Why EIA's Net Crude Import Estimates Are Wrong And May Get Worse

by: Robert Boslego


Crude export ban lifted.

WTI – Brent spreads respond.

Exports based on defective model.

Crude supply/demand will not add up.

EIA audit reveals unknown source of stocks - to be adjusted in 2016.

After much debate, the forty-year ban on crude oil exports was lifted in December. The first crude oil shipment exported from the United States, left from Corpus Christi, Texas before year-end. The crude was sourced from the Eagle Ford Shale of South Texas.

Click to enlarge

The tanker Theo T was loaded Thursday with light crude for ConocoPhillips Co. at NuStar Energy LP's North Beach Terminal in the Port of Corpus Christi. PHOTO: EDDIE SEAL FOR THE WALL STREET JOURNAL

Given how controversial the ban has been, market observers have been anxious to monitor how much crude is being sold into the international market and what impact it will have. In the Energy Information Administration's (EIA) Weekly Petroleum Status Report (WSPR), it reports crude exports (Table 1, line 9, see here).

The lifting of the export ban appears to have had an impact on WTI - Brent spreads. Brent had been trading higher than the "land-locked" WTI. Earlier in 2015, WTI had traded as much as $14 per barrel below Brent. Once the export ban was lifted, WTI has been trading at a premium to Brent. This has been a great development for U.S. oil producers.

In addition, NYMEX futures prices reflect a premium for the balance of 2016.

Note: Based on closing prices January 22, 2016.

The EIA has been reporting the export number to be 500,000 b/d each week for the past six weeks in a row. The implication is that there has been no effect of the export ban being lifted.

But weekly crude exports are based on yet another EIA model. I had recently written about Why EIA's Crude Production Stats Are Flawed and the reason was that their estimates are based on a flawed model. The EIA exports model is also flawed, but for different reasons.

Crude exports are compiled by the U.S. Bureau of the Census and are published monthly. The EIA obtains this data approximately six weeks after the close of the reporting month. Weekly estimates of exports for crude oil are forecast using an autoregressive, integrated, moving-average (ARIMA) procedure. The weekly estimate is updated when a new monthly estimate is calculated.

The ARIMA procedure models a value as a linear combination of its own past values and present and past values of other related time series. The most recent five years of past data are used to obtain the exports forecast.

Because of the change in the export policy, the EIA will need to acquire data for five years based on the above description to use this procedure. However, this model will not produce reliable results because exports depend on price differentials with WTI crude. When the "arb" is open, exports will rise, and vice-versa. An ARIMA model, instead of one with WTI spreads, is pretty meaningless.


The EIA reports weekly crude stock changes based on a survey. Then it tries to "square" the stock change by computing a supply/demand balance. The EIA adds domestic production to net crude imports and subtracts refinery inputs. Because this difference does not equate to the crude stock change it computes, it adds or subtracts a "balancing item" (WSPR, Adjustment, Table 1, line 13).

Crude exports will be misstated, and therefore so will net crude imports. As actual crude exports begin to vary more in the future, this "balancing item" will likely swing widely, making it more difficult for the EIA to detect errors in their estimates, as described below:

Non-sampling errors may arise in the survey estimates from a number of sources including: (1) the inability to obtain data from all companies in the frame or sample (non-response and the method used to account for non-response), (2) response errors, (3) differences in the interpretation of questions or definitions, (4) mistakes in recording or coding of the data obtained from respondents, (5) data timing, and (6) other errors of collection, response, coverage, and estimation."

In June 2015, The Office of Inspector General responded to a complaint alleging that the weekly and monthly data and statistics reported by EIA had been inappropriately altered for the past 10 to 15 years. The complainant asserted that the data had been skewed by the practice of adding data from fictitious companies, called "adjustment companies." In addition, the complainant stated that the EIA inappropriately manipulated its results when the data did not come out to its liking.

The result of an audit revealed that two adjustment figures related to lease stocks data used for weekly reports had not been amended in 30 years. These were adjustments of 10,300,000 and 330,000 barrel estimated figures used to correct for the incomplete coverage of companies that store crude oil on leases. EIA officials replied that they were unable to confirm the validity of these adjustments. In fact, these figures have been included in the WPSR for so long that current EIA officials stated they had neither any idea what the figures were based on, nor how they were originally calculated.

EIA agreed to add an additional footnote (see pages 38 and 39) to the crude oil tables in the WPSR to make the nature of this estimated adjustment more explicit. Additionally, EIA reiterated its plan to stop reporting lease stocks data and to eliminate the estimated lease stocks adjustment in 2016. Stay tuned.


The task of computing statistics for the WSPR is already challenged. The domestic production model has been shown to be deficient, given the changing economics of shale oil and the large swings in oil prices.

With the lifting of the crude export ban, the calculation of exports using current procedures will yield unreliable "net crude import" estimates, further compounding the problem of assessing whether crude stocks make sense each week. The EIA probably needs to change the form EIA-803 to include exports, in addition to imports.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.