Resource Allocation of Available Funding to Programs of Work (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.

11 chapter two Review of LiteRatuRe and PRactice This chapter summarizes the literature review and review of typical agency practices, the findings of those reviews, and their relevance to the overall study. Because of the nature of the topic, resource allocation is addressed in the literature primarily through conference and peer review proceedings, state-of-the-practice reviews, and other documentation of transportation agency experiences. There is a relatively limited body of technical methodological research, academic publications, and reports in technical journals. Thus, although academic research and documentation of practice are included in this review, the findings are heavily oriented toward published documentation of agency experi- ences. The review is organized to present different types of programmatic approaches and resources based on the context of the allocation decision, the approach to allocation, and the methods or tech- niques used to inform the allocation process. Some of the discussion focuses on how practice has developed in addition to the state of practice. This discussion builds on the development of strategic resource allocation as an issue for DOTs, as described in the introduction. The major conferences and peer exchanges addressing resource allocation issues in state DOTs include: • Innovations in Statewide Planning (2005), • Key Issues in Transportation Planning (2006), • Statewide Transportation Planning: Making Connections (2006), • Transportation Planning Capacity Building Program (ongoing), • Chicago Metropolitan Agency for Planning Peer Exchange on Performance-Based Planning (2012), • Performance-based Planning and Programming in the Context of MAP-21 (2015), and • Cross-Modal Project Prioritization (2015). These conferences and peer reviews also contain several presentations by federal agencies and MPOs involved in resource allocation policy and decisions. The literature and state-of-the-practice review focuses mostly on state agencies’ perspectives on resource allocation. PRogRam aLLocations within Business PRocesses Much of the research on programmatic allocation decisions documents programmatic allocations as outcomes of larger planning or investment decision processes. These examples are important because they are indicative of the link between planning and programming. Although some agencies’ programming is determined largely by historic trends or executive direction (Cambridge Systematics 2001; Spence and Tischer 2007; Turnbull 2008), many agencies’ program funding levels are closely constrained by the statewide long-range transportation plan (LRTP) or are incidental to the state transportation improvement program (STIP) project programming. allocations within Long-Range transportation Plans Transportation Research Circular E-C091 provides an example of the variety of contexts for resource allocation. The document summarizes the nine-state peer exchange Innovations in Statewide Plan- ning, which took place in 2005 (Young 2006). Michigan describes its programming process, which includes new efforts at annual investment planning in alignment with the LRTP and annual rebalanc- ing of 11 different highway programs based on strategic priorities. The process increased the role

12 of executive leadership and the transportation commission in adjusting program levels while still recognizing the goals documented in the LRTP. Other states’ contributions at the peer exchange are discussed throughout this chapter. The 2012 study “Trends in Statewide Long-Range Transportation Plans” intentionally constrains its review to the level of planning documents and does not examine resource allocation issues in execut- ing plans (Lyons 2012), despite being an update to FHWA’s 1995 Examples of Statewide Transporta- tion Planning Practices, which includes several discussions of relationships between planning and programming and the inclusion of multiple modal programs in the planning effort (FHWA 1995). “Best Practices in Using Programmatic Investment Strategies in Statewide Transportation Plans,” a report for NCHRP 8-36 Task 67, looks at best practices for focusing the planning process on stra- tegic decisions about programmatic investments and provides guidance for a ten-step programmatic investment strategy (Janik 2007). The report focuses specifically on programming strategies as an explicit objective of the planning process. This provides a unified framework for the statewide planning process that builds on previous research into different aspects of plan development. The report synthesizes best practices from several states and provides strategies that states might apply for their own situation. These examples show how, to a large degree, programmatic allocation decisions and strategies arise from wider investment management efforts enabled through long-range planning or compliance with federal transportation law. allocations through state and other transportation improvement Programs development In these examples, programmatic strategies have evolved through top-down statewide and federal processes, but programmatic strategies also have developed cooperatively from state DOT STIP and MPO TIP prioritization processes. This is especially true in settings in which MPO TIP processes account for a large share of overall transportation outlays. In 2005, New Mexico described its current programming processes as being driven by the MPO and district TIPs without strong consideration of an LRTP (Young 2006) (Figure 2). In 2006, TRB sponsored the major conference Key Issues in Transportation Programming (Turnbull 2008). A major emphasis of the conference was on MPO and state roles in development of the TIP and Figure 2 New Mexico’s project programming process circa 2005: STiP-driven and disconnected from a long-range plan.

13 STIP. One presentation discussed how the increasing degree of financial planning occurring within the STIP process limits the influence of the LRTP and strategic goals in setting program funding. Most participants and presenters were MPO or local government officials, emphasizing the key role that the DOTs’ partners play in resource allocation. MPOs are responsible for a significant portion of existing assets and expected improvement needs in many states. Washington State DOT gave a presentation on how the agency’s resource allocation process is aligned with the LRTP through a 10-year investment plan. Washington is among several states using a 10-year investment plan beyond the STIP horizon to bridge the long-range planning and short-range programming perspectives. Other DOT comments pertained to financing challenges, cost estimation, risk assessment, and responding to public ballot and legislative changes. States that delegate a significant amount of project programming to MPOs and regional plan- ning organizations may struggle to make strategic decisions about program funding levels, given the highly decentralized process. However, many states have succeeded in cooperating with MPOs to accomplish strategic goals, even when most allocation decisions are geographic rather than pro- grammatic (Turnbull 2008). Pennsylvania is a state in which local and regional entities have a strong role in programming, but the state has been successful in establishing structures for coordinating statewide priorities using cross-cutting working groups (Janik 2007). Kentucky’s 2005 peer exchange presentation focuses on how a change to a shorter, policy-based, long-range plan allows the state’s programming to be less constrained by a project-heavy plan and more responsive to scoring criteria during the STIP process (Young 2006). This response and presen- tation is also instructive as an example of how Kentucky previously had a resource allocation process that was constrained by a long-range plan that was updated infrequently. other Program allocation contexts Resource allocation decisions are not constrained to the federally required STIP and LRTP processes. Program allocation decisions arising from the long-range plan and STIP are among the most common contexts in which states execute their resource allocation strategies, but a variety of other processes can affect or be affected by program allocations. NCHRP Synthesis 480: Economic and Development Implications of Transportation Disinvestment reports several cases of unintentional programmatic “disinvestment” over time; such disinvestment has consequences and implications in the wider economy, even if the consequences are not intended or considered by decision makers (Duncan and Weisbrod 2015). Other cases presented by Duncan and Weisbrod (2015) show how funding was reallocated among programs to focus on strategic objectives. Park (2012) compiled the final report “Leading in Lean Times: Maximizing Resources in a Constrained Environment” from a workshop of 13 DOT chief executive officers. Focused on the challenge of growing needs and falling revenues, this series of panels attempted to apply lessons from the private sector to agencies’ strategic initia- tives and business processes. Discussions included how to better align program funding with strategic goals given the inability to fund all desirable functions. These discussions reflect the pressure many agencies feel to realign program priorities, as an intentional change to current programming strategies, without waiting for an LRTP process. NCHRP Synthesis 358 discusses how cost estimation links planning, programming, and delivery together because the accuracy and reliability of cost estimates affect all areas of the DOTs’ responsi- bilities (Anderson et al. 2006). Shifting costs can lead program allocations to diverge from planned splits, which some agencies address by maintaining a second planning document beyond the STIP horizon so that potential projects consistent with the agencies’ strategic priorities are already vetted for amendment to the STIP. Perhaps more common are the situations in which major capital projects come in over budget and delay other projects that have received a funding allocation. This is one scenario in which program allocations may diverge significantly from those expected in the LRTP or STIP, and projects may begin to drive programmatic allocations. Project-focused programming contexts can also be intentional when agencies focus on specific facilities or improvements as their core responsibility. Project-focused programming is a significant

14 focus of performance measure discussions and cross-modal allocation methodologies, as seen in the following sections. These include a discussion of the tools used for allocating resources by build- ing up from projects or comparing overall programs. Overall, states can create their own policies or contexts for making resource allocation decisions and can shape their programmatic investment outcomes through any number of business processes beyond the recurring STIP and LRTP. The case examples in chapter four explore how events such as performance audits and changes in state law can induce states to develop resource allocation that subsequently plays a large role in determining how STIPS and LRTPs ultimately treat resource allocation decisions. PeRfoRmance-Based PRogRamming The passage of MAP-21 advanced discussions of programmatic investment trade-offs and strategies to meet long-term needs through its focus on programming to target specific performance measures. A white paper, Performance-Based Planning and Programming, builds on discussions from a national forum and a workshop on performance-based planning and programming to focus specifically on performance–data-driven programming strategies (Louch 2012). Louch describes performance-based programming as requiring a bottom-up, project-based resource allocation strategy that can identify projects that advance agency goals and performance targets. Louch and other recent guidance, discus- sion, and research focus on the definition of performance measures and setting of targets (Volpe Center 2012; Middleton and Regan 2015). These perspectives see performance-based planning as a linear process. However, there is a significant shortcoming in data and information-related operationalizing performance measurement to make resource allocation decisions during later steps in the performance management process. The FHWA’s Performance-Based Planning and Programming Guidebook suggests the importance of prioritization in achieving performance targets but acknowledges the difficulty in linking this pro- cess to investment priorities in long-range plans (Grant et al. 2013). These documents emphasize the same focus on project prioritization that North Carolina and other states are exploring. This strategy is attractive if the performance impacts of individual projects are sufficiently defined and comparable with all other spending with which they are compared. In some cases, project performance measures may be more intuitive or relevant than are programwide performance measures. However, they may require significantly more data to develop and evaluate during programming cycles. Just as ISTEA led to several conferences on how programming process would change in response to the federal legislation, conferences and peer exchanges have been organized to help states and MPOs work together to address programming challenges, especially through the joint FHWA and FTA Transportation Planning Capacity Building program. Most of these meetings have focused on the definition of performance targets. NCHRP Report 666 also provides guidance on target setting, although this is the explicit purpose of guiding resource allocation (Cambridge Systematics et al. 2010b). The 2012 peer exchange in Chicago discussed resource allocations to program areas based on strategic priorities as a precursor to performance-based prioritization of individual projects (Volpe Center 2012). Associating performance measures with specific programs or strategic goals could offer an additional compromise between an entirely project-driven programming process and resource allo- cation based largely on incremental changes to the previous year’s allocations. The peer exchange highlighted stakeholder input as an important means for setting a high-level programmatic investment strategy (Volpe Center 2012). ex Post analysis and Benchmarking Ex post evaluations and benchmarking (either to specific years or consistent programwide perfor- mance standards) are an emerging and critical aspect of programmatic strategies. No widely published studies exist on states integrating benchmarking with ex post evaluations in the long term as part of regular decision making at the program level. Many performance measures are still used for track- ing conditions and communicating with stakeholders (Grant et al. 2011; Turnbull 2013). To a large degree, this is a function of the limited number of years since adoption of MAP-21’s performance- based planning requirements and the need for ongoing research into consistent or best practices for

15 ex post evaluations. Some studies, including one conducted by the Appalachian Regional Com- mission, have looked at selected programs in an ex post evaluation, yet have been challenged by difficulties in obtaining retrospective performance and asset condition data sufficient to isolate programmatic investment levels in direct relation to system conditions and performance (Wilbur Smith Associates 1998). The Appalachian Regional Highway System was evaluated against a his- toric counterfactual by Jaworski and Kitchens (2016); however, these types of efforts have not been documented at the single-state level. The capacity to relate performance trends directly to program allocations has been developing. Young (2006) summarizes a California DOT presentation describing the early development of per- formance measures for systemwide investment planning and the importance of being able to create a historical profile of programming decision outcomes. California’s was one of the first state agen- cies to employ performance management for planning as a major factor in programming decisions. Minnesota DOT is also publishing scorecards showing program performance trends for measures described in the LRTP. These examples show that development of performance measures is progressing, but consider- able work remains to determine how to most effectively use performance to make allocations to programs. States are experimenting with a variety of ways to use additional data points provided by required performance measures; several states are trying to reconcile how performance can be compared across programs and modes. Whether a state is seeking to achieve cross-programmatic allocations using project performance measures or broader program measures, performance-based programming emphasizes the need for quantitative data and models to inform decisions. PRogRamming acRoss modes One of the greatest challenges facing agencies is the allocation of resources across modes. Because different modes serve distinct markets, lend themselves to different measures of transportation ben- efits, and often are associated with wider policy objectives, it is often difficult to arrive at an apples- to-apples comparison for allocating resources. For example, road performance measures based on standard speed and capacity make little sense for pedestrian infrastructure. In many cases, states do not have discretion to trade federal resources among different modal programs. Furthermore, in many state policy environments, such direct trade-offs can distract from the underlying policy objectives of the overall investment mix. In effect, although some states seek apples-to-apples comparisons, others simply seek a way to understand and articulate the rationale for their investment mix. Some of the earlier research on how states can assess the trade-offs of allocations across modal programs rather than relying on policy-level decisions is offered in NCHRP Project 8-36 Tasks 7 and 7(2), both titled Development of a Multimodal Tradeoffs Methodology for Use in Statewide Transporta- tion Planning (Cambridge Systematics 2001, 2004). The second report adds to case applications carried out by Washington State DOT. Despite a theoretical framework for cross-programmatic comparisons being developed in Phase I, Phase II was not able to successfully compare investments in Washington State ferries around Puget Sound with transferring funding to road programs in the area. The other case application completed was a corridor-level alternative analysis. The conclusion of Phase II was that there was still insufficient understanding of how to include all relevant program costs and benefits in terms of performance or other measurement techniques to compare programs efficiently. During a conference presentation in 2003, in which one focus area was “connections between planning and program delivery,” Minnesota DOT discussed the agency’s new planning process that incorporated a multimodal LRTP and performance measures for all modes (Shunk and Turnbull 2006). The changes were not expected to immediately reduce challenges in programming across modes because funding streams remained siloed and institutional structures remained mode based. Florida also presented on intermodal resource allocation challenges at the time. As one of the chapter three case examples discusses, Florida DOT has been able to decrease the degree to which modal programming is isolated in separate planning functions; however, the agency has not made significant progress in direct comparison of projects across modes. This challenge makes trade-offs between

16 programs difficult to assess. Consequently, progress toward performance goals is largely an issue of optimization within programs rather than more optimal reallocation to achieve goals across programs. Louch (2012) and Grant et al. (2013) reference Minnesota DOT’s continued improvements to planning and programming processes through integration of performance measures and targets in the LRTP and system investment plans, annual capital programs for highways, and the biennial budget request materials. These performance measures mostly affect resource allocations among different highway programs, with other modes using dedicated resources. NCHRP Report 806 explores and provides guidance on “how decision makers at transportation agencies can better analyze and communicate the likely impacts of system performance across mul- tiple investment types” to achieve performance targets (Maggiore and Ford 2015). This effort led to the production of the NCHRP 8-91 Cross Asset Allocation tool, which employs an objective-weighting and mathematical-optimization procedure to consider alternative allocations of resources across organizational units. The tool pools projects for and then maximizes a normalized measure of per- formance to fund programs subject to a given budget. State DOTs had an opportunity to explore this tool during a workshop in April 2014 (Maggiore et al. 2014). In December 2015, North Carolina DOT hosted the Transportation Planning Capacity Building peer exchange Cross-Modal Project Prioritization (Middleton 2015). The peer exchange included DOTs from Delaware, Oregon, and Virginia; MPOs from Rochester, New York, and San Francisco, California; and federal agencies and other North Carolina entities. North Carolina DOT’s resource allocation is based largely on project prioritization, and consequently the agency has been one of the most active in seeking to advance cross-modal comparisons of project outcomes. Because of chal- lenges with normalizing outcomes across modes (such as selecting and weighting quantitative criteria, maintaining local input, and standardizing input and outcome measures), North Carolina continues to rely on program allocations set by the department and stakeholders and then prioritizing projects within programs and between modes. Indiana provides one example of a state attempting to optimize across programs; however, Indiana has looked only at asset management programs on the state high- way system (Bai et al. 2011; Bai and Labi 2012; Bai 2012). As documented in NCHRP 8-36(7), a major barrier to comparison of modal programs is a lack of mode-neutral performance measures that capture not only the different user values but also the dif- ferent wider policy objectives associated with different modes. In an extensive review of multimodal planning and programming, Spence and Tischer (2007) discuss this challenge, as well as modal silos in terms of organizational structure and funding, and how various states are addressing these challenges within benefit–cost, least-cost planning, multicriteria evaluation and other mode-neutral frameworks. These tools are discussed in the following section. modeLs and data foR anaLyzing funding aLLocations Research shows many agencies arrive at program-level investment targets by using models to pre- dict future performance, deficiencies, and associated investment requirements as a basis for man- aging trade-offs (Cambridge Systematics, Inc. 2005; Maggiore et al. 2014). DOTs rely on a variety of models when developing resource allocation plans, including asset management, travel demand, and impact models. These tools help DOTs utilize a variety of types of information to understand the effect on investments. The models may be used to define needs-based resource allocations based on levels of funding suggested by tools or examine specific metrics calculated by models that serve as perfor- mance metrics for programs or projects. Often these tools’ outputs are compared using benefit–cost or multicriteria frameworks to make comparisons among projects and programs that affect different transportation performance measures. asset management or needs models Several models are available from U.S.DOT and private companies to assess investment needs, with the most developed systems available from highway and bridge assets. The most commonly

17 used needs models are the Highway Economic Requirements System (HERS-ST) (FHWA 2009); the National Bridge Investment Analysis System (NBIAS) (FHWA and FTA 2013); AASHTOWare Bridge Management (BrM) software (AASHTO 2013); and privately sold products that are useful for multiple asset classes. Bridge and pavement models have the longest history of use in resource allocation strategies, especially for preservation programs. They utilize a database of existing assets and use patterns for the state DOT’s system along with deterioration rates, project costs, user costs of deteriorated condi- tions, and investment triggers (minimally tolerable conditions). The current generation of models are designed to address specific asset classes or categories and (1) suggest levels of funding to reach user- specified system conditions or (2) forecast system outcomes given as user-defined levels of funding (Robert and Gurenich 2008; Ul-Islam and Hatcher 2008). FTA also uses a needs model called the Transit Economic Requirements Model (TERM) to assess nationwide needs, which are reported to Congress in periodic condition and performance reports along with nationwide results for highways using HERS and NBIAS. In 2012, FTA also released a version of the tool, called TERM Lite, that can be used by individual transit agencies. TERM Lite and several other recent innovations in transit asset management focus on the owners and operators of transit assets as users rather than state DOTs. Transit asset management is a rapidly evolving field with innovations that may spread to state DOTs (FTA 2008, 2010, 2012; Spy Pond Partners, LLC et al. 2012; Paterson and Vautin 2015). Most needs models utilize cost–benefit analysis, including life-cycle asset costs, to measure the impact of different funding levels against the economic cost to users and agencies. Accurate repre- sentation of life-cycle costs supports reallocation of resources among preservation, modernization, and capacity programs. Mizusawa and McNeil (2008) provide a case study in New Mexico, where needs assessment using HERS-ST allowed a pavement maintenance program to be optimized across the state’s facilities. Needs models do not take into account enough asset types to compare the efficiency of different program areas in achieving agencywide strategic objectives. For example, HERS-ST will be able to recommend only highway expansion (or in some cases modernization) as a solution to traffic growth (which may be artificially high because of a constant growth rate), rather than the possibility of bus rapid transit to address the need to move more people in a corridor. States must coordinate the use of several models and interpret outputs to understand interactions among program areas. NCHRP Project 20-57 developed two tools that could combine outputs from bridge and pavement asset man- agement systems to support trade-off analysis among programs over short-term and long-term hori- zons; these are described in NCHRP Report 545 (Cambridge Systematics et al. 2005). Li and McNeil (2011) provide another example of how multiple systems can be combined to assess trade-offs across programs managed by separate systems. Needs models require large data sets of asset conditions and by default use national averages for costs and effects. Improving the local applicability of these agency costs, user costs, and engineering and economic responses over time requires a significant amount of effort on the part of the agency. As an example of how default factors may be insufficient for accurate modeling of needs, when test- ing the cost of bridge closures in NBIAS, all facilities are given a detour length based on the area type (urban versus rural) and functional classification of the roadway the bridge carries. This overlooks the full strategic importance of this bridge, which depends on more accurate criteria regarding the type of truck routes supported or the commodity mix or larger value chain of deliveries supported by the bridge. Liu and Frangopol (2006) propose a network-level model for bridge needs that builds on the standard needs model frameworks. Needs models are used with asset management databases in many states to help develop the scale and scope of preservation programs and in some cases to propose needed levels of capacity invest- ment. They are useful for assessing programs that are structured around classes of facilities and state- wide performance goals. Modifying the performance triggers for investment can suggest optimal allocations, or allocation mixes can be tested for their impact on system performance and cost. Many

18 state DOTs utilize these types of models regularly in developing needs-based program allocations based on performance targets. travel demand models Although travel demand models are not investment management tools per se, they play an increas- ingly important role in how agencies understand likely future demands and consequently investment requirements. These models often determine how an agency understands the size of the market it is serving as a basis for deciding future transportation needs. Assumptions underlying such models profoundly shape the resource allocation discussion. The travel demand models are able to address some areas of transportation need and performance that current needs models do not, including (1) response of the travel system to changing demographics and economics, and (2) more mode- neutral travel performance assessments. An increasing number of state DOTs are utilizing travel demand models to estimate future system performance based on demographic and socioeconomic conditions and a network of facilities that could include multiple modes (Giaimo and Schiffer 2005). Freight-specific demand models are also available, including those developed for the federally available Freight Analysis Framework (FAF) and private data sources. Travel and freight demand models are especially useful to agencies explor- ing the effect of major expansion programs or bridge scenarios that involve multiple closures when a more detailed analysis is desired than is available using a general needs model. Oregon’s 2015 Transportation Options Plan provides an example of how travel modeling can suggest allocations across programs in a funding-constrained environment (Pietz and Becker 2016). Travel demand models are not, by nature, intended as investment management tools and are men- tioned primarily as inputs to other programmatic decision-making resources and paradigms. The scope of travel models (for any mode) is not intrinsically to address life-cycle agency costs or other long-term investment considerations beyond the travel response to improvement or deterioration of the network; thus, such models are always intermediate calculations to ultimate methods by which program-level needs (or benefits of program-level investments) may be ascertained. Travel models have a profound impact on resource allocation decisions because they quantify the size of the trans- portation market in relation to existing or proposed infrastructure. However, when building pro- gram needs estimates from the ground up (project by project), the limitations of these models must be noted. Travel models are most useful for assessing system capacity changes rather than changes to inter sections, or the effects of safety projects, and the like. Typical demand models also do not assess directly the user or social costs or benefits of investment, which may be an important con- sideration in some agencies’ allocation strategies. Standard demand models can provide inputs for benefit–cost calculations or investment models to compare the societal costs of deficiencies across modes under different build scenarios to arrive at a basis for considering programmatic trade-offs. impact models Impact models are used by some states to assess the economic impacts of allocation scenarios. Going beyond needs models or benefit models (which simply quantify and compare the societal value of investment in different programmatic scenarios), impact models assess how the economy may use different outcomes to achieve wider effects in employment, wage income, gross domestic product, or business sales. Impact models can be used to compare the economic effects of an individual project, a suite of projects within a program, or entire programs. In each of these cases, impact models build on information developed previously by means of an asset management model or a travel demand model but can help translate the outputs of those tools to dollars-and-cents measures that may be more useful to policy makers (Lorenz and Weisbrod 2013; Lynch 2000). In addition to factors for the value of transportation changes, these models have several standard inputs, including changes in traffic volumes, vehicle miles traveled, total hours of travel, and travel time reliability. These measures allow the estimation of costs and benefits associated with vehicle operating costs and travel time. Pavement programs may be assessed with a measure of vehicle

19 damage owing to different pavement conditions. These models also include options for assessing the system’s role in providing access to labor, delivery, and supply chain markets and intermodal connectivity. Based on these measures, effects on business costs, household costs, and productivity resulting from changes in business efficiency and agglomeration can be assessed (Zhang et al. 2016). Regional economic impact models can aid in understanding when specific industries may be affected by a program by linking regional economic data with the transportation impacts provided by needs and demand models. The industries directly affected by a program will have indirect and induced effects as their improved or worsened costs affect business partners, employees, and consumers. (Direct and indirect effects are understood as multiplier effects or effects not only on the users of the improved programs but also on their buyers and suppliers in the larger economy.) For example, since 2006, Michigan has included an analysis of economic impacts using an economic impact model when selecting the program mix recommended in its statewide plan, and Idaho uses a different model to determine investment levels in the economic opportunity program area within its strategic initia- tives program (as described in chapter four). Impact models can be a useful communication tool and a valuable decision-making tool when the impact of a program on society is not necessarily clear. For example, using impact models enables an agency to describe how many jobs will be created by investing in one particular mix of program outlays versus another. The Utah Transportation Coalition made significant use of impact models in this way to articulate the value of investing in a recom- mended programmatic mix of transit, highway, and bridge preservation and roadway expansion investments in 2011 (Economic Development Research Group 2012). However, such models can add complexity to the resource allocation process and raise the monetary and time costs associated with evaluating different mixes of investment levels among programs. An additional challenge of impact modeling is that the alternatives for the scenario testing must be well defined based on other programming processes. However, such models could potentially serve an important role in making cross-programmatic comparisons if agencies invest the resources into developing impact model impacts from programs that are not easily comparable in terms of other measures. Benefit–cost and multicriteria analysis Benefit–cost analysis (BCA) provides a framework for comparing programs against one another, although it typically has been used more frequently at the project level. By standardizing the change in travel characteristics or system condition from needs or demand models into simple dollar terms, BCA is a useful tool for making comparisons across dissimilar programs or projects (FHWA 2003). However, BCA relies on the ability to monetize all aspects of a project, which for some societal values, such as equality, is not an easy task (Zerbe 2007). Consequently, there has been growing interest in employing a more flexible framework such as multicriteria analysis (MCA). Weisbrod (2011) writes that MCA “allows for consideration of benefits beyond efficiency or productivity factors,” which are the standard contents of benefit–cost analysis. MCA also allows “consideration of distributional impacts, including local investment and activity shifts deemed to be socially desirable.” Multicriteria frameworks standardize categories of trans- portation project or program effects that would not otherwise be comparable to a normative scale. This allows the inclusion of qualitative, quantitative, and monetary values to be compared using a weighting scheme. Development of appropriate weights is one of the most challenging aspects of MCA. Sadasivuni et al. (2009) present a case study in Tennessee of how MCA can be applied to transportation projects. They use the analytic hierarchy process developed by Saaty (1994) to apply a weighting scheme. The MCA described by Weisbrod (2011) includes job and income growth outcomes. NCHRP Report 786 points to productivity impacts as a potential mode and program-neutral measure. The report presents guidance and a methodology for analyzing the productivity benefits of transport sys- tem improvements. The guidance addresses the needs, availability, and sources of requisite data, agency staff capabilities, and audiences for information regarding productivity improvements

20 (Weisbrod et al. 2014). These reports demonstrate how several states, such as Kansas, Ohio, and Virginia, and MPOs in Boston, Chicago, and San Francisco are documented as using MCA as a basis for resource allocation (either at the project level based on explicitly weighted programmatic values or directly to pools of funding designated for predefined programs). Other states, including Colorado and Oregon, have developed MCA frameworks for projects for use in project evaluation. The Colorado Transportation Analysis Toolkit has been documented in several case studies (High Street Consulting Group and Economic Development Research Group 2013; Weldemicael and Duncan 2016). Oregon developed a spreadsheet-based tool called MOSAIC to identify least-cost solutions during the planning process by evaluating nine categories of transpor- tation system performance that are applied to all projects during the planning process (ODOT 2016). Because of MCA’s flexible nature, the literature continues to include reports of refinements to the analysis. Macharis et al. (2009) present a method for transport project evaluation and appraisal that allows consideration of multiple stakeholder groups that could place different relative importance on quantitative and qualitative criteria. Tsamboulas et al. (2007) present a method for transportation analysis that allows nested weighting of criteria and alternative scaling factors. Most of the applications of BCA and MCA are to specific projects or comparisons across projects during alternative analysis or as a ranking/rating system for project prioritization. These frameworks could be used with program-level needs model outputs if program-level effects were available for enough programs. Multicriteria frameworks also show great promise for the type of performance-based programming in federal guidance and developed for NCHRP Project 08-91 (Maggiore et al. 2014). ResouRce aLLocation in euRoPean states Resource allocation for transportation needs is an evolving topic in many European countries. As with the United States, there is limited academic literature on resource allocation in Europe; however, it is useful to examine the practices of several countries. The way resources are allocated depends on different aspects of the constitutional and legal framework of the countries. Four examples follow that describe countries using long-range plans and annual processes to determine resource allocation splits. France and the United Kingdom have a centralized structure, whereas Germany and Switzer- land are federal states, Switzerland to an extent similar to that of the United States. Some countries use different financing mechanisms for different modes, preventing them from freely allocating the funds across some programs other than by legislative means. The European Union (EU) as a supranational structure of 28 countries (including the United Kingdom) does not prescribe a specific resource allocation to its member states, but by cofinancing transportation infrastructure projects, the EU sets incentives to follow its path. The EU path described in the European Commission’s White Paper on the Roadmap to a Single European Transport Area emphasizes trans- portation projects that support delivering a minimum 60% reduction of greenhouse gas emissions from transport by 2050 (European Commission 2011). The German federal government makes use of its LRTP Bundesverkehrswegeplan to allocate funds to different modes of transportation. In its plan for 2030, endorsed by the national parliament in August 2016, the responsible department deducted the resources for all essential maintenance work and investment in existing infrastructure from the available resources (German Federal Ministry for Traffic and Digital Infrastructure 2016). Less than 30% of resources were allocated to projects for new or extended infrastructure. To approach an efficient resource allocation that is geared to achieve the objectives of a sustainable transportation system, the agency evaluated three scenarios: (1) allo- cation based on the traffic volume associated with each mode (80% for roads, 16% for railroads, and 4% for waterways), (2) consistent with the 2016 budget figures (59% for roads, 38% for railroads, and 3% for waterways), and (3) with emphasis on a more sustainable transportation system, which would be advantageous for rail and waterways (30% for roads, 62% for railroads, and 8% for water- ways). The allocation, as it was decided, considers both the efficiency aspect (benefit–cost ratio) and nonmonetized aspects of sustainable development by choosing an allocation between scenarios (1) and (3), assigning 52% of the resources to roads, 43% to railroads, and 5% to waterways.

21 France makes use of its relatively new LRTP Schéma National des Infrastructures de Trans- port, which follows the principles of sustainable development (French Ministry of the Environment, Energy and the Sea 2011). The plan allocates resources for infrastructure projects within and among the modes of transportation along its strategic goals: 64% of the resources for optimization and pres- ervation. Of the resources for new investments, 74.5% are allocated to the railroads, whereas roads, ports, airports, and navigable waterways share the rest. In Switzerland, the amount available for new investments in the federally owned highway network is fixed at $8.5 billion for new highways and $5.5 billion for extensions within 20 years ($13 billion total for new centerline miles). An ever increasing share of the remaining transportation funds is used for operations, maintenance, and preservation of the existing network, at the expense of projects such as capacity improvements, modernization, and multimodal investments. Switzerland previously reserved the revenues from the gas tax exclusively for roads, but for the last two decades the funds had been made available for nonroad transportation projects in urban areas (MinVG 2016). In 2015, roughly $1.6 billion was spent on the operation and maintenance of the highway network, whereas $0.55 bil- lion was invested in new highways and extensions. During that time, $0.53 billion was dispersed to the cantons to be used for road purposes (Swiss Federal Roads Office 2016). Large Swiss rail infrastructure investments for freight and passenger traffic are financed by gen- eral tax revenues supplemented by a mileage-based heavy goods vehicle levy and a small share of the value added tax. This separate funding channel prevents easy consideration of resource allocation across the major rail and highway programs. Each canton has its own resource allocation strategy for transportation, using funds handed through from the federal government, the canton’s share of the heavy goods vehicle fee, the cantonal vehicle tax and general tax revenues. There are no federal regulations that prescribe a certain resource allo- cation within the 26 Swiss cantons. If a survey were conducted in Switzerland as it was done for the United States in this study, it is likely 26 different approaches would be found. For most cantons the vehicle tax is an important origin of resources, which are earmarked for road expenses. In the United Kingdom, all revenue from transportation fees and taxes is paid to the general fund. The annual budget allocates total funding envelopes from the general fund to the Department for Transport, and the Autumn Statement sets out the department’s spending priorities. The department sets indicative budgets within these envelopes for its programs, which includes an increased devolu- tion of funding to local governments. These budgets are refreshed in line with priorities by an annual corporate planning process. The 5-year capital spending plan is split 30% for roads and 70% for rail. Resurfacing of more than 80% of the national road system is planned, as is significant other preserva- tion funding. Allocation of rail and road budgets to a single agency from the general fund gives the ministerial Department of Transport significantly more cross-modal allocation discretion than most states in the United States or other European countries examined here. Key findings of LiteRatuRe and PRactice Review Overall, the literature review shows that the allocation of a transportation agency’s resources among programs of work often is a reconciliation of long-range planning targets and short-term TIP or STIP priorities at a project level. To the degree that statewide or MPO plans identify and include realistic pro- grammatic targets, such plans can provide a guide to reconciliation of long-term and short-term goals. However, • Year-to-year variations in funding and political and legislative prerogatives often displace such strategies; • The literature clearly shows that needs-based planning models for highway, bridge, and transit programs (at least at the national level) enable some degree of trade-offs among these types of programs, but such models (those showing monetary needs by program for given performance criteria) generally are not applied for freight, bicycle, pedestrian, or other programs; and • Simply monetizing needs (or user benefits), as is done in current models, has been understood by states, as reported in peer exchanges and published reports, to be an oversimplified approach to program trade-offs.

22 Multicriteria processes often have been helpful for agencies in reconciling projects of different programs in a single TIP or STIP process by applying normative weights to reconcile the quantifi- able project benefits to the subjective values of policy objectives. However, such processes have not been well documented as a basis for overall programmatic allocations at higher than the project level. The literature shows that extensive models exist to quantify needs, predict travel market responses, and assess and compare societal benefits of different levels of program funding (in cases in which clearly defined investment scenarios exist for specific transportation networks). Furthermore, models are well developed to describe and compare wider societal impacts of investment in different mixes of programs. However, the application of available models consistently among programs poses an insurmountable challenge to some agencies, especially programs for which use of the facilities often is not well documented (such as bicycle and pedestrian programs), benefits accrue to privately owned portions of the network (such as to freight railroads), or societal benefits of investment may not be readily traced to measurable outcomes (such as highway beautification or scenic byways programs). The literature review suggests a significant gap between the state of knowledge in the field and practi- tioner needs regarding the level of consistency and ready applicability for methods to assess, compare, and use information to comprehensively justify program-level investment trade-offs across the entire scope of an agency’s investment profile. Although proceedings from peer exchanges and other published documents reveal variation in dis- cretion available to agencies in allocating resources, no studies have explicitly addressed how agen- cies obtain or utilize discretion with regard to their programmatic outlays. Table 2 considers these findings in terms of the seven aspects of resource allocation strategy discussed in the introduction. TABLE 2 FINDINGS ON THE SEVEN ASPECTS FROM THE REVIEW OF LITERATURE AND PRACTICE Aspect Review Findings 1. Preservation versus improvement balance Agencies have largely emphasized system preservation despite most reporting significant improvement needs. The need for improvements is supported by models and appears in long-range plans and programming documents. 2. Modal balance State agencies focus mostly on highway assets although the evolution of the policy environment and available tools and technology are increasing their ability and desire to consider other modes. One of the most significant barriers to modal balance currently and historically has been a separation of funding that prevents cross-program allocation even if planning and policy consider multiple modes. 3. Geographic balance The review of practice finds a wide variety of frameworks for geographic allocation of resources. 4. Accountability (transparency versus complexity) The literature shows tools for decision making are becoming increasingly complex as agencies attempt to optimize resources and take advantage of new data. Integrating all of these tools is a major challenge for agencies internally and poses a challenge to providing transparent information to stakeholders. 5. Top-down versus bottom-up A significant number of agencies are working to better integrate their resource allocation and programming processes with their LRTP; however, for a few states this means making their LRTP more project driven. Agencies are using 10-year investment plans and introducing performance-based planning frameworks that more closely link programming to agencies’ long-term goals. In general, the literature and practice review indicates a growing emphasis on creating more top-down resource allocation frameworks. 6. Agency discretion/flexibility versus policy/model-driven consistency Federal policy provides additional flexibility in the use of funds but also increases the use of performance measures that affect agencies’ flexibility to determine program emphasis with complete discretion. Many agencies also have developed more consistent decision frameworks in response to state oversight bodies. There is significant evidence of a desire to improve the quality of models and increase their use. 7. Objectivity versus subjectivity Much of the literature regards the use and development of objective tools, but topics such as multicriteria analysis seek to integrate objective and subjective scoring of projects into a single framework.