The clearest change this week is that the rotation away from traditional mutual-fund wrappers strengthened while aggregate long-term fund flows turned positive. For the week ended 23 September, U.S. long-term mutual funds lost an estimated $19.67 billion, but ETF net issuance reached $39.81 billion. The combined ICI series therefore recorded $20.14 billion of net inflows, reversing the previous week's revised $9.45 billion outflow. Equity products moved from a $12.65 billion combined outflow to a $12.88 billion inflow, while bond products attracted $8.16 billion.[1][2][3]
That is not evidence that the large cash pool has disappeared. Money-market fund assets fell $45.45 billion in the week ended 30 September, to $7.89 trillion, but August's monthly survey still showed $7.928 trillion in money-market assets, 9.9% above August 2025. The latest weekly decline is therefore a flow out of a very large liquidity stock rather than evidence that cash-like holdings have normalized.[4][5]
A second change is visible in official reserves. IMF COFER data released on 30 September show total foreign-exchange reserves rising from $13.10 trillion in the first quarter to $13.22 trillion in the second. The dollar share fell from 57.18% to 56.70%, while the euro share rose from 20.04% to 20.60%. The IMF says exchange-rate effects were limited and that most of the euro increase reflected active purchasing. This is a portfolio reallocation inside official reserve assets, not a transfer of ownership of productive capacity.[6]
ETF issuance more than offset mutual-fund redemptions
The weekly U.S. fund data now show a sharper separation between wrapper choice and aggregate allocation.
| Channel | Week ended 16 Sep | Week ended 23 Sep | What changed |
|---|---|---|---|
| Long-term mutual funds | -$36.72bn | -$19.67bn | Redemptions continued but slowed |
| ETF net issuance | +$27.27bn | +$39.81bn | Creation accelerated |
| Combined long-term funds + ETFs | -$9.45bn | +$20.14bn | Aggregate flow turned positive |
| Combined equity | -$12.65bn | +$12.88bn | Equity flow reversed |
| Combined bonds | +$3.24bn | +$8.16bn | Bond inflow strengthened |
View data
| Indicator / period | Value (US$ billion) |
|---|---|
| Long-term mutual funds | -19.67 |
| ETF net issuance | 39.81 |
| Combined long-term funds + ETFs | 20.14 |
The composition matters because ETF issuance is not the same transaction as mutual-fund cash flow. The ICI combined series is useful precisely because it prevents redemptions from one wrapper from being mistaken for a system-wide withdrawal. August's monthly data reinforce the structural direction: ETF net issuance was $169.3 billion during the month and $1.349 trillion in January-August, compared with $766.9 billion in the same period of 2025.[7]
Cash fell for one week but remains structurally large
ICI reported $7.891 trillion of money-market fund assets on 30 September, down from $7.937 trillion a week earlier. Government funds accounted for $31.57 billion of the decline and prime funds for $15.76 billion, partly offset by a $1.88 billion increase in tax-exempt funds.[4]
View data
| Indicator / period | Value (US$ trillion) |
|---|---|
| 2026-09-16 | 7.921 |
| 2026-09-23 | 7.937 |
| 2026-09-30 | 7.891 |
The monthly comparison is more informative about the stock. Money-market assets were $7.928 trillion at the end of August, up from $7.217 trillion a year earlier. August itself recorded $55.48 billion of net new cash flow after a $57.36 billion outflow in July.[5] The current evidence therefore supports a narrower interpretation than a broad move out of cash: long-term fund demand improved, particularly through ETFs, while liquidity holdings remain historically large.
Mutual-fund redemptions continue -> ETF issuance accelerates -> combined long-term flow turns positive -> money-market assets fall modestly but remain near $7.9tn -> allocation broadens without eliminating the large liquidity buffer
Reserve managers increased euro exposure while the dollar stock was broadly unchanged
The new COFER release is the week's most important official-sector allocation signal. Total foreign-exchange reserves rose by about $120 billion from the first to the second quarter. Dollar claims were broadly unchanged, but their share declined by 0.48 percentage point to 56.70%. Euro claims increased by nearly $100 billion and the euro share rose by 0.56 percentage point to 20.60%. The renminbi share increased from 1.98% to 2.11%, while the yen share fell from 5.43% to 4.96%.[6]
View data
| Indicator / period | Value (%) |
|---|---|
| U.S. dollar | 56.7 |
| Euro | 20.6 |
| Japanese yen | 4.96 |
| Chinese renminbi | 2.11 |
The IMF's decomposition is important. Exchange-rate movements stayed within roughly plus or minus 2% during the quarter, limiting FX translation effects. The Fund attributes most of the euro increase to active purchasing, while the yen decline reflects active selling together with bond-price and currency effects. Reserve-share changes therefore contain both transactions and valuation, but the Q2 euro move has more evidence of active reallocation than the previous quarter's largely valuation-driven shifts.[6]
This does not establish a rapid abandonment of the dollar. Dollar claims were largely unchanged and the currency still represented more than half of reported foreign-exchange reserves. The evidence instead points to incremental diversification at the margin.
Gold still confirms diversification, but September fund data are not yet complete
The latest complete global gold-ETF month remains August. Global gold-backed ETFs added $18 billion, holdings increased by 121 tonnes to a record 4,189 tonnes, and assets under management rose 16% to $615 billion. The World Gold Council separates net fund flows from the price-driven change in assets, so the $18 billion inflow and 121-tonne increase are the cleaner evidence of investor demand.[8]
China provides an official-sector example: the People's Bank of China reported a 20.2-tonne addition in August, taking official gold holdings to 2,387 tonnes after 22 consecutive monthly increases.[9] This is consistent with reserve diversification, but it should not be mechanically combined with COFER because monetary gold is outside the COFER foreign-currency-reserve measure.
Institutional research broadly recognizes the same diversification mechanism, but the evidence classes differ. Goldman Sachs Research describes elevated central-bank gold accumulation as a multi-year trend; this is a research thesis, not a disclosed portfolio position. Vanguard's September fixed-income view favors high-quality credit and neutral U.S. duration while noting competition for capital from sovereign borrowing and AI investment; that is an explicit portfolio view, not evidence that all Vanguard-managed assets share the position.[10][11] The primary flow and reserve data remain the basis for the assessment.
ETF issuance offsets mutual-fund redemptions while cash-like assets remain near $7.9tn
COFER shows nearly $100bn increase in euro claims in Q2, mostly attributed to active purchasing
PBoC reported a 20.2t August addition; gold is outside COFER and must be tracked separately
Dollar reserve claims remain broadly unchanged while other reserve assets gain at the margin
Productive investment remains more selective than liquid portfolio flows
New UNCTAD detail on Latin America illustrates why financial flows and productive capacity should remain separate. FDI inflows to Latin America and the Caribbean excluding offshore financial centres rose 14% to $188 billion in 2025, but the value of announced new investment projects projects fell by roughly one-third. Brazil's inflows increased from $63 billion to $77 billion.[12]
The divergence means a higher recorded FDI flow does not automatically imply a stronger pipeline of new factories, infrastructure or jobs. Existing-company transactions, reinvested earnings and other FDI components can rise while prospective new investment projects commitments weaken. The same distinction applies to this week's U.S. fund data: stronger ETF issuance improves evidence of liquid-market demand, but says little by itself about realized investment in grids, data centres, mining, processing or industrial capacity.
What would change the assessment
The wrapper-rotation thesis would weaken if mutual-fund and ETF flows began moving together for several weeks. A sustained fall in money-market assets combined with persistent positive long-term flows would provide stronger evidence of a broader reallocation out of cash-like instruments. If money-market assets rebound while combined long-term flows turn negative again, this week's improvement would look more tactical.
The reserve-diversification interpretation should be tested against subsequent COFER quarters. A persistent rise in euro and other-currency claims driven by active purchases rather than exchange-rate or bond-price effects would strengthen the case. Stable or rising dollar claims alongside those purchases would continue to describe diversification at the margin rather than a wholesale shift away from the dollar.
For productive investment, the next question is whether announced new investment projects activity stabilizes and whether strategic-sector financing converts into operating capacity. Portfolio flows, reserve allocation and productive investment are related through financing conditions, but they are not additive measures of global wealth movement.
Sources
- Investment Company Institute, Combined Estimated Long-Term Flows and ETF Net Issuance, 30 September 2026: https://www.ici.org/research/stats/combined_flows
- Investment Company Institute, Estimated Long-Term Mutual Fund Flows, 30 September 2026: https://www.ici.org/research/stats/flows
- Investment Company Institute, Estimated ETF Net Issuance, 29 September 2026: https://www.ici.org/research/stats/etf_flows
- Investment Company Institute, Money Market Fund Assets, 1 October 2026: https://www.ici.org/research/stats/mmf
- Investment Company Institute, Trends in Mutual Fund Investing: August 2026, 29 September 2026: https://www.ici.org/research/stats/trends_08_26
- International Monetary Fund, Currency Composition of Official Foreign Exchange Reserves: 2026 Q2, 30 September 2026: https://data.imf.org/en/news/imf%20data%20brief%20september%2030
- Investment Company Institute, Exchange-Traded Fund Data: August 2026, 29 September 2026: https://www.ici.org/research/stats/etf/etfs_08_26
- World Gold Council, Gold ETF Flows: August 2026, 9 September 2026: https://www.gold.org/goldhub/research/gold-etfs-holdings-and-flows/2026/09
- World Gold Council, China gold market update: official buying accelerated in August, September 2026: https://www.gold.org/goldhub/gold-focus/2026/09/china-gold-market-update-official-buying-accelerated-august
- Goldman Sachs Research, Gold Is Forecast to Climb as Central Banks Buy the Precious Metal, 28 August 2026: https://www.goldmansachs.com/insights/articles/gold-is-forecast-to-climb-as-central-banks-buy-the-precious-metal
- Vanguard, Active Fixed Income Perspectives — September 2026, 22 September 2026: https://advisors.vanguard.com/insights/article/series/active-fixed-income-perspectives
- UN Trade and Development, Latin America pulled in more capital in 2025. Its project pipeline tells a different story, 30 September 2026: https://unctad.org/news/latin-america-pulled-more-capital-2025-its-project-pipeline-tells-different-story
Data cutoff: 4 October 2026, 23:59 UTC.