Identification and Evaluation of the Cost-Effectiveness of Highway Design Features to Reduce Nonrecurrent Congestion (2014)

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.

53 A methodology was developed for conducting a life-cycle benefit–cost evaluation for the design treatments considered in this research. The method uses expected improvements in travel time, travel time reliability, and safety to estimate mon- etary benefits of treatment installation and compares those benefits to the expected costs of implementation and main- tenance of the design treatment. This section describes the methodology for determining the values of these benefits and costs and describes the calculation procedure to estimate the final benefit–cost ratio. A spreadsheet tool to implement this methodology is described in Chapter 2 under Analysis Tool. Overview of Life-Cycle Benefit–Cost analysis Methodology The life-cycle benefit–cost analysis methodology is intended to obtain two measures that compare the benefits and costs of design treatments expressed in monetary terms: the benefit– cost ratio and net present benefits. These two measures are defined by Equations 6.1 and 6.2: ( )=Benefit–Cost Ratio 6.1B C ( )= −Net Present Benefits 6.2B C where B is the present value of treatment benefits ($), and C is the present value of treatment costs ($). These measures can be used to assess whether a specific design treatment has positive net benefits for application at a given site (i.e., if B/C > 1 or B - C > 0), and they can also be used to compare the cost-effectiveness of alternative treat- ments. Any specific treatment is evaluated over its service life (i.e., the period of time over which the treatment will con- tinue to provide benefits without renewal, reconstruction, or replacement). When alternative treatments with differ- ing service lives are compared, that comparison needs to be conducted over multiple renewal cycles for one or both treatments. The analysis period is typically the least common multiple of the service lives of the design treatments being compared. For example, comparison of a design treatment with a 10-year service life to a treatment with a 15-year ser- vice life would need to be conducted with a 30-year analysis period (i.e., three life cycles for the first treatment and two life cycles for the second treatment). This comparison of treat- ments over multiple life cycles is why this type of analysis is referred to as life-cycle benefit–cost analysis. The costs of design treatments are determined by combin- ing the initial implementation or construction cost and the annual maintenance cost, as shown in Equation 6.3: ( )= IC AMC (USPWF) 6.3C + where C = design treatments costs ($); IC = implementation or construction cost ($); AMC = annual maintenance cost ($); and USPWF = uniform series present worth factor. The uniform series present worth factor is defined by Equation 6.4: ( ) ( ) ( )= + − + USPWF 1 1 1 6.4 i i i n n where i is the minimum attractive rate of return or discount rate, expressed as a proportion (e.g., i = 0.04 represents a 4% discount rate); and n is the service life of design treat- ment in years. The benefits of design treatments in the life-cycle benefit– cost analysis combine both traffic operational and safety benefits, as shown by Equation 6.5: ( ) ( )= +AOB ASB USPWF 6.5B where AOB is the annual traffic operational benefit ($), and ASB is the annual safety benefit ($). C h a p t e r 6 Life-Cycle Benefit–Cost Analysis

54 Equation 6.5 is suitable for the current assessment tool, which is based on constant traffic volumes. A potential enhancement of the tool would allow the user to specify an annual percent age growth in traffic volume. Equation 6.5 would then be replaced by Equation 6.6: B i j j j= n jAOB ASB 1 1 6.6 1 ∑( ) ( ) ( )= + +   where AOBj = annual traffic operational benefit for year j ($); ASBj = annual safety benefit for year j ($); and 1/(1 + i)j = single-amount present worth factor for year j. The annual traffic operational benefit for a design treat- ment is determined as shown by Equation 6.7: D + V N Lk k k d k AOB VOT VOR 6.7 1 24 ∑ ( ) ( ) ( )= ∆ ∆σ = where ∆Dk = change in annual traffic operational delay due to the design treatment during hour k (vehicle hour); VOT = value of travel time ($/vehicle hour); ∆sk = change in the standard deviation of travel time during hour k; VOR = value of reliability ($/vehicle hour); Vk = traffic volume on facility during hour k; Nd = number of days per year (= 250 days); and L = roadway segment length (mi). This approach to assessing the value of travel time and reliability is based directly on the current state of knowledge about the value of reliability. It may be appropriate to update this approach as the state of knowledge evolves. In addition, a possible enhancement of the tool could incorporate addi- tional operational benefits, such as vehicle operating cost, including and fuel cost, savings and reduced emissions. The annual safety benefit for a design treatment is deter- mined as shown by Equation 6.8: ASB NReduction CC NReduction CCFI FI PDO PDO= ( ) ( )+ + DSB CC DSB CC DSB CC FSI FSI MI MI PDO PDO ( ) + ( ) + ( ) 6.8( ) where NReductionFI = predicted number of fatal-and-injury crashes per year to be reduced by the congestion-mitigation effect of a design treatment; NReductionPDO = predicted number of property- damage-only crashes per year to be reduced by the congestion-mitigation effect of a design treatment; CCFI = crash cost savings per fatal-and-injury crash reduced ($); CCFSI = crash cost savings per fatal-and-severe- injury crash reduced ($); CCMI = crash cost savings per minor-injury crash reduced ($); CCPDO = crash cost savings per property- damage-only crash reduced ($); DSBFSI = annual number of fatal-and-severe- injury crashes reduced as a direct safety benefit of the design treatment; DSBMI = annual number of minor-injury crashes reduced as a direct safety benefit of the design treatment; and DSBPDO = annual number of property-damage- only crashes reduced as a direct benefit of the design treatment. The crash severity levels used in the benefit–cost analysis are derived from the KABCO scale of crash severity levels (K = killed; A = incapacitating injury; B = nonincapacitating injury; C = possible injury; O = no injury; and U = injured, severity unknown) for which the Federal Highway Administration has developed crash cost estimates (9). Severe-injury crashes, as this term is used in the benefit–cost analysis, are equivalent to incapacitating injury crashes (also known as A-injury crashes in the KABCO scale). Fatal and severe-injury crashes are combined in the benefit–cost analysis because, if fatal crashes were considered alone, the ran- dom occurrence of a single fatal crash might influence the analysis results too strongly. Minor-injury crashes include both noninca- pacitating injury crashes (also known as B-injury crashes) and possible-injury crashes (also known as C-injury crashes). The safety benefits from the congestion reduction effects of the safety treatments are represented in Equation 6.8 by the terms NReductionFI and NReductionPDO. The methodology for deriving these terms has been presented in Chapter 5 in Equations 5.12 through 5.25. Each of the individual terms of the life-cycle benefit–cost methodology is discussed below. Implementation or Construction Cost The implementation or construction cost for a design treatment is the initial one-time cost to install or construct that treatment. This input to the assessment methodology is provided by the user. Highway agencies generally have good information on the cost of implementing treatments. Annual Maintenance Cost The annual maintenance cost for a design treatment is the recurring yearly cost of maintaining the design treatment

55 in place. Depending on the nature of the treatment, these costs could be incurred by either highway agency maintenance forces or contractors and could be either recurring costs to keep the treatment in repair or per incident costs to deploy the treat- ment or restore it after use. Annual maintenance costs are sup- plied by the user as an input for the assessment methodology. Minimum Attractive Rate of Return or Discount Rate The minimum attractive rate of return or discount rate (i in Equation 6.4) represents the time value of capital invested in design treatments to reduce nonrecurrent congestion and improve reliability. The discount rate is used to reduce future costs and benefits to their present values so they can be com- pared on a common basis. The suggested default value of the discount rate is 7%. This value of the discount rate was chosen on the basis of Office of Management and Budget Circular A-94 (10), which specifies a real discount rate of 7% for analysis of public investments. Circular A-94, which has been U.S. government policy since 1992, has been reis- sued within the last year with the discount rate provision unchanged. Service Life The service life (n in Equation 6.4) of design treatments varies over a broad range from five (or fewer) to 20 years (or more). It is possible that, due to traffic volume growth, some design treatments may lose their effectiveness in reducing non recurrent congestion before the end of their physical life. Such treatments may be considered to become functionally obsolescent. This possibility should be considered in choos- ing the service life for a treatment. Change in Annual Traffic Operational Delay The change in annual traffic operational delay for a specific design treatment during a specific hourly time-slice (∆Dk in Equation 6.7; also referred to as ∆LIk in Equation 4.39) is computed with a procedure documented in Chapter 4. ∆Dk is derived directly from the area between the treated and untreated travel time index (TTI) curves using the approxi- mation shown in Figure 4.7 and Equation 4.40. Once the treated and untreated TTI curves have been established for a design treatment, the computation of ∆Dk using the proce- dure based on Figure 4.7 and Equation 4.40 is performed in the same way for every design treatment. The methods for determining the treated TTI curves vary by design treatment and are illustrated in Chapter 4 under Quantifying Design Treatment Effects on Reliability by Using the Cumulative TTI Curve. Change in the Standard Deviation of Travel Time The standard deviation of travel time for a specific design treat- ment during a specific hourly time-slice (∆sk in Equation 6.7) is computed with a procedure documented in Chapter 4 under Change in Variance. ∆sk represents the difference between the standard deviations of the treated and untreated TTI curves, like the example curves shown in Figure 4.4. The stan- dard deviation of either the treated or untreated TTI curve can be determined with the approximation shown in Figure 4.8 and Equation 4.42. ∆sk is then determined as the difference between those standard deviations, as shown by Equation 6.9: ( )∆σ = σ − σ 6.9untreated, treated,k k k where suntreated,k = standard deviation of travel time (h) for the untreated condition, derived from an untreated TTI curve like that shown in Figure 4.4; and streated,k = standard deviation of travel time (h) for the treated condition, derived from a treated TTI curve for a design treatment, like that shown in Figure 4.4. Values of travel time and reliability This section presents the approach used in the Analysis Tool to quantify the value of reliability. Figure 6.1 illustrates a typi- cal travel time distribution curve shown by Warffemius (11). The distribution is skewed with a relatively long tail toward higher travel times, which is typical of data for unreliable conditions. The mean travel time shown in Figure 6.1 rep- resents the travel time for the average motorist. The differ- ence between the mean travel time and the ideal or free-flow travel time (labeled in the figure as the travel time without delays) represents the average delay to motorists under the prevailing conditions. Value of Travel Time and Delay In economic studies, the value that a person places on his or her time spent traveling can be determined based on revealed preference or stated preference studies. In a revealed prefer- ence study, the subjects indicate the trade-offs they are willing to make between time and money through real-life decisions. These types of studies can be very difficult to set up and mea- sure. In a stated preference survey, respondents are presented choices that help researchers determine their willingness to trade money for time and vice versa. These studies are much

56 results of studies on the value of travel time and delay reduc- tion. Based on a review of these studies, Concas and Kolpakov (12) made the following recommendations concerning the value of travel time and delay reduction for use in benefit– cost studies: • Personal travel time (including commuter travel) should be valued at 50% of the prevailing wage rate. • On-the-clock paid travel (e.g., commercial vehicle driver) should be valued at 100% of the driver’s wages plus benefits. • The use of the national average wage rate is recommended as the basis for determining the value of time unless reli- able information on the earnings of particular users of a transportation facility is available and these earnings are significantly different from the national average. The most recent available estimate (May 2009) of the national average wage rate from the Bureau of Labor Statis- tics is $20.90/h (27). The default value used for the value of travel time in the Analysis Tool is $15.68/h. Users may replace this value with any value considered more appropriate for their local condition. Value of Reliability Warffemius (11) makes the case that variability (i.e., the vari- ance or standard deviation of the travel time distribution) is a useful measure of reliability. The greater the variance or simpler to conduct, and it is assumed that respondents’ stated preferences would be close to their revealed preferences in most cases. In transportation benefit–cost studies, the value of travel time and, thus, the value of delay reduction is typically con- sidered to be a percentage of the prevailing wage, with dif- ferent percentages assigned to the various trip types. The primary division of trip type is between work trips and non- work trips. Work trips are those that are conducted on the job, in which the cost to the employer is the total of the wage and benefits of the driver, plus the same for any employee passengers. Freight trips, as a subcategory of work trips, may have additional costs per hour if the freight is time sensitive, as in the case of perishable goods. Generally, nonwork trips are valued at a lower rate than work trips. For nonwork trips, the value of time may be greater for the driver than for the passenger, since the passenger could participate in other activities while in the car and is not required to dedicate their time to the task of driving. Passengers who are children also have a lower value of time, since their time cannot be converted into wages. Nonwork trips may also be catego- rized by trip purpose, such as commuting to work, commuting from personal errands, and leisure trips, as these may all have different values. Because the value of time will vary from person to person and from trip to trip, simplifying assumptions need to be made to determine travel time savings benefits for a specific treat- ment on a given roadway, as users and trip types will be diverse over the life cycle of the treatment. Table 6.1 summarizes the Figure 6.1. Typical example of travel time distribution curve used to estimate delay and reliability. Source: Warffemius (11).

57 Copley et al. defined the reliability ratio explicitly as the “value of 1 minute of standard deviation”/“value of 1 min- ute of travel time.” The method for estimating the standard deviation of travel time presented previously in this chap- ter under Change in the Standard Deviation of Travel Time can be used to implement this concept. Warffemius (11) states that the average travel time and its variation (i.e., standard deviation) can be presented in stated preference surveys in such a way that these attributes are not correlated. As a consequence, the economic benefits of travel time savings and reliability improvements can be added together without the risk of double counting. This supports the combination of these two types of benefits by addition, as shown in Equation 6.8. standard deviation of the travel time distribution, the greater the unreliability of travel times. Warffemius indicates that the value of reliability can be expressed as a multiple of the value of travel time, with that multiplier referred to as the reliability ratio, as shown by Equation 6.10: ( )= ρVOR VOT 6.10 where r is the reliability ratio. Warffemius indicates that Copley et al. (28) estimated the reliability ratio as equal to 1.3 on the basis of a stated preference survey among commuters in Manchester, United Kingdom, who used their car as solo drivers on their journey to work. Table 6.1. Results of Studies on the Value of Travel Time and Delay Reduction (11) Study Year Data Used VOT Estimate Becker (13) 1965 40% of wage rate Beesley (14) 1965 Data from survey of government employees in London, United Kingdom 31% to 50% of wage rate Lisco (15) 1967 20% to 51% of wage rate Miller (16) 1989 Survey of multiple route choice models 60% of gross wage (on average) Small (17) 1992 Values derived from multiple mode choice transportation models 20% to 100% of gross wage; 50%-reasonable average Waters (18) 1992 Travel data from British Columbia, Canada 50% to 100% average wage rate for personal travel, depending on LOS; 120% to 170% of average wage rate for commercial travel, depending on LOS Waters (19) 1996 Travel data from 15 commuting studies in North America 40% to 50% of after-tax wage rate (mean: 59% of after-tax wage rate; median: 42% of wage rate) Calfee and Winston (20) 1998 Data from National Family Opinion survey covering commuters from major U.S. metropolitan areas 14% to 26% of gross wage; 19% of wage-average estimate Small and Yan (21) 2001 Data on commute travelers on SR-91 in California Average VOT is $22.87/h, or 72% of sample wage rate Brownstone and Small (22) 2003 Travel data from ETC facilities in HOT lanes on SR-91 and I-15 in Southern California VOT saved on the morning commute: $20 to $40 per hour, or 50% to 90% of average wage rate in the sample U.S. DOT (9) 2003 Estimates are based on multiple sources of data 50% to 120% of the wage rate depending on type of travel (personal versus business); 50% of wage rate for personal local travel; 100% of wage rate for commercial local travel Small et al. (23) 2005 Travel from SR-91 in greater Los Angeles, Calif., area, collected over 10-month period in 1999 to 2000 Median VOT is $21.46/h or 93% of average wage rate Tseng et al. (24) 2005 Data collected in June 2004 for Dutch commuters who drive to work two or more times per week Mean VOT for all travelers: 10 euros/h (approximately $12.10/h) Litman (25) 2007 Results are drawn from multiple travel time studies 25% to 50% of prevailing wage (for personal travel) Tilahun and Levinson (26) 2007 Data from stated preference survey of travelers on I-394 in Minneapolis–St. Paul, Minn., area $10.62/h for MnPass (ETC system) subscribers who were early/on-time $25.42/h for MnPass subscribers who were late $13.63/h for nonsubscribers who were early/on-time $10.10/h for nonsubscribers who were late Note: VOT = value of time; LOS = level of service; ETC = electronic toll collection; HOT = high-occupancy toll; DOT = Department of Transportation.

58 reliability of a morning commute depends on the importance of arriving at a certain time: some jobs have set start times for which late arrivals can have significant consequences, but other jobs have flexible start times and a late arrival has a much smaller impact. The reliability of the evening commute has a lower value for most people because the arrival time at Table 6.2 presents a broader set of research results that have quantified the reliability ratio. As with travel time, travel time reliability is valued differently depending on the trip type and the person making the trip. For example, when driving to the airport to catch a flight, reliability is highly valuable, since unexpected delay can result in a missed flight. The value of Table 6.2. Results of Studies on the Value of Reliability Study Authors No. of Respondents Trip Type Reliability Ratio Copley et al. (28) 167 Mostly work commutes 1.3 Black and Towriss (31) 354 Car travelers 0.79 Small et al. (32, 33) NA Commute to work 1.3 Halse and Killi (34) 505 Shippers 0.68 Parsons Brinckerhoff (35) NA High income (60K+) To work 0.8 From work 0.6 Nonwork 0.4 Low income (<60K) To work 1.0 From work 0.3 Nonwork 0.2 NA Trip distance— work related 5 mi 1.88 10 mi 0.94 20 mi 0.47 Trip distance— nonwork related 5 mi 2.02 10 mi 1.02 20 mi 0.51 Black and Towriss (31) NA Car trips to and from work 0.55 All trips in sample 0.70 Asensio and Matas (36) NA NA 0.98 Noland and Polak (37) NA Commuting 1.27 Bates et al. (38) NA NA 1.1 Ghosh (39) NA NA 1.17 Yan (40) NA NA 1.47 Small et al. (33) NA NA 0.65 Bhat and Sardesai (41) NA NA 0.26 Hollander (42) NA NA 0.10 Tilahun and Levinson (26) NA NA 0.89 Carrion-Madera and Levinson (30) NA NA 0.91 Hensher (43) 198 Long distance (<3 h) 0.57 Small et al. (33) 5,630 Commute 3.5 Lam and Small (44) 332 Male 0.66 Female 1.4 Small et al. (23) 1,155 Commute 0.91 Brownstone and Small (22) 601 Commute 0.4

59 Chapter 4 discusses assumptions that can be applied for various treatments, summarized as values of pi in Table 4.8. Only treatments that eliminate crashes (Classes IIA and IIB) have these direct safety benefits. Other treatments that reduce crash incident duration or otherwise reduce crash conse- quences would not have such benefits because they do not reduce the number of crashes. Direct safety benefits may be used, if desired, to supplement the congestion-related effects on safety. Crash Costs (CFSI, CMI, and CPDO) Most highway agencies assign a cost savings to crashes reduced for each level of crash severity, based on either their own experience or on published values from the U.S. DOT or the National Safety Council. The benefit–cost analysis method- ology developed by the research team uses the default values shown in Table 6.3, which were taken from recent U.S. DOT data (9) as adapted for use in SafetyAnalyst (45), but agen- cies are free to replace these values with values from other sources, such as the Insurance Institute for Highway Safety, or with their own agency’s values, as appropriate. home is less important that the arrival time at work. The reli- ability of personal errands or leisure trips is also expected to be less valuable than a morning commute, because it is expected the arrival time is much less important for these trips. Freight trips may have a very high value of reliability, especially when delivery logistics are based on just-in-time deliveries, and late arrivals can have an impact on produc- tion. A review of the value of reliability was conducted at the SHRP 2 Reliability Workshop on the Value of Travel Time Reliability and Cost–Benefit Analysis (29). An overview and meta-analysis on this topic completed in 2010 is provided by Carrion-Madera and Levinson (30). As accounting for travel time reliability is a relatively new concept in transportation benefit–cost analyses of roadway improvements, agencies are less likely to have developed reli- ability values than travel time values specific to their roadways and drivers. The default value for the reliability ratio used in the Analysis Tool is 0.8, but agencies are encouraged to use values appropriate for their own state or metropolitan area, if available. Costs of Crashes Crash Cost Reduction Due to Congestion Reduction (NReductionFI and NReductionPDO) The crash cost reduction due to congestion reduction has been estimated based on the crash rate–traffic density rela- tionships presented in Chapter 2 under Evolution of Research Approach for Traffic Operational Analysis and summarized in Equations 5.12 through 5.25. Crash Cost Reduction Due to Direct Safety Benefits of Design Treatments (DSBFSI, DSBMI, and DSBPDO) Some design treatments have direct safety benefits apart from their potential congestion reduction effects (i.e., they reduce crashes even when installed on uncongested facilities). Table 6.3. Default Values of Crash Costs by Severity Level Severity Level Cost Savings per Crash Reduced ($) Fatal and severe injury 1,908,000a Minor injury 51,000b Property-damage only 4,000 a Weighted average crash cost based on costs of $5.8 million for fatal crashes and $402,000 for incapacitating-injury crashes. b Weighted average crash cost based on costs of $80,000 for nonincapacitating-injury crashes and $42,000 for possible-injury crashes.