.
The second force is global supply-chain realignment:
Trade fragmentation → manufacturing diversification → foreign direct investment → new Asian production capacity → increased intermediate-goods and finished-goods exports.
Vietnam is the clearest case, but Korea and Japan also participate through overseas investment, machinery exports and the construction of new regional supply networks.
These two forces should not be confused.
AI infrastructure is a technology-led capital-expenditure cycle. Supply-chain realignment is a geopolitical investment cycle.
They are occurring simultaneously, and together they are providing parts of Asia with an economic buffer that did not exist in the same form during earlier global slowdowns.
Decoupling—or a New Form of Dependence?
The evidence does not support complete Asian decoupling.
South Korea and Taiwan remain dependent on investment decisions made by U.S. hyperscalers, semiconductor designers and data-center operators. Japan’s equipment cycle depends on semiconductor capacity expansion in Korea, Taiwan, the United States and elsewhere. Vietnam’s manufacturing expansion depends on foreign capital and access to U.S., European and Asian export markets.
These economies are not becoming independent of the global economy.
They are becoming less dependent on one particular version of the global cycle.
The traditional cycle moved through broad consumer demand, European and U.S. manufacturing, inventory accumulation and conventional capital expenditure.
The new cycle is narrower but much more capital-intensive. It is centered on AI computing, semiconductors, data centers, electrical infrastructure and the physical reorganization of global production.
This is not full decoupling.
It is a selective re-coupling of Asian manufacturing to AI infrastructure and geopolitical supply-chain investment.
The Current Verdict: Partial Decoupling
Of the three initial hypotheses, the data currently support partial decoupling.
Asia is not experiencing a conventional region-wide boom, and the official G20 has not yet entered a confirmed downturn. But Korea, Japan, Taiwan and Vietnam are displaying stronger manufacturing, export and investment momentum than the European economies whose leading indicators weakened in June.
The divergence is also broader than semiconductors alone.
Korea’s non-semiconductor exports are rising. Taiwan’s traditional industries are beginning to recover. Japan is benefiting from corporate investment and production equipment. Vietnam is attracting manufacturing FDI while expanding electronics, machinery and industrial production.
At the same time, the strongest growth remains concentrated around AI infrastructure, advanced electronics and relocated manufacturing capacity. Domestic consumption and traditional industries are less uniformly strong, while export orders and machinery data continue to show volatility.
The most accurate conclusion is therefore not that Asia has broken free from the global business cycle.
It is that a new investment cycle has become powerful enough to offset part of the old one.
Whether that offset becomes a durable structural shift will depend on three tests.
First, AI capital expenditure must remain strong enough to sustain semiconductor demand after the current capacity expansion.
Second, the recovery must continue spreading from chips and servers into machinery, electrical infrastructure, transportation, consumer industries and domestic investment.
Third, Asian exports and production must remain resilient for longer than the historical lag that normally follows a global turning point.
Until those conditions are met, the distinction remains open.
Asia may be decoupling.
But it may also be leading—and becoming dependent on—a different global cycle that traditional indicators are only beginning to capture.
Isn't it worthwhile to observe?
Related Analysis
Productivity, Employment and the Possibility of a Higher U.S. Speed Limit
How AI investment, productivity and employment may be reshaping the U.S. economy’s non-inflationary growth potential.
Explore More Macro & Data Analysis →