(John Kemp is a Reuters market analyst. The views expressed are his own)
* Chart 1: tmsnrt.rs/2bVNOQh
* Chart 2: tmsnrt.rs/2bVO2Xm
* Chart 3: tmsnrt.rs/2bVNHnH
* Chart 4: tmsnrt.rs/2bVOqVK
* Chart 5: tmsnrt.rs/2bXyrnO
By John Kemp
LONDON, Sept 1 (Reuters) - U.S. motorists used a record volume of gasoline in June as good weather and cheap fuel encouraged a strong start to the summer driving season.
The volume of gasoline supplied to domestic consumers averaged 9.664 million barrels per day, an increase of 273,000 bpd compared with the same month in 2015 (tmsnrt.rs/2bVNOQh).
Consumption beat the previous record of 9.640 million bpd set in July 2007, according to monthly estimates published by the Energy Information Administration (EIA) on Wednesday.
Gasoline consumption has been rising strongly since 2013, and especially 2014, reversing the previous downtrend since 2007.
Strong economic growth and increased employment have coupled with a substantial decline in gasoline prices to stimulate record fuel consumption.
Gasoline demand shows a clear influence from the weather in the first half of 2016 and mirrors statistics on traffic volumes and national park visits (tmsnrt.rs/2bVO2Xm).
Park visits are one proxy for leisure-related journeys, which boost traffic volumes and fuel use, especially during the summer driving season (tmsnrt.rs/2bVNHnH).
Unusually warm weather in February and March spurred large increases in driving, park visits and fuel consumption. Cooler temperatures in April and much of May brought slower growth.
Temperatures climbed pleasantly above normal in June and were accompanied by faster growth in traffic volumes, park visits and gasoline consumption.
The EIA has also implemented a new system for calculating weekly petroleum exports and consumption that should significantly improve data quality.
The agency has always estimated the volume of gasoline and other petroleum products supplied to the domestic market by analysing the volume of fuel that disappears each week and month from primary reporting sources.
The volume of each product supplied is equal to domestic production plus imports minus exports and minus stock changes (product supplied = production + imports - exports - stock change).
The agency collects data on production, imports and stocks directly through compulsory weekly and monthly surveys sent to the industry.
But data on exports comes indirectly from the shippers’ export declarations (SEDs), now renamed electronic export information (EEI), made to U.S. Customs.
Customs data is analysed and compiled by the U.S. Census Bureau and published only monthly with a delay of six weeks after the month in question.
The EIA has always used the monthly export numbers to calculate its monthly values for the volume of petroleum products supplied to the domestic market.
But as a result of the publication lag, the agency has been forced to extrapolate weekly export numbers based on the last available monthly figure from the Census Bureau adjusted by seasonal trends in previous years.
The system worked well when the volume of fuel exported was small and changed little from one week or month to the next.
But as export volumes have climbed and become more variable the potential for estimating errors in the weekly export numbers has increased.
Errors estimating exports lead to equal and opposite errors estimating domestic consumption through the formula employed by the EIA.
However, under a new system agreed with U.S. Customs and the Census Bureau, the EIA is now receiving raw EEI export data and using it to calculate export volumes in near-real time.
The use of raw EEI numbers should significantly improve data on exports and by extension on the volume of products supplied to domestic customers.
The EIA has been developing the system since 2013 and trialling it internally to validate its accuracy (“Statistical methodology of estimating petroleum exports using data from U.S. Customs and Border Protection”, EIA, August 2016).
The new methodology should lead to welcome improvements in the accuracy of weekly statistics on fuel exports and consumption.
But it is important not to overstate the errors that occurred under the old system or the improvements that will be made under the new one.
Some analysts have obsessed over the small differences between weekly and monthly estimates for U.S. consumption of gasoline and other refined fuels.
In reality, the differences belong to a category Nobel physics laureate Max Planck described as phantom problems “void of all meaning”.
“All that mental work and effort was expended on a mere nothing” (“Scientific autobiography and other papers”, Planck, 1950).
Differences between the weekly and monthly consumption numbers have been small and the two series have generally tracked each other with a very good degree of accuracy (tmsnrt.rs/2bVOqVK).
For example, weekly data showed U.S. gasoline consumption averaging around 9.760 million bpd during June compared with a monthly estimate of 9.664 million bpd.
The difference between the two estimates was less than 100,000 bpd, or below 1 percent of domestic gasoline consumption.
Weekly consumption estimates have been a fairly good predictor for the monthly demand estimates published later.
In any event, errors in estimating consumption have been matched by equal and opposite errors in estimating exports.
To the extent that oil traders, investors and analysts are interested in gasoline or oil demand, the errors offset one another.
In the first half of 2016, U.S. domestic gasoline consumption was generally overestimated but U.S. gasoline exports were underestimated.
In the vast majority of cases, gasoline was exported to foreign customers who used it, rather than into long-term storage (storage options are generally cheaper and more convenient in the United States).
In the case of gasoline, U.S. refiners significantly boosted exports to Mexico and other destinations in Latin America, reflecting limits on local refiners’ ability to meet growing demand there.
If gasoline consumption was marginally weaker at home than the weekly numbers implied, consumption in Mexico and elsewhere was stronger by an offsetting amount.
The problems with the old methodology affected the allocation of consumption between domestic and overseas consumers but not the overall level of demand.
The bottom line is that during the first half of 2016 the United States consumed a seasonal record volume of gasoline AND exported a record volume of gasoline and blending components to customers overseas (tmsnrt.rs/2bXyrnO). (Editing by Dale Hudson)