Why H.8's September Loan Average Differs From September 30

The Federal Reserve's 30 October 2026 H.8 release reports two different September loan balances in Table 2. The seasonally adjusted monthly figure is $14,100.0 billion. The Wednesday 30 September figure is $14,141.3 billion. Both describe loans and leases in bank credit across US commercial banks, in billions of US dollars.

H.8's September monthly loan level is a pro rata average of Wednesday observations, not the balance on 30 September. Subtracting the average from the final Wednesday gives $41.3 billion using the rounded published cells. That difference cannot be labelled lending added in the final week: its starting point is an average across the month.

This article was prepared on 31 October in Brisbane after Friday's scheduled NYSE core-equity session. The 30 October release's newest weekly observation is 21 October, so its publication date must not be used as the loan measurement date. No stock price, option quote or market reaction is used here.

Wednesday balances receive calendar-day weights

The Fed's H.8 methodology defines weekly levels as estimated balances at Wednesday's close of business. It constructs monthly levels from those weekly observations using a pro rata average. A monthly cell is therefore a calculated estimate based on Wednesday observations, rather than a collection of separately observed daily balances.

The technical explanation allocates each seven-day block ending on a Wednesday according to how many of its days fall within the month. Multiply each Wednesday balance by that day count, add the weighted balances and divide by the number of days in the month.

For September 2026, the block ending Wednesday 2 September contains two September dates, the 1st and 2nd. The blocks ending on the 9th, 16th, 23rd and 30th each contribute seven days. The weights are 2, 7, 7, 7 and 7, totalling September's 30 days. Giving every Wednesday the same weight would change the convention.

A fictional September makes the difference visible

Consider one invented bank-loan series with the following Wednesday balances. All figures are fictional billions of US dollars. They illustrate the weighting calculation and do not reconstruct the Fed's actual September estimate.

The weighted sum is 3,700 billion-dollar-days. Dividing by 30 gives approximately $123.33 billion. An equally weighted average of the five balances gives $120 billion; the final Wednesday balance is $140 billion.

In the fictional example, day weighting gives $123.33 billion, compared with a $120 billion simple average and a $140 billion final Wednesday balance. Those three answers come from the same invented observations. Each calculation gives a different role to the dates.

The final balance is higher because the invented series rises through the month. Different fictional paths could leave the average above the final Wednesday. An average-to-final comparison alone also cannot identify how much lending occurred during any particular week.

Our earlier claims-window example explains an equally weighted four-week average that changes when an observation leaves the window. H.8 adds a calendar-day allocation. The PCE measurement guide addresses another separate step, expressing changes over different periods at annual or monthly rates.

Preserve one release vintage

A correct weighting formula can still fail to reproduce a published cell if its inputs come from different versions of the series. The Fed explains that quarterly benchmarking updates estimation weights using newer bank reports. Revisions to underlying reports and seasonal factors can also alter past estimates. Its technical example warns that displayed weekly and monthly values may differ from published data because of revisions.

Use the monthly and weekly cells from the same dated release when comparing them. A reconstruction needs all relevant Wednesday observations, their calendar-day weights and the same seasonal-adjustment basis. The fictional calculation above establishes the method; it does not certify that five historical real-world cells have been recovered and reconciled.

The H.8 release calendar describes a usual Friday publication time of 4:15 p.m., with an exception for federal holidays. A usual schedule does not establish the actual publication second of the 30 October report. The dated HTML header establishes the vintage used here.

Keep the banking measure separate from option pricing

Interpreting the loan average correctly does not establish how a bank's shares or options should trade. The Options Industry Council's pricing guide identifies the underlying price, strike, time remaining, implied volatility, dividends and interest rate among premium inputs. This banking comparison estimates none of them. The implied-volatility guide explains the separate pricing concept.

The SEC's options-risk discussion warns that buyers can lose the entire premium and some writers face unlimited potential loss. Assignment can require the seller to fulfil the contract, including funding a purchase or delivering shares where applicable. Exercise, settlement and broker funding requirements need contract-specific checks. Concentration can increase portfolio losses; tax treatment requires separate advice.

OIC's liquidity explanation connects entry and exit conditions to bid-ask spreads and available size. Transaction costs reduce results, and a banking release cannot guarantee an executable option price. Options are unsuitable for some investors.

Options Matrix Pro publishes this general education and offers commercial options-analysis software. This article is not personal investment, financial, legal or tax advice and recommends no position or transaction. Sources were checked on 31 October 2026 Australia/Brisbane; the current-release address will roll forward, and the estimates can be revised.

If a research record labels H.8's September monthly cell as a 30 September closing balance, correct the field to a pro rata monthly average and keep the dated Wednesday observation separately.