Are Foreigners Selling US Assets (Part II)

Dramatic USD selling took place in April, but there has been differences by FX cross, and we are now arguably six months into an asset allocation shift, the intensity of which may be peaking

Aug 12, 2025

Careful flow analysis suggests an allocation shift away from the dollar may have peaked and will likely be a less dominant force in H2 compared to H1 2025.
This does not necessarily mean that USD is set for recovery; rather, traditional fundamentals, such as the cyclical state of the US economy (and Fed policy) will now again play a bigger role in potential further USD weakness.

Introduction

Since March, we have been hyper-focused on a potential asset allocation shift out of US assets, with STRATEGIC implications for the dollar.

This is something we have been tracking day and night for Exante Data’s institutional clients. But we also did publish a public substack on the topic in April (see here), with the following key observations:

First, we highlighted that April saw highly unusual foreign selling of US fixed income (and likely driven by the private sector.
Second, we highlighted that Japanese investors had been sellers of international fixed income (and likely US fixed income) during April.
Third, we highlighted that anecdotal information from CIOs in our network, as well as signals from options markets pointed to dramatic USD hedging demand (put buying), especially in April.

In this update, we evaluate the structural flow situation with a look to trends in H2, given that almost four months have passed since our last update. We also present more information on global equity flow dynamics (which have been important) and discuss some concrete metrics of hedging flows

The equity flow dimension: From US exceptionalism to ?

Before we turn to the topics covered in the April Substack (see the sections below), we will go into some detail with the trends observed in cross border equity flows.

First, we will note that we came into 2025 with an incredibly strong trend in US equity inflows.

The big picture history is illustrated below.

Doc.com era (1999-2021): US stocks accounted for a high proportion of all cross-border stock flow in the dot.com period (specifically around 34%)
BRICS/global growth era (2005-2007): US stocks accounted for a lower proportion of all cross-border stock flow, as capital moved to international stocks (specifically 24%)
Post-GFC period (2009-2014): The US share of global cross-border equity flow recovered to 34%.
The period of US exceptionalism (2017-2024): The US share of global cross-border flows spiked to 43%.

This is the big picture narrative using official BoP data. But such data is lagged (only available to Q1). Hence, to get a granular feel for the trends in 2025, we have to use alternative data. The below charts look at monthly portfolio flow proxies from EPFR, specifically zooming in on cross-border equity flows from the Eurozone into the US and from various Asian economies into the US

Both Eurozone and the Asian flows have changed dramatically in 2025.

The Eurozone flow peaked right after the US election (November) and turned notably negative in March-April, and have tentatively recovered since then.

The Asian flow peaked in January, and has weakened more gradually to seemingly bottom in June.

Relatedly, note that the the dramatic move in the TWD also happened later than the gap in European currencies, in May-June, vs March-April.

A more detailed look at cross-border equity flows is presented in the heat-map below, which conveys the same message as the chart but with country specific details available.

Note that some non-Eurozone countries, such as Sweden, arguably showed even more front-loaded responses to the policy shock from Trump II, and that equity flows there have also been looking better (albeit not fully normalized) in June and July.

And now back to the topics covered in April, drawing on the real-time analytics from Exante Data’s data platform (specifically the USD Allocation page / analytics).

Bond Update: Private Sector Flows Normalizing…

US bond flows looked very bad and were a complete outlier in April. But as our proprietary tracking shows, the relative bond flows are substantially normalized in May-July, based on proxy flows from mutual funds and ETFs.

Relatedly, Japanese bond flows, which are globally important, and looked shaky in April, have recovered, with flows specifically destined to the US looking better in May, and overall FI flows looking better in June and July also, as shown in the two charts below.

The first chart illustrates that after soft flows into the US in Mar-April, May looked better, and June was neutral.

The second chart, looks at a more timely proxy of overall Japanese (asset) flow in the BoP. The recent negative readings, in May, June and July shows that Japanese investors have resumed sending capital into global FI markets (negative influence on Japan’s BoP), after some temporary repatriation in March-April.

Official Flows: Not the main thing (never too dislocated)

We argued back in April that our leading indicators of official flow into US fixed income looked ‘normal’ and that official selling was likely not the main factor in the USD sell-off at the time.

Here we simply show the more formal data (via TIC) we have gotten since, which confirms the narrative that March-May (the last three data points in the chart) looked fairly normal, and certainly not sending any clear signal of a regime shift in official sector behavior as some/many had speculated during April.

Color on hedging flows

Finally, we want to provide a bit of color on the hedging dimension. This is important since FX markets are generally more liquid than securities markets (and extreme example of this would be corporate credit vs FX).

We argued back in April that significant FX activity was taking place, and this was also obvious from option indicators are the time.

The chart below shows option skew (risk reversals) normalized by vol levels. The shift in April is clear, with option shew moving from favoring USD calls (for EUR, TWD and) to favoring USD puts (and for JPY moving more deeply to favoring USD puts). We can see this downward shift for all key crosses. But we can also see that USDJPY skew has mostly normalized (which is partly the case for USD vs GBP also). Meanwhile, EUR and TWD skew is still severely dislocated from what was the norm in 2022-2024, signaling still persistent demand for USD hedging.

In April, the real-time option indicators were one of the only objective indicators of hedging demand. But several months have now passed, and there are ways to measure hedging demand and positions.

The chart below shows one chart from our proprietary data collection, aggregating hedge ratios of US domiciled fund with a focus on international bond exposure. It shows how US investors from 2021 to end 2024 increasingly hedged their non-USD exposure (with a hedge ratio > 95%( by end 2024). However in Q2, we have seen this hedging shift, with more appetite for non-USD (open) exposure, with the hedge ratio dropping from >95% to < 90%.

Conclusion

This year has seen dramatic Dollar moves, with the USD underperforming notably versus traditional drivers (yield curve moves) on a global basis in March, April, May and June (see chart below).

But the USD underperformance partially reverted in July, and this movement (less one-sided depreciation pressure on the USD) is consistent with the signals from our flow analysis.

First, bond flows have generally normalized after dramatic weakness in April.
Second, a lot of currency hedging activity has already taken place in April-May, and hedging demand may be moderating, at least in some USD crosses.
Third, equity flow weakness (essentially loss of US market share in global cross-border flow) has moderated during July.

All told, 2025 has seen an important USD asset allocation shift, and serious question marks around the USD bull trend that has been in place for most of the past 10-11 years have appeared. But precise and holistic flow analysis also indicates that the allocation shift may have peaked and that the force from this factor will be less dominant in H2 compared to H1 2025.

This does not mean that the USD is necessarily set for recovery. But it does mean that traditional fundamentals, such as the cyclical state of the US economy (and relatedly Fed policy) will play a bigger role in facilitating potential further USD weakness.

Money: Inside and Out

Discussion about this post

Ready for more?