Guidebook for Air Cargo Facility Planning and Development (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.

64 Cargo forecasts are generally undertaken as part of an airport’s master planning activity, as part of an environmental assessment, to accommodate facility improvements, or in response to unforeseen demand or expectations of the local business community. They are then used to assist planners in the identification of future cargo facility and apron requirements. 5.1 Data Sources Virtually all U.S. airports track total cargo volume, as well as subsets of cargo such as freight (including express) and mail, on a directional (inbound and outbound) basis. Commonly, these data sets are managed by airport accounting departments and are compiled from monthly opera- tions reports used to settle landing fees and satisfy other carrier reporting requirements. Whether disseminated publicly or not, this data is kept by the airport on a carrier-level basis, which can be organized into market share by individual carrier or type (all-cargo versus belly). For those airports to which it is applicable, cargo will also be organized into domestic and international increments. In addition to tonnage data, monthly airline reports provide critical inputs related to monthly frequencies and aircraft types. There is no single standard for how or if airports generate public reports from this and other data. While the web page of the Port Authority of New York & New Jersey contains extensive monthly data sets pertaining not only to airport operations but also to customs entries by country and commodity, for example, other airports may include noth- ing more than a single entry for total annual cargo in their public reports. Almost all member airports report annual tonnages to ACI-NA, which publishes a top 50 data set by year on its web site and a more extensive set for members only. However, U.S. airports are not compelled to join ACI-NA, and major cargo airports, such as UPS’s regional hub in Rockford, IL, will not be found in ACI-NA’s statistics. Air cargo tonnages are typically reported by airport commissions and are reported to the public on an annual basis, but monthly reports are useful to isolate seasonal trends. While it is uncommon for carriers to report weekly or daily tonnage numbers, planners can use second- ary references (such as OAG’s Cargo Flight Guide) or request carrier schedules to record flight operations in peak-period analysis; this is critical where aircraft parking ramps are at a premium. Because of the limitations of individual references, it is advisable to use multiple sources of pri- mary and secondary inputs. OAG’s Cargo Flight Guide, for example, does not include schedules for integrated carriers and may also identify flights by ACMI (aircraft, crew, maintenance, and insurance) carrier rather than client. A variety of institutional sources are commonly used to calibrate individual airport fore- casts, including forecasts by Boeing, Airbus, IATA, and the FAA. These are detailed later in this chapter with an assessment of the strengths and weaknesses of each source. For specialized C H A P T E R 5 Air Cargo Forecasting

Air Cargo Forecasting 65 facilities—such as cold storage facilities—airport planners may seek trade data originating with U.S. Customs and Border Protection (CBP) that can quantify monthly and annual tons by com- modity type for both import and export shipments cleared at the local customs port. Much trade data can be accessed at no cost from the U.S. Census Bureau and through subscriptions to governmental sources such as the BTS TranStats service. Secondary commercial providers also sell packaged reports that often blend public and proprietary sources. There is no substitute for the unique perspectives obtained through original interviews and surveys with on-airport cargo-related tenants and key off-airport constituencies. The former may include local station managers as well as corporate property managers and route plan- ners who commonly have distinctive insights into carriers’ intentions for the local market. Off- airport constituencies may include freight forwarders, trucking companies, and major shippers (manufacturers and distributors) of time-sensitive commodities. Area economic developers may also provide insights and data characterizing the local origin-and-destination market. It is essential that the planner preparing the air cargo forecast understands the cargo market(s) under consideration in terms of commodities being shipped and received, as well as the economics of modal alternatives and nearby competing airports. 5.2 Methodologies 5.2.1 Time-Series Trend Analysis One of the most common forms of statistical analysis is the discrete time series, which observes phenomenon through regularly spaced intervals (contrasted with the continuous time series, which records an observation at every instant of time). This analysis can be organized to mea- sure trends that may be extended to forecast future values. To be used as a predictor, time-series analysis requires confidence that the period to be forecasted will be much like the period from which the trend multiplier (usually a CAGR) was derived. (CAGR = (ending value ÷ starting value)1/(number of years) - 1). CAGR provides a smoothed rate of return describing yield on an annu- ally compounded basis. One of its weaknesses is that it does not reflect volatility, which can be substantial from one year to another, but rather creates the illusion that there is a steady growth rate. For many years, a 20-year horizon was the accepted time frame for forecasting. Clearly, the early years had the greatest credibility and the most distant years the weakest. Airport activity has been volatile as the airline industry has been affected by uncontrollable factors such as escalating fuel prices, economic swings, and labor issues. Longer historical periods are still often preferred, but the beginning and ending years of the time series should be closely scrutinized for the effect that anomalous years can have on trend analysis. While it is customary to use increments such as decades in a time series, a 10-year time series initiated with the extraordinary losses in 2002 would likely miss common peak years (use- ful in gauging historical capacity) from the late 1990s through 2000. On the other hand, a longer time series must be qualified in terms of applicability because the industry itself has changed so greatly since the 1990s. In fact, the demise of former all-cargo ten- ants such as Airborne Express, BAX Global, Emery Worldwide, and Kitty Hawk has left sprawl- ing vacancies at on-airport cargo facilities. In many predominantly domestic air cargo markets, market shares of FedEx and UPS have risen from around 50% 20 years ago to over 90% in 2012. Such market consolidations may have the twin effect of emptying multi-tenant buildings of folded former legacy carriers while leaving the surviving dominant carriers more likely to have required single-tenant (stand-alone) facilities dedicated to their individual operations. The ulti- mate outcome is a dearth of prospects to fill vacancies.

66 Guidebook for Air Cargo Facility Planning and Development In international gateways, gains in international cargo tonnage have at least partially masked losses in domestic cargo. Total cargo tonnage may have changed very little in the course of 20 years, but the carrier composition may have changed dramatically. Similar changes may have transpired in the mix of belly cargo market share versus freighter share. Forecasts created in the last 10 years have typically assumed that the bottom of the air cargo demand cycle had already been found and that recovery would begin immediately. Unfortu- nately, many forecasts based on that assumption have proven overly optimistic to date. Even if only to provide a contrast, time-series analysis remains a useful planning tool. If con- cerns exist related to anomalous years of data, multiple analyses can use a variety of beginning and ending years. Regardless of the interval, charting market deterioration since past peaks illu- minates how much facilities capacity may exist from an airport’s past peak demand. The ideal use for trend analysis has been described as a mature industry experiencing relatively consistent, gradual growth—a description that contrasts greatly with the recent experience of the U.S. air cargo industry. 5.2.2 Regression Analysis (Econometric Modeling) Regression analysis is a statistical technique for estimating relationships between a dependent variable and one or more independent variables. Regression analysis helps explain how the typi- cal value of the dependent variable changes when any one of the independent variables is varied while the other independent variables are held fixed. Regression analysis is widely used for prediction and forecasting. Regression analysis is also used to understand which independent variables are related to the dependent variable and to explore the forms of these relationships. In restricted circumstances, regression analysis can be used to infer causal relationships between the independent and dependent variables, with the critical caveat that correlation does not by itself prove causation. The dependent variable of air cargo growth may be associated with such independent variables as jet fuel prices, GDP, composite leading indicators (CLIs), and population—customarily using a combination of time-series and growth curves. Most U.S. airports only serve local or regional O&D markets. Therefore, cargo growth may track closely with local or regional economic attri- butes, so reliable functional relationships may exist between an airport’s cargo growth and the area GDP, income, and population growth. However, at international gateways, air cargo growth may be at least as influenced by economic conditions in O&D countries as by local economic con- ditions since air cargo is often trucked great distances across several states, or across the country in many instances, to these international gateway airports. Econometric modeling (such as multiple regression analysis) is often perceived as more effec- tive with broadly defined markets (countries and entire continents) in which multiple factors influence aggregate growth and other variables may be held constant. A flaw is this assumption that supply is unconstrained, which contrasts starkly with individual airports, where cargo capac- ity is constrained by the hub-and-spoke systems of carriers and limited aircraft fleets. The ability of carriers to shift capacity between airports as well as between air transport and other modes (particularly trucking) poses substantial risks to the assumption of unlimited capacity supply that meets graduated demand. U.S. airports have experienced extraordinary growth (Greensboro, NC, with FedEx) and losses (Des Moines, IA, with UPS) attributable to network adjustments by integrated carriers that seemingly had nothing to do with local cargo demand generation. Like time-series analysis, regression analysis is a useful tool to evaluate historical relationships between cargo growth and other econometric elements. However, it is an imperfect (wildly so in some circumstances) predictor of future trends—not least because of its assumption of

Air Cargo Forecasting 67 unlimited capacity supply—and, therefore, should be considered only one of several potential analytical tools. 5.2.3 Market Share Analysis Market share analysis compares local activity levels with a larger entity, most commonly in comparisons between a particular airport and its regional traffic or with total national traffic. His- torical data is used to establish the ratio of local airport traffic to total national traffic—customarily using source data from the FAA Aerospace Forecasts document for national data. Much like the preceding methodologies, market share analysis has limitations as a predictor. Most obviously, this methodology assumes that the proportion of activity that can be assigned to the local level is a regular and predictable quantity. As has already been established, the U.S. air cargo industry remains in the midst of a prolonged period of contraction that has touched most airports but not equally. For example, as in Figure 5-1 depicting Hartsfield-Jackson Atlanta Inter- national Airport, some gateways were able to offset some domestic losses with international gains. Indications in late 2012 from the two dominant integrators suggested that near- to medium- term domestic fleet utilization strategies may favor up-gauging aircraft size but serving fewer U.S. markets by air, while expanding the utilization of trucks for domestic feeder service. The impact may negate organic air cargo growth at many small and medium-sized markets or, con- versely, may support growth at strategically located airports that can potentially serve as access points to multiple, possibly larger, markets. All of the preceding suggest that imperfections exist in market share modeling as it pertains to projecting local airport trends relative to regional and national growth. Market share analysis at the individual airport level is integral to understanding how the mar- ket has evolved and, therefore, may indicate potential direction going forward. At the individual airport level, market shares of international and domestic as well as belly cargo versus freighter cargo are essential for facilities planning since this analysis informs judgments about future demand for freighter positions and other related considerations. Carrier market share—possibly through the prism of ground handlers possibly serving multiple carriers in a common warehouse and ramp space—is necessary for calculating the individual utilization rates of cargo facilities. In summary, market share analysis is an essential piece of air cargo analysis at the individual airport level, but as a predictor of future relationships between local and national trends, it must be qualified. Source: Airports Council International, Webber Air Cargo analysis. The star represents peak total tonnage. 100,000 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 – 200,000 300,000 400,000 500,000 600,000 700,000 800,000 900,000 1,000,000 Total Domesc Internaonal Figure 5-1. ATL annual cargo (metric tons): CY 2000–2011.

68 Guidebook for Air Cargo Facility Planning and Development 5.2.4 Institutional Forecasts The introductory section cited several sources of institutional forecasts commonly used by airport planners and others to calibrate local cargo forecasts. For the vast majority of U.S. air- ports, only domestic cargo is materially significant since international shipments of local O&D cargo will either be trucked or flown on a domestic segment to a gateway, with trucking increas- ingly likely and therefore having a negligible impact on the feeder airport’s cargo totals. Forecast- ing inbound and outbound domestic cargo (and related translation into freighter operations) will suffice. However, at international gateways, directional (import and export) forecasts will often be segregated by region (for gateways with multiple transcontinental routes), although often a composite international multiplier entails the international market share and growth rates of each individual segment. Using institutional forecasts is not a substitute for other methodologies but more accurately is a surrogate for the labor involved. Entities such as Boeing and Airbus perform intensive econo- metric modeling (GDP and fuel prices, to name two independent variables) to inform their biennial forecasts—typically with budgets and other resources well beyond the means of air- port planners and even most consulting firms. In fact, the FAA often cites the Boeing forecast, in particular, for use in its own efforts. However, just as the U.S. economy is composed of regional economies that may little resemble one another—the Rust Belt versus the Farm Belt, for example—local airports may not conform precisely to national economic expectations. If such institutional forecasts are used as the basis for individual airport forecasts, adjustments should be made to recognize local conditions. The latest Airbus effort is their “Global Market Forecast: 2012–2031.” It is an integrated doc- ument entailing both passenger and cargo forecasts—contrasted with Boeing, which releases separate reports. Passenger forecasts are available from both sources and may be of particular use in incremental considerations of belly cargo capacity versus freighter demand. Both Airbus and Boeing forecasts are available as free downloads from the respective corporate websites. Signifi- cantly, both companies’ cargo forecasts project growth rates in terms of revenue-ton-kilometers [1 ton of revenue-producing cargo flown 1 mile (Boeing) or kilometer (Airbus)], which clearly puts a premium on longer-haul segments (such as those over the Pacific), while airport cargo forecasts are typically expressed in cargo tons and flight operations as a derivative forecast. A significant liability of the Airbus forecasts has been that detailed cargo forecasts have been produced in less reliable intervals—not surprisingly for a manufacturer that has struggled com- petitively in the freighter market. A significant advantage is that Airbus’ market forecasts tend to be segmented into much smaller sub-continental groupings, allowing more precisely delin- eated pairing of routes and markets. For international gateways with diverse networks of direct destinations, this advantage is invaluable. For airports where only domestic or perhaps only modest international service is offered, planners may use either (or both) the Airbus or Boeing forecasts for guidance. Whether one source is more conservative than the other is only evident on a segment-by-segment basis but not as a whole. Boeing’s “World Air Cargo Forecast 2014–2015” is the latest biennial installment of the cargo-specific document. Like the Airbus version, the Boeing 20-year (through 2033 in the latest installment) forecast is compiled from econometric models and airline interviews undoubtedly enriched by Boeing’s dominance in the freighter market and resultant access to the insights of the world’s dominant freighter operators. While the Boeing forecasts are not as narrowly strati- fied as those of Airbus in terms of market segmentation, it has the significant advantage of a timely production schedule and relatively uniform structure over time—facilitating the reuse of forecast templates by airport planners. Perhaps the Boeing cargo forecast’s greatest virtue is that it is unlikely to meet any critical opposition since it is so ubiquitous in the efforts of the FAA and

Air Cargo Forecasting 69 others. Clearly, the popularity of Boeing’s forecast derives in large part from the acceptance of its methodology and its history of reliability. The FAA Terminal Area Forecast (TAF) provides summary historical and forecast statistics on passenger demand and aviation activity at U.S. airports based on individual airport projections. The TAF model can be accessed from the FAA website, and model users can relatively easily generate their own forecast scenarios. The principal input for the TAF is the FAA Aerospace Forecasts, which are developed from econometric models intended to explain the relationships and emerging trends for all major seg- ments of air transportation. Typical of econometric models, the FAA forecasts assume uncon- strained capacity. They also assume no further contractions of the industry through bankruptcy, consolidation, or liquidation. These assumptions are likely to change after forecast publication, and the FAA provides cautions accentuating the recent unpredictability of commercial aviation. Both the FAA Aerospace Forecasts and the TAF are repositories of economic data that may be useful in conducting regression analyses. They also possess forecasts for passenger activities use- ful in considerations of potential belly capacity available for cargo. The air cargo element of the FAA Aerospace Forecasts [in revenue ton-miles (RTMs)] assumes that security restrictions on air cargo transportation will remain in place and that most of the shift from air to ground transportation has already occurred. Finally, the forecasts assume that long-term cargo activity will continue to be tied to economic growth. While obviously uncer- tain, these assumptions are defensible. The forecasts of RTMs are based on models linking cargo activity to GDP, with domestic cargo RTMs linked to real U.S. GDP as the primary driver and international cargo RTMs based on world GDP growth (adjusted for inflation). Distribution between belly and all-cargo carriers is forecasted on the basis of historic trends in market shares, changes in industry structure, and market assumptions. IATA produces an annual cargo-specific forecast that is stratified into more narrow market segments than any of the preceding forecasts. Its liabilities include that the detailed version must be purchased (unlike those previously cited), and it is only completed in 5-year increments. It should be noted that IATA only forecasts in 5-year increments due to the belief that forecasts beyond that horizon are so seriously compromised as to be virtually meaningless. That is an assessment with which many industry observers agree. Potentially among the most illuminating sources of forecasts would be the air carriers, which commonly develop in-house forecasts with 5-year increments being common for traffic and 5- to 10-year increments for fleet forecasts. Particularly at hub airports, where a single carrier has a commanding share of belly cargo, and at the many airports where FedEx and UPS may have combined market shares in excess of 90%, carrier forecasts would be invaluable. Unfortunately, these forecasts are considered commercially sensitive and are rarely shared with airport opera- tors or their consultants. However, the preferred collaborative process of developing forecasts should present the opportunity to at least test the airport’s own forecasts against perceptions of the carrier-tenants. Moreover, the carriers will typically provide input into operations forecasts related to fleet expectations for the near- to mid-term. 5.2.5 Operations Forecasts Airports’ cargo operations forecasts are principally derived from tonnage forecasts. As much as tonnage is a critical input for planning warehouse capacity, operations are critical for plan- ning ramp capacity. Airport planners need as much feedback as possible related to carriers’ fleet and route plan- ning. While the gauge of aircraft is critical to calibrate aircraft capacity, it is also critical to know

70 Guidebook for Air Cargo Facility Planning and Development how much of the payload is dedicated to the local market. If the aircraft continues to other cities to build/break loads before returning to the hub, partial loads decrease throughput anticipated for the warehouse and may shorten the time the aircraft will be on the ground. A thorough understanding of airline schedules may allow airport planners to maximize the use of aircraft ramp positions by getting multiple turns on a single position when schedules are com- patible. Moreover, a carrier may be able to double or triple its local tonnage without adding another operation if its current payload dedicated to the local market is small. These considerations are particularly important at international gateways such as ATL and Dallas/Ft. Worth Inter national Airport (DFW), where international freighter operators commonly share multi-stop service with other gateways. Clearly, this information must also be reconciled with the actual capacity of each ramp position in terms of the maximum gauge of aircraft that can be accommodated. Airport planners can extract current fleet and flight operations data from landing reports and flight schedules from proprietary sources such as OAG Cargo Flights (www.oagcargo.com). Industry-wide fleet information can also be gained from Airbus and Boeing, as well as from secondary sources such as Air Cargo Management Group’s Cargo Facts (www.cargofacts.net). Both Cargo Flights and Cargo Facts are available on a subscription basis. No matter how cred- ible the secondary sources, interviews with cargo carriers (and handlers where applicable) are indispensable for verifying potentially outdated secondary sources as well as for gaining unique forward-looking insights into prospective future operations on a specific market basis. In order to derive operations from tonnage, airport planners must first determine the market share presently transported by passenger carriers (therefore not contributing to freighter opera- tions) and then make assumptions about future trends regarding that distribution. The FAA Aero- space Forecast provides such forecasts for both domestic and international cargo on a national airport system basis. Once that belly cargo has been deducted from total cargo to isolate the tonnage that specifically drives demand for freighter operations, planners must make assumptions about the carriers’ pay- load limits that would trigger either additional frequencies or a change in gauge of aircraft. Again, it is also critical to know how the local market is presently served by the carriers—as a stand-alone destination or as part of a multi-stop routing—in order to evaluate how much capacity is avail- able before another frequency would be required. Unlike passenger service, which mostly is daily, freighter service at many U.S. airports may occur on weekdays, with perhaps partial service on weekends. Consequently, airport planners may use an annual standard of 282 annual cargo days (5.5 days/week), adjusting according to local schedules, which may only have weekday (5 days/week or 260 days/year) or alternatively full calendar (7 days/week) service. Operations will typically be forecasted on a three-tier basis compatible with tonnage forecasts on low-, base-, and high-case scenarios. Additional matrices can easily be formed to create alternative forecasts on the basis of a range of load factors. While the approach just described for deriving operations from tonnage is appropriate for air- ports served by a variety of carriers, planners at airports with relatively modest cargo operations may opt for a simpler approach comparable to the market-share methodology described earlier. Applied to operations, the approach would entail simply calculating the tons/operation that the airport has recently experienced and then applying that average to future tonnage forecasts. On an applied basis, airport planners may combine the tons/operation with the airport’s number of ramp positions (recognizing variable capacity) and aircraft turns per day per position in order to determine total ramp capacity in tonnage terms.