Data Quality and Verification

1
Data Quality and Verification

• Office of Highway Policy Information
• Federal Highway Administration

2
Why Verify

• Verification versus Reviews
• MF reviews look at process
• Verification looks at specific numbers
• Old system is hand-entry of the data
• Nobodys perfect
• Motor Fuels team checks and re-checks
• Errors still possible

3
Correspondence

• Verification memo
• Dated
• Response due April 15
• Virtually assured data will change
• Second round
• If no changes in the first round FYI
• If changes in the first round re-check

4
How to evaluate

• Compare State 551M monthly to the MF-33GA (gas
  and gasohol) and the MF-33SF (diesel and other
  special fuels)
• Sum all the months in the year and check against
  the MF-21
• Gasoline Total Consumption column
• Special Fuel Private and Commercial Highway Use
  column
• Note there are reasons these dont always match

5
Table MF-21- Motor Fuel Use

• Summary of the FHWA annual analysis
• Draft attribution data
• Available for review and correction
• Becomes the MF-27 (finalized)
• As MF-27, used to attribute federal highway
  revenue to the States

6
MF-21 A Closer Look - I

• Starting point two-thirds over from the left
• Special fuels column
• Data taken from State data
• Removed all non-highway gallons
• Thats it for special fuels
• Gasoline
• Remember gasoline/gasohol combined
• Start at column Total Consumption agrees with
  total on MF-33GA

7
MF-21 A Closer Look - II

• Next column to the left Losses
• Depends on State legislation
• None
• Flat percent losses
• Actual losses
• Complicated combinations
• Limited to one percent of total consumption plus
  actual losses
• New procedures in place

8
MF-21 A Closer Look - III

• Non-highway Use
• Two non-highway types
• Private and Commercial
• Public
• Two treatments
• PC either State or model data
• Public is all model results
• Breakdown by PC type in Table MF-24

9
MF-21 A Closer Look - IV

• Public on-highway use
• Two types
• Federal civilian use
• State, County, Municipal, and Native American
  government use
• All data generated from FHWA models
• Note Total Highway Use column is combined
  gasoline\gasohol used in attribution

10
Summary

• Verification
• Is different than Process Review
• Final State data for attribution
• Should not substitute for checking the data
• Month-by-month
• With future techniques

11
Modeling Data

• Why FHWA models data
• Consistency among States
• Data not otherwise available
• Model off-highway gasoline
• PC gasoline off-highway
• Public gasoline off-highway
• Relatively small impact

12
Modeling Data

• Gasohol Model
• Relatively large impact
• Why an issue
• Federal/State definition differences
• States dont report by federal definition
• Attribution impact
• Federal revenue
• Incentives

13
Modeling Gasohol

• How the current model works
• IRS reports total gasohol revenue
• States providing reasonably good gasohol data are
  allocated their share of gallons derived from IRS
  data
• States with no gasohol are allocated zero gallons
  of gasohol
• States not in either of these categories are
  allocated the remaining gallons derived from

14
Modeling Gasohol

• Regression analysis
• Dependent variable number of ethanol gallons
  consumed
• Independent variables
• Ethanol plant proximity
• Producers incentive amount
• VMT in State non-attainment areas proportionate
  to State VMT
• Blenders incentive amount
• Total gasoline consumed in the State

15
Modeling Gasohol

• Sum the estimated amounts derived from the
  regression calculation
• Derive each estimated amounts percentage of the
  total estimated
• Multiply these percentages by the IRS remaining
  control total
• Derive the distribution for each States three
  levels of gasohol

16
Modeling Gasohol

• Model issues
• Over emphasizes proximity to gasohol production
• Conversion from MTBE to ethanol will spread
  gasohol use
• Model needs improvements

17
Modeling Gasohol Summary

• Significant impacts
• State data necessary but not sufficient
• Needs improvements before re-authorization