the feasibility of modernizing the interstate highway system via toll finance

lable at ScienceDirect

Research in Transportation Economics xxx (2014) 1e9

Contents lists avai

Research in Transportation Economics

journal homepage: www.elsevier .com/locate /retrec

The feasibility of modernizing the Interstate highway system via tollfinance

Robert W. Poole Jr. *

Reason Foundation, 140 W. Tropical Way, Plantation, FL 33317, USA

a r t i c l e i n f o

Article history:Available online xxx

Keywords:Interstate highwaysToll roadsFinancial feasibilityRoad pricingMileage-based user feesReconstructionAll-electronic tollingTruck-only lanes

JEL classification:H4 e public financeH440 e publiceprivate partnershipsH540 e infrastructureR410 e highwaysR420 e toll roadsR480 e tolls

* Tel.: þ1 954 587 9426 (office), þ1 310 292 2386 (E-mail address: [email protected].

http://dx.doi.org/10.1016/j.retrec.2014.04.0070739-8859/� 2014 Published by Elsevier Ltd.

Please cite this article in press as: Poole, R.Transportation Economics (2014), http://dx.d

a b s t r a c t

This paper addresses two related problems: how to pay for reconstruction and modernization of theaging Interstate highway system and how to deal with political opposition to increased use of directcharging for highway use. It proposes introducing tolling only on individual corridors as they arereconstructed and (if needed) widened. First, the cost of reconstructing all existing Interstate lane-milesis estimated, using FHWA HERS unit cost data. The cost of adding lanes, where indicated by projectedgrowth in light and heavy vehicle travel, is estimated using HERS unit costs. The revenue that could begenerated by an inflation-adjusted per-mile tolling system (to be implemented via all-electronic tolling)on this rebuilt and modernized system is estimated. Finally, the net present value of toll revenues iscompared with the net present value of modernization cost as an initial estimate of financial feasibility.The analysis is carried out for each of the 50 states and the District of Columbia.

� 2014 Published by Elsevier Ltd.

1. Introduction

The Interstate Highway system is America’s most valuable set ofhighways. With just 2.52% of total lane-miles (Federal HighwayAdministration, Table HM-60), the urban and rural Interstates ac-count for 24.4% of all vehicle miles of travel (VMT) (FederalHighway Administration, Table VM-1). But this system is notadequate for the 21st century.

Its major corridors were mostly opened during the 1960s and1970s. With a typical design life of 50 years, most will needreconstruction between 2010 and 2030. While some have beenwidened to accommodate traffic growth, others will likely needwidening when they are rebuilt for another 50-year period. In themost truck-heavy corridors, widening might take the form of truck-only lanes, while in the most-congested urban areas, a growingnumber of long-range transportation plans include networks ofvariably priced express lanes. Over 100 major interchanges

mobile).

W., Jr., The feasibility of modoi.org/10.1016/j.retrec.2014.0

constitute serious bottlenecks, mostly on urban Interstates(Cambridge Systematics, Inc., 2008).

This paper seeks to estimate whether reconstruction and se-lective widening of the Interstates could be financed via per-miletoll revenues. It first develops a cost estimate for reconstructingthe existing Interstate system, both rural and urban, based on unit-cost reconstruction figures from the Federal Highway Administra-tion (FHWA). Next, it develops estimates of traffic and potential tollrevenue on the reconstructed Interstate systems of each state, overa 35-year period. The net present value of toll revenues in 2010 isthen compared with the 2010 reconstruction cost estimates as aninitial indication of toll-finance feasibility.

Because some of the projected traffic volumes are beyond thecapacity of the existing lanes, the analysis next identifies specificInterstate corridors which will need widening, based on level ofservice (LOS) standards. The cost and time-frame for these laneadditions are estimated, using FHWA lane-addition unit costs. Thewidening analysis also uses data from FHWA’s Freight AnalysisFramework to identify corridors where truck volume warrantstruck-only lanes as part of the widening, and cost estimates forthese additions are included.

ernizing the Interstate highway system via toll finance, Research in4.007

mailto:[email protected]

www.sciencedirect.com/science/journal/07398859

http://www.elsevier.com/locate/retrec

http://dx.doi.org/10.1016/j.retrec.2014.04.007



R.W. Poole Jr. / Research in Transportation Economics xxx (2014) 1e92

The net present value of the reconstructed and widened systemis then compared with the previously estimated net present valueof toll revenue as an indication of the toll-feasibility of financing theoverall project via per-mile tolls.

2. Summary of previous research

In the early 2000s, the National Cooperative Highway ResearchProgram conducted a study of future options for the Interstatesystem. Task 10 of this project defined a case for modernizing thesystem (PB Consult, Cambridge Systematics, Pisarski, & Heanue,2007). It was premised not only on the network layout itself be-ing 70 years old and the pavement beginning to exceed its 50-yeardesign life, but also on new factors like global economic integration,metropolitan congestion, and post-industrial geography.

The study foresaw a slowdown in the growth rate of VMT andthe likely topping out of VMT per capita, and therefore used a 2%annual growth rate for the first 20 years of its 30-year projectionand 1.5% thereafter. Based on those VMTassumptions, it projected arevamped Interstate system widened by 88,600 lane-miles on theexisting 46,800 route miles (plus an additional 84,000 lane-mileson 15,000 route-miles to be added to the system). The wideningwas estimated to cost $1.4 trillion (in 2003 dollars) and the newroutes another $1.74 trillionda total of $3.14 trillion over 30 years.

Reconstruction of worn-out pavement and bridges was notexplicitly dealt with in that project. As a result, a Task 14 was addedto assess the extent to which reconstruction needs were or werenot already being addressed via the investment needs identifiedevery two years by FHWA in its biennial Conditions & Performancereports. The researchers concluded that such needs “appear to havebeen underrepresented in prior work” and “should be the subject ofa follow-on effort to develop appropriate estimating methodolo-gies.” (American Association of State Highway and TransportationOfficials, 2007)

In 2010 AASHTO produced three reports based on the Interstatevision from the 2007 NCHRP report (American Association of StateHighway and Transportation Officials, 2010). Its numbers camelargely from the NCHRP report, but its planning horizon wasextended to 50 years from the previous report’s 30 years. Both theNCHRP and the AASHTO reports used a benefit/cost ratio screen of1.0 in estimating which lane additions and new corridors should beincluded. Neither report identified a funding plan for the $3 trillionupgrade, though both mentioned that tolls and congestion pricingcould help generate funding.

Several others have examined the need for Interstate modern-ization and the possible use of toll revenues to finance it. In 2011the former vice chairman of the National Surface TransportationPolicy & Revenue Study Commission co-authored a paper calling fora national inflation-adjusted toll on all Interstate system users topay for restoration, expansion, and modernization of the system(Schenendorf & Bell, 2011). The electronically collected toll reve-nues would be deposited in a new account in the Highway TrustFund and would provide grants solely for the Interstate program.Themodernizationwould be carried out on a cash basis, rather thanusing toll revenue bonds as is the normal practice for toll facilities.

Regan and Brown published a paper that year on tolling as awayto pay for Interstate modernization. Their top-down estimate of thecost to reconstruct andmodernizewasbetween$1.3 and$2.5 trillion(Regan&Brown, 2011). In a 2012paper, Graells suggested that about10,000 route-miles of rural, inter-city Interstates with heavy trucktraffic could be rebuilt and modernized via toll financing, based onan average toll (car þ truck) rate of 15 cents per mile, yielding anestimated $22 billion per year in revenues (Graells, 2011).

The above papers all made national-level estimates based onhighly aggregated data. There appears to be no previous research

Please cite this article in press as: Poole, R. W., Jr., The feasibility of modTransportation Economics (2014), http://dx.doi.org/10.1016/j.retrec.2014.0

aimed at quantifying the cost of Interstate reconstruction and se-lective lane additions on a state-by-state basis, or on estimating tollfeasibility state-by-state. This paper may thus be the first attemptto do so.

3. Reconstruction cost

The starting point for this analysis is unit cost data from FHWA’sHighway Economic Requirements System (HERS). Appendix A-4 inthe U.S. DOT’s 2010 Conditions & Performance Report providestypical costs per lane-mile for various types of improvements. Thedata used for this portion of the analysis came from the columnheaded “Reconstruct Existing Lane.” These unit costs are nationalaverages in 2008 dollars. They were adjusted for inflation, using theBureau of Labor Statistics CPI calculator, to provide costs as of 2010,the base year for all calculations.

Second, they were adjusted for state-specific variations, usingdata from R. S. Means Company’s Heavy Construction Cost Datavolume. This volume lists tables of cost factors for several cities ineach state. An average of the “site construction” (which includeslabor) and “concrete” (as a proxy for material) was used to give ahighway construction cost index for each locale. For a state’s ruralindex, the cost factor for a randomly selected small city in that statewas used. For the urban index, the figure for the state’s largest citywas used. As an example, the state adjustment factor for rural In-terstates in Alabama was .812 (meaning its cost is 81.2% of thenational average). Hence, the 2010 national HERS rural unitreconstruction costs were multiplied by this factor for Alabama.

3.1. Rural Interstates

State-specific data on lane-miles of rural Interstates were ob-tained from FHWA Highway Statistics Table HM-60, available on-line. This table does not break down lane-miles by individualInterstate routes, so the reconstruction calculations dealt with eachstate’s rural Interstate system as a whole. The spreadsheet for ruralInterstates included 49 states, excluding Delaware and the Districtof Columbia (which have no rural Interstate miles in HM-60).

TheHERS rural cost estimates are given separately forflat, rolling,andmountainous terrain. FHWA provided a listing of the fraction ofrural Interstate route-miles in each category for each state. Thosefractions were entered in the spreadsheet for each state and used tocreate a weighted average reconstruction cost per lane-mile. Thiscomposite unit cost was multiplied by the state’s adjustment factor(from R.S. Means), and that unit cost was multiplied by the numberof lane-miles to provide the estimated total rural Interstate recon-struction cost, in 2010 dollars. Those numbers ranged from a low of$101 million for Rhode Island to a high of $7.84 billion in California.Sensitivity analysis found that without the state-specific adjust-ments, the total cost would be 8% higher.

Since the premise of this study is the use of all-electronic tollingto provide the revenue stream, the reconstruction cost must alsoinclude outfitting each reconstructed corridor with the gantriesand tolling equipment needed for AET. Electronic tolling expertDaryl S. Fleming estimates the cost of equipping rural highways forAET at $250,000 per mile (Fleming, 2013). After including this costfor each state, the total reconstruction cost is $148 billion, which is5.4% more than reconstruction cost alone.

3.2. Urban Interstates

For urban Interstates, the HERS reconstruction unit costs areprovided for four different-sized urban areas: small urban, smallurbanized, large urbanized, and major urbanized. Table HM-60provides lane-mile data only for the total of urban Interstate


R.W. Poole Jr. / Research in Transportation Economics xxx (2014) 1e9 3

lane-miles in each state. FHWA does not provide the fraction ofurban lane-miles in each of the four size groups for each state,which therefore needed to be estimated.

The database for the Urban Mobility Reports of the TexasTransportation Institute provides freeway lane-mile data for 101urban areas, listed in four size groups as small, medium, large, andvery large. Those four size groups were used as proxies for the fourHERS urban categories. The 101 urban areas were grouped by state,and the fraction of TTI-reported freeway lane-miles in each TTIgroup was used to produce a weighted average urban HERSreconstruction unit cost for each state. After adjustment by thestate cost adjustment factor, this number was multiplied by theurban Interstate lane-mile total from HM-60 to yield a total urbanreconstruction cost figure for each state. These ranged from a low of$315 million in Vermont to a high of $59.2 billion in California.

For urban Interstates more tolling gantries are needed than forrural Interstates, because the former have far more on-ramps andoff-ramps. Fleming estimates a typical urban AET equipment andinstallation cost of $2.5millionper route-mile (Fleming, 2013). Afteradding those costs, the total cost of reconstruction came to $441billion, which is 10.4% higher than the cost of reconstruction alone.

3.3. Initial conclusion on reconstruction costs

The estimated cost of reconstructing the rural Interstate systemis $148 billion in 2010 dollars, or $1.20 million per lane-mile. Whilethe overall cost is large, the unit cost seems surprisingly modest.This may be due to the relatively small fraction of rural Interstatemiles in mountainous terrain, as well as the lack of need to acquirenew right of way for reconstruction. It may also be due to the HERSunit cost figures not reflecting the full reconstruction typicallyneeded at the end of a pavement’s useful life, a question that hasnot yet been resolved (American Association of State Highway andTransportation Officials, 2007).

Reconstructing urban Interstates is estimated to cost $441billion or $4.78 million per lane-mile, which also seems low, giventhe high cost of new expressway lanes in urban areas. But as withthe rural Interstates, this baseline cost is only for reconstructingexisting lane-miles and does not include any costs for new lanes orright of way. Moreover, it is an average of states with mostly smalland medium urban areas where reconstruction averages less than$2 million per lane-mile, and the small number of states with largeand very large metro areas whose typical reconstruction costsaverage $4e$7 million per lane-mile.

Overall, reconstructing the entire existing Interstate highwaysystem, without adding any new lanes or new routes, is estimatedto cost $589 billion in 2010 dollars, based on HERS unit costs. Nearlyhalf of this would be spent in just eight states: California, Texas,New York, Illinois, Georgia, Pennsylvania, Michigan, and Florida.

Recent Interstate reconstruction and widening projects inseveral states have reported costs larger than those derived in thispaper. It is difficult to compare the reported costs of actual projectswith the HERS construction and reconstruction costs, because astate DOT’s reported total project costs typically include design,environmental studies and mitigation, and other factors that arenot actual “construction” or “reconstruction,” per se. For thisreason, the costs in this paper are likely to underestimate totalproject costs for the reconstruction and widening of Interstates.

4. Estimating traffic and revenue

4.1. Rural Interstates

The basic approach to projecting toll revenueswas adapted froma more-detailed analysis of toll-financed reconstruction of


Wisconsin’s Interstates (Poole, 2011). That study used the principleof “value-added tolling,” under which tolls would be applied to aspecific Interstate corridor only after it was reconstructed andmodernized (similar to the “used and useful” principle in publicutility regulation, under which an electric utility cannot include anew facility in its rate base for billing purposes until it is actually inuse). In that study, some of the reconstruction also involved laneadditions, using detailed corridor-by-corridor data provided by theWisconsin DOT. The present 50-state reconstruction analysis wascarried out at a higher level of abstraction, treating each state’srural (or urban) Interstate system as a whole. In this more-aggregated look at basic toll feasibility, the calculation is done asif the entire set of rural Interstates were reconstructed by 2020 andtolling began that year. Since the estimated costs and projected tollrevenues are discounted to 2010 for this analysis, reconstructionover a (farmore realistic) longer periodwould not have a significantimpact on the results. This analysis is presented as an exercise totest basic feasibility, not as a proposed reconstruction plan. Thereader may prefer to think of this model and time frame beingapplied to the state’s first Interstate reconstruction project (e.g., theproposed toll-financed reconstruction and widening of I-95 inNorth Carolina).

Traffic (VMT) is projected starting with state-specific 2010FHWA data (from Highway Statistics Table VM-4) for light vehiclesand heavy vehicles. As in the Wisconsin study, once tolling begins,some diversion of traffic off the Interstate is assumed, since somefraction of vehicles that previously used the Interstate will declineto do so when it is tolled. A diversion rate of 10% was used for lightvehicles (cars, vans, pickup trucks, etc.) and 20% for heavy vehicles.The 10% diversion rate in the Wisconsin study drew upon recenttraffic and revenue modeling for the proposed toll-financedreconstruction and widening of I-70 in the Midwest (WilburSmith Associates, 2010). In that analysis, Wilbur Smith Associates(WSA) selected 10% as a reasonable light-vehicle diversion rates for“modest” tolls of around 5¢/mile (higher than the 3.5¢/mile used inthis study). In that same study, WSA assumed 30% diversion fortrucks, based on national average truck tolls of 20¢/mile. Since thisstudy’s baseline rate for trucks is 14¢/mile, a diversion rate of 20%was used.

The traditional approach of assuming a diversion rate for a tolledversus non-tolled highway will become less relevant in future de-cades during which Interstate reconstruction takes place if, as ex-pected, states proceed to phase in some form of mileage-based userfees for all their roadways (to replace fuel taxes). In that environ-ment, all roads will carry some form of explicit charge per mile,with the charge being higher for premium facilities such as In-terstates. This change, over several decades, will likely reduce theextent of “diversion” in the traditional sense, because drivers willknow that alternative routes also carry a per-mile charge (ratherthan being perceived as “free”).

The baseline toll rates initially selected for this studye 3.5¢/milefor light vehicles and 14.0¢/mile for trucks, both in 2010 dollars e

were chosen as potentially being in the right ballpark to pay forreconstruction. Those rates are below the national average (2010)toll rates of 4.9¢/mile for cars and 19.9¢/mile for trucks on long-distance toll roads, compiled for the previously noted Wisconsinstudy. For a 45-year projection to 2054 (10 years’ construction from2010 through 2019, followed by 35 years of tolling), both toll rateswere adjusted annually by an assumed CPI increase averaging 2.5%per year.

State-specific annual VMTgrowth rates for light vehicles and forheavy vehicles were based on a methodology recently developedby the US DOT’s Volpe Center (Pickrell, Pace, West, & Hagerman,2011). For light vehicles, these annual growth rates are drivenmostly by population growth and ranged from a low of .3%


Table 12010 Urban per-mile toll rates, by urban area size.

Peak rate % of VMT Off-peak rate % of VMT

CarsSmall urban areas $.05 30% $.035 70%Medium urban areas $.06 40% $.045 60%Large urban areas $.075 50% $.055 50%Very large urban areas $.10 60% $.070 40%

TrucksSmall urban areas $.20 35% $.14 65%Medium urban areas $.24 40% $.16 60%Large urban areas $.30 40% $.18 60%Very large urban areas $.40 35% $.20 65%


(Connecticut) to a high of 2.2% (Arizona). The truck VMT growthrate (average of single-unit trucks and combination trucks), drivenmore by state economic growth, ranged from a low of 1.8% (Wis-consin) to a high of 3.4% (Arizona). Car and truck VMT was pro-jected year by year for each state, using these growth rates from2010 through 2019. Starting with 2020, the projected VMT numberfor each year was adjusted downward by the selected diversionrates, and these adjusted volumes were used from 2021 through2054, the last year in the projection.

The gross toll revenue in each year was calculated, for cars andfor trucks, by multiplying the CPI-adjusted toll rate for that year bythe adjusted VMT for that year. The net toll revenue for that yearwas then defined as 85% of the gross toll revenue. This reflects twoadjustment factors. The first is an estimated 10% of revenue devotedto highway maintenance, an assumption typically used in pre-liminary toll road traffic and revenue studies. It is a realistic esti-mate, by reference to a recent NCHRP report that puts typicalannual highway maintenance costs at $14,000 per lane-mile(NCHRP, 2012). The traffic and revenue spreadsheet developed inthis paper for a randomly selected state, Indiana, shows that 10% ofthe gross toll revenue in the first year of rural Interstate tollingwould be $65 million. At $14,000 per lane-mile for the 2865 rurallane-miles in Indiana, the maintenance need would be $40.1million, well below $65 million.

Another 5% of gross revenue was assumed for the cost of tollcollection based on all-electronic tolling using a streamlined busi-ness model, as derived in a 2012 study by Fleming, McDaniel,Grijalva, and Sanchez-Ruiz (2012). This peer-reviewed study useddata from several small toll facilities that have adopted AET alongwith a low-cost business model. The authors used data from theseleading-edge operations to develop a generic model for AET thatcan be applied to larger toll facilities. The streamlined businessmodel includes minimizing the use of billing (versus pre-paid ac-counts) and charging service fees to cover the additional costs ofnon-transponder transactions. Under this kind of business model,transponder collection costs are the best overall indicator of netcollection costs.

Each state’s spreadsheet includes the net toll revenue collectedfrom cars and trucks for each year. The final step was to computethe net present value (NPV), as of 2010, of the net toll revenuestream from 2020 through 2054. A 6% discount rate was used, to beconservative. (The higher the discount rate, the lower the NPV.) Thesum of each year’s discounted revenue number is the NPV of totalrevenue, as of 2010. The NPV of toll revenue is not an estimate ofbonding capacity. But by comparing the NPV of revenue to the 2010reconstruction cost estimate, one obtains a general estimate ofwhether the assumed toll rates would be sufficient to support thereconstruction and maintenance of the Interstates in question. Thelarge majority of states had an NPV of revenue greater than the2010 cost of reconstructing their rural Interstates.

4.2. Urban Interstates

The traffic and revenue projections for urban Interstates werecarried out via a similar set of state-specific of spreadsheets, but thetolling assumptions were more complex. Although the focus of thisstudy is on paying for reconstruction, America’s urban congestionproblem suggests that if urban Interstates are to be tolled forreconstruction, the toll rates should be higher during peak periodsthan at other times of day, to reduce the extent of peak-periodcongestion. Moreover, rebuilding urban Interstates will be farmore costly than rebuilding rural Interstates, so higher toll rates arejustified on that basis, as well.

Therefore, a peak/off-peak toll regime was assumed, with ratesin all cases higher during peak periods, and with higher rates in


larger, more-congested urban areas than in smaller, less-congestedones. Table 1 shows the toll rates used for this exercise and thefraction of the VMT charged the peak and off-peak rates. This set ofassumptions is intended as a proxy for the kind of variable pricingsystem that would be tailored to the specifics of each urban area.Note that the peak periods are assumed to be longer in duration thelarger the size of the urban area, and that toll rates also increasewith the size of the urban area, since congestion is more intense, onaverage, the larger the urban area. Since some trucks (but by nomeans all) have the ability to shift their trips to off-peak times ofday, this behavior was assumed for the larger urban areas.

As with the urban cost spreadsheets, it was necessary tocompute a weighted-average car toll and truck toll for each state,depending on how many lane-miles of urban Interstate each statehas in each of the four urban-area size categories. The same urbanfreeway lane-mile data from TTI were used for this purpose. Maine,Montana, North Dakota, South Dakota, and West Virginia have nourban areas listed in the TTI database, so their urban Interstateswere assumed to be in the “small urban” category.

Using the derived car and truck toll rates, the last step was tocreate a traffic and revenue spreadsheet for each state that hasurban Interstates. Although the toll rates are higher than for therural case, the same diversion rates of 10% for cars and 20% fortrucks were used. The three-part rationale is that drivers in urbanareas generally have higher values of time and trip-time reliability(and hence greater willingness to pay), that variable tolling offersdrivers a value proposition of less congestion in exchange for thetoll, and that there are typically few or no uncongested expresswayalternatives. Gross toll revenue was first calculated and thenreduced by 15% to cover maintenance and other operating costs,including all-electronic toll collection. The total net toll revenue foreach year was discounted using the 6% discount rate, enabling thetotal NPV of net toll revenue to be calculated, base-lined to 2010.Only seven states had an NPV of revenues less than 100% of urbanInterstate reconstruction costs.

4.3. Initial conclusions on toll-financed reconstruction

These results suggest that modest toll rates (lower than thosecharged on most established toll roads and much lower than onnew toll facilities) would be sufficient to pay for the capital andoperating costs of a reconstructed Interstate system in most states(using the definition of “reconstruction” embedded in the HERSunit costs). The large majority of states had percentages between150% and 250%, with a small number of outliers. The seven mostproblematic states were Alaska (24%), Montana (43%), New York(85%), North Dakota (69%), South Dakota (60%), Vermont (43%), andWyoming (77%).

This is a surprisingly positive result. When the Interstate systemwas being planned, many advocated a toll-finance model similar to



what had already been used to construct many eastern super-highways that were later incorporated into the Interstate system(e.g., Pennsylvania Turnpike, New York State Thruway, Ohio Turn-pike, etc.). But concerns over low traffic and hence insufficient tollrevenue in the southern, mountain, and western states led to thefederal fuel tax/Highway Trust Fund model being adopted instead.But major economic and demographic shifts over the last 50 yearshave dramatically altered traffic volumes and hence the prospectsfor toll finance of the system’s reconstruction.

The tables of costs and revenues include existing tolled In-terstates for those states which have them. Because the tollingmechanism is already in place on those Interstates, some of thesefacilities may have been partially reconstructed (due to a tollagency’s ability to generate revenues for expansion being greaterthan that of a state DOT) and may therefore need less-extensivereconstruction than is assumed in the numbers for those states.

5. Interstate widening

5.1. Which Interstates need widening?

Some of the traffic projected above for high-growth statescannot be accommodated with the existing number of lanes. Insuch cases, the ratios of NPV of revenue to reconstruction costsderived previously are unrealistically high, because those revenuesare based on more traffic than those Interstates can handle in theirpresent configuration. Hence, the next step was a preliminaryanalysis of where additional lanes are needed to accommodate theprojected traffic.

For this analysis, the simplified method used for the recon-struction analysis, which treated each state’s rural Interstates as asingle project (and did likewise for urban Interstates), was nolonger suitable, since it is highly likely that some corridors carrysignificantly more traffic than others. FHWA provided a spread-sheet listing each Interstate highway within each of the 50 statesand DC, giving for each (e.g., I-10 in Alabama) the number of route-miles, lane-miles, and 2010 average daily VMT. From this infor-mation the average number of lanes of each and the 2010 daily VMTper lane-mile were computed. In the initial analysis, only one-digit(I-5) and two-digit (I-95) Interstates were included; the three-digitfacilities that are solely urban in nature were addressed at a laterstep. The initial set included 237 state-specific one- and two-digit,mostly rural Interstates.

This information was scanned for rural Interstates whose 2010traffic was at or above Level of Service C. Althoughmany state DOTsuse LOS D or even LOS E as the threshold for rural lane additions,the “value-added tolling” premise of this study was interpreted tomean that toll-paying customers should receive better service thanprovided on non-tolled highways. Whereas LOS A and B aredescribed by traffic engineers as providing drivers with a high levelof comfort, LOS C is described as offering drivers “some tension,”with LOS D’s driver comfort defined as “poor.” (TransportationPlanning Branch, 2011)

Florida DOT’s LOS C standards were used for initial screening ofthese mostly rural Interstates (Florida DOT, 2009). LOS C for ruralexpressways equates to about 12,500 daily VMT per lane-mile. Theanalysis used 12,000 DVMT per lane-mile to identify one-digit andtwo-digit rural Interstates as widening candidates. The next stepwas to estimate traffic volume growth for each of the 237 corridors,at 10-year intervals: 2010, 2020, 2030, and 2040. Since the ruralInterstate traffic growth rates used in the previous traffic and rev-enue analysis were on a statewide basis, and no comparable growthrates were available for individual Interstate facilities, the statewideannual growth figure for light vehicles (which is consistently lowerthan the growth rate for trucks) was used and applied to the 2010


VMT for each of the one-digit and two-digit Interstates in that state.This produced estimates of daily VMT per lane-mile for 2020, 2030,and 2040 for each facility, which were used for screening purposes.

A review of that spreadsheet identified 96 corridors whereDVMT/lane-mile exceeds the 12,000 threshold. Using thisthreshold, 41 need more capacity as of 2010, 5 more need wideningby 2020, another 27 by 2030, and another 23 by 2040.

A similar analysis was carried out for the urban portions of two-digit Interstates and all those with three-digit designations. Moststate DOTs use a different standard for urban Interstate congestionthan for rural Interstates, often accepting LOS E conditions as athreshold for lane additions. Based again on the premise of value-added tolling, a higher level of service than current practice wasadopted, with LOS D as the urban lane-addition threshold. Thepreviously cited FDOT handbook’s Table 1 provides LOS D volumesfor urbanized area freeways, which average 18,676 daily VMT perlane-mile. DVMT/ln-mi of 18,000 was used as the threshold for laneadditions. Following a similar procedure to that used for rural In-terstates, DVMT per lane-mile was projected at 10-year intervalsafter the actual values for 2010. This analysis identified 97 urbanInterstates, in 30 states, that exceed or will exceed the LOS Dstandard during the 30-year period.

5.2. Estimated widening cost

The data on the 96major (one- and two-digit) Interstates and onthe 97 urban (three-digit) Interstates made it possible to producevery preliminary cost estimates for those widening projects.FHWA’s table of HERS representative lane-addition constructioncosts provided the unit costs. Because portions of the long-distanceInterstates pass through urban areas, accurate cost estimationrequired the use of unit costs for both rural and urban lane-additions. The rural lane-addition costs are given for flat, rolling,and mountainous terrain. Adjusted to 2010 dollars, they are,respectively $2.251 million (flat), $2.462 million (rolling), and$7.597 million (mountainous) per lane-mile.

The urban lane-addition costs are more complex, since the tablelists both “normal” and “high” costs for each of the four size cate-gories of metro areas. The midpoint of those two figures for eachsize category was used. After adjusting for inflation, those 2010 unitcosts per urban lane-mile were:

Small $4.448 millionMedium $5.725 millionLarge $11.178 millionVery Large $29.717 million.

For the urban Interstates, the costing is straightforward. Aspreadsheet was used to identify howmany lanes need to be addedto reduce the 2040 DVMT/lane-mile to 18,000 or below. The resultswere that 48 require two new lanes, 23 require four, 10 require six,and 16 require eight or more lanes, with California and Texas ac-counting for most of the six- and eight-lane cases. In two Californiacases (I-405 and I-605), even eight lanes would not suffice. Becausesix-lane and eight-lane additions are highly unlikely to be possibleinmajormetro areas, the calculations for those cases were based onadding just four lanes, and assuming that more-aggressivecongestion pricing would be used there to deal with the remain-ing congestion.

Since the spreadsheets indicated in which decade the futurelane addition projects should be carried out, the 4% annual con-struction cost inflation factor from theWisconsin study was used toestimate the build-year cost, and the 6% discount rate was used toobtain the NPV of that cost as of the base year 2010 (to make itcomparable with the 2010 reconstruction costs estimated


Table 2Estimated toll feasibility of Interstate reconstruction and widening.

State NPV ofreconstructioncost ($M)

NPV ofwideningcost ($M)

NPV oftotal cost($M)

NPV of nettoll revenue($M)

Revenue/Cost ratio

Alabama $6368 $411 $6779 $13,743 203%Alaska $4772 $0 $4772 $1140 24%Arizona $11,301 $8332 $19,633 $22,329 114%Arkansas $3858 $1998 $5856 $11,321 193%California $70,275 $117,509 $187,784 $124,438 66%Colorado $8670 $2275 $10,945 $13,646 125%Connecticut $6045 $3866 $9911 $9649 97%Delaware $1824 $299 $2053 $2298 112%D. C. $578 $544 $1132 $819 72%Florida $22,006 $13,914 $35,920 $49,190 137%Georgia $25,646 $19,250 $44,896 $46,556 104%Hawaii $1049 $666 $1715 $1631 95%Idaho $3537 $1904 $5441 $4649 85%Illinois $36,610 $20,465 $57,075 $47,295 83%Indiana $12,024 $4650 $16,674 $25,076 150%Iowa $4439 $1405 $5844 $8331 143%Kansas $6079 $0 $6079 $7720 127%Kentucky $7091 $3183 $10,274 $15,337 149%Louisiana $7183 $1665 $8848 $15,545 176%Maine $2175 $0 $2175 $2430 112%Maryland $9700 $7547 $17,247 $20,872 121%Massachusetts $17,812 $20,070 $37,882 $20,387 54%Michigan $23,498 $4227 $27,725 $26,501 96%Minnesota $9265 $3950 $13,215 $13,150 100%Mississippi $3922 $191 $4113 $7293 177%Missouri $12,676 $6358 $19,034 $24,595 129%Montana $6905 $0 $6905 $2940 43%Nebraska $2564 $2490 $5054 $4788 95%Nevada $4803 $2304 $7107 $7478 105%New Hampshire $3331 $0 $3331 $3211 96%New Jersey $19,075 $12,714 $31,789 $22,693 71%New Mexico $5605 $2787 $8392 $10,591 126%New York $37,480 $12,924 $50,404 $31,790 63%North Carolina $11,266 $9407 $20,673 $29,607 143%North Dakota $3080 $0 $3080 $2118 69%Ohio $20,549 $9610 $30,159 $36,854 122%Oklahoma $5469 $1861 $7330 $11,761 160%Oregon $6176 $1930 $8106 $11,144 137%Pennsylvania $24,154 $10,013 $34,167 $31,032 91%Rhode Island $1328 $386 $1714 $2197 128%South Carolina $5301 $2493 $7794 $13,281 170%South Dakota $3946 $0 $3946 $2370 60%Tennessee $10,364 $7065 $17,429 $29,396 169%Texas $42,149 $47,419 $89,568 $95,648 107%Utah $8013 $1890 $9903 $15,163 153%Vermont $2913 $0 $2913 $1260 43%Virginia $13,605 $9605 $23,210 $29,966 129%Washington $15,805 $12,872 $28,677 $22,673 79%West Virginia $5336 $180 $5516 $5456 99%Wisconsin $6500 $1174 $7674 $10,704 139%Wyoming $5058 $0 $5058 $3888 77%Totals $589,178 $393,743 $982,921 $973,950 99%


previously). For each long-distance Interstate, it was necessary toestimate what fraction is rural and what fraction is urban, in orderto use the correct construction cost for each portion. Spreadsheetswere produced for each state that has Interstates that needwidening, with the final result of each being the net present valueof the state’s widening costs.

5.3. Truck-only lane candidates

A separate data set was obtained from FHWA’s Freight AnalysisFramework (FAF), focused on truck traffic on one-digit and two-digit Interstates. It was organized by Interstate route, breakingdown each one into its separate segments in each of the statesthroughwhich it is routed. For each segment, the database gives thelength in miles, daily truck VMT in 2007, and projected daily truckVMT in 2040.

The FAF uses sophisticated modeling techniques driven by his-torical data and forecasts about goods movement not only viahighway truck but also via other modes. It therefore provides amore detailed look at future truck traffic on specific Interstatecorridors in individual states. The FAF 2040 daily truck VMT figureswere used as a cross-check on the overall 2040 traffic projectionsused in the previous widening analysis.

For each of the 96 corridors in the previous lane-additionsspreadsheet, the projected 2040 total traffic was compared withthe FAF 2040 truck projection, to yield an estimate of the fraction oftraffic constituted by trucks in that year. Interstates with a pro-jected 2040 truck fraction of 40% or greater were selected for truck-only lanes. The larger 237-corridor spreadsheet was also scannedfor corridors with a high fraction of truck trips in 2040, which led toadding 16 more truck lane corridors.

Several multi-state truck corridors were evident, in which aparticular Interstate serves multiple contiguous states that all havehigh projected 2040 truck volumes and percentages:

� I-40, from California through Tennessee, encompassing sevenstates;

� I-70, from Pennsylvania to Missouri;� I-80, from Ohio west through Nebraska;� I-81, from Tennessee north through Pennsylvania.

Other possible multi-state truck-lane corridors include I-10from California to Mississippi, I-30 in Arkansas and Texas, and I-65in Tennessee, Kentucky, and Indiana. Single-state truck-only lanesmight be warranted for I-76 in Colorado, I-84 in Idaho, I-69 inIndiana, and I-71 in Kentucky.

Some would argue, on both safety and capacity grounds, that alower threshold than the 40% truck traffic used here should definecorridors warranting truck-only lanes. Mannering and Washburnestimate that an expressway with just 20% heavy trucks canaccommodate much lower traffic flows than one used only by cars;they estimate reduced vehicle throughput up to 40% due to thatvolume of trucks (Mannering & Washburn, 2013).

The identified truck-only corridors were included in the indi-vidual widening spreadsheets for each of the 42 states with Inter-state corridors needing lane additions. The truck-only lane caseswere modeled as two lanes in each direction, for operational rea-sons. From these 42 spreadsheets, the NPV of widening cost wasdetermined, and that NPV was added to the NPV of reconstructioncost for each state. The NPV of widening costs totaled $394 billion.That is lower than some of the ballpark estimates cites in previoustop-down studies for two reasons. First, the criteria for lane addi-tions was based on LOS criteria and state-specific VMT growthrates, rather than a modest benefit/cost ratio of 1.0 and historicalVMT growth rates. Second, previous studies totaled up


reconstruction costs in construction-year dollars, rather than dis-counting future costs to present value.

6. Overall financial feasibility estimate

Table 2 provides a state-by-state summary comparing the pre-viously estimated NPV of revenue with the NPV of both recon-struction and widening.

Overall, for all 50 states plus DC, the estimated NPV of recon-struction and widening cost was $982.9 billion, and the estimatedNPV of toll revenuewas $973.95 billion, using the standard baselinetoll rates everywhere. Hence, the NPV of revenue equaled 99% ofthe NPV of cost, a surprisingly positive result. Thirty states had NPVof revenue greater than NPV of cost, once widening costs areincluded. (Nine of those had ratios much higher than 100%, sug-gesting that toll rates lower than the baseline rates would suffice.)



Another nine had ratios in the 80e90% ranges, which suggests thatwith somewhat higher toll rates than the baseline rates, theirrevenue/cost ratios could exceed 100%. And another nine mostlyurbanized states had ratios ranging from 54% to 79%. Only threestates had ratios less than 45%.

How do the estimated costs of Interstate modernizationcompare with cost figures on Interstate highway investment needspresented in the most recent FHWA Conditions and Performancereport? Chapter 8 of that report presents two improvement sce-narios (Federal Highway Administration, 2012). Exhibit 8-6, “In-termediate Improvements,” presents results for each category ofhighway if all improvements with a benefit/cost ratio greater thanor equal to 1.5 were carried out. Its estimate for Interstates is $34.9billion per year, which is $9.6 billion per year greater than actualcurrent Interstate investment. Exhibit 8-7, the “Improvement”scenario, covers all investments with a B/C ratio greater than orequal to 1.0. For this scenario, the added investment is $14.7 billionper year. Thus, the C&P report suggests that all cost-effective im-provements to the Interstate system could be achieved via anadditional investment of $14.7 billion per year over a long period.

To compare this to the estimates in Table 2, a spreadsheet wascreated covering the same 35 year period as the tolling modeled inthis studyd2020 through 2054. $14.7 billion in current dollars wasassumed to be raised in each of those years (adjusted using thesame 4% annual cost-increase factor used in the widening costanalysis), and the results discounted to 2010 using the same 6%discount rate used in the prior NPV calculations. The NPV of thisinvestment stream was $339 billion, compared with the NPV of$974 billion shown in Table 2. This suggests several things. First, theC&P investment needs estimates may well not include the kind ofreconstruction recommended in this paper, since the C&P esti-mated improvement cost is so much less. Second, the C&P estimatemay not reflect the improved LOS criteria for lane additions used inthis paper. Third, even if an annual increase in non-toll revenue of$14.7 billion per year could be put in place by 2020, it wouldgenerate far less than the tolling approach outlined in this study.

7. Conclusions and policy implications

The main conclusion of this study is that reconstructing theInterstate highway system via toll revenue finance is considerablymore economically feasible than has been generally believed. Un-like the situation when the Interstate Highway system began in1956, the country’s economic growth and shifts in population andgoods-movement trends since then appear to make it feasible nowfor most Interstates to be reconstructed and widened using tollfinance, at reasonable toll rates.

The alternative to toll finance would be very large increases infuel tax rates. Just to pay for the $589 billion reconstruction costestimated in this study (ignoring the needed widening) wouldrequire an additional 40 to 50 cents per gallon over a period of 10years. And that increasewould apply to all gallons used, whether onthe Interstates or on local streets and roads. Thus, a motorist driving15,000 miles per year, in a vehicle getting 25 miles/gallon, wouldpay an additional $270 per year based on a 45¢/gallon increase. Bycontrast, assuming the same motorist drives 5000 of her 15,000annual miles on Interstates, she would pay $175/year for thosemiles at the baseline toll rate of 3.5¢/mile.

Several important implications of these findings are worth abrief discussion.

7.1. Political feasibility

The previous analysis was based on the “value-added tolling”principle of only introducing tolling when a corridor is


reconstructed (and in some cases also widened). While there is agrowing consensus on the need to transition from a highwayfunding system based on per-gallon fuel taxes to one based on per-mile user fees, there is no consensus on how tomake the transition.This report suggests that one part of that transition could be toconvert the entire Interstate system to per-mile all-electronictolling over several decades, using existing AET technology.

One argument against tolling often made by auto clubs andtrucking groups is that charging fuel taxes and tolls on the samehighway represents “double taxation.” If fuel taxes remain in placefor some years after the reconstruction and modernization of In-terstates, the average amount that motorists and truckers pay infuel taxes for the miles they drive on tolled Interstates could berebated. Electronic toll collection makes it feasible to calculate fueltax rebates, since the vehicle’s identity is known. That makes itpossible to identify the owner to whom a fuel-tax rebate is owed,estimate the number of gallons consumed by that category ofvehicle while driving that length of tolled Interstate, and therebycalculate the fuel tax payment to be reimbursed electronically. Thatwould appear to be a relatively simple software operation, though arebate program would probably add some additional operatingcosts.

7.2. Transition to mileage-based user fees

Much of the policy discussion on replacing per-gallon fuel taxeswith mileage-based user fees (MBUFs) in recent years has assumedthat a single system must be used to keep track of all miles that avehicle drives on any category of road or highway. But many stateDOTs and transportation planners envision the per-mile fees onurban expressways being variable, as an important tool forcongestion-reduction. To do that, the MBUF equipment on thevehicle must be able to distinguish between congested express-ways and ordinary streets and roads. And that, in turn, impliesknowingwhere the vehicle goes, raising potential privacy concerns.

That conflict could be avoided by implementing MBUFs via atwo-part system, as implicitly assumed in this paper. Basic per-milecharges could be levied via a simple, low-tech system intended torecord all miles driven in a state, without regard to location or typeof roadway. This could be as simple as annual odometer readings ora mileage-only device that plugs into the vehicle’s diagnostic port.Some estimates are that a basic rate of 1.5e2.0¢/mile would sufficeto replace current fuel tax revenues, depending on the state.

Premium charges would apply to limited-access highways e

expressways and Interstates. And this can be done, as proposed inthis paper, with current state-of-the-art all-electronic tolling (AET).AET makes use of low-cost transponders supplemented by license-plate imaging, and is already widely used statewide in California,Florida, and Texas as well as in the 15 E-ZPass states in theNortheast and Midwest. This kind of AET handles variable pricingvery well; it does not require the use of costly GPS equipment.

Implementing toll-financed Interstate reconstruction using AETcould thereby be the first major step toward replacing fuel taxeswith MBUFs, since the Interstate system already handles 25% of allVMT. It would be logical to extend the AET system to all otherlimited-access highways in the state, such as other urban express-ways and other limited-access inter-city highways. The remaininghighways, arterials, and local streets would be covered by the basicannual per-mile charge, as described above.

7.3. Project procurement and publiceprivate partnerships

Reconstructing and widening of a state’s Interstate highwayswould be carried out as a large set of projects over a period ofseveral decades. Each of those projects would likely be a



“megaproject”di.e., costing a half billion dollars or more. Tolledmegaprojects lend themselves to being developed as long-termconcessionsda form of publiceprivate partnership.

Transportation megaprojects have a poor track record, in termsof cost overruns, schedule slippage, and over-optimistic projectionsof traffic and revenue (Flyvbjerg, Bruzelius, & Rothengatter, 2003).Under a toll concessionmodel, firms compete for the right to detail-design, finance, build, operate, and maintain a toll facility for a longenough period to have a reasonable prospect of making a return ontheir equity investment. In such agreements, the risks of costoverruns, late completion, and traffic and revenue shortfalls can beshifted to the concession company (Poole & Samuel, 2011).

Many states have state toll agencies which already operatevarious tolled facilities and are experienced in financing toll pro-jects. Toll agencies increasingly use design-build procurement, soas to limit bond-buyers’ exposure to cost overruns and latecompletion. But traffic and revenue risk remains with the tollagency. State PPP enabling legislation often permits state tollagencies, in addition to state DOTs, to make use of toll concessions,and this makes sense in cases where the toll agency has reached thelimits of its bonding capacity or where the risks of a particularproject are higher than the agency and its bond-buyers arecomfortable with.

States without an experienced toll agency should make use oftoll concessions for their Interstate reconstruction and wideningprojects, both for risk-transfer reasons and to take advantage of theexperience of toll concession companies.

7.4. Federal or state program?

This paper’s initial feasibility results suggest that most statescould toll-finance the reconstruction and modernization of theirexisting Interstate highways. Just a handful of states had somecombination of high costs and low traffic that would appear tomake toll financing difficult.

Thus, one key question is whether the toll-financed recon-struction and modernization of the Interstates should be a federalprogram like the original 1956 program, with a federal tollproviding funds for redistribution to the small number of problemstates. The case for such redistribution is far weaker than it was in1956, given the changes in population distribution and goodsmovement since then. In the new era of fiscal constraint, the federalrole might be more limited, such as ensuring nationwide all-electronic tolling interoperability and setting uniform design andperformance standards for the modernized 21st century Interstatesystem.

The largest obstacle to launching toll-financed reconstruction ispermission from Congress, which could occur in the next reau-thorization of the federal surface transportation program in 2014.Current federal law prohibits tolling “existing” lanes on Interstatehighways, which has been interpreted as banning tolling to pay forreconstruction (i.e., replacement) of worn-out existing lanes.

Only three states have permission, under a pilot program, to usetoll finance to reconstruct a single Interstate highway each: Mis-souri, North Carolina, and Virginia. None has reached politicalagreement on actually doing this, to date. But as the reality of thecost of Interstate reconstruction and modernization sinks in, andthe low cost and convenience of all-electronic toll collection be-comes better understood, elected officials may catch up with publicsentiment that is already receptive to tolling as better than (or lessbad than) increases in transportation taxes to pay for major newinvestments in highway infrastructure (Zmud, 2008).

Expanding the pilot program to all states would increase thelikelihood of an initial pathfinder state developing the politicalconsensus to adopt the toll-financed, value-added tolling model for


reconstruction and modernization. The demonstration value of thisfirst mover could be very large, just as the successful imple-mentation of the first priced managed lanes (on SR 91 in OrangeCounty, California in 1995) became the role model for dozens ofsubsequent managed lanes projects across the country.

8. Caveats and limitations

The cost and revenue estimates in this study are the firstbottom-up, corridor-by corridor estimates of reconstruction andwidening for all U.S. Interstates. They are intended as a first-cut,50,000-foot view of the situation, to gain a reasonable estimate ofthe cost of reconstructing existing Interstate lane-miles and addinglanes in corridors where plausible estimates of VMT growth war-rant (including freight corridors where high fractions of trucktraffic justify truck-only lanes). In addition, this study is the firstserious effort to estimate the extent to which moderate, inflation-adjusted tolling could finance the costs of reconstructing andwidening the Interstates.

The results are heavily dependent on assumptions, approxima-tions, and choices. Such choices include the selection of 2.5% as theaverage inflation rate for the next 40 years, 4% as the averageconstruction cost inflation, and 6% as themost appropriate discountrate for the net present value calculations. Changes in any of thosenumbers would affect the quantitative results.

The revenue estimates also depend critically on the state-by-state projections of traffic growth (VMT). The author believes therates used, derived from recent analytical work at the Volpe Center,are reasonable. For light vehicles they range from a low of .3% peryear in Connecticut to a high of 2.2% in Arizona, while truck VMTgrowth ranges, from a low of 1.8% per year (Wisconsin) to a high of3.4% (Arizona). If significantly lower VMT growth rates were used,traffic and revenuewould be less, but sowouldwidening needs andcosts. The revenue projections depend critically on the toll ratesbeing indexed to inflation. The finding that most states couldfinance Interstate reconstruction and widening via tolling dependson inflation-indexed toll rates.

Data limitations in some cases required the use of simplifyingassumptions. Since the VMT projections were available by state butnot by individual Interstates, the statewide VMT growth rate wasapplied to every Interstate within that state. That is unlikely to bethe case, but no better data were available.

The estimated reconstruction and widening costs are likelyoverstated to some extent, since they do not reflect reconstructionand widening projects that may already have been completed orare currently under way. (Thus, a small portion of the estimatedreconstruction and widening costs may already have been incurredin a few states and should be omitted from our totals.) But in someother cases, urban widening costs may be understated for verylarge urban areas, where the calculations used the average of theHERS “normal” and “high cost” lane addition costs. In addition,there is some question within the highway community aboutwhether the “reconstruction” cost figures in HERS reflect total (orextreme) reconstruction or more moderate heavy repaving. If theHERS numbers do not reflect total reconstruction, the reconstruc-tion cost estimates in this paper will understate the true cost ofreplacing all Interstates over the next several decades.

Somewhat higher unit costs would be required for truck-onlylanes (due to the more durable pavement and higher-weight-capacity bridges needed) but were not explicitly estimated in thewidening calculations. On the other hand, the likely higher tollrates that could be charged to longer combination vehicles (LCVs)were also not modeled, somewhat underestimating the revenuesfrom the truck-only lanes. And for the very large urban areas wherelane additions were held to just four, and higher congestion tolls



were assumed necessary, the resulting higher toll revenues werenot estimated.

References

American Association of State Highway and Transportation Officials. (August 2007).Developing a process to assess potentially underestimated highway reconstructionneeds in the U.S. DOT conditions and performance and AASHTO’s bottom line re-ports (a scoping study). NCHRP project no. 20-24(52), task 14 final report.

American Association of State Highway and Transportation Officials. “Unlockinggridlock,” April 2010; “Unlocking freight,” July 2010, and “Connecting rural andurban America,” August 2010.

Cambridge Systematics, Inc.. (Nov. 12, 2008). Estimated cost of freight involved inhighway bottlenecks. Office of Transportation Policy Studies, Federal HighwayAdministration.

Federal Highway Administration. (March 2012). Selected capital investment scenarios(Chapter 8). 2010 Conditions and performance. U.S. Department ofTransportation.

Federal Highway Administration, Office of Highway Policy Information. Table HM-60, 2008 lane-miles Accessed 19.02.13.

Federal Highway Administration, Office of Highway Policy Information. Table VM-1,vehicle miles of travel Accessed 19.02.13.

Fleming, D. S. (Oct. 18, 2013). Estimated per-mile costs of highway AET equipage (PhD,PE, analysis prepared for Robert Poole).

Fleming, D. S., McDaniel, T. L., Grijalva, R. L., & Sanchez-Ruiz, L. A. (November 2012).Dispelling the myths: Toll and fuel tax collection costs in the 21st century. Policystudy no. 409. Reason Foundation.

Florida DOT. (2009). 2009 FDOT quality/level of service handbook.


Flyvbjerg, B., Bruzelius, N., & Rothengatter, W. (2003). Megaprojects and risk. Cam-bridge University Press.

Graells, J. (Summer 2011). Rebuilding interstates: a basic to-do list for project de-livery. Tollways, 9(1).

Mannering, F. L., & Washburn, S. S. (2013). Principles of highway engineering andtraffic analysis (5th ed.). John Wiley and Sons.

NCHRP report 722Assessing highway tolling and pricing options (Vol. 2) (2012 (p.256). Transportation Research Board.

PB Consult, Cambridge Systematics, Pisarski, A., & Heanue, K. (May 2007). Futureoptions for the national system of interstate and defense highways. NCHRP projectno. 20-24(52)dtask 10 final report.

Pickrell, D., Pace, D., West, R., & Hagerman, G. (Nov. 15, 2011). Developing a multi-level vehicle miles of travel forecasting model. Volpe National TransportationSystems Center. Submitted to the Transportation Research Board.

Poole, R. W., Jr. (September 2011). Rebuilding and modernizing Wisconsin’s interstateswith toll financing. Policy study no. 398. Reason Foundation and WisconsinPolicy Research Institute.

Poole, R. W., Jr., & Samuel, P. (February 2011). Transportation mega-projects and risk.Policy brief 97. Reason Foundation.

Regan, E., & Brown, S. (Spring 2011). Building the case for tolling the interstates.Tollways, 8(1).

Schenendorf, J., & Bell, E. (2011). Modernizing the U.S. surface transportation system.Association of Equipment Manufacturers.

Transportation Planning Branch. (Oct. 14, 2011). Level of service D standards forsystems level planning (p. 2). North Carolina DOT. Updated.

Wilbur Smith Associates. (June 2010). I-70 dedicated truck lanes feasibility study,attachment a, sketch level traffic and revenue study.

Zmud, J. (2008). Compilation of public opinion data on tolls and road pricing. NationalCooperative Highway Research Program synthesis 377. Transportation ResearchBoard.


http://refhub.elsevier.com/S0739-8859(14)00008-0/sref1















































the feasibility of modernizing the interstate highway system via toll finance

Documents