KenMacro Desk · Week Ahead

Labour held, so US CPI now decides the Fed trade for gold, the dollar and Treasury yields

Last week the labour market was supposed to crack, and it did the opposite. That hands the argument inside the Federal Reserve straight to inflation, and it means Friday’s US CPI report, with PPI and the European Central Bank the day before it, is the last real evidence anyone gets before the September Fed meeting on the 15th and 16th. What that number does to rate expectations is what moves Treasury yields, and what Treasury yields do is what moves the US dollar and gold. So this is the chain I am watching this week, and these are the levels I am watching it against.

All times London. The Fed meets on 15 and 16 September, with a Summary of Economic Projections.

Where the market actually finished

Friday gave us the starting point, and it is worth being precise about it because the headline reaction and the closing marks were not the same thing. Payrolls beat by a mile, the dollar and yields jumped, and then a good deal of that came back before the settled prints.

The dollar index settled at 99.16, up about a quarter of a per cent on the day. Look at the week instead of the day and the picture inverts: it closed at 99.68 on 28 August, fell every session into a 99.00 close on Thursday, and Friday’s payroll rally recovered barely a sixth of that. The strongest labour print of the year could not get the dollar back to where Wednesday left it.

The 2 year Treasury yield finished at 4.37 per cent and the 10 year at 4.78 per cent, on the Treasury’s own official close. Across the whole week the 2 year added three basis points and the 10 year added five, which is a mild steepening rather than a front end repricing. If Friday had genuinely rewritten the Fed path, that relationship would normally run the other way.

Gold fell roughly 1.2 per cent, to somewhere between 4,419 and 4,430 dollars depending on which vendor’s cut you take, and the range it travelled getting there is the interesting part. Friday ran from 4,368.53 up to 4,514.11, and Thursday had already topped out at 4,510.60. Gold has now been rejected twice in two sessions from just above 4,510, and it settled almost exactly on the 4,429.04 level this desk published a week ago.

Brent settled at 96.28 dollars, up close to 8 per cent on the week, and WTI at 91.48 dollars, up almost 10 per cent. Those are the corrected settlement figures from the exchange rather than the 92.68 print that circulated in one wire wrap on Friday afternoon.

Monday is quiet, and quiet is not the same as irrelevant

Monday 7 September is Labor Day in the United States, with the NYSE and Nasdaq closed. That makes it a shortened US week and it means very little cash market information, but it does not mean nothing happened.

Two things carry into the open. OPEC+ meets online on Sunday, and at the time of writing Saudi Arabia, Russia, Iraq, Kuwait, Algeria, Kazakhstan and Oman are expected to leave October production policy alone while the group works out its 2027 baselines, with a capacity review due at the end of September and the next full ministerial not until 29 November. And on Saturday US forces struck three Iranian tankers, the Downy off Kharg Island, the Stark 1 near Jask and the Kylo in the Gulf of Oman, after Iranian ballistic missiles were fired at a US carrier and a destroyer.

A US official described that as a new tanker for tanker policy. For anyone trading macro this is not filed under oil and left there. Higher energy prices raise headline inflation, headline inflation shapes what people expect from central banks, those expectations set bond yields, and yields decide the relative appeal of the dollar, of gold and of everything that competes with a risk free rate.

The July CPI report already shows that channel running. The energy index was up 14.7 per cent over twelve months and gasoline 24.6 per cent, even though both fell on the month. That is what a live supply shock looks like inside an official inflation statistic, and it is why oil is the first chart on this desk right now rather than the last. If you want the mechanism in full, I have written it up separately in how geopolitical risk moves oil.

The Fed argument has become very simple

Two speeches give you the whole framework for this week. At Jackson Hole on 28 August, Kevin Warsh described an economy at full employment with inflation he was not prepared to look through, and pointed at PCE inflation running 3.7 per cent over twelve months. He committed to a discipline rather than to a decision, which is a polite way of saying the data would decide it.

The number underneath that is worth pulling out. The twelve month PCE figure is 3.7 per cent, but the six month annualised figure Warsh cited is 4.1 per cent. The recent run rate is faster than the annual rate, not slower. That is the opposite shape to the one a disinflation argument needs.

Then on 3 September, Christopher Waller gave the market the other branch. He said recent data suggested we are finally seeing some signs of disinflation, and that if this continues in the data due over the next two weeks he would be inclined to support holding the target rate where it is. His phrasing was blunter than that in the room: give disinflation a chance, we can wait one meeting.

But he attached a condition, and the condition is this week. If inflation comes in hot, he said, he would consider a rate hike, and he judged that policy is currently only slightly restricting demand, so it may not take much acceleration to nudge him into supporting tighter policy. The data he was talking about is Thursday’s PPI and Friday’s CPI.

It is also worth remembering where the committee already sits. The target range is 3.50 to 3.75 per cent, and the July meeting held it by nine votes to three, with Hammack, Kashkari and Logan all preferring a hike at that meeting. Three hawks are already on the record. Waller is describing the conditions under which he stops being the fourth.

Thursday: PPI is the first test, and the ECB gets in front of it

The Bureau of Labor Statistics releases August producer prices on Thursday 10 September at 08:30 New York, 13:30 London. PPI measures prices received by domestic producers, so it is not CPI, but it gives an early look at whether cost pressure is building further upstream.

July showed final demand unchanged on the month and still 4.7 per cent higher over twelve months, with the measure excluding food, energy and trade services up 0.4 per cent on the month and also 4.7 per cent over the year. The question to ask on Thursday is not whether it beat or missed. It is whether it changes what the market expects on Friday.

If PPI runs hot and the 2 year yield pushes higher with hike pricing behind it, that is information. If PPI runs hot and the 2 year refuses to move, that is also information, and arguably the more useful of the two. The number tells you what happened. The reaction tells you what the market thinks it means.

The ECB lands 15 minutes before PPI, and nobody is scheduling around that

Thursday is not only a US day. The European Central Bank announces at 14:15 Central European time, which is 13:15 in London, with updated staff projections, and Christine Lagarde’s press conference starts at 14:45 CET, or 13:45 London.

Put those next to US PPI at 13:30 London and the sequence is awkward on purpose. You get the ECB decision, then fifteen minutes later a US inflation print, then a press conference that runs straight over the top of the PPI reaction. All 65 economists in the Reuters poll expect a 25 basis point rise to a 2.50 per cent deposit rate, and around 91 per cent think that is where the deposit rate ends the year.

So the hike itself is not the event. The projections and the guidance are, because a unanimously expected move is already in the price. And if you are looking at the dollar on Thursday afternoon, do not assume every tick is American. The dollar is a relative price, and for half an hour that afternoon the other side of it is doing more of the talking.

Friday: CPI gets the final word

August CPI is released on Friday 11 September at 08:30 New York, 13:30 London. July headline was 3.4 per cent year on year, easing from 3.5 per cent, and core, which strips out food and energy, rose 0.2 per cent on the month for 2.5 per cent over the year. Consensus clusters tightly at about 2.4 per cent for annual core in August, with the Cleveland Fed nowcast near 2.38 per cent, so a print in line with expectations is already assumed rather than hoped for.

This is where Waller’s argument gets tested rather than debated. If inflation keeps cooling, the case for raising rates immediately becomes much harder to make even after a labour report that strong. If it comes in hotter than expected, the Fed is looking at resilient employment, sticky inflation and elevated energy prices at the same time, and that is a far cleaner hawkish combination than anything it has had this year.

One thing to hold in mind while you read the release. Core CPI excludes energy by construction, so a hot oil market shows up in headline first and only reaches core later, through freight, packaging and production costs. A soft core print in an energy shock is not the same evidence as a soft core print in a calm market. If you want the underlying series rather than the headlines, the desk keeps a live page on US CPI inflation.

The CPI decision tree

This is the piece I would actually have open while the number prints, because the useful work is done before the release, not after it.

The reason I lay it out this way rather than picking a side is that I am not trying to guess the number. I am trying to know, in advance, what each version of the number would oblige me to do, so that the decision is already made when the volatility arrives.

The chain this desk actually watches

The KenMacro transmission chain. CPI enters on the left. The gold candle is the last link, not the first.

CPI itself is not the trade. The trade comes from what CPI changes. If the data moves Fed expectations, and those expectations move the front end of the Treasury curve, money starts repricing across currencies, equities and commodities, and only then does it reach the chart most people are staring at.

By the time the gold candle moves, four of those steps have already happened. That is the difference between seeing a move and understanding one, and it is why the 2 year gets more of my attention than the metal does. There is a fuller explanation of why yields and the dollar sometimes disagree in why the dollar can fall when US bond yields rise.

Trump’s rate demand adds a contradiction worth pricing

There is a political layer this week too. After Friday’s jobs report the President renewed his demand for sharply lower rates, saying the United States should have the lowest rate of any country in the world, and threatened to stop trading with countries running surpluses against the US if the Fed does not move.

Leave the politics alone and look at the mechanism, because the contradiction is the interesting part. He wants lower interest rates. More aggressive trade restrictions disrupt supply chains and raise the cost of imported goods, which adds to inflation. And additional inflation pressure is precisely the thing that makes it harder, not easier, for the Fed to justify cutting.

It also puts Warsh in an uncomfortable position, since he is the President’s own appointment and spent Jackson Hole sounding hawkish. For markets the question is narrow: does this stay rhetoric, or does it become policy capable of moving actual trade flows and prices? Only the second one belongs in a forecast.

Oil is the wildcard with no scheduled release time

Here is the reason I do not want anyone reading Friday’s CPI in isolation. Brent finished last week above 96 dollars, with the US and Iran exchanging strikes and shipping through Hormuz badly impaired.

The scale of that disruption is easy to underestimate. Crude and petroleum liquids through the Strait averaged 4.9 million barrels a day in the second quarter of 2026, against 21.6 million a day in the fourth quarter of 2025 before the conflict started. Traffic is running at less than a quarter of where it was, and six commodity vessels transited on Wednesday against a ten day average near thirteen.

Oil reaches markets through several channels at once. It lifts headline inflation, squeezes household spending, raises transport and production costs for business, and feeds longer term inflation expectations. If the shock persists it eventually produces the combination nobody wants, which is weaker growth alongside firmer prices, and that is the point at which the stagflation conversation stops being theoretical.

So CPI is the scheduled catalyst this week. Oil is the unscheduled one, and it does not wait for 13:30.

Gold: this is not a case of CPI up, gold down

Gold closed Friday around 4,430 dollars after the jobs report lifted yields and the dollar. The retail version of this week says hot CPI means gold down and cool CPI means gold up, and it is not good enough, because it skips the only question that matters.

You need to know why yields are moving. If CPI runs hot and the 2 year rises because the market is pricing a more aggressive Fed, with the dollar strengthening alongside it, gold is facing a straightforward monetary policy headwind and the level to defend is 4,346.16.

But if longer dated yields rise because the market is worried about inflation persistence, fiscal credibility or an energy shock, and the dollar weakens while that happens, that is a completely different regime with the same arrow on the chart. Same direction, different driver, different trade. Last week’s five basis point move in the 10 year against three in the 2 year is a small example of exactly that distinction.

Note too what gold did not do. It was rejected from just above 4,510 twice in two sessions, on Thursday and again on Friday, which gives us a live ceiling to work against rather than a guess. Between 4,510 and 4,346 is the range this week has to resolve.

The dollar: Friday found the rate advantage again, but CPI has to confirm it

The index ended Friday at 99.16 after payrolls revived expectations of a September hike, and the intraday behaviour told us more than the close did. The dollar jumped harder than where it finished, and hike probability climbed on the print before easing back.

The pricing itself needs a caveat rather than a decimal place. CME’s FedWatch had a September hike at roughly 62 per cent after the release, against about 49 per cent on Wednesday, while other vendors quoted anywhere from the mid fifties to the mid sixties for the same meeting on the same afternoon. The direction of travel is reliable and the exact number is not, so it gets published here as a band.

What that tells us is that the employment report strengthened the hawkish case without finishing the argument. Inflation still has to confirm it. For the cleanest bullish dollar branch this week I want to see all four links move together: hot inflation, higher Fed pricing, a 2 year through 4.39, and the index confirming it. If one of those refuses, that refusal is the signal and I would rather investigate it than chase the headline. There is more on the index itself in the DXY dollar index forecast, and the front end has its own live page at US 2 year Treasury yield.

UK traders: Friday starts six hours early

For anyone trading sterling, Friday begins well before the American session. The ONS publishes the July monthly GDP estimate at 07:00 London alongside the rest of the activity data.

The June report had the economy growing 0.3 per cent on the month, with output up 0.4 per cent across the three months to June, and that three month rate had slowed from 0.6 per cent in the May period. Services did the work at 0.5 per cent over three months while production was flat.

That gives cable an early catalyst six and a half hours before US CPI, and the obvious trap is treating the pair as though only one economy exists. Cable closed Friday at 1.3521, still comfortably above the 1.3427 to 1.3431 base this desk published last week, so the structure has not been threatened. It is simply the weakest of the majors I follow, and it will be pushed around from both ends on Friday.

What I am watching, and in what order

I am not going into Friday trying to guess a single number. I am going in already knowing the branches, because that is the only preparation that survives contact with a volatile release.

The evidence on the table is that US labour is resilient, that Warsh thinks inflation is still too high and has the six month PCE run rate to point at, that Waller thinks enough disinflation may be emerging to justify waiting but only if the data confirms it, and that oil is sitting at a level which keeps energy inflation firmly in the conversation. Three FOMC members already voted for a hike in July. The Fed meets four days after CPI lands, with new projections.

So the screen is short: Fed pricing, the US 2 year, the US 10 year, the dollar index, gold, and Brent. Then three questions, in this order. What does the market currently believe? What could make it change its mind? And where does the money go if it does?

I do not need to predict the next candle. I need to know what would make the market change its mind.

The bottom line

Last week asked whether the labour market was starting to crack, and the answer was no. Payrolls came in at 162,000, above every forecast in the Reuters poll, unemployment held at 4.1 per cent and the previous two months were revised up by 55,000 combined, with July flipping from a loss to a gain. That gave the Fed room to concentrate on prices.

Now inflation gets the next word, and it gets it twice, on Thursday and again on Friday. Hot, and the hawkish case gets cleaner, hike pricing rises, the 2 year presses its 2026 high, the dollar finds support and gold has a harder environment to work in. Cool, and Waller’s patience argument strengthens, pricing falls back, the front end eases, dollar support fades and gold gets breathing room.

But do not trade the arrows blindly. Watch the reaction, because the reaction is where the information is, and the lesson from the last two weeks has not changed.