Winners and Losers: Distributional Impacts of Highway User Fees
B. Starr McMullen
Lei Zhang
Kyle Nakahara
Oregon State University
Case Study: OregonProposed change in highway user charges: From a Gasoline Tax to Vehicle mile fee
Purpose of Tax: To Collect Road User Fees (We will not consider congestion fees here)
Intent: Revenue Neutral FeeIntent: Revenue Neutral Fee
VMT tax set at $.012/mile to replace $.24/gallon gasoline tax ($.012 = $.24/20 mpg)
Background
Oregon legislature has realized the futility of trying to fund highways with the current 24 cent/gallon tax
No political support for raising the tax Given trend towards more fuel efficient vehicles, fuel tax
now serves as a road user fee as light vehicle road damage is more related to miles rather than fuel consumed
Suggestion by legislatively appointed Road User Fee Task Force (RUFTF): Replace gasoline tax with a Vehicle Mile Tax (VMT)
Who Gains and Loses From This Change ?
Distribution of Costs:
Between Income Groups
Between regions (urban/rural)
Identification important for decisions regarding revenue distribution
Regressive, Progressive, and Proportional or Neutral Tax/Fee Structures
A regressive fee takes a greater percentage of income from lower income groups and higher income groups pay a smaller percentage of income
A progressive fee means that higher income
groups pay a progressively higher percent
of their income in fees
In a proportional or neutral fee structure all income groups pay the same percent of their income in fees
Average Oregon Household Expenditures With Oregon Gasoline Tax of $.24 (2001, with average Oregon gasoline price of $1.46/gallon)
Income group
Number of Households
Average HH expenditure with gas tax $.24/gallon
Average HHexpenditure
as % of income
Average tax revenue
with gas tax of
$.24/gallon
Tax as % of income under
$.24/gallon gas tax
1 39 658 6.63% 108 1.09%
2 75 918 4.09% 151 .67%
3 65 1174 3.17% 193 .52%
4 62 1595 3.06% 262 .50%
5 40 1859 2.75% 3066 .45%
6 67 1993 1.81% 328 .30%
Average HHincome
9935
22433
37038
52096
67500
109962
Average Household taxes under Gasoline tax of $.29/gallon (Using 2001 Average Oregon price of $1.46/gallon)
Income group Number of households
Average tax revenue with
gas tax of $.29/gallon
Tax as % of income under
$.29/gallon gas tax
Average change in Tax
revenue for $.05 increase in tax/gallon
($)(as %of income)
1 39 131 1.32% 23 (+.23%)
2 75 182 .81% 31 (+.14%)
3 65 233 .63% 40 (+.11%)
4 62 317 .61% 55 (+.11%)
5 40 369 .55% 64 (+.10%)
6 67 396 .36% 68 (+.06%)
What has the rise in gasoline prices done to the incidence of overall gasoline expenditures?
Average 2001 gasoline price in Oregon was $1.46/gallon with a $.24/mile tax – and this was regressive
What has happened as gasoline prices have risen – let’s use a gasoline price of $2.64/gallon including $.24/mile gas tax
Average Oregon Household Gasoline Expenditures as a Percent of Income With Oregon Gasoline Tax of $.24
Income group
Number of Households
With an Average
gas price of $1.46/gallo
n(2001)
With an Average
gas price of $2.64
Change in incidence
1 39 6.63% 11.99% +5.36%
2 75 4.09% 7.40% +3.31%
3 65 3.17% 5.73% +2.56%
4 62 3.06% 5.54% +2.48%
5 40 2.75% 4.98% +2.23%
6 67 1.81% 3.28% +1.47%
Average HHincome
9935
22433
37038
52096
67500
109962
The SUITS INDEX The Suits Index, bounded between -1 and 1 A value of -1 implies the lowest income
group bears the entire burden of the tax; A value of 1 implies the highest income
group bears the entire tax burden. A value of 0 implies the tax is proportional.
Suits Index= -0.17623
Suits Index based on Oregon static model
Income group Accumulated income (%)
Accumulated tax (%)
1 2.179938882 5.3804729982 11.64535438 19.793825033 25.18959838 35.771774594 43.36117767 56.478710525 58.5510772 72.046961286 100 100
•The change from a Gas Tax to a VMT will result in an increase in the cost per mile of driving to some; a reduction in the price of driving to others
• Price Increase: Vehicles with MPG >20• Price Decrease: Vehicles with MPG< 20• No Change: Vehicles with MPG =20
Objections raised to change to VMT
1. It will be regressive: Households in lower income groups will be the “losers,” higher income households will gain or lose less
2. Rural areas will lose from the change in policy
3. This policy will not encourage use of high fuel efficiency vehicles
Is the Proposed Change in User Fee Structure a Regressive Change?
Static Analysis: Assumes that behavior is not affected by a change in fee structure; each driver drives exactly the same amount with each vehicle as before the fee was implemented
Dynamic Analysis: Tries to account for the fact that consumers will change driving behavior in response to the change in the price of driving that the tax change causes
Static Analysis: Impact of a Change from a $0.24/gallon gasoline tax to a $.012/mile VMT (2001)
Income GroupNumber of
Households
Average HH Expenditures
including $.24/gallon tax
Average HH Expenditures including $.012/mile VMT
Change in Expenditure (as % of income)
1 39 658.90 666.72 7.81 (+.07%)
2 75 917.84 923.03 5.19 (+.02%)
3 65 1169.61 -4.40(-.01%)
1595.10 1595.33
0.23(<0.01%)
5 40 1858.85 1833.51
-25.34 (-.04%)
6 67 1992.60 1986.60 -6.00(<.01%)
624
1174.01
Static Analysis: Impact of a Change from a $0.24/gallon gasoline tax to a $.012/mile VMT (2001) cont.
Suits Index= -0.22542
Suits Index based on Oregon static model
Income group Accumulated income (%)
Accumulated tax (%)
1 2.179938882 7.5455912252 11.64535438 26.456692223 25.18959838 42.126194474 43.36117767 59.059735435 58.5510772 72.501016076 100 100
Alternative Policy Scenarios:
Alternative Policy 1: Gas Tax of $.24/gallon for vehicles with < 20 mpg; VMT of $.012/mile for vehicles with mpg > 20 mpg
(2001 gas prices)
Alternative Policy 2: Step fee: a. MPG < median MPG pays 2 cents/mile; b. between median MPG and 20 MPG pays 1.5
cents/mile; c. MPG >20 pays 1 cent/mile
Static Model Alternative Policy 1: Gas Tax of $.24/gallon for vehicles with < 20 mpg ; VMT of $.012/mile for vehicles with mpg > 20 mpg (2001 gas prices)
Income Group Average Expenditures
Under Gas Tax
($)
Average Mixed Policy
Expenditures ($)
Change in Expenditures (as
% of income)
1 658.90 675.36 16.46 (+1.66%)
2 917.84 935.01 17.17 (+.08%)
3 1174.01 1191.91 17.90 (+.05%)
4 1595.10 1623.57 28.47 (+.05%)
5 1858.85 1881.25 22.40 (+.03%)
6 1992.60 2023.68 31.08 (+.03%)
Suits Index = -0.18493
Alternative Policy 2: Step fee: a. MPG< median MPG pays 2 cents/mile; b. between median MPG and 20 MPG pays 1.5 cents/mile; c. MPG>20 pays 1 cent/mile
Income group Number of Households
Average HH tax revenue
@24 cents/gal
Average tax revenue step
fee
Average revenue change
1 39 108.31 123.60 15.29 (+1.54%)
2 75 150.88 176.66 25.78 (+1.15%)
3 65 192.99 223.88 30.89 (+.08%)
4 62 262.21 305.32 43.11 (+.08%)
5 40 305.56 380.27 74.71 (+1.11%)
6 67 327.55 404.11 76.56 (+.07%)
Suits Index = -0.16165
DYNAMIC ANALYSIS:
Once behavior changes by the consumer are considered
(movement along the demand curve), the relevant
measure of the change in welfare for consumers is the
change in consumer surplus (CS)—not simply the
change in tax revenue (TR)
For a tax increase, consumers may end up paying less in
taxes, but they may do so by driving less—and that
involves another loss
To get dynamic response, we need a model that take into account the behavioral responses – which may differ by income group and by location
We first use an OLS model – this gives an estimation of changes in vehicle use (i.e. changes in household vehicle miles traveled)
Dynamic Model 1: Ordinary Least Square (OLS) Regression
OLS MODEL VARIABLES Our OLS model is based on the following equation
M = f(Pf,I,U,C,SUB,CHILD,WORKER,MALE)
Where M is the total annual miles driven by the household, Pf is the fuel cost per mile under the gasoline tax I is annual household income U is a dummy variable, = 1 if urban, = 0 else C is the number of vehicles the household owns. SUB = 1 if the household has more than one type of
vehicle, = 0 else CHILD number of children WORK number of workers MALE = 1 if respondent male, = 0 else
OLS Results: Dependent Variable – Annual Household Miles (n=339)
Variable Name Coefficient Standard Error T-Statistic
Constant -17.72 6.25 -2.84
Fuel Cost -8.76 2.39 -3.67
Income 2.21 0.61 3.60
Fuel Cost * Income
0.72 0.24 3.05
Fuel Cost * Substitution
0.44 0.40 1.09
Urban -0.16 0.10 -1.67
#Vehicles 0.54 0.13 4.18
Vehicle Substitution
1.39 1.05 1.32
Male Head 0.17 0.09 1.94
#Worker 0.21 0.05 3.95
#Children 0.04 0.04 0.91
Italicized variables are logarithmic
Changes in Consumer Surplus, Tax Revenue and Welfare by Income with change to VMT = $.012/miIncome Group Average
Change in Consumer
Surplus
Average Change in Tax
Revenue
Average Change in
Welfare
1 -7.93 5.31 -2.61
2 -6.88 6.52 -0.36
3 9.97 -4.51 5.46
4 -2.58 8.61 6.03
5 30.44 -13.81 16.63
6 13.69 -2.78 10.91
Average Changes in Consumer Surplus, Tax Revenue and Welfare by Location ($/Household)
Impact of a Change from a $0.24/gallon gasoline tax to a $.012/mile VMT: Alternative Policy 1 and Alternative Policy 2
Suits Index= -0.133 with gasoline tax
Suits Index =-0.142 with VMT = 1.2 cents/miles
Suits Index = -0.145 with Alternative Policy # 1
Suits Index = -1.111 with Alternative Policy #2
Suits Index based on Oregon OLS model
Conclusions
1. The Dynamic Model Comparison of the change in tax revenues makes the policy impact appear less regressive than the static model (See Suits Indices)
2. However, is the appropriate measure the change in tax paid or the change in consumer surplus?
3. ALL of the VMT-fee policy scenarios have a considerably smaller impact on incidence than the increase in gasoline prices in recent years caused by external forces
4. Different VMT-fee structures have different impacts on incidence or equity, but the difference is not large
CONCLUSIONS/REMARKS (cont’d.)
If we go to VMT, very small impact: likely to have little impact on driving relative to recent increases in gasoline prices
Policy question: If we don’t go to VMT and can’t pass higher gasoline taxes, how do we fund roads?
Alternative funding sources? Local options taxes : sales taxes, local
gasoline taxes Higher Registration fees From General Funds BondingDo we really want greater reliance on non-user
fees?What effect would these taxes have on
regressivity relative to the VMT and gas taxes?
Other Possibilities for a VMT1. VMT-fees may also be designed to achieve
sustainability objectives, such as reducing fuel consumption, reducing greenhouse gas emissions, and encouraging the ownership of greener vehicles
2. Congestion pricing based on VMT-fee technology
3. VMT for congestion and environment in addition to gasoline tax?
Questions and Comments
This research was funded partially by ODOT and OTREC. The author would like to thank Alan Kirk, Jim Whitty, Betsy Imholt, Becky Knudson, Brian Gregor, Jack Svadlenak, Satvinder Sandhu, and Anthony Rufolo for their assistance. The authors are solely responsible for the opinions expressed here.
Contact Information
B. Starr McMullen Lei Zhang541-737-1480 [email protected] [email protected]
Additional Slides
Conventional Wisdom
The static model will overestimate the impact of a
tax increase, underestimate the impact of a tax
decrease
Static model assumes that the change in tax
revenues paid is the only impact that a tax
change will cause – a direct transfer from
consumer to the Government
Change in Consumer Surplus with Demand Response
Price (p)
Pvmt
Pgas
Qvmt QgasQuantity (q)
Total Change in Revenue for an Increase in Price: B-A
B
A
Price (p)
Pvmt
Pgas
Qvmt QgasQuantity (q)
Total change in revenue to agency fromPrice increase = B-A;Price decrease = A-B
Elasticity by Income group – OLS based on Average Income
Income Group Average Income Elasticity w/ SUB Elasticity w/o SUB
1 $9,055.90 1.7577 2.2125
2 $21,983.11 1.128 1.5828
3 $36,899.07 0.7603 1.2151
4 $51,952.61 0.5174 0.9722
5 $67,394.80 0.3326 0.7874
6 $106,043.36 0.0108 0.4656