Click for next page ( 16


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 15
Alternative Approaches to Heady Vehicle Taxation Applications Manual - Estimahng VMT-Related Revenue 3.0 Estimating VMT-Related Revenue This chapter presents procedures for using data on vehicle characteristics and usage to estimate the revenue that should be generated by registration fees, fuel taxes' and weight- distance taxes. The procedures produce estimates of both total revenue and revenue obtained from individual vehicle classes. The estimates of revenue by vehicle class are required for evaluating equity (discussed In Chapter 6~; while the estimates of total revenue that should be produced are of potential value for evaluating evasion (discussed in Chapter S). The latter estimates are also needed to evaluate the expected productivity of a proposed new tax (such as a weight- distance tax). However, the productivity of existing taxes is estimated more directly and more accurately by adjusting current revenue for forecast changes in vehicle use and for any proposed changes in tax rates. Chapter 4 contains furler information about estimating the productivity of new and existing taxes. AD the revenue-eshmation procedures presented in this chapter require estimates of VMT by vehicle class. Unfortunately, the most commonly used methods for developing these estimates produce significant biases in the estimates of truck VMT. The first section of this chapter discusses these biases and ways of reducing or eliminating them. Procedures for using VMT estimates and other data on vehicle characteristics to estimate the revenue that should be generated by registration fees, fuel taxes, and weight-distance taxes are presented In Sections 3.2, 3.3, and 3.4, respectively. 3.] Estimating Truck VMT For most heavy-vehicle taxes, the revenue collected is at least partly dependent on the VMT of these vehicles. For these VMT-related taxes, estimates of VMT by vehicle class are required for producing estimates of: Revenue forecasts for proposed new taxes; Tax payments by vehicle class (for equity analyses); and Revenue that should be collected (for comparison with actual revenue collected for use In evasion analyses). These estimates of VMT by vehicle class are usually derived from traffic classification counts. Unfortunately, the development of truck VMT estimates from traffic classification Cambridge Systematics, Inc. 15

OCR for page 15
Alternative Approaches to Heavy Vehicle Taxation Applications Manual - Estimating VMT-Related Revenue counts is complex, and the most commonly used procedures for developing these estimates frequently produce significant biases. In particular: I. The "Scheme F" classification algorithm used by most automatic vehicle classifiers (AVCs) produces significant undercounts of We number of six-tire vehicles. In a recent test, I] of I? AVCs undercounted these v`?hirlo~ he mire than on -"rant! a-~ _~_1~e _r .1 ~ ~? ~. . ~ , ~ ~ ~ ~ _v r ~ ~` L ~, ~ ~ In or anise AVIS unaercountect these vehicles by more than 50 percent. The most commonly used procedures for deriving estimates of truck VMT from classification counts generally produce upward biases in these estimates and these overestimates can exceed 25 percent.2 3. For several classes of larger vehicles, and particularly for multi-trailer combinations, additional upward bias may exist In the classification counts, and hence in the resulting VMT estimates.3 However, if aU heavy-vehicle classes are considered as a group, the resulting upward bias is likely to be fairly small. The first problem can be addressed by modifying the AVC classification algorithms to reduce or eliminate the undercounting. We recommend that AVC vendors be required to provide algorithms that produce reasonably accurate counts for aD major vehicle classes and, if practical, that vendors be responsible for on-site AVC testing and algorithm calibration. Separate calibrations should be required for different types of locations at periodic intervals (e.g., once every three years), because the wheelbases of autos, pickups, and vans have been changing significantly. States should assume responsibility for overseeing the testing and calibration of AVC systems and other data collection systems and for enforcing performance standards for these systems. ' B.A. Harvey, et. al., Accuracy of Traffic Monitoring Equipment, prepared for FHWA.by the Georgia Tech Research Institute, Atlanta, Georgia, 1995, Tables 4, 7, 13, 14, 17, 21, 24, 27, 30, 32, 35, and 38. 2 Herbert Weinbla~, "Using Seasonal and Day-of-Week Factonng to Improve Estimates of Truck VMT," Transportation Research Record 1522, Transportation Research Board, 1996, pp. 1-8. The most commonly used procedures use weekday classification counts, without any seasonal or day- of-week adjustment, as the basis for distributing estimated annual average daily traffic (AADT) among vehicle classes. In urban areas and in most rural areas, truck traffic is much lower on weekends than on weekdays. Also, outside of urban areas, automobile traffic is often higher on weekends. Accordingly this use of weekday distributions for apportioning AADT among vehicle classes usually produces substantial overestimates of truck AADT and VMT. (The overestimates are highest where truck traffic drops to particularly low levels on weekends and lowest where the weekend drop is modest and where no drop occurs. The latter locations have significant amounts of truck traffic win both origins and destinations that are relatively distant; while the former locations have little or no such traffic. On a few roads, far from the origins and destinations of truck traffic, a weekend-like drop may occur in He meddle of Rho -wand r~cl~ltina in n~cciLl" underestimates of truck AADT.) ,- ~o ~ r~-~ 3 Roger D. Mingo, Improving FHWA Travel Estimatesfor Combination Vehicles: Evaluation of the VM- 1 Table, R. D. Mingo and Associates, prepared for Oak Ridge National Laboratories and FHWA, June 1994; and Roger D. Mingo, Evaluation of FHWA's Vehicle Miles of Travel Estimates for Heavy Vehicles, R. D. Mingo and Associates, prepared for the Association of American Railroads, Washington, DC, April 1991, pp. 6 and 20. 16 Cambridge Systema tics, In c.

OCR for page 15
Alternative Approaches to Heavy Vehicle Taxation Applications Manual - Estimating VMT-Related Revenue The second problem can be addressed by implementing improved procedures for using traffic counts to derive estimates of annual average daily traffic (AADT) by vehicle class and of VMT by vehicle class. Such procedures should use seasonal and clay-of-week adjustment factors for truck counts that reflect variations in truck traffic (which tenets to vary significantly by day of week) rawer than factors that only reflect variations in total traffic (which exhibits much less day-of-week variational A generalized version of such a procedure that is now being used by the Virginia Deparknent of Transportation (VDOT) is presented In Transportation Research Record 1522:5- This procedure has been implemented as an optional capability of the TRADAS 2 system for analyzing traffic data.6 The third problem is substantially less significant than the second problem, ant} it probably warrants less attention. It occurs because, when vehicles are closely spaced, there often is a tendency for AVCs to mistake two or more vehicles as a single vehicle. AVCs do not work well when vehicle speeds are changing. However, AVCs usually are not installed near intersections or on upgrades where such speed changes are common and where vehicles frequently become closely spaced. Nonetheless, problems of closely spaced vehicles can occur on any road, ant! they are common on many urban roads. One somewhat imperfect possibility for addressing the problem of closely spaced vehicles is to collect data on the extent to which larger vehicles are overcountect when traffic is heavy, and to use these data to mollify classification counts collected under these conditions.7 We expect that procedures to acIdress the second problem will be implemented by most states in the next few years. With such procedures in place, estimates of statewide VMT of trucks with three or more axles will be substantially unproved over the estimates that are currently available. If procedures for addressing the third problem are also Implemented, it should be possible to produce estimates of statewide truck VMT that have no obvious biases and that probably have errors that are no greater than one or two percent. 4 If reliable estimates of AADT by vehicle class are needed at an individual site, data requirements become even more stringent. For such purposes, day-of-week factors should be developed only from data collected at that site; i.e., at every site used for short-duration classification counting, at least one classification count of at least seven days duration should be collected, and this count (or counts) should be used to develop day-of-week factors for application to any shorter-duration counts collected at the site. (Mark E. Hallenbeck, Results of the Empirical Analysis of Alternative Data Collection Sampling Plans for Estimating Annual Vehicle Loads at LTPP Test Sites, Chaparral Systems, Inc., anti Washington State Transportation Center, prepared for FHWA, draft, June 1997.) s Herbert Weinblatt, op. cit., pp. 4-7; and Cambridge Systematics, Inc. Virginia State Traffic Monitoring Standards, and Cambridge Systematics, Inc., TMS Algorithms, prepared for VDOT, Richmond, Virginia, 1995. 6 Chaparral Systems Corporabon;7he Traffic Data System. TRADES, Santa Fe, New Mexico, 1996. Another possible means of addressing He Bird problem is to avoid using AVC classification counts that are collected during periods of heavy traffic. Replacement classification counts can be Obtained from some combination of manual counts and/or He imputation of counts from other manual and/or AVC counts. Sophisticated imputation procedures Hat are appropriate for this purpose are described in Cambridge Systematics, Inc., THIS Algorithms, op. cit., and have been implemented in TRADAS (Chaparral Systems Corporation, op. city. Cambridge Systematics, Inc. 17

OCR for page 15
Alternative Approaches to Heady Vehicle Taxation Applications Manual - Estimating VMT-Related Revenue Unto improved estates of heavy-truck VMT are available, the revenue~stimaUng procedures presented in this chapter will be of little value for estimating evasion. However, for the purpose of equity analyses, the procedures can be used with currently available VMT estimates to produce usable attributions of revenue to truck classes. Whatever biases may exist in the estimates of revenue by truck class based on inadequate estimates of truck VMT will carry over to have similar effects on estimates of cost responsibility by truck class based on the same inaccurate estimates of truck VMT. The resulting estimates of equity ratios will not be affected significantly by these offsetting inaccuracies In VMT. For estimating revenue for a new VMT-related tax, the currently available truck VMT estimates will probably have to be adjusted using practical procedures that are less complex than those referenced above. If this must be done quickly without any new field data collection, the Truck Inventory and Use Surveys (TIUS) can be used in a manner similar to that described in the next section. More accurate adjustments, however, will often require some new vehicle classification counts to make adjustments for differences between weekend and weekday traffic. Procedures for developing these revenue estimates are presented in the remainder of this chapter. The use of estimates of potential revenue developed from currently available and improved estEmates of heavy-truck VMT for evaluating adequacy, equity, and evasion of heavy-vehicle taxes is discussed further in Chapters 4, 6 and S. 3.2 Registration Fees Estimates of registration-fee revenue by vehicle class are required for equity evaluations (discussed in Chapter 6~; and good estimates of the reg~stration-fee revenue that should be obtained from heavy trucks can provide a useful ~nclication of the extent of registration-fee evasion resulting from misreporting of mileage by state (discussed In Chapter S). The following discussion of estimating reg~stration-fee revenue focuses primarily on estimates for use in equity evaluations, the most important use of these estimates. The registration fees that are or should be paid to a given state by vehicles registered under the International Registration Plan (IRP) reflect the portion of the vehicles' VMT Hat occurs in that state. Accordingly, a state's receipts of registration fees from heavy vehicles is, in part, dependent on the in-state VMT of these vehicles. The procedure presented below is appropriate for estimating total registration fees that should be paid by all vehicles and by vehicles in each of several classes. For the purpose of an equity evaluation, the procedure should be applied with balanced attention to all vehicle cIasses.9 However, for evaluations of evasion, the principal focus should be on trucks ~ U.S. Bureau of the Census, Truck Inventory and Use Survey, quinquennial. 9 Errors in VMT estimates (such as discussed in Section 3.1) will often result in some inaccuracy in Me attribution of registration-fee revenue (and, in particular, an underallocation of revenue to (Footnote continued on next page...) 18 Cambridge Systematics, Inc.

OCR for page 15
Alternative Approaches to Hearty Vehicle Taxation Applications Manual - Estimating VMT-Related Revenue with three or more axles (FHWA Classes 6-13), because most of the VMT-related registration revenue is in these heavier vehicle classes, and because of the difficulty involved in estimating VMT accurately for six-hre two-axle trucks. The procedure consists of five steps: 1. Obtain VMT by FHWA vehicle class; 2. Partition the FHWA classes into additional classes, as appropriate; 3. Allocate VMT into the partitioned classes; 4. Estimate the number of "fuD-~ne equivalent" (FIR;) vehicles in each vehicle class; and 5. Resinate total registration fees paid bv vehicles In each class. ~J The first step involves obtaining He estimates of annual VMT by FHWA vehicle class that ah states develop. The 13 FHWA vehicle classes are listed in Exhibit 3.~. In the seconc! step, the FHWA classes are partitioned to distinguish separately any other unportant classes of vehicles. Motorcycles are often combined with passenger cars anc! classes of trucks are sometimes combined to eliminate configurations that are less common, such as four-or-more-axle single-unit bucks and seven-or-more-axle combi- nations. Sometimes the FHWA classes are reduced to as few as five classes (autos, four- tire trucks, six-or-more-tire single-unit trucks, and combinations). For the purpose of an equity analysis, these visual classes should be partitioned by registered GVW. The GVW classes may be limited to two or Tree weight classes each for single-unit trucks and combinations, or broken down into as many as 20 to 30 weight classes, depending on the quality of the data and computer resources available Revenue classes shouIc! also be distinguished. As a minimum this should be a division of all vehicle classes to distinguish vehicles paying full fees from those paying reduced fees. Usually it is desirable to break the former class down into apportioned vs. non- apportioned vehicles. Often it is also desirable to break the latter class into two or more, such as: Weight-fee exempt (typically, commercial vs. noncommercial); Registration-fee exempt (typically, publicly owned vehicles, which are also exempt from weight fees); Fuel-tax exempt (typically, buses of various types and Federally owned vehicles); and Reduced-fee, nonexempt (typically, farm trucks and selected other local resource haulers), which may consist of several subcategories win different rates. two-axle satire trucks). However, these inaccuracies will have similar effects on both He revenue attribution and cost allocation results for each vehicle class, and so the effect on equity ratios will tend to be offsetting. Cambridge Systemat~cs, Inc. 19

OCR for page 15
Alternative Approaches to Heavy Vehicle Taxation Applicators Manual - Estimating VMT-Related Revenue Exhibit 3.1 FHWA Vehicle Classes 1. Motorcyclesi 2. Passenger Cars2 3. Over TwmAxie, Four-Tire Single Unit Vehicles2 4. Buses (with Six or More Tires) 5. Two-AxIe, Six-Tire, Single Unit Trucks 6. Three-Axie, Single Unit Trucks 7. Four or More Axle Single Unit Trucks 8. Four or Fewer Axle Single Trailer Combinations Five-Axie Single Trailer Comb~nabons 10. Six or More Axle Single Trailer Combinations 11. Five or Fewer Axle Multi-Trailer Combinations 12. Six-Axie Multi-Trailer Combinations 13. Seven or More Axle Muld-Trailer Combinations ~ Optional. 2 Classes 2 and 3 may be combined. Source: FHWA, Traffic Monitoring Guide, February 1995, Section 4, Appendix A. 20 Cambridge Systen~tics, Inc.

OCR for page 15
Alternative Approaches to Heavy Vehicle Taxation Applications Manual - Estimating VMT-Related Revenue The revenue class divisions need not be used in the cost-allocation process if it is assumed that operating weights and travel patterns (annual mileage and split by functional class of highway) are the same among revenue classes for any given vehicle class. On the other hand, it is often desirable to retain more axle-configuration divisions for the cost- allocation process, but to eliminate these divisions for the revenue-attr~bution process. This recognizes that cost responsibility is likely to differ significantly among axle configurations within a given GVW class, but that the tax structure may not have such divisions. With the above possible exceptions, for purposes of an equity analysis, the Step3 allocation of VMT among the various vehicle and revenue classes should be performed in the same way as it is performed in the corresponding cost analysis. The VMT allocations performed for these two purposes should be performed using TIUS data to adjust and augment the available VMT data for each vehicle class. The TIUS data used for this purpose should be data for vehicles registered in the state; however, because this involves making inferences from data obtainer} from a small sample of vehicles, it is often necessary to supplement Me state's TIUS data with data from adjoining states, or, in a few cases, to use national data. TIUS data should be used to verify and possibly to modify the splits among some of the lighter vehicle classes for which nearly all annual vehicle mileage is in the base state. TIUS data also should be used to fit curves for annual mileage vs. maximum GVW and for percent of miles out of state vs. maximum GVW, both of which tend to increase with maximum GVW. TIUS data do not exist for passenger vehicles and for publicly owned vehicles. Therefore, these relationships cannot be developed for these vehicle classes from the TfUS. Common practice is to make the simplifying assumption that all annual mileage of these vehicles is in the base state, and to assume the same relationship between annual mileage and GVW for publicly owned trucks as is developed for commercial vehicles from TIUS data. In order to perform much of Me subsequent analysis, it is necessary to build a vehicle population file that is fully compatible with the VMT file in terms of the various class breakdowns. Registration data must first be supplemented by aciding Federal vehicles based in the state from Highway Statistics data and estimates of any other vehicles excluded from registration data files (e.g., some types of publicly owned vehicles). Breakdowns of registration data into some vehicle classes will require the use of TIUS data (e.g., for GVW breakdowns of some vehicle classes), and possibly other sources (e.g., splits between four-tire and six-or-more Are trucks). Once these relationships and adjustments are developed, they can be used to compare statewide VMT from traffic counts with VMT obtained from the product of registration data and annual miles per year. Any difference by vehicle class may be due to: errors in the count-based VMT estimates (partially corrected for by the above adjustments); net flow across the state's borders of vehicles based in state vs. those based out of state; or inaccuracies in building up some of the more approximate relationships from different data sources. This comparison is essential as a reasonableness check, and it usually will indicate a need to make some adjustments to achieve reasonable estimates of traffic flows across the state's borders In each direction for each vehicle class. Cambridge Systematics, Inc. 21

OCR for page 15
Alternative Approaches to Heavy Vehicle Taxation Applications Manual - Estimating VMT-Related Revenue Step 4 involves separate analyses for vehicles based In state and those based out of state. For ~n-state vehicles, the principal task is to estimate the average registration fee paid per registered vehicle, and for out-of-state vehicles the principal task is to estimate the number of vehicles In each class that would oav tl~e same total re~i~tr~tinn raven ''" i! ~11 Weir mileage was In state. ~_ _ _O ~, _^ A_ ~ For ~n-state vehicles, an analysis must be made of the amount of partial-year registrations in each major vehicle class. For most vehicle classes, registration revenue normally exceeds the product of a) vehicle registrations as of a given date; and by the vehicle reg~stration-fee rates. Most of the excess occurs because light vehicles pay full-year registration fees even if they are brought into the state or first purchased during the year. On Me other hand, some states do not purge registration files frequently or quickly to eliminate scrapped vehicles, so that vehicle registration totals may be slightly higher than actual vehicles in operation. For heavy vehicles, most states have provisions for partial- year registration, usually on a per-month basis, at rates that are no higher than the fuD- year rate, so that registrations tend to rise during Me spring and to fall during autumn after the harvest. The required analysis involves: a) determining exactly what practices are followed for partial-year registrations and for purging of scrapped vehicles; by comparing registration totals and new registrations for each month of the year for each major vehicle class with the registration totals used in Me preceding steps (usually registration totals as of the end of a specific month); and c) estimating the ratio of Fl Es to registrations for each major vehicle class based on the above analyses. For out-of-state vehicles, FTEs are estimated by dividing VMT In each class by the average annual mileage in the class. The average annual mileage in each of these classes should be significantly higher than for the equivalent in-state classes because a large portion of out-of-state vehicles are owned by long-haul for-hire interstate carriers. The average annual mileage for these vehicles should be a properly weighted average of TIUS average annual mileages of interstate vehicles In other states. The truck weight survey described in Section 6.4 can be used to develop the appropriate weights for estimating average annual mileages from other states. Finally, for each vehicle class, Me number of in ~ vehicles (from Step 4) is multiplied by the annual registration fee for the class to produce an estimate of total registration fees paid by vehicles in each class (Step 5~. 3.3 FuelTaxes Estimates of fuel-tax revenue by vehicle class are required for equity evaluations (discussed in Chapter 6~; and good estimates of the revenue that should be received from Me diesel-fuel tax can provide a useful Indication of the extent of evasion of this tax (discussed in Chapter S). Also, if (as In Oregon) a significant numnber of vehicles are exempt from fuel taxes, VMT-based estimates could be needed of the revenue that would be derived from any proposed extension of the fuel tax to vehicles that currently are exempt (Chapter 41. However for this. nuance if VAT inf~rm~n from lATPi~,ht_~liCt=~= 22 --it r a--- ~-~ ~- Cambridge Systematics, Inc.

OCR for page 15
Alternative Approaches to Heavy Vehicle Taxation Applications Manual - Estimating VMT-Related Revenue tax reports is available, it win produce better results than- VMT estimates derived from traffic counts.~ The procedure for allocating fuel-tax revenue to vehicle classes is similar to the procedure for estimating reg~stration-fee revenue obtained from vehicles with three or more axles. For fuel taxes, the five steps are: I. Obtain VM] by FHWA vehicle class; 2. Partition the FHWA classes into other classes as appropriate, and further partition these classes by fuel type; 3. Allocate VMT Into the partitioned classes; 4. For each class, estimate average fuel efficiency; and 5. For each class, use the VMT and fuel efficiency estimates to estimate total in-state fuel consumption. The first, third, and fifth steps of this procedure are essentially the same as the corresponding steps of the procedure for estimating reg~stration-fee revenue from heavy trucks. However, the fourth step and one aspect of the seconc! and third steps are somewhat different. The fuel-type destinations between (at least) gasoline and diesel fuel developed in Steps ~ and 3 are important because of fuel-efficiency differences, tax rate differences in many states, ant} substantial differences in evasion rates. For non-apportioned vehicles and apportioned vehicles based in state, the split can usually be accomplished using registration data (possibly using TIUS data to adjust for variations in VMT per vehicle by fuel type). For apportioned vehicles based out of state, the split shouIcl be assumed to be identical to He split for similar apportioned vehicles based in state. For all truck classes, the Step 4 estimates of fuel efficiency should be obtained by taking a weighted average of TIUS estimates of fuel consumption rates for all vehicles in the class (total gallons consumed divided by total VMT) anc! inverting to obtain miles per gallon. The fuel efficiencies obtainer! for each class shouIc! be reviewed to assure that they are consistent with those obtained for similar classes and Hat small samples for some classes do not result In unreliable fuel~fficiency estimates for these classes. In order to avoid use Has discussed in Section3.1, estimates of heavy-truck VMT derived from traffic counts using current procedures genera-fly have significant upward biases and will result in significantly overestimating fuel-tax revenue. On the other hand, VMT information provided by motor carriers win wei~t-distance tax reports may reflect underreporting of VMT by some carriers. However, since similar underreporting by interstate carriers may also affect revenue from a proposed fuel tax, using VMT info~-`,~ation~from weight-distance tax~reports is likely to produce a better estimate of revenue that would be produced by an expanded fuel tax. Since apportioned vehicles generally operate more miles per year than nonapportioned vehicles, they are more likely to use diesel fuel. A recent analysis of California data shows that 99.9 percent of apportioned combinations use diesel fuel but only 96.9 percent of nonapportioned combinations. For single-unit trucks, on an overall basis (not controlling for registered weight), the percentages are 76 percent and 7.2 percent, respectively. Cambridge Systematics, Inc. 23

OCR for page 15
Alternative Approaches to Heavy Vehicle Taxation Applications Manual - Estimating VMT-Related Revenue of unreliable fuel-effic~ency estates, MUS data for some sets of vehicles classes that should produce very similar fuel efficiencies may be combined in order to produce a single (more reliable) estimate for the entire class. A nonlinear curve of fuel efficiency vs. GVW should be developed, carefully fit to the TIUS data for each fuel type. The curve fining process should take into account the sample size for each data point, at least judgmentally. Experience suggests that a graphically fit smooth curve will provide the most satisfactory relationship, as distinct from a nonlinear mathematical function. The fleetwide average in-use fuel efficiency of automobiles may be obtained from Highway Statistics, and automobiles may be treated as if they all use gasoline. AVCs are not capable of accurately distinguishing between automobiles (FHWA Class 2) and four-tire trucks (FHWA Class 3~. For this reason, several states now produce only a single combined VMT estimate for FHWA Classes 2 ant] 3; and the quality of separate estimates clepends upon the care with which the AVCs have been calibrated. States that do not have separate estimates of Class 2 ant! 3 VMT may split total VMT In the two classes on the basis of the total number of vehicles in these classes or they may further adjust this split to reflect estimated differences In the average annual mileages of the two vehicle cIasses.~3 3.4 Weighi-Distance Taxes States that have a weight-distance tax generally have access to ah the information neecled to determine the revenue produced by aB vehicle classes of interest for equity evaluations, and so further estimates generally are not required for this purpose. However, good independent estimates of the revenue that shouIc! be receiver! from this tax are requires! for estimating evasion. Also, states that are considering implementing such a tax require estimates of the amount of revenue that such a tax would generate. If good estimates of VMT by vehicle class are available, the procedure for estimating the revenue that should be produced by a weight-distance tax is straightforward: 1. Obtain VMT by FHWA vehicle class; 2. Allocate VMT into the weight-distance tax classes; 3. For each tax class, multiply VMT by the applicable tax rate; and 4. Sum He estimates of tax revenue across ah tax classes. i2FHWA, Highway Statistics, annual, Table VM-1. i3 Estates of ~ per vehicle may be obtained from special surveys ~ Me state or from analyses done in nearby states Mat are considered to have similar usage patterns. A review of national data for recent years suggests Mat use of VMT data may result in a modest increase or decrease in the Class 3 share of VMT (relative to the share produced by using only registration data). 24 Cambndge Systematics, Inc.

OCR for page 15
Alternative Approaches to Heady Vehicle Taxation Applications Manual - Estimating VMT-Related Revenue If the only available VMT estimates are believed to have significant upward biases (for reasons discussed in Section 3.~), the resulting revenue estimate is likely to be too high. This estimate win not be useful for escorting evasion. However, it is possible to adjust this estimate to produce a revenue estimate that is reasonable for estimating the yield of a proposed new tax. One simple approach is to multiply the VMT-basect revenue estimate by the ratio of actual revenue from the cliesel-fue! tax to an estimate of revenue from the diesel-fuel tax derived using the same VMT estimates. This adjustment factor win combine two acljus~anents to tEte weight-distance~tax revenue estimate: an adjustment to eliminate the bias in the VMT estimates; and a further adjustment reflecting the combined effect of ah types of fuel-tax evasion. The first of these adjustments clearly is desired. The second one also is desirable, but it may reflect somewhat more evasion than would actuary occur under a weight~istance tax.~4 lathe size of Me adjustment factor is influenced by some types of evasion (such as the use of untaxed or undertaxed fuel) that do not occur for weight-distance taxes. Cambridge Systematics, Inc. 25

OCR for page 15
Alternative Approaches to Heady Vehicle Taxation Applications Manual - Estimating VMT-Related Revenue 26 Cambridge Systematics, Inc.