Fifty-Third Week Storage Report Causes Historical Confusion
Though somewhat unremarkable in that it featured a 151 Bcf withdrawal that was within market expectations, the weekly storage report released last Thursday by the Energy Information Administration (EIA) drew attention from industry analysts and data crunchers alike because it was a calendar anomaly -- it was the 53rd weekly storage report for 2004.
In a normal year, there are 52 weekly observations and this makes for neat comparisons versus both last year's data and the oft-quoted five-year average analog. The report released Thursday, however, did not conform to that framework as it included data through Friday Dec. 31 -- the 53rd Friday in the calendar year of 2004.
Though the fact that there are not exactly 52 weeks in a year might seem like a small detail in the tracking of natural gas storage inventories, it makes the task of contrasting Thursday's report versus historical figures quite difficult. A week ago Thursday, the EIA reported that storage inventories as of Dec. 24 stood at 2,849 Bcf, which was 230 Bcf more than the year-ago tally and 358 Bcf more than the five-year average figure. But in the report issued last week, those differentials had narrowed significantly. At 2,698 Bcf, storage as of Dec. 31 stood only 79 Bcf more than the EIA's year ago comparison and 279 more than its five-year average figure.
At the heart of the situation is the EIA's practice of lining up historical comparisons beginning with the first report (based on survey end date, which is typically a Friday) of each year. In the rare years when there are 53 Fridays, the last report of the year receives special EIA treatment. Prior to 2004, the last time there were 53 Fridays in a year was 1999.
"Given that a year consists of 52 weeks plus one or two days, some years contain 53 Fridays," the EIA stated in an appendix to its methodology released Thursday. "This circumstance occurs only once every few years, so generally there is not a sufficient number of data observations in a given five-year period to average only observations for the 53rd week."
Without 53rd week data, the EIA went to an estimated year-end proxy for calculating the five-year average comparison. "As an approach to establish the average value when a 53rd Friday occurs, which occurs on either December 30 or 31, the five-year statistics are based on estimated values for corresponding dates during the five-year reference period. The calculations are based on a linear interpolation of the working gas inventory values between the last week of each year and the following week," the EIA note continued.
But the problem does not end there. Because the EIA starts each comparison year with the first survey end date (typically a Friday) in that year, you can wind up with weekly comparisons that are off by nearly a calendar week. Such will be the case in this Thursday's report, when data for the week ending Jan. 7 will be compared against data from previous years when the survey week ended Jan. 2 or Jan. 3 instead of Jan. 9 or Jan. 10.
However, a more profound problem may be not with this apparent misalignment, but rather that once released, the 53rd week data is not used in any subsequent historical comparisons published by the EIA. Moreover, under its current framework, the EIA will, for the rest of 2005, publish year-ago figures back 53 weeks rather than the usual 52 week jump in a typical year. It is this 53-week jump that is largely responsible for the apparent decline in the year-on-year storage surplus to just 79 Bcf from 230 Bcf the week prior.
Citigroup analyst Kyle Cooper -- who computes his own five-year average figure -- said, "I don't use the EIA five-year average for that very reason. I take the closest calendar day to the [EIA report ending date] for each of the previous years.... You could say that this is splitting hairs when you choose one report because it is three days off rather than another because it is four days off, but it is better than the six-day [misalignment] in the EIA methodology. Their way is not wrong, it is just a different way of looking at the data," Cooper reasoned.
EIA's Bill Trapmann defended the Administration's current methodology, while allowing that it is subject to continual review. "We considered a number of different ways of handling the historical data at the early stages of the report. In the end, we went with the way that would allow for transparency...and a methodology that would allow us to order the weeks in a simple fashion.
Trapmann noted that while the EIA reported the comparison to last year and the comparison to the five-year average along with its weekly updates, it also provides the market with the raw data, "so that the market can conduct its own analysis."
Following a barrage of public criticism over an erroneous EIA storage number and the subsequent handling of a large revision late last year, the EIA on Friday released its call for comment on its storage data revision policy (see related story this issue). And while noting that the solicitation does not specifically call for comments on the year-on-year and year-on-five-year comparison, Trapmann indicated Thursday that the administration is always accepting suggestions on any facet of the storage report, regardless of the vehicle in which they are communicated.