Two Tapes Disagreed With the Disinflation Story on Friday, and Both Were Energy and Rates
The week that ended Friday, August 14 produced the most consistently soft run of US inflation and demand data in months, and two markets declined to take the hint. Treasury yields rose. Crude oil rose more. Neither move is large enough on its own to overturn a disinflation narrative built on three government releases, but both moved in the direction the releases argued against, and they moved in the same week. That is worth writing down carefully, including the parts that are contested.
Start with the case for disinflation, because it is real. The Bureau of Labor Statistics published the July Consumer Price Index at 8:30 a.m. Eastern on Wednesday, August 12, under embargo number USDL-26-1378. The all-items index rose 0.1% on a seasonally adjusted basis after falling 0.4% in June, and was up 3.4% over twelve months, down from 3.5%. Core CPI, all items less food and energy, rose 0.2% for the month and 2.5% over the year. Shelter rose 0.1% and the BLS said it accounted for roughly two-thirds of the monthly all-items increase.
Producer prices followed on Thursday, August 13. The BLS headline on that release reads that the producer price index for final demand was unchanged in July, seasonally adjusted, with services advancing 0.2% and goods falling 0.7%. The same release puts final demand up 4.7% over the twelve months to July, against a 4.9% consensus recorded by Trading Economics. The core figures need their labels attached, because the providers do not agree. The BLS's own core — final demand less foods, energy and trade services — rose 0.4% in July and stands 4.7% above a year earlier. Trading Economics' page reports core producer prices at 0.2% against June and 4.2% over twelve months, describing that series in its body text as excluding food and energy while labelling it in its own components table as excluding food, energy and trade services. Those cannot both be the same measure, and this article uses the BLS figures wherever it needs a core.
Then Friday's consumer print, and here the primary document is available. The Census Bureau's advance monthly retail trade report for July, released Friday, puts sales at $763.6bn, down 0.6% from June — with a sampling margin of error of plus or minus 0.4 percentage points, which is worth carrying, because it means the decline is real but not large against the survey's own noise. June was revised to a 0.2% gain that Census itself flags as not statistically significant. Nonstore retailers fell 2.2%, motor vehicle and parts dealers 1.8%, gasoline stations 0.9% and electronics and appliance stores 0.5%; clothing rose 1.9% and food services rose 0.5%. The 0.1% consensus the print missed comes from Trading Economics, as does the control measure — sales excluding food services, auto dealers, building materials stores and gasoline stations, which feeds the GDP calculation — reported down 0.4%. Census does not publish that control aggregate in the advance report at all; the exclusion it does publish, stripping out autos and gasoline alone, is down 0.2%. Three releases, three misses to the soft side.
Now the first tape that disagreed. To say whether Friday's yields rose or fell you need a baseline that is not itself a screen quote, and one exists. The Federal Reserve's H.15 Selected Interest Rates release, dated August 14, 2026, carries daily constant-maturity data through Thursday, August 13: the 2-year Treasury at 4.15%, the 10-year at 4.63% and the 30-year at 5.21%, with the federal funds effective rate at 3.63%. Friday's own official marks do not publish until the following business day. That gap is precisely why the weekend picture depends on commercial providers, and why they need auditing.
Against that 4.63% Thursday baseline, every Friday 10-year quote this desk could fetch sits higher. Advisor Perspectives' Treasury Yields Snapshot for August 14 gives the 10-year note at 4.68% and the 2-year at 4.17% at end of day. Trading Economics' US 10-year page shows 4.70% on August 14, up 0.05 percentage points. A markets wrap published on Yahoo Finance and timestamped 5:30 p.m. Eastern Friday states in its body text that the 10-year yield fell 0.05% to 4.69%. The three levels span two basis points. The three implied directions do not agree at all.
There is a simple test for this, and it is worth spelling out because it decides the argument. A provider's stated level minus its stated change should reproduce the prior session's mark. Trading Economics' 30-year quote of 5.26% on August 14, up 0.046 points, implies a Thursday level of 5.214% — which rounds to the 5.21% the Fed's own H.15 published for August 13. That reconciles. The Yahoo wrap's 10-year at 4.69% after a fall of five basis points implies a Thursday close near 4.74%, eleven basis points above what H.15 recorded. That does not reconcile, and it is the direction claim, not the level, that fails. This desk is not calling the 4.69% level wrong. It sits comfortably inside the 4.68%–4.70% cluster the other providers report. What cannot survive contact with the Fed's own series is the sign attached to it. The defensible published line is that the 10-year finished Friday somewhere around 4.68% to 4.70%, up several basis points from Thursday; that the 2-year was near 4.17%, up about two basis points from H.15's 4.15%, though that leg rests on a single provider; and that the 30-year was near 5.26%, up about five. Readers should treat a single decimal on any of those as false precision until Monday's H.15 lands.
The second tape that disagreed was crude, and here the provider problem is worse by an order of magnitude. Trading Economics' WTI page shows $82.40 a barrel on August 14, up 1.42% or $1.15. That figure passes the reconciliation test cleanly: $82.40 less $1.15 is $81.25, which is exactly the previous close Investing.com lists for the September-dated WTI contract. Two independent providers agreeing on Thursday's settlement is the strongest single anchor available in this piece.
It is also where the agreement stops. Forbes Advisor's oil page quotes WTI at $81.84, expressly stamped 8:25 a.m. Eastern on August 14, and Brent at $87.47 in the same morning update but without a time stamp of its own. The page is explicit that these are intraday levels rather than settlements, noting that settlement prices are the official closing levels set at the end of each trading day. Fortune's August 14 oil piece states that at 6:30 a.m. Eastern that day oil was priced at $89.53 a barrel with Brent as the benchmark. Trading Economics' Brent page headlines $88.50, up 1.42 or 1.63%, while its own body text on the same page says Brent was at around $87 a barrel on Friday. Read that sequence again, because it is the reportable part. Two providers quoted Brent on the same Friday morning, and the one that stamped its quote at 6:30 a.m. sat $2.06 above the one carried in an 8:25 a.m. update. A third provider disagreed with itself by about a dollar and a half within a single page. None of these is a settlement. This publication does not have a settlement price for Friday's crude from a settlement source, and so will not print one. WTI finished the week somewhere around $82 and Brent somewhere around $88, with providers spread across roughly a dollar on WTI and two on Brent depending on what hour of the day their screen was captured.
The weekly move is the number that actually matters for the thesis, and it comes from one provider. Trading Economics reports that crude extended gains to more than 5% on the week, and that Brent gained nearly 5%, attributing the move to the United States increasing economic pressure on Iran to reopen the Strait of Hormuz. The same page cites a maintained naval blockade of Iranian ports, an International Energy Agency warning of a deeper global supply deficit, and Houthi targeting of Saudi Arabia's Jazan refinery. A second provider corroborates the magnitude. Oil & Gas 360, in a report timestamped 15:30 UTC on August 14 — roughly 11:30 a.m. ET, before the close — wrote that WTI was up "around 5% this week" and that Brent was "trading up around 5% this week," while quoting intraday levels of $80.91 for WTI at 09:30 ET and $86.84 for Brent, both below Trading Economics' figures and both explicitly intraday rather than settlements. That piece attributes the year's support to "the effective closure of the Strait of Hormuz," noting that roughly a fifth of the world's oil and liquefied natural gas moved through the waterway before the Iran war began in late February, and that Iran says it has closed the strait to commercial shipping while the United States says sailings continue. So the weekly percentage is now two-provider; the specific Friday narration that follows remains Trading Economics' alone.
Why does a 5% week in crude matter to a 0.1% CPI print? Because of the gap between a monthly rate of change and a level, and July's report shows it starkly. The BLS put the energy index down 1.5% for the month — and up 14.7% over twelve months. Gasoline fell 2.9% in July and is up 24.6% from a year earlier. Headline CPI at 3.4% against core at 2.5% leaves a gap of about nine-tenths of a percentage point, and energy is doing most of that work. July was disinflationary because energy fell that month, not because energy is cheap.
The pass-through channel is measurable and it has not been measured yet. The Energy Information Administration's weekly gasoline and diesel update, with a gasoline release date of August 11, put the US regular grade average at $4.006 a gallon for the week ended Monday, August 10 — down 7.3 cents from $4.079 the prior week and up 88.8 cents from a year earlier. On-highway diesel was $5.257, up $1.503 year over year. The West Coast averaged $5.075 and the Gulf Coast $3.543. Note the date. That reading covers the week before crude's move, so the retail series has not yet registered any of it; the next weekly print is due the week of August 17.
There is a footnote in Friday's retail sales report that runs the same way. Gasoline station receipts are a nominal dollar figure. They fell 0.9% in July in significant part because pump prices were falling in July. If the past week's crude move sticks through August, that component mechanically reverses, flattering the headline retail sales number while doing nothing good for real household consumption. The same arithmetic that made July's report look weak can make a later one look stronger for a reason nobody should welcome. The producer price data carries a similar level-versus-rate warning, and a sharper one than the headline suggests. Final demand was unchanged in July, which is the disinflationary headline, and goods prices fell 0.7% on the month. But final demand is still up 4.7% over the twelve months to July, and the BLS core measure — final demand less foods, energy and trade services — did not merely fail to fall. It rose 0.4% on the month, and sits 4.7% above a year earlier. A monthly zero at the headline, resting on a 4.7% annual rate and a core that rose, is a deceleration rather than an arrival. Energy is the input that can put the monthly number back on the board without the core measures moving at all.
Households appear to have noticed something the CPI did not. The University of Michigan's preliminary August Survey of Consumers put the Index of Consumer Sentiment at 51.0, against 55.2 in July's final reading and a consensus of 55.0, with current conditions at 51.8 and expectations at 50.6. Year-ahead inflation expectations rose to 4.3% from 4.2%, while the 5-to-10-year measure held at 3.3%. The survey's own site carries survey director Joanne Hsu's summary that sentiment "fell about 8% this August, ending two consecutive months of improvement," and that while views of personal finances saw only minor declines, expected business conditions sank 11% for the short run and 17% for the long run. Yahoo Finance's wire account of the release separately quotes Hsu saying that "although the early-month weakening in sentiment was pervasive across various demographic groups, notably large reductions were seen among older consumers, lower-income consumers, and those without a college degree." This desk reached the survey's summary page but not its full PDF tables, and flags that second quotation as read in the wire report rather than in the primary release. Sentiment collapsed and one-year inflation expectations went up, in the same survey, in the week CPI cooled.
Trading Economics' own explanation of the 10-year's move on Friday reaches for both of those threads. Its page notes that Michigan year-ahead inflation expectations rose in August, a fifth month above the 4% level, and ties the selling at the long end to concern about Federal Reserve complacency on inflation — concern it describes as recently exacerbated by the surge in energy prices from the war in the Middle East — with softer producer prices and weak retail sales failing to stop it. That is a coherent story. It is also one day of trading, and one provider's narration of it. Yields can rise on a soft data day for reasons that have nothing to do with oil — supply, positioning, a Friday afternoon with thin books before a weekend of Middle East headline risk.
The honest counter-case deserves equal space. A crude rally driven by demand strength would be one thing; this one is described by its own source as supply-driven, and supply-driven energy shocks are contractionary as well as inflationary. Weak retail sales and a 51 sentiment print are disinflationary signals in their own right. A single week's move in a commodity that has ranged from roughly $55 to $118 over the past year is not a trend, and readers should be suspicious of anyone — including this article — who assembles two days of price action into a thesis. What can be said is narrower: on Friday, the rates and energy tapes did not corroborate the data, and that disagreement is the thing to watch rather than to resolve.
Rate expectations moved decisively toward the data during the week, though the well-sourced figures are already stale. Citing the CME Group's FedWatch Tool, which derives probabilities from 30-day fed funds futures, a Motley Fool analysis published August 12 reported a 67% chance of a 25-basis-point September hike as of July 31, falling to 44.4% by August 7. Trading Economics' August 14 page put markets at around a 35% probability of a September increase against 55% a week earlier, but does not name a source for those figures, so they are reported here as that outlet's characterization rather than as a FedWatch reading. No post-close Friday FedWatch figure was obtainable, and none should be read as current.
The strategist commentary on the record predates all three of this week's releases and should be dated accordingly. In a J.P. Morgan Wealth Management note published August 5 on Chase's insights site, chief investment strategist Phil Camporeale said: "The combination of a slower-than-expected normalization of supply chains around the Strait of Hormuz and market questioning of inflation-fighting credibility after the July FOMC meeting has lowered the bar for a rate hike in September." The same note stressed that a move would not be seen as the start of an aggressive tightening cycle but rather a measured step aimed at reinforcing the Fed's inflation-fighting credibility, and put market-implied September odds at roughly 65% at that time — a figure the subsequent data has substantially undercut.
The calendar decides this, not the weekend's commentary. Minutes of the July 28–29 FOMC meeting are scheduled for release Wednesday, August 19, per the Federal Reserve's own meeting calendar. That calendar also resolves a question worth stating plainly: the September 15–16 meeting is marked with an asterisk, meaning it will be accompanied by a Summary of Economic Projections. The BLS release notes the next CPI is scheduled for Friday, September 11 — the last inflation reading before the committee convenes. In between, the EIA's weekly gasoline series will show whether crude's move reached the pump, and Monday's H.15 will settle what Friday's yields actually did.
Equities, for their part, largely ignored all of it. Per the Yahoo Finance markets wrap timestamped 5:30 p.m. Eastern Friday, the S&P 500 closed at 7,785.76, down 13.23 points or 0.17%; the Dow Jones Industrial Average at 53,732.41, down 107.58 or 0.20%; the Nasdaq Composite at 26,729.16, down 0.28%; and the Russell 2000 at 3,068.42, up 15.57 or 0.51%. The Cboe Volatility Index finished at 14.25, down 2.60%. Advisor Perspectives noted that the latest Freddie Mac Primary Mortgage Market Survey put the 30-year fixed mortgage rate at 6.67%, unchanged from the prior week; that write-up gives no release date for the survey itself, so the reading should not be taken as Friday's.
One last caution about screens, offered because this week supplied a textbook example. Gold on Friday was quoted at 4,432.00, up 0.26%, in the market widget of the same wrap whose body text said gold rose 0.24% to $4,373.48 — a gap of more than $58 inside a single article. The widget labels its quote GC=F, a gold futures ticker; the body text attaches no instrument to its number at all. Yahoo Finance's Friday live blog, published at 6:26 a.m. Eastern before the opening bell, showed gold at 4,405.10 under that same GC=F label, alongside S&P 500, Dow, Nasdaq and Russell 2000 futures rather than any closing level. That the two GC=F readings differ is unremarkable, since they are eleven hours apart. What cannot be reconciled is either of them against the unlabelled figure in the body text. No gold price is printed in this article, because the only quotes that labelled their instrument cannot be squared with the one that did not.
This article is news and analysis, not investment advice. It does not recommend the purchase or sale of any security or commodity, and readers should consult a qualified professional about their own circumstances.
