Estimation Monthly Ethanol Consumption in the U.S.

1
Estimation Monthly Ethanol Consumption in the U.S.

• Carol Joyce Blumberg, Mathematical Statistician
• Michael R. Conner, Survey Statistician
• Susan Lord, Survey Manager
• Petroleum Division
• Office of Oil and Gas

2
Disclaimer

• This is a working document prepared by the Energy
  Information Administration (EIA) in order to
  solicit advice and comment on statistical matters
  from the American Statistical Association
  Committee on Energy Statistics. This topic will
  be discussed at EIA's Fall 2008 meeting with the
  Committee to be held October 23 and 24, 2008.

3
General Purpose of this Study

• Analysts (both inside and outside EIA) want quick
  access to how much ethanol is blended weekly into
  motor gasoline.
• However, EIA weekly surveys do not collect the
  volume of ethanol.
• It is only collected on EIA monthly surveys
• So, the big question that motivated this research
  is whether the methods used by analysts to
  estimate weekly volumes of ethanol are feasible.

4
Definition of Ethanol

• Ethanol, also called ethyl alcohol, pure
  alcohol, grain alcohol, or drinking alcohol, is a
  volatile, flammable, colorless liquid. It is a
  psychoactive drug, best known as the type of
  alcohol found in alcoholic beverages and in
  thermometers and its molecular formula is
  . (http//en.wikipedia.org/wiki/Ethanol)

5
Definition of Fuel Ethanol

• Fuel Ethanol--An anhydrous alcohol (ethanol with
  less than 1 water) intended for gasoline
  blending (EIA Glossary at http//www.eia.doe.gov/g
  lossary/index.html)
• Note 1 EIA studies Fuel Ethanol. But sometimes
  we say/write Ethanol when the context is clear.
• Note 2 Fuel Ethanol is considered a type of
  oxygenate.

6
Units of Measurement

• The units of measurement for this study are
  thousand barrels per month, which are the units
  used in the surveys and for reported Product
  Supplied.

7
Method 1 Definition

• Estimated Ethanol Consumption (using Product
  Supplied definition)
• (Net Production of Ethanol Imports of Ethanol)
  (Exports of Ethanol Stock Changes for
  Ethanol)
• Note The monthly data needed for Method 1 are
  available 60 days after the end of the month of
  interest

8
Some Notes on this Definition

• This method of estimating Ethanol Consumption is
  published in Monthly Energy Review
• The components of this definition are all public
  data available through Petroleum Navigator
• Net Production and Gross Production are the same
  for Ethanol
• Most of this consumption of fuel ethanol is
  reported as inputs by refiners and blenders. The
  balance is captured by an adjustment.

9
Ethanol Adjustments To Balance Supply and
Disposition

• Adjustments will result from errors in ethanol
  supply (production and imports), but most of the
  errors we discover are on the disposition side.
• We are unable to identify fuel ethanol exports in
  Census data. Hence, fuel ethanol exports are
  zero in the adjustment calculation.
• Some blending terminals are missing from the
  blender survey.
• Some blending terminals are out of scope.
• Less than 50,000 barrels total capacity
• Shipments and receipts limited to rail and truck.

10
U.S. Annual Fuel Ethanol Consumption (in thousand
barrels per month)
11
Surveys Used to Collect Monthly Ethanol Data

• Stock Change
• EIA-810 Monthly Refinery Report
• EIA-811 Monthly Bulk Terminal Report
• EIA-812 Monthly Product Pipeline Report
• Production and Stock Change
• EIA-819 Monthly Oxygenate Report
• Imports
• EIA-814 Monthly Imports Report

12
Examples of Population Definitions

• EIA-819all facilities that produce (manufacture
  or distill) oxygenates (including MTBE plants,
  petrochemical plants, and refineries that produce
  oxygenates as part of their operations located in
  the 50 States and the District of Columbia). In
  July 2008 there were 154 respondents, with some
  respondents reporting for more than one facility.

13
Population Definitions--Continued

• EIA-814 each Importer of Record (or Ultimate
  Consignee in some situations regarding Canadian
  imports) who imports crude oil or petroleum
  products into the 50 Statesthe District of
  ColumbiaPuerto Rico, the Virgin Islands and
  other U.S. possessions including into Foreign
  Trade Zonesin the 50 States and the District of
  Columbia, and from Puerto Rico, the Virgin
  Islands and other U.S. possessions into the 50
  States and the District of Columbia.
• Note Each shipment is recorded. In July 2008
  there were 30 shipments imported of Fuel Ethanol
  by 13 different companies.

14
Method 2

• Method 2 uses Production volumes of Motor
  Gasoline (with Alcohol) collected at the Weekly
  level separately for Conventional and
  Reformulated gasoline.
• Collected on a stratified random sample of
  facilities using two weekly surveys
• EIA-800 Weekly Refinery Report
• EIA-805 Weekly Terminal Blenders Report

15
Example of Sample Definition

• EIA-805Weekly Terminal Blenders Report
• Respondents are operators of blending plants
  selected by the EIA. Companies are selected into
  the EIA weekly sample according to a procedure
  that assures coverage of 90 percent of each
  information element.

16
Method 2--Continued

• Data are then converted to weekly estimated
  production totals
• Published in Weekly Petroleum Status Report
  (WPSR)
• The weekly volumes are then converted to
  Monthly-from-Weekly volumes as explained on next
  few slides.

17
Weekly Data Collection

• Data for the weekly surveys are collected on
  Mondays
• Reporting period begins at 701 a.m. EST on
  Friday and ends at 700 a.m. EST on the following
  Friday.
• For research purposes, however, we define a week
  as being from a Friday through the next Thursday.

18
MFW (Monthly-from-Weekly) Totals

• The MFW estimates are weighted totals of the
  weeks that contain the days of a certain month.
  For example, for July 2006, the weighted total
  was
• 6(data reported for 7/7/06)
• 7(data reported for 7/14/06)
• 7(data reported for 7/21/06)
• 7(data reported for 7/28/06)
• 4(data reported for 8/4/06)

19
Calendar for July 2006
20
An Advantage of Method 2

• Method 2 is preferred by analysts because the
  data needed are available within 12 days of the
  end of the month, as opposed to the 60 days with
  Method 1

21
Reformulated versus Conventional Gasoline

• Reformulated gasoline Finished gasoline
  formulated for use in motor vehicles, the
  composition and properties of which meet the
  requirements of the reformulated gasoline
  regulationsunder Section 211(k) of the Clean Air
  Act.  It includes gasoline produced to meet or
  exceed emissions performance and benzene content
  standards
• (EIA Glossary--http//www.eia.doe.gov/glossary )
• Conventional Gasoline Non-reformulated gasoline

22
Method 2 Definition

• Estimated Ethanol Consumption
• kMonthly-from-Weekly Reformulated Gasoline
  (with alcohol) volume Monthly-from-Weekly
  Conventional Gasoline (with alcohol) volume
• where k is a some constant between .09 and .10,
  depending on the analyst

23
Initial Questions

• What is the best value of k to use?
• Once you determine k, how close are the estimates
  given by Method 1 and by Method 2? If possible,
  determine which is better.
• How well do the estimates of volume, from the
  better of Method 1 and Method 2, match with total
  capacity per month for all ethanol facilities?

24
Important Aside--Percentage of Ethanol (Alcohol)
in Gasoline

• For most states, the percentage of ethanol in
  both reformulated and conventional gasoline is
  about 10
• California has only reformulated gasoline, which
  has about 6 ethanol
• Minnesota and a few other areas have small
  amounts of E-85, which is conventional gasoline
  with around 85 ethanol

25
Answering Question 1

• Question What is best value of k to use?
• Method Because of the disparity among
  conventional gasoline (with alcohol) and
  reformulated gasoline (with alcohol) in the
  expected percentage of ethanol, I proposed
  fitting a no-intercept model
• Method 1 Ethanol volume

26
Values Expected for Estimated ßs

• We would expect that would be somewhere
  around .086 since California (at 6 ethanol) in
  2007 had 34.94 of the reported gross production
  of the volume of reformulated gasoline (with
  alcohol), with the rest of the production at 10
  ethanol
• We would expect that would be a little bit
  bigger than .10 since some states have small
  volumes of conventional gasoline at about 85
  ethanol.

27
Results

• Estimated Ethanol Consumption
• .07473(MFW for Reform. Gas)
• .12993(MFW for Conv. Gas)
• using 42 months of data from January, 2005 to
  June, 2008
• R2 .9973

28
Actual by Predicted Plot
29
Graph of Residuals
30
Discussion of Residuals

• Notice that there seems to be a change in the
  predominant sign of the residuals around
  observation 20 (August 2006)
• So, we re-estimated the regression coefficients
  using only the last 23 months.

31
Regression Equation for Last 23 Months

• Estimated Ethanol Consumption
• .07134(MFW for Reform. Gas)
• .14068(MFW for Conv. Gas)
• using 23 months of data from August, 2006 to
  June, 2008
• R2 .9990
• Notice that these coefficients are further away
  from what we expected than when we used all 42
  months

32
Actual by Predicted Plot
33
Graph of Residuals
34
Conclusion so Far

• It appears that either the MFW volume of
  Reformulated Gasoline is too high OR volume of
  Conventional Gasoline is too low OR both.
• So, we now suspected the problem was in the MFW
  volume of Method 2 and not in the volumes of
  Method 1 as both internal and external analysts
  have suggested. We also suspect the problem is
  in the Conventional Gasoline volumes from looking
  at the data.

35
Next Step

• Second author pointed out that there are
• also unpublished data related to ethanol
  consumption.
• They are the volumes of Reformulated Gasoline
  (with alcohol) and of Conventional Gasoline (with
  alcohol) based on the monthly surveys used to
  produce Petroleum Supply Monthly and Petroleum
  Supply Annual, which are Censuses of appropriate
  companies.

36
Method 3

• Ethanol Volume
• Volumes here are for Gasoline (with Alcohol)
• Volumes are the final (revised) values

37
Values Expected for Estimated ßs

• We would expect that would be somewhere
  around .086 since California (at 6 ethanol) in
  2007 had 34.94 of the reported gross production
  the volume of reformulated gasoline (with
  alcohol), with the rest at 10 ethanol.
• We would expect that would be a little bit
  bigger than .10 since some states have small
  volumes of conventional gasoline at 85 ethanol.

38
Results

• Estimated Ethanol Consumption
• .08608(Month. Prod. Supplied for Reform. Gas)
• .10110(Month. Prod. Supplied for Conv. Gas)
• using 42 months of data from January, 2005 to
  June, 2008
• R2 .9995

39
Actual by Predicted Plot
40
Graph of Residuals
41
Possible Explanation for the Pattern in the
Residuals

• No data are collected on exports of fuel ethanol
  by either EIA or Census Bureau
• Census Bureau does collect data on exports of all
  types of ethanol combined
• Hence, in computing Product Supplied (Method 1)
  no exports are subtracted out
• But, exports of fuel ethanol appear to have only
  begun recently (anecdotal information)
• So, recently the Product Supplied volumes may be
  a little higher than they should be because of
  lack of a Census Bureau code for exports of fuel
  ethanol.
• Therefore, pattern in the residuals is as would
  be expected.

42
Final Conclusions from Regression Analyses

• The Product Supplied volumes of ethanol reported
  in the Monthly Energy Review (that is, Method 1)
  and unpublished Product Supplied volumes of motor
  gasoline with alcohol (that is, Method 3) appear
  to be extremely consistent.

43
Conclusions for Regressions (continued)

• Method 2, even the improvement of using separate
  multipliers for the two types of gasoline
  (reformulated and conventional), appears to be
  inconsistent with the other two methods (Method 1
  and modified Method 3).

44
Next Question

• This was our third question originally, but was
  modified based on results from the regression
  analyses
• How well do the estimates of ethanol volume using
  Method 1 match with facility capacities?

45
Methodology

• 1st Compared at the respondent level
  (unpublished confidential data) the reported
  ethanol volumes produced with the respondents
  capacities as known by EIA. Identified
  facilities/respondents with production greater
  than capacity by more than 5 consistently.
• 2nd Investigated the identified discrepancies by
  various methods described on next few slides

46
Determining Facility Capacities

• We do not ask for these capacities on any of
  EIAs surveys.
• Must find the capacities and changes in capacity
  by doing research
• Research is done by the third author using a
  variety of methods

47
Sources of Capacity Data

• Websites
• Renewable Fuels Association (RFA)
• http//www.ethanolrfa.org/industry/locations/
• ICM,Inc (design construct ethanol facilities)
• http//www.icminc.com/
• State websites (both government and ethanol
  associations)
• Respondent/Facility websites
• DTN Ethanol Center website (marketing information
  site)
• http//www.dtnethanolcenter.com/

48
Sources of Capacity Data (continued)

• Journals/Industry Publications
• Ethanol Producer Magazine
• DTN eNewsletter
• (Subscribe at http//www.dtnethanolcenter.com/)
• Personal Contacts by Email and Telephone

49
Sources of Capacity Data (continued)

• Other
• Comparison with a capacity list kept by Tancred
  Lidderdale of EMEU (Office of Energy Markets and
  End Use). Note The EMEU list is based heavily
  on third authors filebut, there are some
  instances where it has more up-to-date
  information.
• Air permit ceilings (maximum volume that
  facilities are permitted to produce)

50
Partial Answer to Question 3

• By using the above sources of information, we
  found that indeed the capacities of many
  facilities had increased. Reasons included
• Expansion
• Technology Improvements
• Tweaking
• Note Stated Capacity is a soft number since it
  is the average maximum barrels per month under
  assumed usual conditions. The number of days of
  operation, feedstock quality, etc. determine each
  months capacity deviation from Stated Capacity.

51
Final Conclusions

• The volumes reported in Monthly Energy Review are
  consistent with other monthly data.
• Once we complete our investigations of the
  discrepancies at the respondent level between
  production and capacity, we will have Total
  Capacity gt Total Production.

52
Recommendations

• 1. Create one of more new tables (in Petroleum
  Navigator) showing the following details
• A. Fuel Ethanol Supplied
• B. Refinery and Blender Production of gasoline
  with ethanol (broken down by reformulated and
  conventional gasoline)
• C. Adjustments
• D. Total Fuel Ethanol Consumption as presently
  reported only in Monthly Energy Review

53
Recommendations (continued)

• Reconsider methods of adjusting gasoline
  production in the weekly and monthly published
  volumes to make the two data systems more
  consistent. As part of this, make use of
  available weekly data on gasoline blending
  components when calculating the weekly
  adjustments.

54
Recommendations (continued)

• 3. Begin collecting weekly data directly on fuel
  ethanol (as is already planned for 2010)
• 4. Perform annual comparisons of fuel ethanol
  plant capacity data from both EIA and non-EIA
  sources with EIA fuel ethanol production data to
  identity cases where additional follow-up may be
  needed with survey respondents

55
Recommendations (continued)

• Add the collection of company/facility capacities
  to the present EIA-819 Monthly Oxygenate
  Report.
• These recommendations should be part of the
  on-going broader discussion concerning fuel
  ethanol data requirements.

56
Thank you

• Contact Person
• Carol Joyce Blumberg
• Petroleum Division (EI-42)
• Office of Oil and Gas
• Carol.Blumberg_at_eia.doe.gov
• Phone [phone]
• Fax[phone]