The drain paused — the stagflation didn't

Where we are

The market regime reads YELLOW, and has since August 3, when it softened from RED on the U Mich consumer sentiment rebound to 55.2. That reading deserves a warning label: the RED/YELLOW threshold sits at 55, so the headline colour is 0.2 of a point from flipping back. A routine revision would do it. This edition treats the colour as provisional and readers should too — a colour change from this position is a data artefact until proven otherwise, not a market reversal.

The business cycle is a steadier read. FALL — stagflation — has been confirmed by 21 consecutive daily reads, and the weekly series has shown FALL since July 20. That is the longest uninterrupted stretch in this quadrant all quarter; the June whipsaw (to FALL on June 1, back to SUMMER on June 8) has not repeated. The cycle is nonetheless sitting on the SUMMER/FALL boundary as well: the growth score reads neutral rather than negative, so the quadrant would turn on a small upward move in growth. Confidence in the reading is moderate — the direction is clear, the margin is not.

One disagreement inside the engine is worth naming rather than smoothing over. The growth indicator pillar reads green, while the growth cycle score reads neutral. These are different measures — one reads the level and trend of the growth family directly, the other scores three components into the quadrant axis — and when they disagree, that is information: growth data is firm enough to look healthy in the level, not firm enough to pull the cycle out of stagflation. Inflation remains the dominant risk factor, with core PCE and core CPI both still rising and only 5-year breakevens pointing the other way.

The drain paused

Measured the same way at both ends — the Fed, ECB and Bank of Japan on a constant basis — global liquidity fell from $18.11T to $17.64T over twelve weeks, a 2.6% contraction. That remains the standing trend. But the most recent week broke from it: $17.42T to $17.64T, a rise of $220B and the largest single-week increase in the window. It was also enough to move the liquidity classification off "draining" for the first time since June 15.

The composition is what makes it interesting. The Fed was flat, at $6.75T against $6.74T a week earlier — US net liquidity has barely moved all quarter. The entire increase came from the two banks that had been doing the draining: the ECB added $70B to $6.84T and the Bank of Japan added $140B to $4.05T. Over the full twelve weeks those two are still down $360B and $130B respectively. Last week they stopped subtracting and started adding.

One week does not reverse a quarter, and this is a level series that moves with exchange rates as well as with policy — a $220B weekly swing across two non-dollar balance sheets can be revaluation as easily as intent. The honest statement is narrow: the drain paused, the twelve-week trend is still negative, and it takes a second and third week like this one before the trend itself is in question. Two further caveats belong in the open. Six of the thirteen weekly rows are reconstructed point-in-time rather than recorded live, so the early-window week-to-week detail is less reliable than the recent end. And the headline series flipped basis twice in this window — China dropped out of coverage on June 29 and re-entered on July 6 — which moves the headline number by the size of the PBoC without anything real happening; every figure above is on the constant three-bank basis precisely so that coverage cannot be mistaken for a move.

China itself is unchanged since the last edition. The PBoC balance sheet on file is still the June publication, now 69 days old, at $7.32T. The July print has not appeared and the release watch has not flagged one. The four-bank aggregate stands at $24.96T.

Everything eased except the price of money

Financial conditions flipped from neutral to loose this week, after holding neutral since June 22. Two pillars moved on the same day: the dollar eased from restrictive to neutral, with the broad index down 0.8% on the week to 119.70, and global liquidity moved off draining. Volatility is quiet at 15.15, down from 17.09 a week ago and down 2.7 points over the quarter. High-yield spreads narrowed from 284 to 271 basis points, investment grade from 80 to 78. Four of five pillars improved.

The fifth did not. The 10-year real yield is 2.44%, up from 2.40% last week and up 31 basis points over twelve weeks, and it remains classified restrictive — the only pillar that has not eased at any point in this window. The composite direction reflects that: the level reads loose, the direction still reads tightening. Both are true, and the pair is the whole picture. Positions financed at these real rates get no relief from calm credit or a softer dollar; the cost of carry is set by the rates pillar, and the rates pillar has not moved.

This also resolves — for now, and in credit's favour — the tension flagged in the August 2 edition. That edition noted the unusual pairing of rising real yields with flat high-yield spreads, and asked which side would move first. Over the past week credit did not converge toward rates; it went the other way, with spreads narrowing 13 basis points while real yields rose 4. Credit markets are pricing no stress at all. That is a resolution of the near-term question and a sharpening of the longer-term one.

What to watch

• Whether the liquidity turn extends. One week is a data point; three consecutive weeks of ECB and BoJ balance-sheet expansion would be the first genuine change in the dominant force of this quarter. A single week that reverses is revaluation noise.
• Consumer sentiment, on August 14 and again at month-end. At 55.2 against a threshold of 55, the next print — or a revision to this one — decides whether the market regime reads YELLOW or RED, and neither outcome would be a statement about markets.
• The July PBoC publication, due around August 15 and now the second month running that this sits on the watch list. The June print showed China easing against a European and Japanese drain; that drain has now paused, which makes the Chinese contribution less pivotal than it was a week ago, not more.
• The real yield. It is the one pillar that has moved in one direction all quarter, and the only one currently classified restrictive. Conditions cannot be durably loose while the price of money is not.
• Data quality is clean this week: all 27 sources report current, nothing stale, no release alerts outstanding. Where a figure above is reconstructed rather than observed, it is labelled as such.

The frame, not the trade

There is a specific way to be wrong this week, and it is to treat two threshold crossings as two pieces of news. The regime softened to YELLOW because a survey moved 0.2 past a line. Conditions flipped to loose because two pillars improved on the same day while the direction underneath still reads tightening. Neither is a structural change, and both would read as one in a headline. Meanwhile the thing that actually changed — $220B of balance-sheet expansion out of Frankfurt and Tokyo after a quarter of contraction — crossed no threshold at all and produced no label.

The question that puts to a real book is about what you are anchored to. If a position was sized for draining liquidity, one week of expansion is not permission to resize — but it is the first evidence against the premise, and it is worth knowing now whether you would need two more weeks or six to change your mind, before the weeks arrive and the answer becomes retroactive. If a position depends on financing costs easing, note that the rates pillar has moved against that thesis for twelve straight weeks while every other pillar has moved for it. And if the stagflation quadrant is what your defensive positioning is built on, the reading is 21 days confirmed and sitting on a boundary — durable enough to plan around, close enough to the line that the invalidation is worth writing down rather than assumed. That is the exercise: not what the labels say this week, but which of your positions would have to be reconsidered if the label you are relying on is the one that moves next.