Why a 3.3% Flash Estimate Is Not a Price

On 1 September 2026 Eurostat, the statistical office of the European Union, published its flash estimate of euro-area inflation for August 2026. Annual inflation, measured by the Harmonised Index of Consumer Prices, is expected at 3.3 per cent, up from 2.9 per cent in July.

The components are not one story. Energy is expected at 14.3 per cent, up from 10.3 per cent in July. Services are expected at 3.0 per cent, down from 3.3. Non-energy industrial goods and food, alcohol and tobacco are both expected at 1.2 per cent. Every August 2026 cell in the release is flagged “e” for estimated. The complete HICP for August is scheduled for 17 September 2026. The next flash is scheduled for 2 October 2026.

What the Story Claims

The mainstream reading treats 3.3 per cent as the price of living in the euro area last month. Headlines will say inflation “is” 3.3 per cent and that a 2 per cent target is being missed. The release itself is more careful. It calls the print a flash estimate. It defines annual inflation as the change in consumer prices between the current month and the same month a year earlier, and monthly inflation as the change from July to August, expected at 0.4 per cent for the all-items index.

The basket is weighted. For 2026, services carry 468.2 parts per thousand of household monetary consumption. Non-energy industrial goods carry 252.2. Food, alcohol and tobacco carry 189.4. Energy carries 90.3 — less than a tenth of the basket — and still posted the highest annual rate. Strip energy out and the expected annual rate is 2.2 per cent, unchanged from July. Strip energy, food, alcohol and tobacco and the expected rate is 2.4 per cent, down from 2.5. The office can add those weights. Adding them does not discover what any household paid at the margin.

The implied claim is that a committee of statisticians has now priced August. A weighted average of collected prices is a compilation. It is not a market price.

The Austrian Diagnosis: Calculation Without Exchange

Ludwig von Mises (1881–1973), an Austrian economist who spent his career showing that centrally planned economies cannot work, argued in his 1920 essay “Economic Calculation in the Socialist Commonwealth” that without genuine market prices no one can tell whether resources are being used for more or less valuable ends. That is the calculation problem: a number can be computed without being a price. A price is the bid and the offer that clear a particular exchange. An index is an after-the-fact average of many such exchanges, plus the weights a statistical office chose for this year.

Eurostat’s own notes make the compilation visible. National statistical institutes collect prices. Eurostat compiles the euro-area aggregate. From 1 January 2026 the euro area includes Bulgaria, so the August flash is an EA21 series; data through December 2025 were EA20. Composition changes are incorporated with a chain-index formula. That is honest accounting. It is still not a price. No household in Sofia and no household in Helsinki transacted at 3.3 per cent. Lithuania’s expected August rate is 5.8 per cent. Estonia’s is 1.3. The headline is those national indices averaged with country weights from household final monetary consumption. Averaging does not make the underlying exchanges the same.

The energy print is the same cut in another register. Energy is 9 per cent of the 2026 basket and 14.3 per cent at an annual rate. Services are 47 per cent of the basket and 3.0 per cent. A planner who treats the headline as “the” inflation rate is treating a weighted residual as if it were the next litre of fuel. Mises’s point was never that statistics are useless. It was that calculation which does not rest on exchange cannot tell you which use of a scarce good is more valuable. A flash that moves four tenths because energy jumped cannot tell a baker in Lyon whether to bid for more electricity or more flour. Only the prices he actually faces can do that.

The office will publish the full August set on 17 September. Some cells will be revised. The “e” flags will drop. The methodological fact will not. The HICP remains a harmonised consumer-price index with base year 2025 equals 100, compiled under Regulation (EU) 2016/792. It is comparable across countries by design. Comparability is not a market-clearing price.

The Historical Parallel: A Target Invented on Television

This shape is older than the HICP flash. In 1989 New Zealand became the first country to adopt a formal inflation-targeting regime. Finance Minister Roger Douglas had floated a low range of 0–1 per cent in a 1988 television interview, hoping to anchor expectations while inflation still exceeded 15 per cent. Reserve Bank Governor Don Brash and his team then expanded the range. The 2 per cent figure that later became the global central-bank convention was not discovered in a market. It was a communications device that hardened into a target.

Chapter 2 of Rails to Freedom records that origin and the larger claim around it. Inflation, in the book’s usage, is not “rising prices” as a natural weather system. It is the expansion of money that later shows up in prices, a hidden tax that redistributes from savers to the issuer. A 2 per cent target treats a compiled index as the thing to be steered. Yesterday’s 3.3 per cent flash is the same manoeuvre from the other side: the index is treated as the price, and the gap to 2 per cent is treated as the policy error. Neither the target nor the flash was bid for. Both were posted.

Eurostat’s Statistics Explained page still points to News Release 113/2001 as the first flash estimates. The method has been refined for a quarter of a century. Refinement is not calculation in Mises’s sense. It is a better average of last month.

Why This Matters for Sound Money

Part 2 of Rails to Freedom, “The Bitcoin Bridge,” treats Bitcoin as digital gold: a scarce unit whose supply schedule does not require a statistical office to announce whether the unit held its purchasing power last month. Holders discover the trade-off in the next bid. That is the contrast with a euro whose official inflation number arrives as a weighted estimate, then as a mid-month revision, then as an input to a 2 per cent target that was itself a late-1980s communications choice. Sound money is not a prettier index. It is money that does not need an index to tell you whether it is still money.

What Markets Are Already Doing

Dai is a decentralised dollar-denominated stablecoin generated by the Maker Protocol. The protocol’s smart contracts run on Ethereum, the public, permissionless chain launched in July 2015 whose native asset is ether. Anyone who can post accepted collateral may open a vault, generate Dai, and later repay it. MKR token holders govern the parameters. The MakerDAO site still describes Dai as a currency that “can be used by anyone, anywhere, anytime.”

That is the earned contrast. Eurostat’s 3.3 per cent is a lagged, weighted average of collected consumer prices, flagged as an estimate until 17 September. Dai’s claim to hold a dollar of purchasing power is tested every time a vault is minted, every time a vault is liquidated, and every time Dai trades against other assets on Ethereum. If the claim is wrong, the next trade moves. If the flash is wrong, the mid-month release revises a cell. There is no September auction that subtracts phantom energy from a household in Vilnius that already paid the bill.

This is not a claim that Dai measures euro-area HICP. It is the narrower point the 1 September release invites. A statistical office can compile 21 national indices into one annual rate, chain-link Bulgaria’s accession, and print 3.3 with an “e”. A vault on Ethereum can be undercollateralised for a block, and then it is not, because the protocol sold the collateral. One operation is a bulletin. The other is a price.

Looking Ahead

Watch the 17 September full release for whether the energy cell or the country dispersion moves more than the headline. Watch whether coverage treats the revision as the discovery of August, or as the second pass of an average that was never a price. The 2 October flash will repeat the ritual for September.

Dai will not wait for the next Euro indicator. Vaults will keep being opened and closed on Ethereum. The next liquidation will keep being a function of collateral and debt, not of a 2026 weight of 90.3 parts per thousand. The office will keep compiling. Only one of those operations discovers whether the next unit was worth taking.