Economic and Development Implications of Transportation Disinvestment (2015)

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 Aging Infrastructure Aging infrastructure is broadly identified as a challenge for transportation infrastructure management in the United States (Knowledge@Wharton 2010; AECOM 2011; Transportation for America 2013; U.S. Government Accountability Office 2013). About 75% of the Interstate Highway System is more than 25 years old (Rodrigue 2013a); in some states the major- ity of the road infrastructure is significantly older than that (Figure 3). A 2010 study on the costs of underinvestment and the pressures of growth by AASHTO noted the need for recon- struction and replacement of aging interstate highway infra- structure simply to maintain the same performance levels as before, let alone offer more capacity (AASHTO 2010). Com- pounding the physical deterioration of assets is the concept that older infrastructure may not meet current performance stan- dards. For example, many older interchanges do not comply with current operational standards and therefore create bottle- necks and safety problems (FHWA 2013). Fiscal Constraints Fiscal constraints at the national and state level are increas- ing the incidences of both intentional and passive disinvest- ment decision making. Fuel taxes, the long-time foundation of transportation funding in the United States, are no longer keeping pace with the requirements for system upkeep. Decreased buying power resulting from inflation, reduced VMT, and increased vehicle efficiency have contributed to a crisis in transportation funding (Puentes and Prince 2003). Based on current spending and revenue trends, U.S.DOT estimates that the Highway Account of the Highway Trust Fund will encounter a shortfall before the end of fiscal year (FY) 2014 (Figure 4). State DOTs are faced with both short- term cash-flow issues resulting from an anticipated slowdown in federal reimbursements and long-term fiscal constraints and funding stream uncertainty because of the considerable politi- cal difficulty associated with reforming transportation finance, both at the federal and state level. Climate Change and the Convergence of Risk Factors Yet another issue of prominence in the transportation planning sphere is climate change and its relationship to risk manage- ment. MAP-21 established requirements for a new Risk-based Asset Management Plan to be completed by each state. The This chapter summarizes both the literature review and the review of models and data, their findings, and their relevance to the overall research. LITERATURE ON DISINVESTMENT Factors Necessitating Disinvestment Analysis There are a number of factors that make disinvestment decision making particularly salient to transportation managers—now and in the foreseeable future. These factors include demo- graphic shifts, travel demand trends, the aging of transportation infrastructure, fiscal constraints, the time horizon of analyses, data constraints, and environmental risk factors. Demographic and Demand Shifts From 1970 through the 2008 economic recession, transporta- tion infrastructure has been subject to ever-increasing demand in terms of the VMT on the system. More recently, overall VMT has leveled off in a way that appears to be more than just a short-term trend (Figure 2)—leading to discussions of “Post Peak” transportation planning (Polzin and Chu 2014). Nevertheless, much of the highway system is subject to far greater demand than that for which it was initially designed. For example, the section of I-84 that runs through Hartford, Connecticut, is subject to a daily volume that is more than three times its originally designed capacity (Connecticut DOT 2014a). Moreover, trends such as increasing truck movements on particular subsets of the road network, geographically dif- ferential population growth, and increased urbanization mean that transportation demand may see increasing concentra- tion in certain areas (and thus increased congestion) even as demand may level off or decrease in other areas (U.S. Govern- ment Accountability Office 2008). At the same time, mobility preferences among young people and aging retirees are associ- ated with shifts toward shorter trips and nonautomobile travel (McCahill and Spahr 2013). Consequently, the transportation system in the United States is faced with significant and yet uncertain shifts in demand patterns, with behavioral and eco- nomic changes that may merit reconsideration of the optimal mix of transportation investments. Rather than focusing on a relatively uniform expansion paradigm, demographic and travel demand shifts point to a strategic investment paradigm that places emphasis on efficiently adapting existing or new assets to changing needs over time. chapter two REVIEW OF CURRENT RESEARCH AND PRACTICE

12 plan is to include an assessment of road and bridge assets and their condition, definitions of management objectives and measures, identification of performance gaps, analysis of life-cycle costs and risk, and a financial plan and investment strategies (AASHTO 2012). According to FHWA, “Climate change is one of multiple risks that impact asset management” (FHWA 2014). Risks associated with climate change include acceler- ated asset deterioration from increased variation in tempera- ture, precipitation, flooding, and other severe weather events (Meyer et al. 2010; Strategic Foresight Initiative 2011). In addition, the increasingly understood necessity for resilience planning may begin to put a strain on already limited resources (FHWA 2014). The convergence of multiple risk factors means increased pressure on certain critical infrastructure such as transportation (along with other infrastructure such as water and energy utility networks). Research conducted for the Strategic Foresight Initiative, a collaborative effort of the emergency management community being facilitated by FEMA, highlighted the existence of compounding effects for cities resulting from the convergence of climate change, aging infrastructure, and increasing urbanization of populations, particularly in high-risk coastal areas (FHWA 2014). Simi- larly, the World Economic Forum’s 2013 Global Risk Assess- ment identified the interplay between constrained government resources, rapid urbanization, economic instability, climate change risk, and infrastructure needs as an important risk fac- tor globally (World Economic Forum 2013). In a dissertation examining resilience strategies for critical civil infrastructure systems, Croope framed the issues as follows: Infrastructure systems are critical for sustaining and maintaining a nation’s socioeconomic system. Their importance is underscored by the need to maintain continuity of services. . . . The functional- ity of critical infrastructure systems is continually challenged by the aging process, disasters (both natural and technological), and constrained resources (Croope 2010). The combined influence of demographic and demand shifts, the aging of transportation infrastructure, fiscal constraints, and environmental risk factors means that decision makers are increasingly faced with difficult choices regarding the most effective mix of investments, given limited funding and chang- ing performance requirements. Studies of Underinvestment and Its Consequences In the literature, the concept of disinvestment appears in a fairly broad range of contexts and is described with varying terminology. For many, the conversation on disinvestment begins with an identification of underinvestment and its nega- tive economic consequences and underlying causes. While the issue is globally relevant, awareness of it is particularly strong in the United States and Canada. North American respondents to the World Economic Forum’s survey on global risks were inclined to rate the risks of chronic fiscal imbalances and pro- longed infrastructure neglect as having a higher likelihood of occurring over the next 10 years than respondents in other regions (World Economic Forum 2013). Andrijcic et al. (2013) discusses the deterioration of trans- portation infrastructure in the United States as a slow, ongo- ing process whose socioeconomic implications are becoming increasingly apparent, including “increased economic costs VMT Trends - FHWA FIGURE 2 Trends in vehicle-miles of travel. FIGURE 3 Age of the Connecticut Highway Network.

13 of freight congestion, decreased global competitiveness of the United States, increased travel costs, and reduced safety of travelers.” The authors also note characteristics of the U.S. policymaking process that tends to discourage the type of long- term investment required for system upkeep, such as the short- ness of election cycles, the relative invisibility of benefits from infrastructure investment, and general public misunderstanding of the benefits of proactive maintenance (Andrijcic et al. 2013). The focus on risks to economic competitiveness is com- mon among studies of underinvestment. A joint report by the Eno Center for Transportation and the Bipartisan Policy Center argues that the most dramatic effects of cutting federal funding for transportation (to levels that are in line with reduced rev- enues coming into the Highway Trust Fund) will be economic. The economic consequences will be derived from reduced accessibility—“to labor and jobs as well as markets, goods, and raw materials”—and from the decline of transportation system capacity, even as population continues to grow (Eno Center for Transportation and the Bipartisan Policy Center 2012). Similarly, in a white paper prepared for the American Council of Engineering Companies, AECOM warns that “The consequences of underinvestment in these vital systems are dire, affecting the United States’ global standing as a leader in economic growth, productivity, competitiveness, capital inflow, job creation, sustainability, and lifestyle” (AECOM 2011). In a 2012 analysis of infrastructure investment, the U.S. Department of the Treasury and the Council of Economic Advisors state that “investments in infrastructure allow goods and services to be transported more quickly and at lower costs, resulting in both lower prices for consumers and increased profitability for firms,” and argued that “American transporta- tion infrastructure is not keeping pace with the needs of our economy” (U.S. Department of the Treasury 2012). There are a number of perspectives on what constitutes underinvestment. Most broadly, underinvestment can be defined as investment levels that do not keep pace with over- all growth. For example, Gillen (2012) identified an invest- ment gap for British Columbia by looking at infrastructure and transportation spending as a percent of GDP over time and in comparison with other countries and provinces. Underinvest- ment can also be defined in terms of a gap between funding levels and what is required to prevent deterioration of an asset’s physical condition. Stiff and Smetanin define an infrastructure gap as follows: In general, an infrastructure deficit is the amount of investment required to repair and maintain existing public infrastructure. This includes the immediate funding for required upgrades, and the future investment needed to maintain a minimal level of ser- vice. It does not include investment required to accommodate future growth (Stiff and Smetanin 2010). FIGURE 4 Highway Trust Fund ticker, U.S.DOT.

14 Interestingly, the authors also note that a greater estimated return on public capital, relative to private capital, can also be a sign that public infrastructure capital and private capital are out of balance (Gillen 2012). At a higher level of complexity, underinvestment may be defined according to certain performance standards applied at the asset, corridor, or system level. For example, congestion is a commonly cited performance category that suffers from underinvestment. AASHTO calculates an investment “back- log,” which is defined as the amount of money it would take to bring highways and bridges to a state of good repair, both in terms of condition and performance, is 46% of AASHTO’s 2008 estimated backlog as the result of capacity deficiencies (AASHTO 2010). The ASCE “Failure to Act” series similarly used an approach that accounts for both the effects of physical deterioration and performance deficiencies, such as congestion caused by inadequate capacity (ASCE 2011). “Deficiency” is defined as the degree to which roads and bridges drop below “minimum tolerable conditions” according to U.S.DOT stan- dards. This is significantly different from ideal conditions such as “free-flow” (ASCE 2011). The ASCE series traces the causal relationship from system condition, to performance, to user costs, and finally to eco- nomic impacts in terms of personal income and value added. The types of costs imposed by transportation deficiencies include: • Increased operating costs for vehicles using parts of the system in poor condition; • Costs from vehicle damage owing to deteriorated road surfaces; • Costs of detours to avoid unusable or heavily congested areas; • The added costs of more costly repairs that result from failing to maintain assets in good condition; • Increased costs associated with the additional buffer time that must be added to trips made in congested areas to ensure on-time arrivals and delivery; and • Increased environmental and safety costs from vehicles operating in substandard conditions (ASCE 2011). Although the literature on underinvestment is instructive in terms of identifying issues and risk factors that need to be com- municated in the political arena, where overall funding levels are set, it does not provide adequate decision-support for those faced with managing underinvestment or disinvestment deci- sion making. Moving beyond the underinvestment focus, it is important to recognize that when faced with changing con- ditions, agencies need to be able to make the best informed decisions about both investment and disinvestment strategies. Similar to those cited previously, Aultman-Hall et al. (2010) identified “a longstanding funding gap between iden- tified needs and revenues available.” However, they also go one step further to warn against the “even greater reduction in mobility” that will result from “haphazard disinvestment.” The authors are particularly concerned about the consequences of the funding gap for rural areas and argue that rural America may need to lead the conversation on strategic disinvestment and tradeoffs because “rural areas presently lack the alterna- tive infrastructure or built environment and land-use pattern to be able to strategically disinvest from portions of our roadway network” (Andrijcic et al. 2013). This highlights the importance of considering network performance across modes and geo- graphic areas when seeking to understand the implications of different disinvestment strategies. According to Aultman-Hall et al. (2013), strategic disinvestment would entail a process for prioritization based on the “tradeoffs between maintenance, rehabilitation, expansion, and doing nothing.” Disinvestment in the Literature and Understanding System Performance The word “disinvestment” itself appears in the literature in only a limited number of instances. Nevertheless, as early as 1982, researchers were highlighting the necessity for different types of management approaches as the U.S. highway system transitioned from a paradigm of expansion to a paradigm of maintenance, reconstruction, and in some cases disinvestment (Lee 1982). Lee argues that the data and methods required for this type of management paradigm are different from those required in an era of growing system mileage and capacity expansion. In particular, Lee presents a structure for under- standing critical information needs for sound management practice, as shown in Figure 5: • Arrow (1) represents the effect of improvements (e.g., “overlays, bridges, lanes, shoulders, medians, grading, tunneling, land acquisition, signing, signals, pave- ment markings, maintenance, repair, landscaping, and other construction and operating activities”) on perfor- mance (e.g., the volume-to-capacity ratio or pavement condition); • Arrow (2) represents the influence of use (volumes, vehicle types, vehicle weights, network distribution, etc.) on performance; • Arrow (3) indicates the influence of performance on user costs (e.g., travel time, fuel consumption, vehicle wear, and safety cost); and • Finally, arrow (4) captures the feedback between user costs and demand. Lee noted that the transportation system can be character- ized by certain concentrations of demand (e.g., by time of day, on certain parts of the network) and thus it is even more critical to understand performance in these areas, relative to others, rather than approaching the system from an aggregate or averaged perspective. In setting up a framework that relates investment strate- gies to system performance and user costs, Lee also estab- lished a framework that can, in principle, be used to tradeoff

15 different investment strategies against the costs imposed on users of the system. His understanding of investment included decisions made in setting design standards, noting that in some cases “the costs of overdesign may be just as great as the costs of under-design.” Indeed, the lowering of performance stan- dards to save on cost and (hopefully) more closely approxi- mate actual needs—based on patterns of usage—is a special class of disinvestment situation most often seen in lower- traffic areas in order to redirect funds toward investments with higher returns (Peterson and Marais 1980; Ou 1986; Mercier 1987). A 1998 World Bank report on road deterioration in devel- oping countries addresses the investment-user-cost tradeoff more explicitly. The authors argue that when budgets are con- strained, “the best policy is not simply to reduce all categories of maintenance spending equally” but rather to revise policies and use different maintenance approaches, based on a trade- off analysis between agency costs and the value to users of different maintenance strategies (Harral 1988). Figure 6 pre- sents one example of this type of approach. The line in Figure 6 is the efficiency frontier for a particular example situation, which shows the highest available value for any given level FIGURE 5 Functional relationships between highway costs and benefits. FIGURE 6 Example World Bank Analysis: Net Present Value of Alternate Maintenance Options Applied to a Specific Road Class.

16 of agency expenditure. Under an unlimited-fund scenario, the optimal strategy is indicated by point (I). As funds are reduced, an agency moves left along the frontier. The curve is initially flat, indicating the availability of budgetary savings without increasing costs to road users by a significant amount. After a certain point, however, the consequences of reduced fund- ing become much more destructive. The report also addressed the time dimension of maintenance and deferred maintenance. It argues that, because reconstruction costs are three to five times as much as resurfacing or rehabilitation, no road should be allowed to deteriorate to a condition where it needs partial or full construction, unless it is to be held in that condition deliberately or abandoned: “The failure to maintain roads [is] tantamount to an act of disinvestment, for it implies the sacri- fice of past investments in roads” (Peterson and Marais 1980). Consideration of the time dimension of costs is formal- ized in life-cycle cost analysis. Novick urges the use of life-cycle considerations in infrastructure engineering, par- ticularly to support strategic decision making in situations of constrained resources: The result of deferred maintenance is inevitably substantially higher ultimate life-cycle costs. Conversely and equally clearly, the degree to which any type of transportation facility is well main- tained materially assists in providing better, safer operations with lower life-cycle costs. . . . Effective and timely maintenance . . . minimizes the magnitude and cost of repair and rehabilitation and defers ultimate reconstruction. In other words, effective and timely maintenance reduces life-cycle costs substantially (Novick 1990). He warns that “funding concerns tend to mask an equally important requirement—the need to develop a rational basis for making far-reaching decisions about the required degree of rehabilitation or replacement” (Ou 1986). Novick identifies the need for methods to estimate and communicate: (1) realistic costs to the public of not having a particular piece of infrastruc- ture available, (2) life-cycle costs, and (3) comparative costs of replacement, rehabilitation, reconstruction, and disinvestment. The first of Novick’s defined needs—developing estimates of the cost of not having a piece of infrastructure available— requires a systems-level approach to understanding perfor- mance. Limitations on performance for a given asset cannot be fully understood in isolation, but rather must be under- stood within the broader context of network performance, using analytical approaches that capture diversion within the transportation network. For example, ongoing efforts at the University of Vermont Transportation Center have developed the Network Robust- ness Index (NRI) as a way to identify critical links within a transportation system and to understand performance effects of capacity disruptions (Novak 2010). The NRI (defined in Exhibit 1) is designed to assess the comparative robustness of transportation networks and to improve on localized measures such as volume-to-capacity ratios. It has been tested using a Vermont MPO travel demand model, where link-specific NRIs were used to identify the set of most critical network links. A follow-up work is underway to incorporate trip importance, by trip purpose, into the NRI framework, to support more strate- gic reinvestment and disinvestment decision making (Sullivan and Novak 2014). Fruin and Halbach (1992) performed a network-based analysis of investment and disinvestment in a Minnesota Source: Sullivan, J., L. Aultman-Hall, and D. Novak, Application of the Network Robustness Index to Identifying Critical Road-Network Links in Chittenden County, Vermont, UVM TRC Report # 10-009, June 2010, pg. 3. [Online]. Available: http://www.uvm.edu/~transctr/trc_reports/UVM-TRC-10-009.pdf [accessed April 17, 2014.] EXHIBIT 1 Method for calculating the NRI index.

17 county’s rural road network. The potential for disinvestment arose because a declining number of farms and increased truck size were changing traffic patterns on the local road system. The authors conducted a tradeoff analysis comparing vehicle operating cost and time costs with road and bridge mainte- nance and upgrade costs—with modeling of future traffic on the network and a sensitivity test based on a less intensive crop production scenario (Fruin and Halbach 1992). More conceptually, the World Economic Forum provides a framework designed to help assess system resilience to risk. Resilience is the ability of a system to accommodate, adapt, or recover when certain risks are realized (Mitchell and Harris 2012; McNeil 2013; World Economic Forum 2013). The framework defines five components of resilience; the first three describe system characteristics, the other two relate to resilience performance: 1. Robustness—the ability to absorb and withstand disturbances. 2. Redundancy—the ability to use excess capacity and back-up systems to maintain core functionality in the event of disturbances. 3. Resourcefulness—the ability to adapt and respond flexibly. 4. Response—the ability to mobilize quickly (depends on the capacity to gather information and translate that information into good decision making, in a timely fashion). 5. Recovery—the ability to regain a degree of normality after an event, and to evolve to deal with new or changed circumstances after the manifestation of a risk. These components describe the characteristics of physi- cal systems (e.g., transportation infrastructure itself) and of the people and organizations that manage and interact with the physical system. While generally understood in relation to events or incidents (e.g., natural disasters), some resilience concepts are instructive when considering more long-term dis- investment scenarios. In its recent literature review on Risk- Based Transportation Asset Management, U.S.DOT is careful to define risk broadly to include both “catastrophic failure of an asset” (generally thought of as an event) and the often more gradual or ongoing “failure to ensure desired levels of service” (Proctor and Varma 2012). Finally, a recent study of two highway closures in St. Louis and Appalachia identifies transportation network and economic factors that can act as determinants of the economic impacts of lost system performance (Hodge 2011). The identified fac- tors are also relevant to the impact of closures or restrictions associated with disinvestment and include: • The availability of (and level of information about) alternative routes; • The industry mix in area (how dependent business activ- ity is on pass-by or discretionary visitor traffic); • The mix of traffic (whether the closure affects mostly local trips or long-distance trips); and • The development of mitigation strategies beforehand. Needs-Based Planning and Quantifying the Effects of Unmet Needs by Program “Needs-based” planning is a planning approach that entails deriving target investment levels based on “minimum toler- able conditions” of highway capacity, pavement condition, bridge condition, transit availability, and other performance indicators. A typical needs-based planning study will iden- tify the size of the investment gap for each program and will use models to quantify the performance implications, agency life-cycle costs, and public-user costs of leaving needs unmet. An investment strategy that enables the agency to avoid the life-cycle or user costs of unmet needs is understood to have a benefit when compared with a base case, which fails to address such needs. There is a significant body of research pertaining to eco- nomic analysis of declining transportation investment within the context of needs-based planning. Schroeder et al. (2012) offer a framework that begins by identifying freight infra- structure needs using widely accepted pavement and bridge models, and assesses the comparative economic impacts of unmet needs as the basis for prioritization (using input– output modeling). In 2011, the Arizona DOT completed a statewide plan that compared different investment options with a “business as usual” base case with unmet needs for vari- ous programs (Omer 2011). The Arizona plan presented eco- nomic benefits of different investment levels in terms of the avoided costs of unmet needs in the long term. In 2005, a simi- lar plan in Michigan compared the user costs of unmet needs by program with different improvement cases, using the dif- ference in user costs (and their associated economic impacts, in input–output terms) as the basis for quantifying economic impacts and benefits of additional investment (Wilbur Smith Associates 2005). Also in Michigan, a 2008 study analyzed the comparative impacts and benefits of shifting investment between highway and bridge preservation or expansion pro- grams by assessing needs for each program based on “mini- mum tolerable” future conditions and comparing the relative benefits of investment in each area with a base case in which no investment was made (Fulton et al. 2008). Also in 2008, a Kansas study compared base case conditions (without investment) to future investment case conditions (accounting for changes in pavement condition as well as the associated changes in the routing of traffic) to ascertain economic ben- efits and impacts of investing to meet future needs compared with leaving needs unmet (Kansas DOT 2008). NCHRP Proj- ect 8-36 (67) presents an overview of investment strategies throughout the 50 states, focusing on key decision principles applied by states when considering the economic and per- formance tradeoffs of deciding which needs to meet or leave unmet (Janik 2007).

18 These examples are samples of the current state of the practice in state agencies where agencies assume future needs based on a current understanding of long-term demand, and characterize the economic benefits and impacts of the gap in terms of economic costs accruing to agencies, households, or businesses when such needs are left unmet. It can be noted that in all of the needs-based planning studies to date, needs are presumed to be set based on a singular understanding of future demand—and there has not generally been consideration of different levels of need that may arise from different possible socioeconomic futures. All of the needs-based planning sce- narios assume a static understanding of future economic and demographic conditions and then focus on comparing different investment mixes for meeting the presumed needs of the future economic situation. However, as presented before, it is clear that the uncertainty of future economic conditions (and the likelihood of unmet needs under alternative future economic conditions) makes actual needs and actual risks of underinvest- ment or disinvestment more difficult to understand than the current practice assumes. Learning from Other Disciplines Although this project is focused specifically on highway and bridge disinvestment situations within the public sector regard- ing road and bridge assets, the type of decision and the generic structure of options available for consideration appear in other disciplines. Here we look briefly into approaches used within the private sector to support disinvestment-type decision making. Corporate Divestiture—Insights from a Roundtable Discussion PricewaterhouseCoopers hosted a roundtable of corporate business development executives in 2012 to discuss “strate- gies for managing a successful divestiture” (Pricewaterhouse- Coopers 2012). Divestiture refers to the decision by a com- pany to sell off a portion of the company in order to focus more specifically on activities with higher growth potential or that are more central to an organization’s core mission. Participants in the roundtable recommended a number of factors that could be considered when addressing this type of disinvestment situation. First, a particular activity or business line is ripe for consideration if it experiences “poor performances with declining market share and profit- ability.” Translated into more general terms that also apply within the public sector, disinvestment would be consid- ered if the market being served is no longer as relevant or as strong. Once poor performance has been identified, one must also ask: “why is this business underperforming, and is it worthwhile trying to fix the problem rather than divest?” In the transportation field, this points to the familiar process of developing alternatives and assessing their relative costs and benefits. Next, the recommendations highlight a key question that must be answered as part of the disinvestment assess- ment process: “How critical is this business to the rest of the organization. Do we fully understand the interdependencies and their impact on key stakeholders in the company (custom- ers, suppliers, etc.)?” Performance cannot be considered in isolation. As with roads and bridges, there are network effects and system interdependencies. Lastly, the forum participants caution that disinvestment scenarios should consider the initial costs incurred in the process of disinvestment. Perspectives on disinvestment from the private sector are less complicated than public disinvestment choices because the effects that pub- lic infrastructure outcomes can have on the overall business environment for private firms. Engineering Economics and Replacement Decisions Disinvestment and investment decisions are closely related. In particular, disinvestment situations can come to light in the context of deciding how and/or whether to replace a given asset versus substituting another asset that would likely be used in its place. Given the age of much of the interstate highway sys- tem in the United States, large-scale replacement decisions are becoming an increasingly important type of decision for state DOTs. More broadly, replacement decisions are a com- mon class of decision within engineering economics. It can be generally assumed that an existing asset will be removed at some future time—either when the function it performs is no longer necessary or when the function can be performed more efficiently by a newer and better design (Park 2011). To deter- mine whether an asset should be replaced and what it should be replaced with requires a definition of operational performance requirements. In cases where an asset is deemed essential to operations (meaning that failure of the asset would result in an unacceptable slowdown or shutdown of operations), one must then answer the question “when should existing equipment be replaced with more efficient equipment?” (Park 2011). Embed- ded in this question is the understanding that no asset or piece of equipment lasts forever, that every replacement decision involves at least one alternative option for that equipment’s replacement, and that timing is often a choice. In cases where demand or performance requirements have not changed significantly since the assets initial selection and installation, the driving force motivating replacement is that operating costs nearly always increase as an asset ages. Keeping an asset (the “defender” in engineering economics parlance) usually involves a lower initial cost but higher annual operating costs (which include maintenance and repairs) rela- tive to the replacement option, which costs more upfront but involves lower annual operating costs. Another common cost accounted for in engineering-type analyses is the salvage value of the asset, which is likely to decline over time. In a simple analysis of replacement, one simply compares the net pres- ent value of the future costs for the defender and challenger. In some cases, analyses will also take into account varying assumptions about technology changes in the future, thus rec-

19 ognizing that the type of performance available may not be the same a few years down the road as it is at the time of analysis. Real Options and Flexibility in Decision Making Another analytical construct that can be of relevance to the economics of highway and bridge disinvestment is the concept of a “real option” (Pindyck 2008). The concept of real options stems from financial options theory. “In finance, an option is defined as the right, but not the obligation, to buy or sell an asset under specified terms” (Zhao et al. 2004). The idea behind real options is that any “investment decision can be treated as the exercising of an option” (Pindyck 2008). Firms or agencies have the option to invest (or disinvest), but need not necessarily do so immediately. They can also wait until more information is available—about future conditions such as demand, prices, etc. If an investment involves a sunk cost, there can be considerable opportunity costs associated with investing now rather than waiting. Given future uncertainty, and the full or partial irreversibility of certain kinds of invest- ment and disinvestment situations, a real-options framework adds more insight than a traditional analysis based on the net- present-value of a project’s cash flow. Option theory encour- ages managers to consider options such as: • the option to delay an investment, • the option to stop before completion, • the option to abandon after completion, and • the option to temporarily cease operations. In many cases, a certain amount of investment is required up front to purchase a “real option” that can be exercised at a later point in time. One chooses to make this initial option purchase if the cost is less than the value of flexibility provided by the real option. For example, one may choose to purchase a wider right-of-way for a project than needed for the initial number of lanes to be built, as a way of purchasing the option, but not the requirement, to expand in the future. In general, the availability of an option to disinvest at a later point after initially investing, if demand proves inadequate or costs become too high, increases the net present value of a project under consideration. This is because managers know they will have the option to take advantage of future informa- tion to re-optimize their investment (or disinvestment) strategy at a point in time when uncertainty has been reduced. When a disinvestment option is available, managers become less cau- tious about their initial investment decision and they therefore tend to exercise their option to invest earlier (Keswani and Shackleton 2006). In the case of both investment and dis- investment, increasing uncertainty about future conditions could result in increasingly cautious behavior—favoring a wait-and-see attitude (Bloom et al. 2007). In practice, however, reluctance to disinvest can exceed that predicted by real-options theory, because of factors such as “emotional attachment” and “psychological inertia” (Musshoff et al. 2012). A real-options framework may also be useful for more strategically addressing disinvestment situations. Often, there is a spectrum of available alternatives when considering disinvestment—some more severe or irreversible than others. Depending on the situation, it may be possible to opt for: (1) a gradual disinvestment scenario that is still irreversible, but can be suspended at any point; or (2) a partial disinvest- ment scenario that still maintains the option to restore full performance levels in the future, without incurring prohibi- tive costs to do so. In the first case, managers must consider the tradeoff between: • The flexibility offered by gradual divestment to benefit from possible future positive market developments, and • The greater sale value of a whole firm, relative to the discounted value of partial displaced assets. If a firm chooses gradual divestment, then “the firm holds a bundle of options to sell its partial assets. A marginal sale of assets leaves the options to sell the remaining assets and allows the firm to benefit from their optimal execution in the future” (Gryglewicz 2009). In the second case, managers may choose to make an invest- ment and/or invest in a certain minimum level of maintenance in order to at a later point have the option to restore service without starting over (and thus incurring the greater costs required to start from nothing). This kind of analysis has been commonly performed for power plants and other scalable busi- ness operations: the options available are shutdown, startup, and abandonment, and the key costs considered are called “switching costs,” involved in switching from one operating state to another (Bakke and Viggen 2012). One formalized example of purchasing a real option to enable future reactivation of a transportation service is the Railbanking program authorized by the National Trails System Act. It is “a voluntary agreement between a railroad company and a trail agency to use an out-of-service rail corridor as a trail until a railroad might need the corridor again for rail service. Because a railbanked corridor is not considered abandoned, it can be sold, leased, or donated to a trail manager without reverting to adjacent landowners” (Rails to Trails Conser- vancy 2014). Railbanking allows a private operator to tempo- rarily cease rail operations, while still maintaining the option (at least in theory) to resume service at some future point of time without incurring the prohibitive costs associated with re- acquisition of land. It requires some level of initial investment to establish the agreement with the trail operating entity and to convert the land to its new use; however, this investment is less than both the cost to continue operating and the cost of reacquiring land at a later point in time when demand might be again adequate to justify service. Although the idea of rail- banking falls neatly into a real-options framework, there are

20 political and public perception switching costs associated with transportation infrastructure (even if owned privately) that can be particularly difficult to assess and account for because of the high visibility of these facilities. It is unclear whether a rail corridor, once converted to a public-use trail, will ever be switched back to an operating rail line. Such a switch would certainly require considerable political capital. To summarize, a real-options framework can be useful when managers are faced with future uncertainty and have the ability to consider a spectrum of investment and dis- investment possibilities at various points in time. Manag- ers may choose to wait until more information is available before making an investment or disinvestment choice. Real- options analysis provides a methodology for valuing the flexibility to “wait and see,” and to compare this value with the investment needed to purchase that real option. Apply- ing this valuation methodology to public infrastructure will require consideration of public agency, user (individual and corporate), and external costs associated with each invest- ment and disinvestment option. Applicability to Transportation Disinvestment Decision Making There are a number of issues that make it difficult to trans- fer private-sector or engineering approaches to disinvest- ment situations to the context of public-sector transportation decision making. Most prominently, performance within the private sector tends to be better defined and subject to less uncertainty. Often, performance is reducible to the common denominator of dollars, viewed from a single corporate per- spective with less concern about the incidence of specific costs across different groups (as in the case of transporta- tion decisions that differentially impose costs and benefits on the federal government, state governments, and a diverse set of system users). Transportation operates within a broader socioeconomic context—making its performance both com- plex and open. It is therefore challenging to define perfor- mance targets or minimum tolerable conditions. Although those faced with more straightforward engineering systems may be able to easily define whether an asset is “essential” to system performance, those determinations are not so clear cut for transportation. The safety realm is perhaps the most well-defined. On the other hand, what does it mean for a piece of a transportation network to be economically essential? Moreover, performance needs can be expected to change over time, both because of shifts in demand and technology, and because of the ever-evolving understanding within soci- ety of the aims of transportation. For example, our collective emphasis on environmental sustainability and on livability has changed considerably since the majority of the Interstate Highway System was built. As Zhao et al. (2004) point out, a highway system is subject to both internal and external uncer- tainties in the course of its life cycle, including “changing requirements of users in terms of traffic demand, changing social and economic environment, changes in technology, and deterioration of the highway.” These uncertainties are inter- related; changing social and economic conditions influence demand, which in turn influences deterioration processes. Those deterioration processes can then in turn provide feed- back and deter certain travelers, thus also affecting overall social and economic conditions. The transferability of the issues described earlier will require scrutiny and consideration when developing methods to sup- port strategic disinvestment. Nevertheless, there are core con- cepts from other disciplines that can help structure assessments of transportation disinvestment; namely, the need to acknowl- edge system interdependencies, an emphasis on life-cycle costs (which already appears quite extensively in transportation maintenance management), and the value of flexibility when dealing with future uncertainties. MODELS AND DATA FOR ANALYZING DISINVESTMENT Needs models, demand models, risk models, and impact mod- els all play a role in understanding a transportation disinvest- ment scenario. In all cases, models require data about existing asset conditions, expected deterioration rates, current and pro- jected employment, housing and other drivers of demand, and utilization, as well as per-mile and per-hour factors of user- cost resulting from different demand levels experiencing dif- ferent conditions. Needs Models U.S.DOT and private vendors have developed models to enable agencies to assess highway and bridge investment needs. These models usually begin with: (1) a database inven- tory of existing asset conditions, demands, and factors (such as roadway functional system, climate, terrain, urban, or rural area types); (2) a set of default improvement costs; (3) a set of minimum tolerable future asset conditions; (4) a set of antici- pated per-mile, per-vehicle, or per-hour user costs of falling below minimum tolerable conditions; and (5) a set of assump- tions regarding future demand. Using these inputs, needs mod- els assess: (1) the likely future investment needs, (2) the likely comparative user costs (in dollar terms) if needs are unmet, and (3) the likely agency life-cycle costs (in dollar terms) if needs cannot be met within the most efficient amount of time. Typical data sets used in highway and bridge needs models include: 1. Highway Performance Monitoring System (HPMS) pavement data and 2. National Bridge Inventory (NBI) data.

21 Examples of needs models include the federally sup- ported Highway Economic Requirements System for States (HERS-ST) (FHWA 2009), the National Bridge Investment Analysis System (NBIAS) (FHWA 2010a), as well as the pri- vately syndicated Deighton asset management software. Most of these tools support cost–benefit analysis, comparing the cost of investing in maintaining a given performance standard with the economic cost of failing to make such investments. The greatest strengths of needs models for understanding disinvestment include: • Consideration of both life-cycle and user costs of not investing in assets in comparison with the costs of investing, • Rigorous detail regarding how engineering measures of effectiveness and performance are affected by changing asset conditions over time, and • The ability to assess the sensitivity of user costs and agency costs to different funding levels and different sizes of investment gaps over time. Key weaknesses of needs models for understanding the economic implications of disinvestment include: • Reliance on fixed traffic growth rates to assess future demand can result in an artificially high “needs” pic- ture and lead to overinvestment, and fails to account for network effects of bridge closures or deteriorating asset conditions. • Reliance on large databases and statistical average costs that do not account for risks associated with dis- investment or underinvestment in high-impact “outlier” facilities. • Failure to account for potential changes in infrastruc- ture costs or user costs over the life of the analysis. • The publicly available needs models such as HERS-ST and NBIAS represent a generic “baseline” for predict- ing needs, conditions, and future user costs. More intri- cate systems, while available to states, can be costly to implement and require significant capacity building at the agency level. Needs models are a useful tool for understanding dis- investment because they can provide a ready comparison of different futures assuming different funding levels (or no funding). Although needs models are currently not structured to answer questions about disinvestment per se, their basic computational structure lends itself to analyzing disinvest- ment, at least at the program level. Current needs models sim- ply assess underinvestment (quantifying the economic costs of unmet needs); however, such models may be slightly modified to assess disinvestment (giving the agency savings and user costs of lowered performance targets for programs). However, applying today’s needs models to assess the economic costs of disinvestment ultimately will pose challenges beyond simply assessing the economic implications of changing how needs are defined. Overall, today’s needs models rely heavily on a static pic- ture of future socioeconomic conditions and a static under- standing of cost and demand patterns. Because the models rely on average annual traffic growth rates applied to a standard demographic or business profile of the user population, such models inherently fail to recognize the shifts in demand and user values that make disinvestment necessary. These static assumptions about future demand and use can result in an inflated understanding of needs (and benefits) for some pro- grams to the detriment of other programs that may be more likely to be beneficial in the long term under actual future conditions. In a similar problem, needs models tend to treat similar facilities identically with regard to the user costs of deficiencies. For example, the bridge needs model (NBIAS) applies an average detour length and user cost to bridges based on the area type (urban vs. rural) and functional classification of the roadway the bridge is supporting. The actual loca- tion of the bridge relative to key trade centers and redundant alternative crossings could have significant implications on the “real world” economic cost of a bridge closure (or failure) overlooked by the generalizations of NBIAS. A key area of future research for understanding the eco- nomic implications of disinvestment is in enhancing the value of needs models by integrating them more fully with other types of models used for assessing needs and out- comes, as well as developing new and more rigorous tools for sensitivity testing of future needs assessments. Demand Models Transportation demand models are widely used by MPOs, as well as an increasing number of state DOTs. Typical travel demand models arrive at estimates of future traffic flows by deriving future trips from expected socioeconomic conditions, distributing trips based on likely future development patterns, and assigning trips to appropriate modes or routes. Software packages such as CUBE Voyager, EMM-3, and TRANSCAD are often used to develop these types of modeling applications. For freight movements and truck flows, privately syndicated models (e.g., the IHS/Global Insight TRANSEARCH data set) can provide estimates of commodity flows at the region or county level. Freight demand models underlie the U.S.DOT Freight Analysis Framework, which can be used to visualize and query commodity flows to and from different locations (FHWA 2012a). Demand models are useful for understanding disinvest- ment scenarios because they can show how traffic patterns would be expected to divert if a given facility or system were unavailable as a result of deteriorating conditions. Unlike the static demand assumptions common in most needs models, network demand models can assess and compare the mileage and hours of additional congested and uncongested travel dis- tance and time imposed by lost use of an asset. When linked to land-use models such as UrbanSim (University of California

22 Berkeley and University of Washington 2011) or CubeLand (Citilabs 2014), network demand models can also consider changes in spatial patterns of employment and housing when a transportation network changes. The greatest strengths of demand and network models for analyzing disinvestment scenarios include: • The ability to realistically identify alternate traffic rout- ings and assess changes in travel time and cost when the capacity or connectivity of a facility is lost. • Consideration of likely future employment and residen- tial location patterns that may drive the future demand for both the disinvested facility and likely alternate facilities. • The ability to change socioeconomic assumptions about housing and employment locations as part of a disinvest- ment scenario. Key weaknesses of demand models for assessing dis- investment scenarios include: • They can only address mileage and travel time-based economic implications of disinvestment (they do not assess life-cycle, safety, or other types of user costs). • They do not include any analysis of likely capital costs (or savings) of a disinvestment scenario—they only consider how the scenario will affect network behavior. • They typically only address high-level, regional, or system outcomes (most travel models do not include intersection of micro-level transportation performance outcomes). • They do not implicitly convert travel time and operating cost changes into economic costs or benefits, much less economic impacts. Effectively, demand models are intended as intermediate inputs to economic models. By themselves they are insuffi- cient for any type of economic analysis. However, if they can be used to supplement or complement needs models, they can fill in the gaps that many needs models have for fully assessing disinvestment scenarios. A key area of future research into the economic analy- sis of disinvestment scenarios pertains to the integration of demand models with needs models. Such integration may enable needs models to more comprehensively answer ques- tions about how many users, vehicle-miles, and vehicle-hours of travel may be subject to different types of deficiencies as different links or corridors of a network shift to lower per- formance standards resulting from disinvestment. A key to enabling this progress involves addressing metadata issues such as inconsistencies between the linear referencing sys- tems used in needs model databases (such as HPMS and NBI) and typical geographic information system files used in travel model networks. Current federal initiatives, such as the FHWA’s integration of HERE data with HPMS, are promising directions integrating real-time data with more static information. Another key area for travel demand models is evolving the travel modeling paradigm to regularly consider different possible socioeconomic futures and different associated lev- els of demand. Although travel demand models can assess the demands associated with different land-use and economic scenarios, in practice most models are validated to a single vision of future land-use and economic growth and only used to test different network assumptions for accommodating such growth. Future research into how to most effectively incorpo- rate different socioeconomic issues into travel model scenarios is an important need for assessing the economic implications of disinvestment planning. Risk Analysis Methods As disinvestment becomes more a part of transportation investment management, agencies are likely to employ mod- els for assessing the risks of disinvestment—especially given the uncertainties regarding future demand, potential failure, and costs that may accrue if either a facility continues to be overinvested in and use fails to justify life-cycle costs or, more likely, a facility is disinvested to too far below a per- formance standard, and the new standard proves insufficient leading to failure. MAP-21 legislation recognized and called for the applica- tion of “risk based” planning, and methodological research has identified and tested methods to assess both the risk associated with a demand forecast and the risk associated with estimating financial highway and bridge investment needs. Mehndiratta et al. (2000) present practical applications and challenges fac- ing planners when addressing investment risk, in terms of “real options.” Kruger (2012) offers statistical measures of risk in demand levels associated with different GDP assumptions over different time horizons. Alasad et al. (2014) explored demand risk as it relates to return on investment within the context of public–private partnerships and Maconochie (2010) describes the development and application of highway bridge risk models for use in asset management investment decisions. A typical risk assessment would assign a risk factor to a project based on past agency experience of project outcomes within certain parameters. If the agency is able to track the actual outcomes (or costs) that have accrued from other similar decisions, either to the agency or to users in actual situations where a facility has performed at the lower perfor- mance level envisioned for a disinvested facility, the agency may be able to quantify the likely risk of different types of costs and multiply that likelihood by the magnitude of the cost for use in a traditional cost–benefit analysis. However, agencies often have a lack of suitable case examples, much less databases of disinvestment to realistically base decisions on this type of assessment. Consequently, statistical methods can also be used for agen- cies to assess the likely risks of disinvestment. The second moment method (calculating likely risk based on the standard

23 deviation in cost based on the average values of cost determi- nants) could be applied to user costs in a disinvestment situation to assess risk, as could Monte-Carlo simulations (computerized probabilistic calculations that use random number generators to draw samples from probability distributions). One final risk assessment methodology entails constructing decision trees in which the agency may value all possible outcomes, and then assign probabilities to each possible outcome. Decision trees can be helpful for mapping out the likely economic costs of dis- investment, but are not particularly helpful in determining how the values are to be determined in the first place. Also, decision trees can become very complicated if they are intended to cover all of the possible outcomes for all of the possible determinants of the economic implication of a disinvestment scenario. All three of these statistical methods are far from precise and are likely to pose challenges to transportation agencies deal- ing with the open-ended types of variables that come into play when considering transportation disinvestment. For example, it is much easier to simulate the likelihood that right-of-way for a new road will cost $1 million versus $2 million than it is to simulate the likelihood that (1) demand on a disinvested facility will exceed its forecast, (2) such demand will cause per- formance failures, and then (3) the performance failures will affect the economy. For this reason it is likely that risk analysis will play an important but limited role in disinvestment economic analy- sis, applied only at points in the decision process where there is a manageable range of outcomes for a manageable set of economic performance indicators and when the overall struc- ture of the disinvestment scenario is already largely defined by other types of models and tools. Impact Models Impact models are widely used to assess the economic impacts of investment or disinvestment scenarios. The scenarios can be at the level of individual projects, bundles of projects, or entire programs. Typically, impact models will be applied when the economic cost of the disinvestment outcome is known and translate the cost into earnings, output, employ- ment, GDP, and other economic outcomes using a standard input–output framework. The REMI TranSight and TREDIS models are examples of widely used economic impact mod- els that might be applied to assess a disinvestment scenario (REMI 2013; TREDIS 2014). REMI TranSight and TREDIS are both regional economic impact forecasting and simulation models that are specifically tailored for forecasting the impacts of transportation system changes on the economy of surrounding regions—cities, metropolitan areas, states, or broader regions. Their typical inputs fall into six classes: 1. Traffic Volumes and Vehicle-Miles of Travel Change—representing effects of route diversion and travel distance changes that may result from facility closures and size and/or weight restrictions. 2. Vehicle-Hours of Travel Change—representing effects of speed slowdowns that may result from speed reduc- tions as well as route diversions. 3. Vehicle Damage Change—representing effects of decreased road pavement ratings that lead to more vehicle damages associated with potholes. 4. Travel Time Reliability Change—representing effects of reduced effective capacity on some facilities, as well as concentration of traffic (demand) on other facilities that are kept to a higher standard. 5. Market Access Change—representing effects of shrinking labor markets and/or truck delivery markets owing to reductions in speeds and routing options. 6. Intermodal Connectivity Change—representing options for use of intermodal facilities, ground access routes, or connecting services. The regional economic impact models will translate these six classes of inputs into various measures of change in busi- ness operating costs, household operating costs, and produc- tivity resulting from shifts in business operations technology and agglomeration scale benefits. Although impact models can be helpful for describing the likely economic effects of transportation disinvestment, they can only be used when the disinvestment case has already been largely established through travel demand and needs models. In addition to the private economic impact models, public tools such as those described in NCHRP 2-24 (http:// onlinepubs.trb.org/onlinepubs/nchrp/docs/NCHRP02-24_ Task1LitReview.pdf) assess productivity and other economic outcomes in general terms based on the same types of inputs enumerated earlier. Some of the uses of economic impact models for assessing a disinvestment scenario include: • Understanding which industries will be most affected by the outcome, and the relative magnitude of how industries will be affected by the lower performance standard or alternative facility or program; • Consideration of the indirect and induced effects of dis- investment as the costs of the choice are passed through the economy to buyers, suppliers, and households affected by the business outcomes; and • Consideration of potential feedback loops whereby the costs of disinvestment may cause structural changes in a regional or local economy, further affecting demand pat- terns. (This would entail using an impact model together with a demand model.) Some of the limitations of impact models for assessing a disinvestment case include: • Not showing the initial change in the transportation cost structure, which may be caused directly by the low- ered performance standard or switch to an alternative

24 facility (they only show how this cost affects the larger economy). • Can be costly to implement as a regular part of trans- portation investment decision making and may not be available to all states. • Can be difficult to explain, as they show “multiplier” effects (indirect and induced effects of transportation costs) that may not be intuitively obvious to decision makers based on an understanding of the nature of the disinvestment scenario. Overall, impact models have an important role in under- standing transportation disinvestment scenarios; however, as with risk models and travel demand models, they cannot help the analyst to formulate the options or directly associate the disinvestment scenario with the initial cost borne in the economy. A key area for impact models in assessing dis- investment scenarios will be in establishing the “base case” and “disinvestment case” transportation user costs such that traditional impact methods can then be applied. ASCE Modeling Process: From Needs to Impacts Combining Asset management, Traffic Assignment and Economic Impact Models FIGURE 7 ASCE modeling sequence. Sequences of Models It is most likely that disinvestment outcomes will require arranging needs models, demand models, risk models, and economic impact models into comprehensive sequences in which each model provides one key part of the disinvestment picture. For example, demand models can be used to provide more dynamic estimates of future utilization associated with more realistic economic growth and land-use scenarios to inform needs models (possibly informed by risk models to assess the most likely demands). Subsequently, various possi- ble needs assessments for different possible economic futures can be tested on a variety of investment levels representing dif- ferent likely future economic circumstances, showing diverse investment gaps that then provide the basis for impact analy- ses of possible needs under different socioeconomic futures. Although this sounds like a potential “jumble” of models, the flow of information between models can be very intui- tive, as demonstrated in Figure 7 from the 2011 ASCE Study (ASCE 2011).