September Manufacturing: Why 98.4 / 98.2 Misses 0.3%

September Manufacturing: Why 98.4 / 98.2 Misses 0.3%

The Federal Reserve reports that manufacturing output rose 0.3% in September. Its 16 October G.17 release also prints manufacturing index levels of 98.2 for August and 98.4 for September. Divide those two levels and the result, rounded to one decimal place, is 0.2%.

That difference has a documented explanation. The Fed's industrial-production notes say its published percentage changes use unrounded indexes. A spreadsheet using the printed levels has already discarded information that the official growth calculation retains.

Keep the 0.3% released rate. Label the spreadsheet result by the precision of its inputs rather than silently replacing the source figure.

What the October release prints

The manufacturing row in the release's seasonally adjusted summary uses 2017 equal to 100. These three entries belong to the same October release vintage:

August 2026 manufacturing index: 98.2, marked revised.

September 2026 manufacturing index: 98.4, marked preliminary.

September monthly manufacturing change: +0.3%, also preliminary.

The calculation from the displayed levels is 100 x (98.4 / 98.2 - 1) = 0.2036659...%. That rounds to 0.2%. It is valid arithmetic on rounded inputs. It does not reproduce the calculation from the Fed's underlying unrounded indexes.

Subtracting 98.2 from 98.4 also gives 0.2, but in index points. That subtraction uses a different unit from percentage growth, which divides the change by the earlier level. The near match between those two results here follows from an earlier index close to 100.

A difference between the two rates alone does not establish a publication error.

Two fictional records with identical printed levels

Assume two invented records for a single index across consecutive months. Each record uses the same series, adjustment treatment, base year and release vintage in both months. Round index levels to the nearest tenth for display. Calculate growth from the original inputs, then round that growth to the nearest tenth of a percent. None of the selected numbers lies on a rounding midpoint.

Record A uses an earlier level of 98.16 and a later level of 98.44. They print as 98.2 and 98.4. Growth from the original inputs is 100 x (98.44 / 98.16 - 1) = 0.2852485...%, which prints as 0.3%.

Record B uses an earlier level of 98.24 and a later level of 98.36. Those also print as 98.2 and 98.4. Growth is 100 x (98.36 / 98.24 - 1) = 0.1221498...%, which prints as 0.1%.

Both records would give 0.2% if growth were recalculated from their identical displayed levels. The original-input calculations give different rates because rounding the levels first removes information before division.

These are fictional author calculations, not estimates of the Fed's hidden decimals. Record A demonstrates that the published combination is mathematically possible. It does not reveal which unrounded values produced September's actual result, and Record B does not challenge the released rate.

Keep precision separate from revision

Rounding changes how a value is displayed. A revision changes the estimate. The Fed's methodology notes describe an initial preliminary estimate followed by revisions as more source data arrive. The September reference month therefore differs from the October publication month and from the eventual history a later download may contain.

This article uses August and September as printed together in the 16 October edition. It does not mix a September-release August value with the new September value. It also does not retrieve the full-precision series or claim to reproduce the official rate from those data. A fuller replication would require matching the source's exact series, precision, adjustment treatment and vintage.

The release PDF's first page states 9:15 a.m. Eastern Daylight Time on 16 October 2026, or 11:15 p.m. that day in Brisbane. The official release-date page identifies 16 October as the latest monthly edition. This is the publisher-stated release minute; an independent first-publication second was not measured.

Preparation was at 07:05:15 Australia/Brisbane on Saturday, 17 October. Friday, 16 October was the latest completed relevant Cboe C1 regular options session. Cboe's dated statistics showed a populated 3:15 p.m. Central row, consistent with the 4:15 p.m. Eastern regular-session end. Those observations establish the session boundary only. No market-price response is measured here.

Preserve the source value before researching a contract

A spreadsheet discrepancy can start with display precision. Resolve that issue before using the number to compare releases. A manufacturing-output change alone supplies no company earnings estimate, stock direction, implied-volatility forecast or options valuation.

Contract research needs its own underlying, strike, expiration, multiplier, exercise and settlement terms, plus a timestamped bid, ask and available size. The options-chain guide covers that record. The liquidity guide explains why a displayed midpoint does not promise a fill.

Options Matrix Pro publishes this education and offers commercial options-analysis software. This is general education, not personal financial, legal or tax advice or a recommendation to trade. Buyers can lose their premium. Short positions can produce larger losses, including unlimited loss on an uncovered call. Exercise, assignment and settlement can require funding beyond an option premium; liquidity, transaction costs, taxes and concentration remain separate considerations. Options are unsuitable when the investor cannot tolerate the relevant losses or meet those obligations. Read FINRA's options guidance and the OCC disclosure gateway.

Sources and method

Sources were accessed on 17 October 2026 Australia/Brisbane. The Fed's current October release supplies the released entries; its PDF supplies the printed release minute. The methodology notes explain precision and revisions. The release calendar, Cboe statistics and hours supply date and session context. FINRA and OCC supply risk references. The Fed's current-release links roll forward, so retain the stated October vintage when checking this example later.

For this worksheet, enter 0.3% as the official September manufacturing change and 0.2037% as the separately labelled calculation from displayed index levels. Keep the 16 October vintage beside both.