EIA does not directly survey petroleum consumption; instead, we calculate a similar concept called product supplied from our surveys that span the U.S. petroleum industry. Although we use petroleum product supplied interchangeably with petroleum consumption, the two are not identical.
Consumption is the same as purchases by end users at the final point of sale, such as gasoline from the pump. We survey 1,000 of the more than 100,000 U.S. retail gasoline stations to estimate retail gasoline prices. We gather information about motor gasoline and other petroleum products by surveying a more limited number of respondents in the primary supply chain—refiners, blenders, importers, pipelines, and bulk terminal storage operators—before the products are moved to distributors and retailers. By surveying the entire population of these primary suppliers, we determine the volume of refined petroleum inputs and production, imports, inventories, and shipments available for domestic consumption. Most petroleum products leave the primary supply chain when they are sold to distributors, retailers, and end users and are consumed not long after they are produced and moved through primary suppliers.
We use product supplied as a proxy for consumption because it measures volumes of petroleum products that leave the primary supply chain prior to distribution. Using the example of finished motor gasoline, we define product supplied as:
We calculate that, in 2023, finished motor gasoline product supplied in the United States came from 9.65 million barrels per day (b/d) of finished net production by refiners, blenders, and biofuels plants plus 118,000 b/d of imports. Disposition of finished motor gasoline included a 3,000 b/d increase in primary inventories and 816,000 b/d of finished motor gasoline exports. Disposition accounts for how crude oil and petroleum products are transferred, distributed, or removed from the supply stream, including stock changes, refinery inputs, exports, and products supplied for domestic consumption.
We also include a supply adjustment, a balancing item that reconciles product supplied of finished gasoline with the supply and disposition of its two major components, motor gasoline blending components (MGBCs, derived from petroleum) and fuel ethanol (a biomass-based fuel typically derived from corn in the United States). MGBCs and fuel ethanol are not used in engines individually, but they are blended into gasoline to meet specifications for octane, sulfur levels, and other standards. Because we collect supply and disposition of these products from several surveys, survey and statistical errors lead to natural differences between supply and disposition. The adjustment serves as a balancing item, which reduced finished motor gasoline supply by 7,000 b/d in 2023.
Taken together, 8.95 million b/d of finished motor gasoline was consumed in the United States during 2023 because this was the amount that left the primary supply chain and remained in the United States.
We publish our most definitive set of finalized petroleum statistics in the Petroleum Supply Annual. This report revises any errors or missing data for the previous calendar year. We release this report every August.
Our Petroleum Supply Monthly (PSM) provides comprehensive and more timely petroleum statistics. We release the PSM report on the last business day of the month, and it provides statistics for the month two months before. The PSM report derives its data from surveying all companies that produce, import, transport, or hold in storage any petroleum product in the United States. We do not survey exports but instead receive export statistics from the U.S. Census Bureau.
Our Weekly Petroleum Status Report (WPSR) provides even more timely but less comprehensive information than the PSM because the WPSR surveys fewer respondents. Whereas the PSM surveys the entire primary supply chain, the WPSR surveys the largest respondents only, covering around 90% of the latest monthly volume for each supply source. The WPSR product supplied data can experience large week-to-week swings, partially because of sampling errors from surveying only a subset of the PSM respondents and partially because imports or exports can quickly change depending on when cargoes clear U.S. Customs. We recommend data users analyze the four-week moving average of product supplied to better understand underlying trends in U.S. petroleum consumption.
Because calculating product supplied involves several components, each component can be subject to sampling error, measurement error, and timing differences. Sampling errors can occur for each supply or disposition component and can result in differences between the WPSR estimate and the reported product supplied for the month in the PSM. Differences can also arise because we estimate weekly export data for petroleum products using unedited weekly export statistics from U.S. Customs and Border Protection. Usually, the WPSR product supplied is within +/-2% of the PSM. Over longer periods, these differences can average close to 0%, as they did from 2018 through June 2024.
We rely on PSM and WPSR data for other EIA publications and forecasts. We use the terms petroleum consumption and petroleum product supplied interchangeably in these reports, such as the Monthly Energy Review, Short-Term Energy Outlook, and Annual Energy Outlook, to label the data consistently with other fuel consumption such as natural gas, coal, nuclear, renewables, and electricity.
Absent significant errors, omissions, or natural disasters, we never revise the data published in the WPSR. Historical WPSR estimates of product supplied become less relevant when we publish the PSM because we capture the remaining data that we did not survey in the WPSR. PSM data can serve as a definitive historical benchmark and the WPSR as a near-term estimate of recent history. The unrevised historical WPSR data are still useful, however, particularly for users that wish to understand weekly variability in petroleum market activity among all primary supply chain components.
The Federal Highway Administration publishes monthly data on vehicle miles traveled (VMT) and a motor fuels report that displays motor gasoline sales by state. Although the report on VMT is published with a similar lag as the PSM (about two months), data users would have to make assumptions about vehicle fuel efficiency to determine motor gasoline consumption. The motor fuels report tracks motor gasoline consumption closely with the PSM but can lag between six and nine months. Commercial data providers such as Oil Price Information Service (OPIS) and GasBuddy offer estimates of gasoline consumption from retail gasoline stations they sample.
Principal contributor: Jeff Barron