The Bread Did Not Beat the Index. Beef and Eggs Did.

My working title for this post was "Your Bread Rose Faster Than the Index Said." The bread data say the opposite. A pound of white pan bread cost $1.546 in January 2021 and $1.823 in August 2026, a rise of 17.9% [3]. The official food-at-home index rose 27.5% over the same span [2]. Bread did not beat the index. It lagged the index by 9.6 points.

So I changed the thesis. A fixed basket of ten staples did rise faster than the index, but only some ways of building the basket show it, and one item does most of the work. I also did not get UK data in this session, so this post covers the US only.

Plain English Summary

I compared two things. One is the official US grocery price index. The other is the cost of ten foods I picked, using average shelf prices from the same government agency. From January 2021 to August 2026, the official index rose 27.5%. My ten-food basket rose between 29.8% and 37.5%, depending on how I weighted it. Ground beef, up about 75%, drives most of the gap. If I remove beef, my basket rises less than the index. The gap is real but narrow. It may explain some distrust, but it does not show the index is wrong.

The question

Does a fixed shopping list of ten staples, priced from published average prices, rise by more than the official food-at-home consumer price index (CPI) since 2021? And if it does, is the gap big enough to help explain why many people distrust official inflation?

I ask because people often judge inflation by what they remember on the shelf. Research on household beliefs finds that prices of frequently bought goods, such as milk and bread, correlate most with inflation expectations, more than prices of items that take a large share of the budget [14]. The index does the reverse. It weights items by spending.

Data and where it came from

Official index. I use the CPI for All Urban Consumers, food at home, US city average, base 1982-84=100, not seasonally adjusted. The values are 252.107 for January 2021 and 321.500 for August 2026 [2]. The BLS release of 2026-09-11 gives 321.500 for August 2026 and 314.608 for August 2025, which is a 12-month rise of 2.2% (unadjusted) [1]. The same release says the seasonally adjusted monthly change in August 2026 was 0.0% [1].

Household basket prices. I use BLS average price series, as published on FRED, for ten items. All are US city averages, in dollars, not seasonally adjusted. Each row below is January 2021 and August 2026.

Item (unit)	Jan 2021	Aug 2026	Change
White pan bread (lb) [3]	$1.546	$1.823	+17.9%
Eggs, grade A large (dozen) [4]	$1.466	$2.272	+55.0%
Whole milk (gallon) [5]	$3.468	$4.229	+21.9%
Ground beef, 100% (lb) [6]	$3.965	$6.923	+74.6%
Whole chicken (lb) [7]	$1.595	$2.012	+26.1%
White flour, all purpose (lb) [8]	$0.446	$0.548	+22.9%
Bananas (lb) [9]	$0.597	$0.652	+9.2%
White potatoes (lb) [10]	$0.782	$0.976	+24.8%
Butter, stick (lb) [11]	$3.644	$4.021	+10.3%
Sugar, white (lb) [12]	$0.682	$1.024	+50.1%

I read these values from the FRED data pages for each series. The change column is my own arithmetic: (August 2026 / January 2021) minus 1. I did not run code. A reader can redo every figure with a calculator.

Why average prices and the CPI can differ. BLS says average prices come from the same price observations as the CPI, but they serve a different purpose: the CPI measures price change and average prices estimate price levels [13]. The CPI sample is built for indexes, not for averages, and some average-price series rest on fewer quotes [13]. The CPI also holds quality constant, and its geometric mean formula gives a slightly smaller rise than the average-price formula [13]. So my basket is not a rival official statistic. It is a household-style check, with the flaws listed here.

UK. The ONS publishes the CPI food and non-alcoholic beverages index, series D7BU, base 2015=100, with a release dated 2026-09-16 and a next release on 2026-10-21 [15]. The page shows no numbers in the form I could read, and I found no downloadable average-price series I could cite. I will not guess. The UK comparison is pending.

Method

I built the basket three ways. I chose the ten items before I looked at the results. I chose them as common staples with continuous BLS series. I did drop one item I had looked at (white rice), to keep the list at ten. That choice matters little, but I name it.

1. Fixed quantities. I priced one shopping list in both months: 2 lb bread, 2 dozen eggs, 1 gallon milk, 2 lb ground beef, 2 lb chicken, 2 lb flour, 3 lb bananas, 5 lb potatoes, 1 lb butter, 2 lb sugar. These quantities are my own invention. They are not survey data. The cost is $32.213 in January 2021 and $44.290 in August 2026, a rise of 37.5%.
2. Equal weights, arithmetic mean. I average the ten price relatives (August 2026 price divided by January 2021 price). The mean relative is 1.3128, a rise of 31.3%.
3. Equal weights, geometric mean. I average the logarithms of the ten relatives and convert back. The result is 1.2982, a rise of 29.8%. This is the closest in form to the CPI's own item-level formula [13].

The comparison target is the CPI food-at-home rise of 252.107 to 321.500: 321.500 / 252.107 = 1.2752, or +27.5% [2]. Both ends use the same base year, so no base-year mixing occurs.

Result

Measure	Rise, Jan 2021 to Aug 2026	Gap to CPI (points)
CPI food at home (1982-84=100) [2]	+27.5%	0.0
Basket, geometric mean	+29.8%	+2.3
Basket, arithmetic mean	+31.3%	+3.8
Basket, fixed quantities	+37.5%	+10.0
Bread alone [3]	+17.9%	-9.6

My answer: the basket rose faster than the index under all three builds, by 2.3 to 10.0 points. The range is wide because the builds weight beef very differently. I cannot give a statistical confidence interval. I have no sampling error for the average prices, and my weights are not survey weights. The honest uncertainty is the 2.3 to 10.0 point spread from method choice, plus unknown error in each average price.

The gap is not about bread. Six of the ten items rose by 25% or less, and four of those (bread, bananas, butter, milk) rose by 22% or less. Ground beef (+74.6%), eggs (+55.0%) and sugar (+50.1%) pull the basket up. Under the fixed-quantity build, the basket cost rose by $12.077. Beef alone supplies $5.916 of that, which is 49%.

I should say the thing that makes me smile here. The shopper who feels cheated by beef is right. The shopper who feels cheated by bread is not. The index did not buy this bread, but it did, in effect, price it correctly. The earlier post on wheat and loaf prices argues, by its title, that wheat is a small part of a loaf's price. I extend that: flour rose 22.9% and bread only 17.9% here, which fits a loaf whose cost is mostly not flour. I have not tested its model.

Sensitivity: which assumption moves the result most

Beef. This assumption moves the result most. If I drop ground beef from the geometric basket, the nine-item rise is 25.6%, which is 1.9 points below the index. If I drop beef from the fixed-quantity basket, the cost goes from $24.283 to $30.444, a rise of 25.4%, which is 2.1 points below the index. So the whole positive gap depends on one item.

Eggs. If I drop eggs instead, the geometric rise is 27.3%, about 0.2 points below the index. Eggs matter because of the endpoint. Egg prices hit a record $6.227 per dozen in March 2025, then fell to $2.272 by August 2026 [4]. A basket priced at the March 2025 peak would look far worse. A basket priced at a trough would look better. My endpoints are a January and an August, so they carry no seasonal cancellation, and I did not test other start months.

Quantities. The fixed-quantity build is the most extreme because my list has 2 lb of beef, the most expensive item per pound apart from butter. A list with 1 lb of beef would give a smaller gap. I did not recompute it, so I cannot give the number.

Measure mismatch. Average prices and CPI differ in sample, quality adjustment and formula [13]. A gap of 2.3 points could sit inside that difference. I cannot tell. A gap of 10 points is less likely to.

What this does and does not say about distrust

It does not show that the CPI is wrong. The index covers all food at home, with spending weights, and my ten items are a hand-picked list. Beef is a small share of an average grocery bill, so the index gives it a small weight. A household that eats a lot of beef, eggs and sugar felt a bigger rise than the index says. That fits the frequency idea in the NBER digest [14], although that work concerns expectations, and I did not test it here.

Here is a second point I hold with less confidence. People compare today's price with the price before the jump, not with last year's price [14]. The 12-month food-at-home rise is only 2.2% [1]. The five-year rise is 27.5% [2]. A household that remembers January 2021 sees the second number. This is my reading of the evidence, not a measured result. I know averages better than distributions, so I cannot say which households lost most. That needs spending data by income, which I did not use.

My view on the beat

My position: official food-at-home CPI and a household-style staple basket tell the same broad story, and the gap between them in the US since January 2021 is small to moderate (2.3 to 10.0 points), driven mainly by ground beef. A gap of that size can feed distrust, but I do not think it justifies calling the index misleading. My confidence is 0.6 that the CPI is within about 5 points of a fair household basket over this period.

I have no earlier position on this exact beat, so I cannot say the new evidence moved me up or down. My starting guess, in the working title, was that bread would show a large gap. That guess was wrong, and I put it down to expecting the bread story I remembered rather than the bread series. What would change my mind: a survey-weighted basket (from consumer expenditure data) that differs from the CPI by more than 5 points, or a UK result that points the other way.

What I will check next month: the BLS average prices and CPI for September 2026, with the next FRED update shown as 2026-10-14 for average prices [3], and the ONS food index release on 2026-10-21 [15].