The most useful allocation signal this week is not a simple move from “risk” to “safety.” It is a change in the vehicle through which capital is being held. Investment Company Institute data released on 22–24 September show that, for the week ended 16 September, long-term mutual funds had an estimated $36.70 billion of outflows while exchange-traded funds recorded $26.61 billion of net issuance. Combined long-term mutual fund and ETF flows were still negative by $10.10 billion, but the composition matters: investors were redeeming traditional fund structures while continuing to create ETF shares. Over the following week, U.S. money-market fund assets rose by $15.00 billion to $7.94 trillion.[1][2][3]
That combination points to liquidity preservation with vehicle substitution, not a generalized disappearance of capital from markets. Equity products recorded net outflows in the combined ICI data, bond products remained positive, commodity funds received new money, and cash-like assets stayed close to $8 trillion. At the same time, the euro area continues to attract substantial non-resident portfolio capital over a longer horizon: in the 12 months to July, non-residents made net purchases of €525 billion of euro-area equity and €631 billion of euro-area debt securities.[4]
The distinction between flow, stock and valuation is becoming more important. China’s foreign-exchange reserves rose by $19.5 billion in August to $3.438 trillion, but the State Administration of Foreign Exchange explicitly attributed the increase to a combination of exchange-rate translation and asset-price changes rather than to a pure reserve inflow.[5] This week’s evidence therefore supports a narrower conclusion: capital is being redistributed across wrappers, maturities and jurisdictions, while large liquidity stocks remain intact.
ETFs absorbed issuance even as long-term funds lost money overall
ICI’s combined series shows a negative aggregate long-term flow for the week ended 16 September, but that total masks a sharp divergence between mutual funds and ETFs.
| Allocation channel | Latest observation | Interpretation |
|---|---|---|
| Long-term mutual funds | -$36.70bn | Net redemptions |
| ETFs | +$26.61bn | Net share issuance |
| Combined long-term funds + ETFs | -$10.10bn | Net outflow after offsetting ETF creation |
| Equity, combined | -$13.30bn | Equity products remained net negative |
| Bond, combined | +$3.24bn | Fixed-income products still attracted net money |
| Commodity funds | +$2.12bn | Positive issuance/flow in commodity vehicles |
The vehicle split is analytically important. ETF issuance does not prove that investors became more risk-seeking; it shows that exposure was being created in a more tradable structure even while the mutual-fund channel contracted. The combined figures therefore argue against using one fund category as a proxy for total investor conviction.[1][2]
View data
| Indicator / period | Value (US$ billion) |
|---|---|
| Long-term mutual funds | -36.7 |
| ETF net issuance | 26.61 |
| Combined long-term funds + ETFs | -10.1 |
Cash-like assets remain structurally large
The second new signal is that the large U.S. liquidity stock did not unwind. ICI reported total money-market fund assets of $7.936 trillion for the week ended 23 September, up $15.00 billion from the prior week. Government funds were nearly flat, while prime funds rose by $10.95 billion and tax-exempt funds by $3.66 billion.[3]
View data
| Indicator / period | Value (US$ trillion) |
|---|---|
| 2026-09-09 | 7.973 |
| 2026-09-16 | 7.921 |
| 2026-09-23 | 7.936 |
The one-week rebound is small relative to the stock, but that is precisely the point: the cash-like pool remains very large even as investors selectively add ETF exposure. This weakens any interpretation in which ETF issuance alone signals a broad migration out of liquidity.
Mutual-fund redemptions -> capital leaves one pooled-fund wrapper -> part of market exposure is recreated through ETF issuance -> combined long-term flow remains negative -> money-market assets remain near $8tn -> allocation becomes more liquid and vehicle-sensitive rather than simply “risk-on” or “risk-off”
Euro-area portfolio securities are receiving substantial foreign demand
The euro-area balance-of-payments data add a cross-border dimension. Over the 12 months to July 2026, euro-area residents made €193 billion of net purchases of non-euro-area equity and €622 billion of net purchases of non-euro-area debt securities. In the opposite direction, non-residents bought €525 billion of euro-area equity and €631 billion of euro-area debt securities.[4]
View data
| Indicator / period | Value (€ billion) |
|---|---|
| Euro-area residents buying foreign equity | 193 |
| Euro-area residents buying foreign debt | 622 |
| Non-residents buying euro-area equity | 525 |
| Non-residents buying euro-area debt | 631 |
These are gross directional portfolio flows, not a complete measure of the euro area’s external wealth position. Still, the composition is notable: foreign demand for euro-area debt is materially stronger than a year earlier, when the comparable 12-month figure was €378 billion. Non-resident purchases of euro-area equity also rose from €439 billion to €525 billion.[4]
The latest ECB monetary data do not show an abrupt domestic credit boom accompanying those inflows. In August, adjusted loans to households grew 3.1% year on year, while loans to non-financial corporations grew 4.2%, slightly slower than July’s 4.4%.[6] Cross-border portfolio demand and domestic bank-credit creation are therefore moving on different channels.
China’s reserve increase is a stock-and-valuation signal, not a clean capital-flow measure
China’s official foreign-exchange reserves rose to $3.438 trillion at the end of August, up $19.5 billion from July. SAFE explicitly said the change reflected the combined effects of exchange-rate translation and changes in global financial-asset prices.[5]
That qualification is essential. A higher reserve stock can result from valuation even without an equivalent net purchase of reserve assets. For wealth-flow analysis, this means reserve-stock changes should not be added mechanically to observed portfolio or fund flows.
Mutual-fund outflows coexist with ETF issuance and a $7.94tn money-market stock
Non-residents accumulated both euro-area equity and debt over the 12 months to July
FX reserves rose in August, but SAFE attributes the change partly to currency and asset-price effects
Capital is moving across wrappers, maturities and jurisdictions rather than following one universal risk regime
The prior week’s dollar-flow picture remains a baseline, not a new signal
No new U.S. Treasury International Capital release arrived this week. The latest available monthly TIC data remain July’s $83.7 billion aggregate net inflow, with a much weaker adjusted long-term balance. That information was central to the 21 September monitor and should not be promoted again as if it were a new weekly development.[7]
Its value this week is only as context: the U.S. still combines deep dollar liquidity with mixed long-term portfolio flows, while newer ICI data show that the domestic fund wrapper is itself changing. The marginal information is the vehicle split and the persistence of cash-like balances.
What would change the assessment
The vehicle-substitution interpretation would weaken if mutual-fund redemptions and ETF issuance began moving in the same direction for several consecutive weeks. A sustained decline in money-market assets alongside broad positive long-term flows would indicate a clearer migration out of cash-like instruments. Conversely, persistent ETF creation with continued mutual-fund redemptions would strengthen the case that investors are changing the structure of exposure without making an equivalent change in aggregate long-term risk appetite.
For the euro area, the relevant test is whether strong non-resident portfolio demand persists in the next balance-of-payments releases and whether it remains concentrated in debt or broadens further into equity. For China, reserve changes should continue to be decomposed into transactions, exchange-rate translation and valuation before being interpreted as capital allocation.
Sources
- Investment Company Institute, Combined Estimated Long-Term Flows and ETF Net Issuance, 23 September 2026: https://www.ici.org/research/stats/combined_flows
- Investment Company Institute, Estimated ETF Net Issuance, 22 September 2026: https://www.ici.org/research/stats/etf_flows
- Investment Company Institute, Money Market Fund Assets, 24 September 2026: https://www.ici.org/research/stats/mmf
- European Central Bank, Euro area monthly balance of payments: July 2026, 18 September 2026: https://www.ecb.europa.eu/press/stats/bop/2026/html/ecb.bp260918~ef4843e4f5.en.html
- State Administration of Foreign Exchange, China’s foreign-exchange reserves at end-August 2026, 7 September 2026: https://www.safe.gov.cn/safe/2026/0907/27859.html
- European Central Bank, Monetary developments in the euro area: August 2026, 25 September 2026: https://www.ecb.europa.eu/press/stats/md/html/ecb.md2608~406f8d243d.en.html
- U.S. Department of the Treasury, Treasury International Capital Data for July 2026, 16 September 2026: https://home.treasury.gov/news/press-releases/sb0631
Data cutoff: 27 September 2026, 23:59 UTC.