Overview

Based on the results of the literature review, communications with States, discussions with experts, and review of the Office of Management and Budget (OMB) guidance (13), an appropriate methodology was developed to meet the objectives of this effort given the available information.  

The literature review showed there were no established methodologies to estimate the benefit of collecting roadway data elements for safety. No one State was identified that has collected the exact list of FDE/HSIP on all public roadways within the State.  Therefore, an analysis of collecting all 37 FDE/HSIP to determine the safety benefits was not feasible at the time this investigation was conducted.

An alternate approach was developed to conduct the market analysis.  The costs for data collection were provided from several vendors and one State DOT that had been investigating conducting a similar effort. The number of fatalities and injuries that would need to be reduced in order for the monetized benefits to exceed the costs was estimated to determine the benefits.  The methodology developed is a hybrid of a benefit-cost analysis and a cost effectiveness analysis.

The cost estimations developed for this analysis reflect the additional costs that States would incur based on what is not already being collected through HPMS or not already being collected through other efforts.  At the time of this analysis, the FHWA did not know the extent of data collection practices for all States beyond HPMS requirements.  In order to accommodate a range of data collection practices among the States, the methodology for the analysis was conservatively based on the assumption that all data collection beyond HPMS requirements would be new collection.  Therefore, this analysis of the additional cost to States is most likely greater than the actual cost that would be incurred.  Individual cost estimates would vary by the circumstances in each State.

A location referencing system is already required under HPMS for all Federal-aid highways. In addition, 16 of the 38 FDE/HSIP are also already required for collection under the HPMS for the full extent of Federal-aid highways (1).  Full extent accounts for all Federal-aid highways and ramps located within grade-separated interchanges (i.e., NHS and all functional systems excluding rural minor collectors and locals). Table 3 indicates which of the 38 FDE/HSIP are HPMS full extent elements.

Data Collection Costs

The additional costs identified include the following three sets of data elements:

• A common relational location referencing system.
  • Additional costs would only be incurred on all non-Federal-aid roadways, since HPMS currently requires this for Federal-aid highways.
• The 22 FDE/HSIP that are not required under HPMS.
  • Additional costs would only be incurred on Federal-aid roadways, since 16 of the 37 total FDE/HSIP are already required for HPMS on Federal-aid highways.
• The complete 38 FDE/HSIP.
  • Additional costs would be incurred on all non-Federal-aid highways, since HPMS does not require data collection of these elements on non-Federal-aid roadways.

In order to conduct the analysis, costs were obtained from 12 data collection vendors from around the country. Costs were obtained from the vendors on a per-mile basis along segments, a per-intersection basis for intersections, and a per-ramp basis for ramps.  The costs for developing a location referencing system were estimated per mile.  For the case of traffic counts on segments, an estimate of one count per mile was used to estimate to generate a per mile cost.  These costs included data collection and reduction for integration into a State’s existing system.

Vendors were identified based on the list of vendors involved in the North Carolina and the Transportation Research Board Strategic Highway Research Program 2 (TRB SHRP2) data collection rodeos which were both conducted in 2008.  These rodeos were conducted to test the capabilities of roadway data collection technologies.  Many of the rodeo vendors only collected roadway inventory elements and not traffic counts, so the project team also identified several companies that collect traffic counts to obtain cost estimates.  The (non-traffic) roadway elements are collected using different methods than the traffic data, and, therefore, the costs for each were calculated separately.

The majority of vendors estimated that they would use digital data collection vans to collect the roadway inventory data.  For traffic count data, vendors provided cost estimates based on 48-hour classification counts for segment traffic data, peak hour manual counts for intersections, and technology similar to segment counts to collect ramp data.  The costs provided were averaged to develop estimates.

The analysis was based on information provided by vendors and reflects the methods and costs that would be used if the collection was contracted.   There are other methods of collecting some of these data elements, including extracting the data from existing plans or visual imagery such as aerials or Google Earth.  Some of these methods may be lower in costs, particularly if the cost of agency personnel are not included as part of the costs.

In addition, several State DOTs were contacted to obtain estimates of what the costs would be to collect these “in-house” rather than contract the data collection out to a vendor.  Since the data collection was very specific to this list of FDE/HSIP, the majority of States contacted could not provide an estimate of costs.  Only one State that was considering conducting a similar effort provided cost information. However, that State was only considering the collection on intersections.  The analysis was conducted using the estimates provided by the vendors, acknowledging that these are conservative estimates, and there may be more cost effective methods available (but information for those methods was not available).

Benefits

For the purposes of this analysis, the benefits were calculated by developing an estimate of the benefits needed to exceed a 1:1 ratio and a 2:1 ratio of benefits to costs.  The first step in the analysis was to calculate the cost of a fatality and the cost of an injury.  The 2008 comprehensive cost of a fatality used in the analysis was $6,339,701 and $516,947 for an injury, based on information provided by the National Highway Traffic Safety Administration (NHTSA) (2).  The injury costs reflect a Maximum Abbreviated Injury Scale (MAIS) Level 3 injury.  MAIS injuries are on a scale of 0-5, with 5 being the most severe non-fatal injury.  MAIS Level 3 was chosen as a “mid-point” in this scale.

The future cost of a fatality and injury were forecasted out 20 years to 2031 and then discounted to reflect 2010 dollar values.  A discount rate of 7.0 percent was used.  The benefit estimation assumed that the benefits would not be realized until the data collection is complete: 2021 for Federal-aid and 2022 for non-Federal-aid roads.
An average of the costs of a fatality and the cost of an injury from 2021 through 2032 was calculated.  This calculation provided the cost of a fatality and the cost of an injury, represented in 2010 dollars.

In order to determine the balance of the number of fatalities and injuries that should be used in calculating the benefits, a ratio of the number of fatalities to injuries was calculated using 2009 crash data.  In 2009 there were 33,808 total fatalities and 2,217,000 total injuries (14), equating to a fatality to injury ratio of approximately 1:66.  Using that ratio, the number fatalities and injuries needed to exceed a 1:1 ratio and a 2:1 ratio of benefits to cost was developed for each State, and for each scenario.

Classification of Roadway Ownership

Costs were aggregated to a State level to estimate the reduction in crashes (fatalities and serious injuries) needed to exceed the costs.  The first step in the analysis was to develop a classification of roadway ownership for each State.  Three States were analyzed: an “average” State, a small State (information from Rhode Island was used to represent a small State), and a large State (information from California was used to represent a large State).  To calculate the costs for each State, the mileage, number of intersections, and number of ramps was determined for the Federal-aid and non-Federal-aid roadways.  The mileage was obtained from the FHWA Office of Highway Policy Information (OHPI) (15).  The mileage for the “average” State was calculated using the U.S total (including Washington, DC) and dividing by 51.

There is not yet an estimate of intersections or ramps available through the OHPI.  Therefore, the project team contacted States directly to obtain estimates of the number of intersections and ramps in each State.  In addition to a large State and a small State, Missouri and Ohio were contacted to obtain estimates for the “average” State.  Missouri and Ohio were chosen to represent the “average” State based on their land mass, roadway mileage, and geographic locations.

All of the States except California were able to provide the total number of intersections and ramps in the State.  In California, the total number of intersections was estimated based on the total number of miles in the State. The distribution of mileage of Federal-aid and non-Federal-aid roadways was used to estimate the same distribution for intersections.  This assumes there is the same density of intersections per mile on each roadway set.

The number of ramps provided by the States was applied to the Federal-aid roadways with the justification that there would most likely not be ramps on non-Federal-aid roads.  The distribution of roadway ownership by State, mileage, intersections, and ramps is shown in Table 4.

Table 4. Breakdown of Roadway Ownership.

Aggregated Costs

The aggregated costs were developed for two scenarios:

• Scenario 1:
  • Develop a common statewide relational location referencing system on all public roads that is linkable with crash data (i.e., GIS, linear referencing system, etc).
    • Additional costs would be incurred for developing a referencing system on all non-Federal-aid roadways, since HPMS currently requires this for Federal-aid highways.
  • Collect the FDE/HSIP on all Federal-aid highways.
    • Additional costs would be incurred for collecting the 22 FDE/HSIP that are not required under HPMS on roads since 16 of the total FDE/HSIP are already required for HPMS on Federal-aid highways.
• Scenario 2:
  • Develop a common statewide relational location referencing system on all public roads that is linkable with crash data (i.e., GIS, linear referencing system, etc).
    • Additional costs would be incurred for developing a referencing system on all non-Federal-aid roadways, since HPMS currently requires this for Federal-aid highways.
  • Collect the FDE/HSIP on all Federal-aid highways.
    • Additional costs would be incurred for collecting the 22 FDE/HSIP that are not required under HPMS on roads since 16 of the total FDE/HSIP are already required for HPMS on Federal-aid highways.
  • Collect the FDE/HSIP on all non-Federal-aid roads.
    • Additional costs would be incurred for collecting all 38 FDE/HSIP all non-Federal-aid highways, since HPMS does not require data collection of these elements on non-Federal-aid roadways.

For both Scenarios, the first two initiatives involve developing a statewide relational location referencing system on all public roads, and collecting the FDE/HSIP on all Federal-aid highways.  Scenario 2 adds a third initiative of collecting the FDE/HSIP in all non-Federal-aid roads. 

The base cost for mileage, intersections, and ramps was then disaggregated annually.  The time frames include five years for segments, seven years for intersections, and nine years for ramps on Federal-aid roads; and six, eight, and ten years respectively for non-Federal-aid roadways.  These time frames were selected as reasonable intervals for State and local agency collection of the FDE/HSIP data.

For this analysis, data collection was estimated to begin in 2012 and continue for nine years to 2020 for Federal-aid roadways, and continue for ten years to 2021 for non-Federal-aid roadways.  The analysis assumes an equal distribution of costs over the data collection period.

In addition to the costs of initial data collection, the costs to maintain the data were also calculated.  That is, the costs to update the data as conditions change.  For segment data, it was assumed that five percent of the roadway mileage would be updated annually.  These updates would not be done by re-collecting the data, but rather based on updates from construction/design plans.  It was approximated that updating the inventory would take two hours per mile by an employee earning $20.00 an hour (approximately $40,000 per year).

The intersection inventory would be updated on a three-year cycle for signalized intersections and a five-year cycle for unsignalized intersections.  This assumes that traffic volumes will not change dramatically at unsignalized intersections.  A split of 20 percent of signalized intersections and 80 percent of unsignalized intersections was used to determine the number of signalized and unsignalized intersections.  In addition, the cost of inventory updates was also included in the maintenance costs.

The analysis assumes that a ramp inventory would be updated on a six-year cycle, with counts and inventory updates collected on one-sixth of the ramps per year.

The costs for coding and locating crashes on non-Federal-aid roads was also calculated since States would now have the information they need to locate crashes, which they would not have had previously.  National statistics were obtained from NHTSA to estimate a ratio of fatal crashes to injury crashes (14).  The number of fatal crashes on non-Federal-aid roads was obtained from the NHTSA Fatality Analysis Reporting System (FARS) (16). The ratio of fatal to injury crashes was applied to the number of fatal crashes on non-Federal-aid highways to obtain an estimate of injury crashes on non-Federal-aid highways. Costs for locating and coding these additional crashes were then applied. It was assumed that five crashes could be coded per hour at a cost of $20/hour. These costs only pertain to the costs of coding and locating fatal and injury crashes.  The number of property damage only crashes on non-Federal-aid roads could be not reasonably estimated; therefore, they were not included.

These assumptions on data collection cycles, maintenance, and crash coding were based on standard practices obtained through discussions with several States.

Once the costs for collection, maintenance, and coding of the data were determined, they were summed to establish a total cost per year out to 2032.  This timeframe would allow for the total 10-year data collection period (for non-Federal-aid roads) and an additional 10 years of implementation.  These costs were then discounted using a 7.0 percent discount rate per the OMB guidance to bring the costs back to 2010 values (13).  Once the costs were all in the same value year, an annual average cost during the data collection period and an annual average cost during the maintenance period was calculated.