Guide for Conducting Benefit-Cost Analyses of Multimodal, Multijurisdictional Freight Corridor Investments (2017)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

56 S t e p 6 6.1 Goal Step 6 tasks use the outcomes of Steps 4 and 5 to quantify applicable first-order benefit metrics. 6.2 Tasks Quantify and Value First Order TEE Benefits The TEE measures and metrics to be quantified were addressed in Step 4. They require a mea- sure of travel time/transport cost change (using impedance measures obtained in Step 4) and valuation measures in order to be translated to benefits. The methods to quantify these benefits and value them vary across impacted modes. Appen- dix E provides a high-level categorization of all benefits, valuation methodologies, and the basis to be used. However, the following simple valuation methods and data are commonly used for monetizing benefits. Time-Based Measures Marginal Valuation Factors • Travel time change: Factor cost approach driven by driver/crew costs (driver wage-based value of time by mode). • Vehicle depreciation costs reflecting asset utilization (Table 8) (time dependent variable vehicle costs per hour). • Travel time change (or delays) reflecting cargo/shipper dimension (cargo or freight value of time per hour obtained from stated preference studies) or: – Inventory carrying costs reflecting cargo dimension (discounted capital value of cargo in transit per hour). – Logistics costs reflecting other logistical aspects of cargo such as time sensitivity or perish- ability (logistical costs less inventory per hour or opportunity costs of delayed cargo). • Travel time reliability reflecting cargo/shipper dimension (cargo value of freight travel time reliability per unit of time). Route-Based/Distance Marginal Valuation Factors • Vehicle operating costs (fuel, tires, maintenance) per mile or per ton-mile. • Facility operating costs (e.g., amortized facility wear-and-tear per mile, or per ton-mile). • External costs of maintenance per mile. • External costs of safety/accident costs per mile. • Market prices for fuel/diesel (net of taxes, when build and no-build involve the same mode). Table 8 provides a guide for checking for double-counting errors. Quantify and Value Applicable First-Order Public and Private Metrics and Information Needs

Confusion/Ambiguity Marginal Valuation Bases and MeasurementIssues Sources of Double Counting Travel/ transit time changes (in minutes or hours)— build and do-minimum Measurement issues: Most easily developed for truck freight (highway) using assignment models. Rail assignment is complex. Rail network can use private-domain rail simulation.This works in regional settings but is resource intensive at state and multistate scale. USACE’s Automatic Identification System Data Analysis and Pre-processor (access to USACE) (in port times). When model methods cannot be used due to complexity and scale, network analysis can be used in conceptual stages Double counting can arise by including similar monetized metrics within first-order benefits or by including first-order metrics and related indirect effects in other markets (e.g., land-market- related property value effects). Marginal valuation of travel time or freight transit time savings Travel time: marginal value/cost of time taken to move freight from origin to destination: Operator/driver time cost valued using labor wage rates as factor. Includes labor time in moving freight (driving, loading, and unloading). Equipment time cost: economies of scale in asset utilization. They can be difficult to establish due to privacy but can be a major driver of cost savings. Cargo transit time: Marginal value of the cargo/freight moved itself (logistical cost) is specific to the mode used: Cost of capital in transit: This is simple and is currently part of some models (FHWA Highway Economic Requirement System [HERS], AASHTO Redbook, and World Bank’s Highway Design Model for developing countries). Valued using market interest rates/discount rates. Damage value or depreciation of goods: Value of damage. Time penalties for early/late arrival: Value of reliability. Including each marginal cost only once in the BCA avoids double counting (either as a time-based metric or a cost metric) but not both unless the source is distinct. If all three dimensions are included in costs separately, including time-based reliability as a separate benefit metric will induce double counting. Cost reduction measures in BCA Valuing cost: operating and logistical cost effects of an improvement. They can include labor, equipment, and fuel. This cost can be made more comprehensive to include additional cost of shipping goods for all alternatives compared. Typically expressed in units of cost per hour or per mile based on anticipated changes in time or distance. Each source of cost reduction should be uniquely identified on a per-hour or per-mile basis. Needs careful evaluation if also used in conjunction with time-based metrics. Truck freight reliability Suitable reliability model for different freight modes. There are two types of broad models for valuing reliability: scheduling/schedule delay and reliability ratio. The reliability ratio has become an accepted model to use for truck freight. Percentiles and mean variance measures such as buffer indices of travel time variability can be used. Valuation occurs via reliability ratios. Empirical literature suggests values from 8 to 0.03. Typical to use values of around 1.6 to 2.0. None, as long as measures are able to value only the upper end of the distribution Reliability for other modes For multimodal and intermodal freight, the reliability ratio fails to capture the time sensitivity, safety, and scheduling dimensions. Improved metrics require extensive data on service characteristics of modes. Research still has to establish a value for freight reliability for non-highway modes as well as the applicable class of multimodal shipments. If penalties for late arrival are included, there may be some redundancies in using this separately. Table 8. Ambiguities and methodological issues associated with valuation of Step 5 benefit metrics.

58 Guide for Conducting Benefit-Cost Analyses of Multimodal, Multijurisdictional Freight Corridor Investments Unique Valuation of Subcomponents without Double Counting The monetization of travel or transit time and delay measures can be considered in three distinct parts (building on Table 8): • The utilization of labor as a resource (carrier perspective) for both transit and excess time at terminals in loading and unloading. • The utilization of equipment as assets (asset provider, operator perspectives) of particular value when project elements target delay reduction or improve reliability. • The value of cargo in transit (shipper perspectives). Labor: The Valuation of Transit Time (Operator/Driver/Crew). This is associated with the valuation of time savings benefits to crew/operators, which refers to operator/crew time in all aspects or stages of freight in the freight transport chain from origin to destination. The valu- ation of time savings for equipment operators (rail, barges, and ships) can be approximated by crew time cost savings in various modal operations. The USDOT 2015 guidance provides the most updated values for truck drivers, locomo- tive engineers, rail operators, airline pilots, and engineers. The USDOT values may be used or supplanted by local values available from the Bureau of Labor Statistics or statewide labor data sources when there is discrepancy with national averages. The STB provides railroad-specific wage statistics that can be used for rail studies. USACE published information in its Economic Guidance Memorandums that specified how to determine crew costs based on vessel/barge construction parameters and engine characteris- tics, but that is no longer done. The last published guidance was FY 2004 for shallow-draft vessels (40) and FY 2002 for deep-draft vessels (41). Publicly available data are either very general in nature or specific to one very narrowly defined vessel or barge type. The USACE guidance breaks out base crew costs from operating costs. Crew costs can be appropriately adjusted by using the consumer price index or producer price index for marine operations. Asset Utilization: Valuation of Equipment Time Costs. This refers to monetization of equip- ment time cost (asset utilization). For example, route rationalization by rail can lead to economies in locomotive/train utilization, and cost savings may be associated with a move to a new route. Freight projects can impact system asset utilization in terms of economies of scale or in terms of delay reduction. Such information is not always publicly available, but for railroads it can be inferred from railroad hourly costs which can be useful for valuing delays. This measure is an approximation of the opportunity costs of the equipment in use. For instance, the R1 Schedule 410 can be used to establish cost based valuation measures for both locomotives and freight cars. Similar measures can also be established for barges, and marine vessels and aircraft. In the case of trucks, vehicle deprecia- tion costs or wear and tear costs can be determined on a per hour basis which can be used to account for time dependent vehicle utilization costs. Value of Cargo in Transit: Valuation of Freight Time (VOFT). This refers to the valuation of travel time savings accruing to private sector shippers and operators moving freight. It recog- nizes that a logistics value or cost may be associated with the additional time taken in the context of specific transport chains and cargo types (with and without the project). The literature is relatively scant for VOFT and is limited to freight stated-choice model research. The monetization of cargo can be a logistics cost measure reflecting (at a minimum) inventory carrying costs, reliability, and damage aspects. Freight travel time by any mode is an inclusive concept and typically includes more than just the inventory capital costs associated with freight holding cost and the travel time of the vehicle and driver.

Quantify and Value Applicable First-Order public and private Metrics and Information Needs 59 One approach to valuation is to recognize the inventory (logistic) factors that have an impact on shipper supply chain costs (Figure 10). The literature indicates that inventory cost com- ponents contribute 30 to 65% of the overall annual inventory value for any given cargo. The VOFT could be either a comprehensive cost measure or value per unit of time measure, assum- ing that cargo transit time is valued for three elements for the line-haul move and handling costs at terminals, if applicable: • Inventory capital costs (I): The VOFT with only the inventory capital value of cargo in transit moving from production to destination zones to be used for monetizing transit time and value of damage. The marginal inventory costs (annual) of cargo in transit can be estimated by: 365 24 (3)I H S days hoursi i i ( )( )= × × where Hi = annual unit holding cost per $1 value of inventory. Si = dollar value of a commod- ity (i). Default values of Hi are driven by a discount rate. Winston and Shirley (42) recom- mend 0.15 daily discount rates for perishable cargo, 0.05 for bulk, and 0.10 for other cargo, which can be used for all modes. Within the BCA context, this can be estimated using flow value: (4), ,Change in annual I I Ii i build i nobuld( ) = − • Reliability: The additional inventory costs from unreliability (this requires time-based reliability to be measured first). • Damage: The value of damage as a proxy for risk costs (Figure 10) reflecting damage, spoil- age, or obsolescence (measured in terms of risk or loss per mile or per hour combined with the proportion of cargo that is impacted). This requires the overall proportion of cargo (by weight) that is subject to damage along with the price by weight. • Handling costs at terminals (H): This cost is applicable to account for collection and/or distribution points in order to represent the full transport chain including multimodal movements. The handling cost (at terminal j) is estimated by: (5)H C T Qj j j j= × × Figure 10. Components of inventory carrying costs (shipper supply chain costs).

60 Guide for Conducting Benefit-Cost Analyses of Multimodal, Multijurisdictional Freight Corridor Investments where C, T, Q are all unit cost, time, and output or throughput respectively (unit = per hour, per day) and in the BCA context, this translates to Equation 6: (6), ,Change in annual H H Hj j build j nobuld( ) = − The comprehensive approximation of transportation costs (via the cost per ton-mile for the build and do-minimum scenarios) has been noted in the P&G, FAA, and several guidance docu- ments from USDOT. For highways, USDOT recommends using a mark-up of 15% to account for these costs. If a comprehensive cost valuation measure can be established, then there should be no fur- ther separate consideration of reliability and value of damage to avoid double counting. In practice, the choice of a comprehensive cost metric versus individual metrics is driven by avail- able data and context. Figure 10 can be used as the guiding logic to approximate inventory capital costs. There is no common VOFT to reflect joint end-to-end moves, although indicators are that intermodal cargo may be priced more competitively. In such cases, comprehensive cost assessment using interviews of carriers and a sample of shippers may be a better way to approximate VOFT. Sampling biases may influence these estimates similarly to stated-choice methods; therefore, conceptual studies should consider using intermodal cargo movements combined with shipping costs. Operating Cost Savings and Valuation Measure the operating cost changes of vehicle, locomotive, railcar, vessel, or aircraft use. Vari- able operating costs are functions of distance and time; freight investments can influence one or both, and these can be different across the compared alternatives. Variable operating costs include fuel consumption and power consumption costs, labor costs, materials and supplies, maintenance and repair, and depreciation associated with the use of the equipment (trucks, cars, locomotives, railcars, vessels, or aircraft). Operating costs are also associated with ways and structures, transportation (line-haul crews, switching, dispatchers, and supervisors) and administration. Consider using typical hourly or per-mile rates to approximate operating cost changes, depending on the measure of impedance estimated (time or distance) along with operating costs per hour or per mile, in conceptual stages. This is typically what many tools and conceptual studies attempt to do. Another approach to estimate operating cost savings is to model activ- ity variables like fuel or power consumption directly which can in turn be used prices of these variables. Vehicle/vessel/train-specific equations linking variables that influence operating costs to variables such as volumes, speeds, vehicle-specific characteristics, and terrain can be used to model variable operating costs that vary with time and/or distance. Third, mode-specific simula- tion model tools (e.g., RTC for rail) can also be used to estimate activity measures such as power and energy consumption, time-related performance measures for rail alignment alternatives, and even trip stops. Time delays influence several categories of operating costs including crew, fuel, and asset utilization. Freight trains, for instance, operate with two crew members. Due to restrictions on working hours, train delays may increase crew needs, which lead to additional costs in terms of wages and fringe benefits. Similarly, locomotive operating costs (owned or leased) are time dependent. Fuel use varies greatly according to the type of locomotive, number of locomotives, and operating conditions. As a benefit, include only operating cost categories that can be considered variable across the alternatives being compared. Rail, vessels, and aircraft all have operating costs that are fixed and variable in the context of the investment.

Quantify and Value Applicable First-Order public and private Metrics and Information Needs 61 Data Sources. Two main data sources can be used to estimate operating costs (as a measure of benefit) for rail-related projects on an hourly or per-train-mile or per-ton-mile basis: • Railroads Form R-1 Database Schedule 410: This source is available at https://www.stb.dot .gov/stb/industry/econ_reports.html. STB provides historical data on capital and operating expenses for all Class 1 railroads in the United States. Schedule 410 provides the most relevant information to model operating costs changes as a benefit measure. It also provides capital costs that can be used to approximate the cost side of the BCA. (See Figure 11 for a Sample of the R1 records from STB.) The disadvantage, however, is that the R1 data are available for the entire system, not for a corridor, link, or route. Schedule 410 is available from 1996–2014, with most of the data now available as downloadable spreadsheets. In some earlier years, however, the data are only available in pdf. STB also provides an aggregate historical profile (2006–2014) of commodities transported by any specific Class 1 railroad—the Freight Com- modity Statistics Database, which can be used to support BCA. Schedule 410 provides break- downs of operating expenses by six different categories: – Way and structures. – Locomotive maintenance and repair. Figure 11. Sample of STB R1: schedule 410 showing operating expenses.

62 Guide for Conducting Benefit-Cost Analyses of Multimodal, Multijurisdictional Freight Corridor Investments – Freight car maintenance. – Other equipment costs. – Transportation costs. – Other costs. The R1 database is a system-level database. For the benefit side of BCA, include only those subcategories from the six categories listed that will be variable in the impedance unit measured. The R1 provides the opportunity to compute both per-ton-mile and per- train-hour estimates when cost measures are adjusted by total ton-miles and train hours, respectively. • Uniform Rail Costing System (URCS) tool and database: This source is available at https:// www.stb.dot.gov/stb/industry/urcs.html. The URCS tools and database developed by STB can be used to estimate variable and total unit costs for Class 1 U.S. railroads (see Figure 12). How- ever, the URCS does not properly account for fuel consumption, changes in fuel prices, route characteristics, and rail traffic, and is not always the most appropriate to use in the context of policy or rail service changes. Regardless, there are advantages, which include: – It is both a software tool and a database. – It provides a commodity-specific breakdown of shipping costs. – It provides a per ton-mile cost. – It draws on the R1 database as a source of the costs. – It customizes the costs for route and equipment type used in the transport of cargo. – The tool is maintained by STB and has been updated to include recent 2014 R1 databases. A disadvantage is that it combines capital costs and variable operating costs, and it is not possible to separate the components for use in BCA (i.e., it combines fixed and variable costs). Hence, if these costs are used for determining operating cost savings (on the benefit side), they should represent only the incremental changes in costs. Figure 12. Snapshot of the URCS tool.

Quantify and Value Applicable First-Order public and private Metrics and Information Needs 63 Owens et al. (43) provide their own equations for computing rail fuel consumption (and track maintenance), which could be considered for feasibility studies. USACE closely guards data on the operating costs of vessels and barges, and these costs vary by barge and vessel type. USACE 2002 and 2004 guidance documents (40, 41) can be used to extract operating costs for shallow- and deep-draft vessels (net or gross crew costs) that can be updated using the produce price index for marine industries. With respect to airport cargo improvements, simulation tools discussed earlier also provide the mechanisms for determining operating costs. The best source for most of aviation, including crew costs and operating costs, comes from the FAA guide on economic values (5). Corridor Reliability—Highway Reliability implies different aspects for different modes. Including reliability in the consumer surplus welfare dimension for freight-based projects can be complex. Most of the discussion on metrics in BCA has relied on the moments of the observed travel time distribution (i.e., mean- variance approach) observed over a period of time and/or percentiles (e.g., the buffer index based on the 90th or 95th percentile). Operating costs may proxy reliability implications for on-time arrival or damage value, espe- cially if costs are adjusted by cargo type to include shipper costs (beyond carrier costs). This idea could be applied in the context of other modes to monetize reliability implications. The consid- eration of reliability is resource intensive. If it is done at all, it should be done for all modes to keep comparisons consistent. Four Approaches for Considering Truck Freight Reliability. Four general approaches allow consideration of truck freight reliability in BCA. They are all based on the mean-variance approach relying on means and standard deviations. The methods include data-poor sketch methods and data-rich methods: • Sketch plan methods that approximate before-and-after improvement change without predic- tive equations for the after period. The Highway Freight Logistics Reorganization Tool (HFLRT), which is now part of FHWA BCA.Net, is in this category. It does not estimate reliability metrics but relies on exogenously determined percent increases as inputs to the tool. • Sketch plan methods that approximate before-and-after improvement change with predic- tive equations for the after period. The Strategic Highway Research Program 2 (SHRP2) C11 reliability toolkit developed by Cambridge Systematics (44) is an example that uses a data-poor framework to develop baseline and forecast reliability estimates and relies on the mean-variance approach. • Use of network skims, which are difficult to use since they do not provide continuous distri- butions for developing mean and standard deviations. These allow a full incremental analysis only as a before-and-after comparison. • Data-rich methods that rely on the ability to approximate travel time distributions. These approaches have been discussed and developed by Sage et al. (45) for truck freight. Metrics. Keeping the focus on recurring delay, the most typical metrics used are the percentile-based buffer time and plan time, but they are data hungry. Including the reliability concept in BCA also requires methods to: • Forecast the metric for both the alternatives under examination. • Forecast the metrics over the analysis period. • Have a suitable measure to value reliability. The value on reliability (VOR) measure is typi- cally developed from a stated- and/or revealed-preference modeling framework.

64 Guide for Conducting Benefit-Cost Analyses of Multimodal, Multijurisdictional Freight Corridor Investments Tools for Truck Freight. The literature on VOR for truck freight is quite rich as observed in the range of reliability ratios. A few tools have been developed for considering reliability associ- ated with highway projects, and some of them are specifically suited to corridors, which are a summation of links. Conceptual studies for smaller regional corridors and freight corridors connecting to gate- ways could use tools such as the SHRP2 C11 toolkit. However, for longer corridors at the state and multistate levels (e.g., interstate highways) where there is a need to examine end-to-end reliability, it is usually necessary to use tools such as HFLRT. When it becomes difficult to obtain distributional parameters, consider using delay as a proxy instead of a direct reliability metric or using prior knowledge of a travel time distribution: • Available data sources: Use available data sources such as ATRI’s freight performance measure data and other similar sources such as National Performance Management Research Data Sets to establish measures such as the change in annual hours of delay for all segments. The annual delay (AD) hours for the entire corridor summed over all segments (i) and times of day (tod) in any BCA scenario can be determined as follows: AD L s L s Hi (7)BAU tod i i i i ti = ∑ ∑ −   where L is segment length (i), s is segment speed, st is a threshold speed (e.g., free-flow speed), BAU is the business-as-usual scenario (as an example), and H is the percent of through trucks transporting cargo. Speeds for the alternative scenario can be established using the Bureau of Public Roads function and new volume-to-capacity ratios. Capacity analysis for the effect of improvements could also be undertaken separately. This method is accompanied by several assumptions but may be suited for conceptual analysis for examining cost of delays for end-to-end truck freight trips, which impacts carriers and shippers. Time-dependent operating costs (i.e., operating costs per hour) can be used to value the cost of delay when applicable. If this metric is used, be careful to not double count delay related operating cost savings with other sources of operat- ing cost savings. Such a measure may be sensitive to incident related delays like disruptions. Similar delay measures can be developed for rail, waterways, and air freight, where departures from a threshold arrival time can be used to measure AD. H will refer to the affected portion of shipments per train, vessel, or aircraft. Time-dependant hourly costs will generally include fuel and crew costs, locomotives, railcars, vessels or aircraft costs, and lading costs applicable when there are penalties. • Prior knowledge of distributions: Prior knowledge of travel time distributions like log-normal distributions (left skewed), gamma distributions can be used to approximate metrics like the 90 or 95th percentile times. • Reliability calculation as part of BCA: The reliability benefit can be established using a measure suitable for BCA such as: (8), 160R B B d Q VOT reliability ratiob ij ij Build ij do minimum ij ij( )= − × × × ×− where Rb,ij is the reliability benefit for O-D (i, j), Bij is the reliability metric (in minutes per mile), 60 converts benefits to hours, dij is the distance between origin (i) and destination (j), and Qij is the average of the build and do-minimum volumes (i.e., average of existing and new users). Reliability ratios of commercial trucks ranging from 8 to 0.03 can be used. However, a value of 1.6 or 2 is often used in practice. This measure can be easily extended to other modes by using suitable time-based reliability measures. If the value of the reliability ratio is greater than one, then reliability is more valued for the O-D pair than travel time.

Quantify and Value Applicable First-Order public and private Metrics and Information Needs 65 When using the reliability measure, keep the following factors in mind. • VOR and reliability ratio: The numerical value range of the ratio emphasizes the value of understanding the corridor, its usage and O-D connectivity, and its place within the overall regional network in terms of catering to shipper and industry needs and traded sectors or commodity chains of the economy. This ratio is highly context specific; therefore, spatial transferability of values developed in the literature must be established only after under- standing the base usage in terms of industry reliance. There must be a close linkage/mapping between the commodity transport chains relying on the corridor, the logistical attributes, and the ratio. Sage et al. (45) recommend two approaches to approximate the VOR: – Use a value of 15% as a multiplier or reliability ratio to account for reliability benefits. – Conduct a shipper survey to identify the value of cost reductions. For conceptual analysis, the VOR should be accompanied by sensitivity analysis. The value range obtained must also be guided by industry corridor reliance. For more in- depth studies, this ratio can be developed using either revealed or stated preferences for the corridor. • Estimation of baseline and forecast Bij: Bij must be forecasted over the BCA analysis period and not just for the baseline and do-minimum scenarios. For most conceptual analyses of highway freight, the analyst can use existing toolkits (SHRP2 C11 and the FHWA toolkit) or existing data to understand the travel time distribution for the corridor in the develop- ment of Bij. Sage et al. (45) caution that they are not aware of minimum sample constraints since developing the travel time distribution is a laborious process. It is practical to use data-poor analytic methods to forecast the do-minimum and future values. For other con- texts, NCHRP Report 618 (46) and Sage et al. suggest using data-rich methods that rely on Global Positioning System data to establish the mean of the distribution using the Bureau of Public Roads volume delay function. The Highway Capacity Manual is used to estimate capacity for both alternatives under consideration, while Global Positioning System is used to estimate the mean of the travel time distribution using the Bureau of Public Roads func- tion for both compared scenarios. Variance is estimated for the future periods separately for volume capacity thresholds over and under one separately. One additional requirement is an assumed distribution for the travel time. Sage et al. suggest a gamma distribution. Using these two means, volume-capacity ratio for the two alternatives and variances for the baseline and future, suitable measures for the build and do-minimum (minutes per mile) can be developed for use with Equation 8. • Average volume: Recognize that induced or diverted demand may influence the average volume. Value Other Direct Metrics An asset maintenance impact (e.g., pavement maintenance for highways or changes in track maintenance for rail) is a first-order impact accruing to the facility owner. This category of ben- efits is a component of producer surplus when examined from the point of view of the producer. Highways The 1997 FHWA Highway Cost Allocation Study (47) is often used to quantify asset main- tenance costs using costs on a cents-per-mile basis. FHWA’s HERS has a more complex proce- dure to estimate these costs. (Appendix F provides a table of marginal external costs of highways in cents per mile and per ton-mile.) The determination of changes in maintenance costs for highways has been developed in many tools (see the AASHTO Redbook). Alternatively, simpler methods relying on the changes in distance in miles for the build and do-minimum are used in conjunction with pavement costs in cents per mile to estimate the reduction of pavement maintenance costs.

66 Guide for Conducting Benefit-Cost Analyses of Multimodal, Multijurisdictional Freight Corridor Investments Rail Specific steps must be taken to isolate the investment-specific effects across alternatives on track or locomotive maintenance on the benefit side. A simple way to estimate is to use an estimate of distance traveled and a cost-per-mile estimate. In other words, track maintenance costs can be modeled on the benefit side of BCA if the changes for the build and do-minimum alternatives can be quantified and separated. Databases such as URCS, for instance, include the track maintenance cost directly as part of the operating cost. In such cases, the analyst has to recognize the cost side and benefit side portions of operating costs since URCS blends in fixed and variable costs. Data provided by STB R1 schedule 410 allow a better partitioning of such variable costs. Both the data sources are however aggregate. It is simpler in practice to allocate track maintenance costs to those who own track rights, but locomotives can be owned or leased. Track maintenance costs for railroads are railroad specific and are based on a per-mile system average rate of costs multiplied by the number of cars and locomotives in operation on the route. Waterways and Ports The USACE operating costs also bundle maintenances costs much like URCS. Cargo capacity or throughput is sometimes considered another direct first-order metric if technological, reliability, or productivity enhancements accompany the investments designed to increase effective capacity and throughput at seaports, airports, or intermodal terminals. This can be captured by reduced time to move existing freight by using valuation measures that monetize the value of additional output/cargo movement made possible per unit of time. Identify and Access Data Sources for Valuation Information needs vary based on the stage of the project development process and the specific metric, but the core inputs are from Step 4 forecasts. Advanced stages or special studies may require access to more private data and tools. Table 9 specifies the data needs and tools avail- able to measure and value certain TEE metrics. Using Table 9 as a guide, determine which data sources will be used and be sure the analyst will have access to them. Examine Model and Sources for Performance Metrics and Valuation Measures, and Quantify Benefits When quantifying and valuing benefits, keep in mind that the pursuit of a perfect analysis should not get in the way of performing a useful analysis. In other words, as long as a suitable and objective methodology is used, BCA can still be considered acceptable even if there are data deficiencies but this is accounted for and discussed transparently. Risk analysis is a useful way of addressing input data uncertainties for key variables. This section discusses acceptable approaches and methodologies for all modes. TTS and Valuation for Highways TTS are the most commonly reported TEE metric. The non-highway modes may use mode- use proxies such as transit time, transport cost reduction, or delay reduction. Time savings are valued by using the resource cost method—the driver’s VOT—where the user is the driver or operator. Quantify the Difference in Travel Time. The methodology used for a specific BCA depends on the model used for forecasting flows in Step 5. Different types of demand models can be used with varying levels of sophistication as part of Step 5. Those same demand models provide travel time changes as a result of the assignment process.

TEE Metric(s) Quantification Metrics (Same Mode or Diverted Flow) for No-Build and Build Alternatives Data Needs Valuation Data Needs Public-Domain Data and Tools Private-Domain Tools If Resources Exist and Clearances Obtained Economic and Shipment Data (Public and Private Domains) Travel cost (travel time or transport cost) for impacted modes from transport models/network analysis/ simulation models (from Step 4) and/or external data sources (shipper surveys) Also, reliability Travel or transit time savings (time saved) Transport cost per unit of time and/or per ton- mile saved Delay hours reduction per unit of cargo Buffer indices (truck freight) Measures of on- time arrival or measures of lateness Speed Travel time Distance Volumes—current and forecasts (e.g., average annual daily traffic, vehicle miles traveled [VMT], units, tons, and carloads) Directional flows Peak and off- peak, if needed Capacity Vehicle mix Vehicle occupancy Commodity flows (O-D matrix) Commodity mix Intermodal cargo volumes, if applicable Global positioning system Resource value of time (VOT) (dollars/hour labor time)— USDOT guidance VOFT per hour or VOFT computed as separate logistical components to degree possible Market interest rate Cost metrics per unit of time or distance Reliability ratio for trucks Value of freight transport reliability or, if not available, a cost proxy Conceptual analysis: Freight Analysis Framework (FAF) flows FAF routable database Payload data Available simulation models by mode Existing travel demand models, if available, for truck projectsa Existing commodity flow models (for cross-modal shifts) Network analysis Oakridge National Laboratories routing tools Mode-specific tools/models/equations to quantify effects. FAA Terminal Area Forecasts (http://aspm.faa.gov/main/ taf.asp) Railroad performance measures (http://www.railroadpm.org) Feasibility: Simulation models by mode Travel demand models Custom travel models Public-domain mode-specific tools/models/equations to quantify effects FAA Terminal Area Forecasts Customized models (e.g., CUBE) Private-domain tools (several) by mode (rail and airports) Restricted tools from USACE for ports and waterways Global Insight models Private-domain mode-specific tools/models/ equations to quantify effects (Table 7) Association of American Railroads (AAR) models (proprietary Intermodal Competition Model) Commodity flows and trade data: BTS/Research and Innovative Technology Administration STB Public Waybill U.S. Maritime Administration U.S. Coast Guard USACE Waterborne Commerce Statistics Lock Performance Monitoring System State departments of transportation (DOT) Local agencies Port Import/Export Reporting Service (Journal of Commerce) USACE Waterborne Commerce Statistics STB Waybill restricted data sets Commodity Flow Survey restricted data sets Global Insight’s Transearch Intermodal Association of North America (IANA) Table 9. Direct (first-order) TEE metrics and public and private data sources. (continued on next page)

TEE Metric(s) Quantification Metrics (Same Mode or Diverted Flow) for No-Build and Build Alternatives Data Needs Valuation Data Needs Public-Domain Data and Tools Private-Domain Tools If Resources Exist and Clearances Obtained Economic and Shipment Data (Public and Private Domains) Diverted flows and TEE metrics (for projects with diversion) Diverted volumes (carloads and truck Annual Average Daily Traffic) Reduced truck delay hours in landside access Directional flows diverted by corridor segment VOT (dollars/hour labor time) from USDOT guidance VOFT per hour Conceptual: Table 7. Accepted Mode Diversion Models/Tools Feasibility and investment grade: Mode choice equations calibrated for context RAILNET (rail) Rail Traffic Controller (rail) Terminal models (air) RAILSIM (Systra) Train Energy Model FAF3 IHS Global Insight’s Transearch IANA Fuel and/or other operating costs Fuel reductions (gallons) Operating cost reductions (dollars) Travel/process time data needs Fuel efficiency by (vehicle) class (miles/gallon and ton-miles/gallon) Fuel price (dollars/gallon) Operating costs by vehicle class (dollars/VMT and dollars/hour) All stages: URCS American Trucking Research Institute AAR State agencies FRA STB R-1 reportsb Private data and models AAR Transportation Technology Center, Inc., models and RTC Several aviation private-domain simulation models USACE IANA Highway/track maintenance and repair costs (AMR) Highway/track maintenance and repair cost reduction in dollars per length of infrastructure Length of project (miles) FHWA Highway Cost Allocation, 1997 values updated to base period Cost/mile or cost/ton-mile for other modes Highway/track maintenance and repair costs (dollars/VMT, dollars/ton-mile, or dollars/mile) (Appendix E) All stages: BTS/ Research and Innovative Technology Administration FRA USACE AAR Other federal agencies State DOTs Local agencies Post-processor tools Simple extrapolation procedures Private data and models IANA aSome regions and states do not have travel models or commodity flow models. States and regions also vary in the types and models they have available for planning studies. bThe R-1 Schedules 410 and 415 (Class I Railroads) are freely available as Excel files. The same information can be acquired in Portable Document Format (pdf) or Excel format by purchasing the analysis of Class I Railroads from AAR (https://www.stb.dot.gov/econdata.nsf/f039526076cc0f8e8525660b006870c9?OpenView&Start=1&Count= 300&Expand=1#1). Table 9. (Continued).

Quantify and Value Applicable First-Order public and private Metrics and Information Needs 69 Induced demand sketch plan adjustments can be made on the cost side by relying on travel cost elasticities in order to forecast changes in travel speeds. The elasticities applied to general- ized costs can be used for developing revised travel costs. See Appendix G for more information on this approach. Determine the Value of Driver/Operator Travel Time and Apply It to Changes in Travel Time. TTS are valued using the factor cost approach using in-vehicle VOT. USDOT pub- lishes valuation parameters for monetizing labor, adjusting for trip type and trip length (48). This approach splits applicable wage adjustments by assigning percentages of the wage rates for: • Work-business (100% of the wage, based on hourly cost to the employer). • Non-work travel including walk, wait, and transit time (100% of wage). • Intercity travel (70%). • Local travel (50%). For all other modes, 100% of the wage rate is recommended. Use USDOT resource values and percentage allocations for VOT unless the local values differ significantly. In such cases, local values could be used. Use Average Vehicle Occupancy Factors to Convert Vehicle Hours to Person Hours. For trucks, use the guidance in FHWA’s HERS State Version (49). (In the most recent edition, occu- pancy rates are 1.025 for single-unit trucks and 1.12 for combination trucks.) Some states develop occupancy factors for multiple vehicle categories. Use these factors when available. FHWA’s Surface Transportation Efficiency Analysis Model (STEAM) is the only public-domain tool that allows for trip tables and travel costs for seven modes: auto, truck, carpool, local bus, express bus, light rail, and heavy rail (according to FHWA). It also allows the user to incorporate special circumstances or new modes by modifying these parameters. Tools such as STEAM are best designed to work directly with outputs of demand models and are driven by the rule of half. Beyond carriers (accounted for in driver time), the travel or transit time savings for truck freight also accrue to private parties such as shippers who move cargo in trucks. A VOFT by mode (truck, rail, waterways, and air) is needed to monetize the value of freight time to shippers. When this is not possible, analysts have sometimes used other proxies such as freight cost (transport cost) reductions. A corridor project is a sum of links and may include nodes such as terminals as well: • For corridors, understand the through demand. The corridor travel time can be taken to be the sum of link times and distances, and standard crew-wage-based VOTs are used (USDOT). • For links and nodes (the project category includes terminals or inland logistics facilities or distribution centers), loading and unloading time must be considered. The VOTs now must also consider ground staff wages, which are not considered in USDOT guidelines. Travel Time Savings for Rail, Waterways, and Airports, Simulation modeling is best suited for obtaining forecasts of travel cost changes for ports, airports, terminals, and small scale geographies for rail networks since they are quite detailed in their input requirements. Inland waterway projects and marine port projects follow USACE’s P&G and evaluate transit times and vessel delays for very specific contexts. USACE modeling efforts rely heav- ily on proprietary, confidential information with highly restricted access for projects under

70 Guide for Conducting Benefit-Cost Analyses of Multimodal, Multijurisdictional Freight Corridor Investments the purview of USACE (channel deepening/widening and inland navigation lock and dam projects). Rail simulation models are much better suited to provide measures such as travel time and speeds. Simulation models such as Rail Traffic Controller (RTC, developed by Berkeley Simulation Software), Systra’s rail simulator RAILSIM (50), and FRA’s General Train Move- ment Simulator (GTMS) (51) can be used to quantify actual transit time changes and train performance measures for use in BCA. The first two are often used for evaluating network rail capacity improvements along with grade crossings and separations, while the third is more oriented toward safety aspects and evaluating avoided costs or losses of disruptions in rail networks. RTC. The RTC simulates train movements over a rail corridor, train delay statistics (the pri- mary measure of capacity adequacies), and various train operating statistics, such as average speed (velocity). The model can be used to evaluate the sensitivity of corridor performance to traffic level and mix, evaluate benefits from capital improvements, design a train schedule, and identify bottle- necks or capacity constraints and options for removing them. Almost all the U.S. Class 1 railroads have standardized RTC for their use and have prepared infrastructure and operations data for entry into RTC for much of their route network. The advantages of the RTC package for freight railroads are that it is specifically designed for North American freight operations and fully accounts for the characteristics of such operations. It has been adapted to user experience to make it the leader for U.S. freight operations. The RTC also provides activity measures such as changes in fuel consumption that can be used for measuring operating cost changes with respect to fuel. Some states such as California rely exten- sively on RTC for managing and analyzing rail capacity and service improvements. Therefore, understanding the models available and used in the areas is a helpful way to decide which model to use to evaluate travel time and operating cost changes. The disadvantages of the RTC package are the complexity in use of scales larger than the regional scale. The model can be cumbersome and resource intensive for multistate studies. Therefore, we recommend RTC’s usage if the inputs for incremental analysis can be easily devel- oped or in more advanced analysis. Otherwise, we recommend using network analysis. RAILSIM. RAILSIM is more suited for contexts where passenger rail is involved instead of freight. If analysts already have access to RTC, then the model can be used to provide time and activity statistics. For very large-scale applications (state or multistate), it is best to use public- domain or data sources and network analysis to approximate transit time or transport cost changes, at least in conceptual stages. However, due to the lack of more precise estimates, the inputs from aggregate public-domain sources (and some private-domain data that are publicly available) should be subjected to sensitivity analysis. GTMS simulates rail operations to support risk assessments and safety and operational analy- ses. It is suitable for safety evaluation of plans for new products and systems related to Positive Train Control (PTC). Similarly, for airport and airside improvements, at least four different simulation models could be used, including Total Airport and Airspace Modeler (52), Airfield Capacity Model, FAA’s Airspace Simulation Model (SIMMOD), and the Massachusetts Institute of Technology’s Runway Capacity Model (LMI), as noted in the literature. FAA Terminal Area Forecasts are also often used in airport BCAs. Simulation models are more data and time intensive than spread- sheets such as FAA’s Airfield Capacity Model and Airport Delay Model, so conceptual studies could use FAA tools rather than simulation models.

Quantify and Value Applicable First-Order public and private Metrics and Information Needs 71 6.3 Models, Tools, and Methods Acceptable models, tools, and methods are discussed extensively in Chapter 2 of the technical report. Each of the tools discussed have their pros and cons, the specific metrics they include or exclude, the methods they use, and the philosophy associated with calculations and treatment of transfers. This section discusses specific types of tools that can be used to quantify benefits for one or more impacted modes. Steps and Tools A multimodal corridor BCA may require an integrated assessment; the reliance on more than one set of tools may be needed. The tools are summarized in Table 10. Each step calls upon spe- cific tools—Step 3 uses freight databases, Step 4 forecasting uses models and network analysis tools for behavioral forecasts, and Step 5 uses pre-packaged tools to support the calculation of TEE user benefits and other benefits. Table 10 includes tools used to support forecast of travel times, costs, impedances, and delays as well as post-processor tools. Interpolation, Extrapolation, and Updating of Valuation Parameters and Operating and Maintenance Costs TEE benefits are derived from volume estimates based on interpolated volumes and gen- eralized costs between the base or reference year and the end of the analysis period for all scenarios. Most often, forecasts are available for only two points; in such cases, the interim values should be interpolated. If, on the other hand, the analysis period is longer than available for forecasts (e.g., from demand models), the volumes and benefits should be extrapolated. Operating and maintenance costs may be assumed to grow with inflation (on the cost side and benefit side). VOT parameters may be adjusted using growth rates over the analysis period. Producer Surplus Determination Producer surplus to freight providers and operators comprises changes in asset maintenance costs. If user fees or fares are involved, then the benefits associated with fares or fees are also part of producer surplus. They can be measured as marginal change in revenues minus the change in marginal costs for freight operators based on new–diverted users alone. It is easy to see why this leads an analyst to revisit the assumption of considering tolls and user fees as transfers in multi- modal projects. Table 10 presents a summary table of TEE direct metrics for a straightforward truck highway project. Avoid Double Counting There is the potential for double counting between first-order TEE metrics and indirect effects reflecting impacts that occur in other secondary and tertiary markets (temporally spaced out) such as land/property and labor markets. As a general rule, indirect effects (e.g., land use effects and fiscal impacts) should not be included in BCA because they are captured in economic impact analysis or fiscal analysis, and they are considered capitalization of the first-order TEE benefit (time savings). However, those effects can be major motivations for intermodal facilities. In such cases, they should be accounted for separately. The exception to this rule is when market failures such as imperfect competition in transport and related markets affect the shadow costs (e.g., wages) that are used for valuing time. In such cases, adjustments are recommended to the observed VOTs but are to be considered on a corridor-by-corridor basis. The discussion of these adjustments is included as part of Step 7.

72 Guide for Conducting Benefit-Cost Analyses of Multimodal, Multijurisdictional Freight Corridor Investments Highway Rail Inland Waterways Deep-Draft Marine Air HERS RailSim (Systra) more suited for passenger rail Ohio River Navigation Investment Model (ORNIM) National Economic Development benefits HarborSym FAA Airport Airspace Simulation Model (SIMMOD) and Runway Delay Simulation Model FHWA BCA—HFLRT (useful for corridors) RTC suited for Class 1 capacity and operational analysis Other tools with USACE (developed by the Navigation Economic Technologies Program)—Restricted Access Other tools with USACE (developed by the Navigation Economic Technologies Program)—Restricted Access Massachusetts Institute of Technology’s LMI AASHTO Redbook FRA’s GTMS suited for rail safety analysis USACE has been undertaking efforts to develop system models that allow superior analysis of diversion (induced demand) USACE has been undertaking efforts to develop system models that allow superior analysis of diversion (induced demand) Total Airport and Airspace Modeler STRATBencost (private) URCS (operating costs) and STB Report R-1 (Tool) only for shipping costs Non-USACE projects— P&G Economic Guidance documents 2002 and 2004 (crew costs and operating costs) Non-USACE projects— P&G Economic Guidance documents 2002 and 2004 (crew costs and operating costs) FAA Airport Capacity Model MicroBencost Individual companies - - FAA Economic Values Guide Cal-BC Corridor and Cal-BC Network RAIL//NET - - FAA Airport Delay Model FHWA Surface Transportation Efficiency Analysis Model (STEAM)—multimodal (passenger-truck) Rail costing models (intermodal costing model) (Owens et al. [43 ]) - - - Transportation Economic Development Impact System (TREDIS) (private)/MMBCA free. Regional Economic Models- Transight (private). Note: These include a BCA component that allows multiple modes, but they are not strictly multimodal. (TREDIS) (private)/ Regional Economic Models-Transight (private) Transportation Economic Development Impact System (TREDIS) (private)/MMBCA free. Transportation Economic Development Impact System (TREDIS) (private)/MMBCA free. Transportation Economic Development Impact System (TREDIS) (private)/MMBCA free. Several other tools including in-house tools - - GradeDec.Net FRA GradeDec.Net (grade crossings) - - - Note: Few tools discuss the treatment of induced users and use of rule of half (exceptions are CAL-BC and STEAM). Table 10. TEE metrics—public and private post-processor tools and models by mode.

Quantify and Value Applicable First-Order public and private Metrics and Information Needs 73 A related issue is the potential for double counting in first-order and higher-order met- rics; only truly additional gains that are induced from transportation cost changes should be included. Causality is important, but so is the magnitude of gains. As long as higher- order benefits are driven by network effects/network externalities (e.g., induced demand considerations) and additionality considerations, double counting can be avoided. Addi- tionality considerations are also discussed as part of Step 7. However, tools currently in use for quantification of Step 7 benefits associated with logistics and supply chain benefits may double count first-order cargo transit time benefits from reliability. Therefore, examine these individually. Double counting can also arise from considering effects from induced demand such as diverted flows (reductions in agency maintenance costs from diverted flows) and including a reduction in highway capital expenditures as a benefit. Table 11 lists the sample sequential TEE benefit estimation equations. 6.4 Inputs: Recommended Tools and Data Sources A number of tools can assist the analyst with identifying and quantifying first-order metrics: • See Table 9 and Table 10. • Railroads and terminals (direct contact). Benefit Category Equation Equation Number Value of Travel Time Savings (VTTS) Existing users (Annual) (Use average vehicle occupancy only for highway mode) (change in vehicle hours traveled) or affected trips) × change in travel times × average vehicle occupancy × appropriately adjusted VOT (9) VTTS New users (Annual) (Use average vehicle occupancy only for highway mode) (change in demand or volumes) × change in travel times × average vehicle occupancy × appropriately adjusted VOT × 1/2 (10) Change in Reliability Benefits (if applicable) ($) for an O-D pair (ij) (11) Change Annual Inventory Cost Savings Ii: Daily discount rates = 15% perishable, 5% bulk, 10% other cargo (42) (12) Pavement Maintenance Cost Change (highway example) (change in demand or volumes) × change in distance (miles) × marginal external damage cost per mile (13) Change in Fuel Consumption Costs (Fi) (Annual) (highway example) = Fuel consumption rate (gallons/mile) × change in volume × change in distance × days in year × fuel price less taxes (in $/gallon) (14) Cost savings in cargo transfer, if applicable may be considered (15) Rb,ij =Rb,ij ( ) × dij × change in volumeij × VOT Bij – Bij 1 60 Build do–minimum × reliability ratio Change in (annual)Ii = Ii,build – Ii,nobuld Fi Change in (annual)Hj = Hj,build – Hj,nobuld Table 11. Sample sequential TEE benefit estimation equations (conceptual analysis)—highway connector improvements to intermodal yard.

74 Guide for Conducting Benefit-Cost Analyses of Multimodal, Multijurisdictional Freight Corridor Investments • NCFRP Report 22 (2013) for understanding costs for different modes (http://onlinepubs.trb. org/onlinepubs/ncfrp/ncfrp_rpt_022.pdf). • BLS Producer Price Indices or other suitable sources for adjusting and updating for cost inflation for operating cost values used from past studies (operating and maintenance cost statistics). For rail, and aircrafts, the operating costs can also be updated for past years directly by going to published cost schedules discussed in this guidebook. This will be time consuming. The American Trucking Research Institute provides updates on in trucking related operated costs. • See the Excel worksheet in Appendix M (Step 6: Delay Hours and Reliability). • USDOT Revised Departmental Guidance on Valuation of Travel Time in Economic Analysis (https://www.transportation.gov/tiger/guidance). 6.5 Best Practices and Examples Best practices for Step 6: • Discuss all direct first-order benefits, whether quantifiable or not. • Include project elements in the project definition and mapping to benefits. • Maintain transparency in all calculations. • Separate existing freight users and new users in the calculations for TEE user benefits. Apply the rule of half for the estimate of new users from Step 4. • Use VOTs suitable for the context and mode. USDOT provides standard values across modes, but use local values if the corridor regions are significantly different from national averages. • Ensure the valuation of user benefits pays due attention to the multifaceted nature of freight. In other words, the VOFT includes the crew/labor aspect, the cargo logistics aspect, and the vehicle utilization aspect. • Include only project-specific fixed cost portions on the cost side of the equation; include variable costs portions on the benefit side. • Do not double count items in the operating cost per hour or per mile. • Develop benefits that are matched with appropriate post-processor tools or in-house esti- mation methods. Consider principles of proportionality and project scale when selecting methods and tools. • Recognize that when demand models are used for developing forecasts, matrix-based methods are more suited for disaggregative analysis in subsequent steps. Example 1: The National Gateway Corridor provides one example of a multijurisdictional multistate project with a clear delineation of benefits to existing and new users. This project does not consider transit time savings for rail; instead, it uses transit distance change to approximate transport cost reduction valued with operating cost reductions per mile. While details are lack- ing, the assumptions are clearly laid out. For instance, the analysis assumes a generalized cost savings of 30% in using rail over the market-clearing truck rate ($2.02 per unit-mile) because rail and truck service is assumed to be competitive. This cost savings of $.61 per unit-mile is applied to existing service miles for intermodal units. It uses a cross-price elasticity of 0.67 (truck-rail) to estimate a 20% modal shift. The rule of half is applied to time savings benefits of newly generated rail volumes (diverted from trucks). The market-clearing rate is not explained. The inventory cost savings approach is adapted from FHWA HERS. Additional benefits computed include: • Change in pavement maintenances costs. • Congestion cost savings. • Accident cost savings. • Environmental cost savings (including CO2 emissions cost savings). • Fuel consumption savings.

Quantify and Value Applicable First-Order public and private Metrics and Information Needs 75 6.6 Common Mistakes Common mistakes occur when the project team: • Does not use a level of effort that is commensurate with the value achieved. • Does not consider existing and new users or completely fails to discuss new users. A key part of multimodal BCA is its attention to new users and their modal source. • Does not pay due attention to double counting in first-order benefits. This is most likely to occur when operating and/or shipping cost is used for valuation of time, delay, and/or distance. • Assumes that the parameters used for valuation (i.e., VOTs or operating costs) grow in an ad hoc manner. • Double counts benefits. For a truck freight project, for example, the analyst includes the reduction in time or congestion cost savings and includes land-use-related benefits. • Fails to focus only on distance or time variable operating costs/expenses for consideration on the benefit side.