Methodologies to Estimate the Economic Impacts of Disruptions to the Goods Movement System (2012)

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.

33 3.1 A Five-Step Decision/Analysis Process The sequential five-step process shown in Figure 3-1 was developed as part of this research as a comprehensive and practicable framework for evaluating a wide range of freight network disruption events and their many possible economic impacts. The following sections describe each step in turn and break up the flow chart in the figure into its separate and sequential com- ponents. This framework represents the components of an in-depth analysis methodology for analyzing in great detail the economic costs of a transportation system disruption. 3.1.1 Step 1: Define Direct Freight Transportation Network Impacts This initial step identifies the direct, immediate physical effects of a network disruption (see Figure 3-2). This includes identification of the specific transportation facilities affected and the associated modes of transport within, into, and out of the region impacted. The timeframe and the geographic extent of the disruption are also determined. The timeframe here represents the time it takes for the network to return to something resembling its pre-disruption condition. The geographic extent represents the size of the region that is directly impacted by a change in transportation network condition and throughput capacity. The disruption may affect one or more specific highway, rail, waterway, or pipeline network links. It may also take the form of damage to, or loss of, storage space, docking space, or cargo handling equipment at one or more seaports, airports, or other multimodal freight transfer facilities in the affected area. Data Needs An inventory of impacted facilities needs to be assembled and site reports collected in those cases where an assessment is being attempted after a disruption has occurred (as opposed to an anticipatory study of potential disruptions). This may mean searching through various data sources, as well as tracking the timeline associated with the return of specific transportation facilities to either partial or full operational status (or a record of their subsequent loss to future use). In some cases, on-site interviews to obtain the best possible understanding of the nature and extent of the physical and operational disruptions will be extremely helpful. For longer term disruptions, daily situation reports may have been used to identify potential traffic bottlenecks. 3.1.2 Step 2: Identify Current and Future Affected Network Flows by Facility and Link Step 2 identifies current and future affected network flows by facility and link (see Figure 3-3). Various more or less severe impacts on day-to-day operations are possible, from a rerouting of local C h a p t e r 3 Analysis Framework

Intermediate Modeling Step Step 4 SUPPLY CHAIN RESPONSE MODELING INDUSTRY RESPONSE modal diversion other transport response production suspended- inventory production locations shift FLOW ROUTING RESPONSE Model Inputs Step 1 DEFINE DIRECT FREIGHT TRANSPORT NETWORK IMPACTS - “PHYSICAL CHARACTERISTICS” facilities disrupted modes impacted likely time frame of disruption Model Inputs Step 2 IDENTIFY CURRENT AND FUTURE AFFECTED NETWORK FLOWS BY FACILITY AND LINK current volumes by commodity tonnage and value projected volumes by commodity tonnage and value Step 3 DEFINE SUPPLY CHAIN CHARACTERISTICS AND PARAMETERS, BY FLOW TYPE Model Outputs SOCIAL AND PUBLIC SECTOR COSTS - E.G., INCREASED HIGHWAY CONGESTION AND MAINTENANCE, FROM RAIL TO TRUCK DIRECT SUPPLY CHAIN COSTS - NETWORK TRANSPORT COSTS (TRANSPORT COSTS, RELIABILITY COSTS, INVENTORY COSTS) ECONOMIC IMPACTS OF INDUSTRY RESPONSES - LOST JOBS, LOSS IN INDUSTRY EARNINGS, PRODUCTIVITY, ETC. PROBABILITY BANDS DIRECT AND INDIRECT FACILITY IMPACT (E.G., PORT AND SUPPORT JOBS LOST FROM CLOSURE) Step 5 ECONOMIC IMPACT MODELING IMPACTS ESTIMATED OVER TIME (SHORT, MID, AND LONG TERM) AND SPATIALLY (SITE SPECIFIC, LOCAL REGIONAL, NATIONAL) PHYSICAL COMPONENTS OF SUPPLY CHAIN Customer Locations Customer Locations Production Location Production Location Production Location Local Move to Long Distance Mode Long Distance Move Local Move To Distribution Center Local Distribution Centers Local Move to Customers Customer Locations divert to alternative facility Figure 3-1. Five-step decision/analysis tool for studying freight network disruptions.

analysis Framework 35 or through traffic or a reassignment of vessels to other berths or terminals, to shifts in operating times to avoid increased on-network congestion, to a net loss of cargo docking/storage/cargo transfer capacity at ports and land freight consolidation/break-bulk terminals. Longer term disruptions involving major damage to port or other terminal facilities may lead to truck, train, vessel, or aircraft rerouting to other locations and to a resulting drop in local freight traffic volumes. It may also be necessary to account for disruptions to pipeline product flows. Model Inputs Step 1 DEFINE DIRECT FREIGHT TRANSPORT NETWORK IMPACTS - “PHYSICAL CHARACTERISTICS” facilities disrupted modes impacted likely time frame of disruption Figure 3-2. Define direct freight impacts. Model Inputs Step 2 IDENTIFY CURRENT AND FUTURE AFFECTED NETWORK FLOWS BY FACILITY AND LINK current volumes by commodity tonnage and value projected volumes by commodity tonnage and value Figure 3-3. Identify current and future affected network flows.

36 Methodologies to estimate the economic Impacts of Disruptions to the Goods Movement System Data Needs The principal data needs at this step are the following: • A link-node representation of the regional transportation network. • Time series data on regional freight traffic counts (or better still, origin-destination and route- specific traffic movements) from before and after the event. Data on projected/forecast flows over these same network links would also be useful. • Data on the locations and the pre- versus post-disruption throughput and storage capacities of local and regional truck, rail, waterway, and air freight trans-shipment centers in the region. 3.1.3 Step 3: Define Supply Chain Characteristics and Parameters by Flow Type Freight supply chains, which involve moving a commodity from production site to final customers, differ a good deal by commodity transported, including whether the cargo involves shipping raw materials, intermediate, or finished goods. Both physical and financial/institutional adjustments will often be required, typically leading to higher total transaction costs. Physical aspects of supply chains (see Figure 3-4) can vary from simple single mode, direct source-to- market linkages, to multi-link and multimodal linkages involving one or more cargo consolida- tions, breakups, and/or vehicle/vessel transfers. The true costs of disruptions to freight deliveries in the latter case are very difficult to estimate, beyond the effects on the immediately impacted Step 3 DEFINE SUPPLY CHAIN CHARACTERISTICS AND PARAMETERS, BY FLOW TYPE Supp Production Location Production Location Production Location Local Move to Long Distance Mode Long Distance Move Local Move To Distribution Center Local Distribution Centers Local Move to Customers Customer Locations Customer Locations Customer Locations Figure 3-4. Define supply chain characteristics and parameters.

analysis Framework 37 leg in the physical supply chain. However, this may be sufficient for most purposes. Not often captured to date, but potentially costly, are the effects on freight handling costs as well as inven- tory carrying costs from unreliable service delivery times. Also not generally considered are the costs of renegotiating freight deliveries by other means. Shippers, or their freight forwarders or logistics service providers (3PLs), will often need to either change their consignee designations or issue new contracts for both drayage or rail land transfers. Shippers and carriers may have to change their routing documentation on the cargo to maintain ownership and liability insurance. Contract terms between shippers, carriers, and terminal operators may need to be modified. If international cargo is involved, U.S. Customs may need to shift the port of entry, potentially involving multiple cargo releases and re-manifesting of in-bond and other classes of cargo. Such financial costs may be relatively short-lived, but may be significant for shippers and carriers if current contracts cannot be met. Data Needs Data on the physical and financial transaction costs that result from a transportation network disruption are needed, including administrative and other logistics costs that are encountered, above and beyond any direct shipper-carrier negotiated transportation rates. 3.1.4 Step 4: Supply Chain Response Modeling Step 4 models the response of the supply chain to disruptions (see Figure 3-5). This is the most challenging of the five steps in the process, with limited empirical evidence avail- able currently from the literature and with much depending on the specific industrial sector studied as well as the range of logistical arrangements possible by the firms operating within each sector. Figure 3-5. Supply chain response modeling. Intermediate Modeling Step Step 4 SUPPLY CHAIN RESPONSE MODELING INDUSTRY RESPONSE divert to alternative facility modal diversion other transport response production suspended - inventory production locations shift FLOW ROUTING RESPONSE

38 Methodologies to estimate the economic Impacts of Disruptions to the Goods Movement System Data Needs There are the following data needs: • Users can assume average diversion distances and travel times based on knowledge of the network for all commodities (e.g., similar to the DIETT Model procedure). Additional time and cost penalties may be built-in to reflect rescheduling (deferred) shipments and also overtime cost penalties, possibly suitable for the high-level methodology, but this entails substantial error and crude approximations. • Users can assume differential diversion distances and travel times by mode, distance travelled, and product value (e.g., similar to DIETT Model procedure). Additional time and cost penalties may be built-in as above. This approach is possibly suitable for the high-level methodology, but entails error and approximations. Commodity-specific assumptions may reduce the extent of approximation error. • Survey-data-based response modeling of shippers and carriers. Users may assume similar behavioral responses to those collected from previous surveys and reported in the literature, and matched to similar disruption situations. Highway network rerouting and truck load con- solidation models make similar assumptions in a planning situation to real world trucking responses, such as share diverted, share rescheduled, etc., based on matching characteristics. Modeling Possibilities Depending on the geographic extent and duration of the network disruption, analytic modeling will be needed to address the following topics: • Freight flow delivery responses – Short-term rail, barge, truck, and air freight traffic rerouting models, including the possible rerouting of trans-continental landed trade or ocean vessel movements; – Short-term mode shift models from, for example, barge to rail or to truck; – Short-term source-market restructuring of origin-destination flows; – More permanent restructuring of origin-destination flows. • Industry responses – Inventory draw downs, – Reduced or suspended production levels, – Shifts in the location(s) of production. Depending on their duration and extent, any shifts, re-locations, suspensions, or delays in production may have significant impacts on carrier, shipper, and receiver profits. More permanent forms of origin-destination freight delivery channel restructuring could have even greater impacts. 3.1.5 Step 5: Economic Impact Modeling This step models the economic impacts of network disruptions (see Figure 3-6). The extent and depth of economic impact modeling will depend heavily on Step 4—i.e., what is assumed about supply chain responses to disruptions, or what is estimated based on more in-depth modeling. For the high-level methodology presented in Chapter 5, relatively simple and straightforward assumptions and rules of thumb are developed and applied. Figure 3-7 illustrates the basic concept of the types of assumptions and methodologies that could be employed at each stage of the analysis process, for both a high-level methodology and an in-depth methodology. As seen in the figure, economic impacts for the high-level methodology would focus on simple rules of thumb and short-term, direct impacts. Rules of thumb would include unit transportation and inventory costs, jobs lost per container for direct facility disruptions, and “elasticities”

analysis Framework 39 that could express the average relationship between increases in transportation costs or delayed delivery costs, and industry output and sales revenues. The types of data, analysis tools, and the outputs will necessarily differ depending on whether the high-level methodology or the in-depth methodology is being used. A high-level meth- odology may not allow detailed analyses of supply chain or industry behavior responses. An in-depth methodology should include a more rigorous and nuanced assessment of this type of supply chain behavior. For example, for the supply chain response component in a high-level methodology, one could use simple route choice or modal diversion curves or sketch modeling, whereas more detailed logistics or industry response modeling would be used for an in-depth methodology. Data Needs For the high-level methodology, input data requirements would be kept to a minimum (e.g., no sophisticated network modeling of freight flows would be conducted). Instead, reason- able assumptions would be made about length of detours or diversions, shipping delays, and modal diversions. These, in turn would provide simple measures to which economic impact rules of thumb could be applied (e.g., if we know that transportation costs for a particular commodity will Model Outputs SOCIAL AND PUBLIC SECTOR COSTS - E.G., INCREASED HIGHWAY CONGESTION AND MAINTENANCE, FROM RAIL TO TRUCK DIRECT SUPPLY CHAIN COSTS - NETWORK TRANSPORT COSTS (TRANSPORT COSTS, RELIABILITY COSTS, INVENTORY COSTS) ECONOMIC IMPACTS OF INDUSTRY RESPONSES - LOST JOBS, LOSS IN INDUSTRY EARNINGS, PRODUCTIVITY, ETC. PROBABILITY BANDS DIRECT AND INDIRECT FACILITY IMPACT (E.G., PORT AND SUPPORT JOBS LOST FROM CLOSURE) Step 5 ECONOMIC IMPACT MODELING IMPACTS ESTIMATED OVER TIME (SHORT, MID, AND LONG TERM) AND SPATIALLY (SITE SPECIFIC, LOCAL, REGIONAL, NATIONAL Figure 3-6. Economic impact modeling.

40 Methodologies to estimate the economic Impacts of Disruptions to the Goods Movement System increase by a given amount, we can estimate, using cost-industry output elasticities, the impacts to industry output, final sales, and from this, impacts on sales, taxes, employment, and earnings). For the in-depth methodology, the simple approach is broadened to incorporate either freight network modeling or detailed statistical or panel studies of supply chain responses. A broad database of observed responses and data from previous surveys and literature may be assembled and could be matched to similar disruption situations. This includes any survey- based responses from impacted shippers, carriers, freight handlers and forwarders, and receiv- ers. Detailed shipment and other supply chain handling and transaction cost data are required for pre- and post-disruption conditions, by principal cost element (freight rates, delivery time penalties, insurance rates, fuel costs, etc.). Data on inter-industrial linkages, typically in the form of inter-industry by commodity use-and-make tables, is needed to capture indirect impacts. The economic impacts would then be expanded in the in-depth methodology to use more sophisticated economic analysis tools and models, including I-O-based tools and gen- eral equilibrium economic impact models that typically involve an econometric or statistically calibrated component. For long-distance, inter-regional impacts, the FHWA’s national com- modity flow database (the FAF) will need further disaggregation down to county, or perhaps even lower, levels of spatial resolution. No standard method currently exists, although there have been a number of recent efforts to do so (see Southworth 2009), while others are ongoing (e.g., NCFRP 20). IHS Global Insight’s Transearch database (see http://www.ihs.com/products/ global-insight/industry-analysis/commerce-transport/database.aspx) offers a popular propri- etary county-level multiple commodity flows matrix. However, at the time of this writing, all such matrices must rely on a combination of shipper and carrier survey plus secondary data DISRUPTION IMPACT METHODOLOGIES -TOOLKIT Freight Data Assessments Supply Chain Responses Economic Impact Methodologies Economic Impact Metrics High Level Methods In Depth Methods Regional Input Output Models - -Direct transportation and inventory costs, supply chain losses plus multiplier impacts -Losses in Regional Economic Activity from Port and other Iintermodal Facility Disruptions High Level Freight Flows Data basic commodity and O-D data -FAF -Commodity Flow Survey Dynamic Modeling Tools: Econometric Models, I-O Interregional Models -Long run Industry location, outpout or other response models (e.g., PRISM model) Basic Route and Modal Diversion Modeling Assumptions -Diversions based on alternative routes, modal options, and capacity Freight Network Models National, Regional, International Trade -Least cost and time path diversion analysis -Logistics response models -Industry response models National I-O Model -Direct transportation and inventory cost impacts from diversion plus multiplier impacts -Economic losses from MAJOR facility disruptions, such as major seaports examples: -MARAD Port Econ. Impact Kit -IMPLAN National Level and Sub - National Region Impacts -GDP, employment, income -Static Impacts over time - no behavioral responses National, Sub-National Region, Metropolitan and Corridor Impacts -GDP, GRP, employment, income, fiscal impacts, -Dynamic impacts over time - supply chain and industry response impacts included Detailed Freight Flows Data (if available) -MPO or statewide freight models -TranSearch Data -Detailed adjustments to FAF or other high level freight data Figure 3-7. Disruption impact methodologies toolkit: high-level and in-depth methods.

analysis Framework 41 sources (e.g., county or zip code area population and economic activity reporting) and a number of flow modeling assumptions. Modeling Possibilities The possibilities are organized into the following three groups of costs or impacts. 1. Social and public sector costs • Changes (increases or decreases) in network maintenance costs, based on net regional, mode-specific VMT changes – Changes from traffic rerouting will have little impact on overall routine highway main- tenance costs since regional VMT will not change much, but there can be a transfer of impacts from one location. or possibly from one state to another, in cases of widespread diversion incidence – Changes are more significant where there are modal diversions and truck volumes fall significantly across the entire network • Increased network congestion on undamaged facilities – Crude estimates can be based on additional VMT under congested conditions and based on diversion assumptions or actual diversion analysis, using sources such as the Texas Transportation Annual Mobility Report – Within a travel demand model where additional levels of link congestion can be estimated and measured using metrics such as VMT or VHT under congested conditions (LOS D,E,F) 2. Changes in externalities, such as emissions, based on net regional changes in mode-specific vehicle miles of travel – These impacts may also need to be weighted to reflect increased overall levels of congestion from truck reroutings (i.e., if volumes are diverted from LOS C facilities to other facilities, which then operate under LOS F) emissions and congestion costs would both experience net increases 3. Direct supply chain costs • Increased direct truck transport costs from diversion, based on vehicle mile and vehicle hour changes – Given truck VMT and VHT changes, trucking cost factors can be readily obtained from numerous benefit-cost analysis studies and sources • Increased inventory carrying costs, calculated based on changes in ton hours by commodity type, value of the commodity, and assumed time value of money (inventory costs) – Increased direct truck transport costs from diversion can lead to increased inventory costs calculated based on changes in ton hours by commodity type, value of the commodity, and assumed time value of money (inventory costs); the DIETT Model, for example, includes algorithms and look-up tables for this – Possible increases in costs due to investments made in maintaining more resilient (more flexible, with more redundancy) supply chains. Assumptions may also need to be made about the share of shipper costs borne by/passed on to the carrier or to beneficial owners of cargo (BFOs). Assessments should also reflect net regional impacts, such as potential increases in local trucking activity in some areas at the expense of other areas (i.e., potentially no net change in costs or benefits, but a change in the spatial distribution of impacts). Measures include producer and consumer surplus. Inter-Industry and General Equilibrium Impacts These are industry impacts due to increased transport costs and time, reduced reliability, reduced supply chain efficiency, increased warehousing, and other supply chain modifications. This analysis may include both direct and indirect impacts due to the effects on firms’ abilities to satisfy final demands for their products and services. I-O modeling is the most common approach

42 Methodologies to estimate the economic Impacts of Disruptions to the Goods Movement System to use here (not necessarily using the complete U.S. framework). More elaborate inter-regional and iteratively designed general equilibrium models may be useful for more detailed studies over protracted periods of time. Metrics include changes in employment, income, value added, business income, and gross regional product. The ability to model industry-specific impacts is limited by the extent to which commodity-specific flows and transport plus other transaction cost increases can be estimated in Step 3 above. Considerations include the following: • Direct industry impacts represent that share of the overall transport cost increase borne by each industry sector (e.g., manufacturers, big box retailers, agricultural and other raw materi- als producers, and transportation, warehousing, and distribution service providers). Increases reflect increased transportation costs and increased inventory costs, which combined result in increased overall logistics costs, reduced reliability that can be monetized, supply chain inef- ficiencies, loss of JIT capabilities, etc. • Increased transport and supply chain costs are described above in Step 3, but further model- ing can be applied, such as via the PRISM regional economic impact model, to derive, not just direct, but also indirect and induced impacts due to the effects on firms’ outputs and abilities to satisfy final demands for their products and services. PRISM utilizes an I-O model (IMPLAN) to estimate these, combined with assumptions and information specific to each industry/commodity impacted. • Metrics include changes in employment, income, value added, business income (lost sales), and gross regional product. • The ability to model industry-specific impacts is limited by the extent to which commodity and NAICS-specific freight flows and transport cost increases can be estimated in Step 3 above. • Assumptions can be made about the extent to which increases in transport and other logistics costs (due to diversions, more road time, more VMT, and increased overtime hours) are passed forward to shippers versus absorbed by the trucking companies and operators. Both producers and other consumers of freight transport, as well as the freight transport providers themselves (i.e., the truckers), are sectors within the economy, and each sector can be treated differently in terms of inter-industry impacts. For some major firms, those providing in-house transportation, costs are a direct impact to the businesses. • Potentially useful models include PRISM, TREDIS, REMI, and IMPLAN, which model the static inter-industry relationship structure via I-O models and, in some cases such as REMI, are more general equilibrium models. • Simplified I-O-based models are more suitable for the high-level methodology in estimating the unit changes in economic impact due to changes in the transportation system. REMI or similar GE models are less suitable for a high-level methodology due to very high implementation costs and complexity of modeling approaches.