How to Read Commodity Supply, Demand and Inventory Balance Sheets

Align unit and period, then close the quantity identity

A commodity balance sheet accounts for where available material came from and where it went. A global table commonly places beginning stocks and production on the supply side, with consumption and ending stocks on the use side. Country tables add imports and exports; processed-product tables may include input, yield and loss. A calendar year, crop year, marketing year, week and daily rate are not interchangeable, so read the header and footnotes before comparing rows.

If a table does not close, check unit, period, geography, rounding, raw-versus-processed material and the sign on trade. Energy analysis may infer a global stock change from production minus consumption; agricultural tables often calculate ending stocks as the residual between supply and use. An implied change is not the same as observed tank inventory. It can include statistical discrepancy and unobserved regions.

Normalize absolute stocks by use, capacity and season

An inventory figure of 100 is impossible to classify without the consumption rate. Agriculture often uses ending stocks divided by annual use; energy analysts use inventory divided by daily demand as days of cover. Storage capacity, minimum operating inventory and usable quality also matter. High aggregate stocks may provide little buffer if they are remote, off-specification or trapped behind limited withdrawal capacity.

Use several measures and compare each with its own history and seasonal band.

Year-over-year comparison aligns season better than the previous week, but a distorted base year can mislead. A five-year average can lag structural growth or new capacity. Place the level, annual change, normal deviation, historical percentile and rate of change together, and state which comparison supports the conclusion.

Do not mix estimates, forecasts and final observations

Commodity statistics combine surveys, administrative records, models, satellite observations and trade data, then revise them. Yield before harvest is a forecast; export sales, inspections, customs and vessel loadings refer to different stages. EIA weekly petroleum estimates and monthly data differ in coverage and method. Replacing every old value with the latest revision evaluates past decisions using information that was unavailable at the time.

Retain the value date, covered period, release date, vintage and retrieval date. For forecast-error work, store the initial forecast, pre-event update and final estimate as separate series. A decline in the forecast from last month is different from an inventory level below last year, even when both are described as “lower.”

Do not translate surplus and deficit mechanically into price direction

A surplus tends to build stocks and a deficit tends to draw them, but prices react to the surprise relative to expectations and to the system’s ability to adjust. A small surplus can be bullish relative to an expected large one. A deficit can have limited effect when stocks are ample, substitution is easy or supply can respond quickly.

Price also helps restore balance. Lower prices may remove high-cost supply and stimulate use; higher prices can encourage conservation, substitution, output and stock release. The response is delayed when crops require a planting season, mines require development or LNG needs infrastructure. Supply, demand and price therefore form a feedback system rather than a one-way chain.

• Expectation gap: distinguish the published result from what the market had already discounted.
• Buffer: assess usable inventory, spare capacity, substitution and policy stocks.
• Adjustment time: separate daily or weekly demand from seasonal and multi-year supply.
• Location: a balanced world total can coexist with a shortage in one region or grade.

Replace a point forecast with quantity ranges and disconfirming evidence

Recalculate a base case alongside supply-upside, supply-downside, demand-upside and demand-downside cases. Express each in quantities rather than adjectives: production plus or minus 2%, demand plus or minus 1%, an export bottleneck or an inventory-estimation error. Observe how ending stocks and stocks-to-use change. Do not stack every uncertainty in its maximum direction without evidence that they can occur together.

Write disconfirming conditions in physical variables rather than “the price fell.” Acreage, yield, refinery utilization, vessel loadings and inventory bands can each define a threshold that forces the balance to update. Price is both an outcome and a summary of new information, so an inconsistency between quantity scenarios and market spreads is a reason to investigate, not proof that either source is correct.

Turn a balance sheet into an updateable research ledger

1. 01Freeze definitions

   Record commodity, grade, geography, period, unit, source and seasonal adjustment.

2. 02Retain the prior vintage

   Keep initial, revised and final values instead of overwriting history.

3. 03Reconcile the identity

   Close supply and use, explaining discrepancy, rounding and implied stocks.

4. 04Normalize the level

   Add stocks-to-use, days, capacity, seasonal band and historical percentile.

5. 05Update cases and triggers

   Keep quantity assumptions, next releases, update thresholds and unresolved gaps.

The EIA board in Macro Research Workbench can place inventory levels and changes beside production, trade, refinery inputs and utilization. Financial Templates Hub can preserve versions, assumptions and review dates across USDA or EIA releases. Start with free inspection, then use Core for recurring saving and export or Global for custom data and point-in-time research only when that workflow is required.