MnrAq

230 Third IMO GHG Study 2014
A series of reports, entitled Accuracy of Petroleum Supply Data, exists for EIA statistics that identify types of
error that may exist in US energy statistics (Heppner & French, 1996-2008; Heppner & Breslin, 2009). These
include:
1 Sampling error (difference between the sample estimate and the population value): this arises because
“surveys are administered to samples of the monthly populations to reduce respondent burden and to
expedite the turnaround of data” (Heppner & Breslin, 2009).
2 Non-sampling error (two types):
a
b
Random: “on average, and over time, values will be overestimated by the same amount they are
underestimated. Therefore, over time, random errors do not bias the data, but they will give an
inaccurate portrayal at any point in time” (Heppner & Breslin, 2009).
Systematic: “a source of bias in the data, since these patterns of errors are made repeatedly.”
The series of reports by EIA identified specific sources of uncertainty (non-sampling errors) that may include:
1 insufficient respondents coverage of target population;
2 nonresponse;
3 response error; and
4 errors due to lack of survey clarity.
The EIA report identifies imports and exports as statistics with greater uncertainty, similar to IEA. “Because of
the irregularity of imports for crude oil and petroleum products, the magnitude and range of percent errors
for both the MFW [monthly-from-weekly] and the PSM [petroleum supply monthly] imports numbers can
be expected to be much larger and wider than for production and stocks” (Heppner & Breslin, 2009). No
discussion assessing the accuracy of marine fuel statistics (domestic or international) is provided by EIA in
these annual reports. However, fuel totals are expected to exhibit similar or greater uncertainty to imports, for
reasons that IEA has identified in the QA/QC discussion.
For the Third IMO GHG Study 2014, the consortium specifically reviewed the 2009 report by Heppner and
Breslin, because it was the most recent such report we had obtained, and because it reported the US imports
percentage error for distillate and fuel oil in 2007 – a common year for both Second IMO GHG Study 2009
and Third IMO GHG Study 2014. (Each of these reports presents a running series of five years’ data, so this
report reported percentage error statistics on imports for 2003–2007.)
For US residual fuel oil imports, the EIA 2007 monthly-from-weekly (MFW) “range of percent errors was
57.38, ranging from -28.72 to 28.66 percent.” This error is much larger than the range of percentage errors for
production, or stocks, or even crude oil imports, which are all in the order of 10% or less. For example, “the
2007 range of the MFW percent errors [for fuel oil production], ranging between -5.16 and 3.86 percent, was
9.02”, and “the 2007 range (2.02) of the PSM percent errors [for fuel oil stocks], ranging from -1.84 to 0.18
percent, was the smallest range over the 5-year period”. The percentage error in monthly and annual statistics
for US distillate fuel imports was smaller than fuel oil imports, but bigger than error ranges for distillate
production, stocks, etc.
Analysis of US statistics provided two insights into our analysis of potential uncertainty in global top-down
inventories for shipping. First, imports and exports are confirmed as important sources of uncertainty even
for a nation with very good statistical data on its energy balances. Second, uncertainties surrounding different
fuel types can be dissimilar. We do not take any of the specific US calculations on percentage error to
represent global statistical error, nor do we imply that the analysis done by EIA represents IEA percentage
error. Moreover, we recognize that maritime bunkers (indeed international bunkers for aviation and marine)
are unaddressed in the US evaluation of accuracy of energy data. Combined, these two insights provide
independent evidence that import and export statistics can jointly contribute uncertainty in energy balances,
also identified as a potential uncertainty by IEA.