Microeconomics of Reight Car Supply and Demand Douglass Wm. List Catherine S. Cook George F. List 3 The second part of the paper describes how a comparable process is used to estimate a demand curve from traffic records common on most freight railroads. Methods for handling such issues as the definition of demand and demand segmentation are reviewed. The section concludes with examples of actual demand curves . constructed using the outlined methodology. • Abstract To deal effectively with freight car investment . decisions, knowledge of relevant demand and supply conditions is indispensable. Critics of microeconomics argue that although important variables that influence demand and supply can be isolated by economic theory, they can never be measured accurately empirically. This paper offers a methodology for defining and quantifying a railroad's demand and supply for freight cars in classic microecono'mic terms, and shows how that knowledge can be used in the capacity investment decision process. The third section of the paper demonstrates how • the demand and supply curves interact, how the curves can be adjusted to reflect broader market realities, and how they can be. used to guide management.decision-making. Drawing on specific examples, the authors show how this approach to investment can lead to 'fundamental insights different from those often reached by other evaluation processes. . . . The first part of the paper describes how to construct 'a supply curve for freight car capacity. The relative economics of all sources for freight car capacity are explored, including existing servicer able equipment, potential repair program candidates, foreign equipment, leased equipment, and new cars. The paper is illustrated throughout with • data drawn from actual railroad situations. Management Consultant, Baltimore, Maryland. Formerly Vice-President, Marketing and Strategic Planning, CSX Equipment, President, Planning Associates, Cincinnati, Ohio. Formerly Director, Strategic Development, CSX Equipment. Assistant Professor, Civil Engineering, Rensselaer Polytechnic Institute, Troy, New York. 1 Introduction Decisions regarding the addition or deletion of freight car capacity are fundamental to a railroad company. Often involving millions of dollars, these decisions are usually made at the highest levels of the organization after extensive involvement of many managers and analysts. Despite the importance of these decisions, many managers remain frustrated that they are often made in the context of inadequate information information that fails to portray the key aspects of such a major capacity decision with adequate richness as to its merits, dimensions, and potential ramifications. In the paper which follows, we will show how the classic economic concepts of supply curves and demand curves can be used to evaluate and communicate key aspects of car capacity situations. This approach to evaluating car capacity issues was initially developed as part of the equipment planning process of CSX Equipment. A primary objective was to make more explicit the relationships between marginal costs and marginal revenues as-they relate to equipment investment. Its conceptual origins lie in a broad base of published and unpublished thinking on microeconomics and financial analysis in general, as well as the application of these concepts to the specifics of capacity investment, transportation, and railroad freight cars. Several of the more significant references in these areas are noted at the conclusion of the paper. So as to preserve an orderly presentation of the concepts, we begin by considering a single railroad as a "closed" system. Freight cars do notflow in and out of the system, and we can create standalone, internal demand and supply curves for this system without concern for broader markets. We subsequently explore how the framework can be adjusted to reflect market realities better, and how the concepts can be applied in the capacity investment decision-making process. Definition of Car Type Quantity demanded and supplied depend on a number of factors, not the least of which is the shippers' ability to substitute another car type for the one in question when market prices change. Some car types, such as jumbo covered hoppers, have close substitutes, such as medium cube covered hoppers and even'open top hoppers fitted with lids. A change in the price of jumbo covered hoppers, the prices of the substitutes remaining ' constant, can be expected to cause a certain amount of substitution. A fall in the price leads shippers to use more jumbos, and arise in the price • leads shippers to desire more of the substitutes. To a great extent, the responsiveness of quantity demanded and quantity supplied to price depends on how widely or narrowly the car type is being defined. There are very few close substitutes for "box cars", for example, butseveral alternatives to 50 foot, 100 ton box cars with cushion underframes. In our analysis, we assume for simplicity that there is no substitution, among car types. That is, the railroad cannot divert the demand for one car type to another. Constructing a Supply Curve for Freight Car Capacity We begin our discussion by looking at the supply curve for freight car capacity. We start with the supply curve because it is the easier to conceive, and influences the structure of the demand curve. Structural guidelines The supply curve is the relation between the amount of capacity supplied and theprice. For any given price we need to ask what amount of capacity the railroad will choose to supply. First, we define the appropriate capacity unit, such as "car-days", "car-months", or "car-years" along the • x-axis, and "$/car-day" or other cost per unit of capacity along the y-axis. Second, we identify the time period relevant to the investment decision. The pertinent time period will vary depending on the certainty of forecasted, demand and the flexibility of the railroad's repair, purchase, lease and build capabilities. We are primarily concerned with capacity that can be brought on-line within that time frame, keeping in Supply Curve Car Type A Dollars «o y per car per day New Cars <•• '•— Medium Repair — Light Repair v : •y V l 0.0 ',, , ..v-? ..* -. l -1 l l l l i 3.0 6.0 9.0 12.0 l l 15.0 i ^ ,' , i l l l .18.0 21.0 i 24.0 i 1 27.0 1 -% i s : l l 30.0 - : \1 33.0 36.0 39.0 Cumulative Equivalent Cars (OOOs) Figure A Freight Car Capacity Supply Curve mind that we are as interested in potential capacity as we are in existing capacity. Hence, a one-year analysis should consider only how many new cars could reasonably be purchased and installed within that time frame, not the infinite number of new cars which could be purchased in the longer term. To create the supply curve, blocks of capacity are arrayed from left to right in order of ascending cost, as shown in Figure A. The process for estimating the cost of each block of capacity is explored in more detail below. Existing serviceable fleet Determining how much capacity is represented by a given set of cars requires some thought Ultimately, we need to ensure that the measure of capacity used in the supply curve is consistent with that used in the demand curve. We favor the use of a measure like "available car days" as a unit of capacity since it is easier to manipulate consistently with data from the demand side. To measure the "available" car' days from a set of cars, total car days (number of cars times total time period) must be reduced for unavoidable out-of-service time required for normal maintenance and other unavoidable time between loads not captured by the demand analysis. This is the least costly segment of the fleet to use. Its cost can be modelled as the basic maintenance costs to keep these cars in service: brake shoe replacement, wheel replacement, and the like. Such costs typically run on the order of $25-50 per carper month. Existing unserviceable cars Railroads usually possess an inventory of cars which require substantial repairs to return to service. These cars are typically listed as "(heavy) bad- order" equipment In general, repair to these cars is a more economical way to add capacity to the servicea"ble fleet than the acquisition of new cars. For purposes of the supply curve, we price this latent capacity as an annuity. First, meaningful repair categories may need to be defined, based on extent and cost. For each group of cars with similar repair needs, we estimate: the cash cost of the repair, the expected serviceable life of the car following repair, the periodic cost of keeping the car in service, and the residual value of the car when it falls out of service. Standard discounted cash flow techniques can be used to convert this stream of cash flows into an equivalent periodic Discount Rate: Full Life Cycle Basis 15.0% New Car Cost: . $35,000 Anticipated Life (years): Repair Schedule: Year 8 Year 15 Year 23 30 8,650 10,100 10,100 ves appear to function just as we would expect from microeconomic theory, with prices on available cars hovering just below the level that would make it attractive to invest in additional capacity. New cars New car construction is usually at the far right side of the supply curve. There are two distinctly different methods for valuing new car construction, with profound implications for capacity investment decisions. On some railroads, it is typical to look at the cost of new car capacity on a full life-cycle basis. A complete economic life cycle of cash flows is estimated that includes not only new car acquisiService tion costs and normal operating Life Cycle costs, but also a provision for Basis majorrepairs several times over 15.0% the life of the car (say in years 8,15 and 23 for a car that is to $35,000 last 30 years). This entire stream . of cash flows is then reduced to 8 a single periodic payment' 0 0 0 An alternative approach is to treat the new car in precisely the same manner as an unserviceable car being repaired- We do not 854 854 Basic Maintenance (per year): assume that the car will automat2,000 4,500 Salvage Value: ically be repaired at the end of its . initial service cycle. Instead, we $(37,082) . Net Present Value (before taxes): $(45,052).. simply assign a residual value to $(6,862) $(8,264) Equivalent Annual Cost: the "hulk" which will remain at the end of the first service cycle. These two alternative apFigure B Estimating the Cost of New. Car Capacity proaches to estimating the equivalent periodic cost of new payment This periodic payment is the cost as-. car capacity are shown in Figure B. signed to this source of capacity in the supply curve. Working through these two different approaches with numbers, one discovers that the first approach In effect, the "annuity" approach is an estimate of yields a dramatically lower estimate of periodic cost the lease rate'we would have to pay to get the for new equipment than the second approach. In specified capacity .from an external investor. Minieffect, the automatic repair assumption creates mum free market lease rates for equipment are set "cheap" future capacity that subsidizes the cost of in precisely this manner. Investors contemplating the first service cycle through the discounted cash rebuilding cars to place them on the market look at flow analysis. these very same cash flows to determine whether current market rates will support the financing costs Which approach is correct? That depends largely of the anticipated rebuilding. Market rates themselon investor behavior. Some freight car owners are so intent on maximizing asset productivity that they can be expected to meet downstream repair needs almost automatically'. These owners can "take the long view" and effectively plan for an ongoing cross-subsidy from older cars to newer ones. Most railroads, however, behave differently. Cars which fall out of service are accumulated until there is a clear need to repair them. The decision to expend funds on major repairs is subject to extensive internal review. In such situations, any assumption of automatic downstream repairs when acquiring new cars results in a gross underestimation of the true cost of the additional capacity. The single cycle approach is also somewhat more realistic if one believes there are reasonable risks of premature technological obsolescence. From a purely economic perspective, one would argue-that new cars should be viewed as having a single service cycle with a probability distribution of residual values. The ."worst case" scenario is that the car will only go one service cycle because lack of demand or technological obsolescence precludes repair for a second cycle. Immediate repair andretum to service is in fact the "best case" scenario in most.instances. Observations about the supply curve Following the principles outlined above, we can construct a supply curve for freight car capacity as shown previously in Figure A. If the analysis has been done correctly,- the total number of cars owned should be accounted for. Discrepancies may be due to ambiguous car type .definition and /or inaccuracies in record keeping. The general shape of this curve is significant We have a large region over which capacity is almost free. 'Capacity then moves up steeply in price, ending in a region where no additional capacity can be obtained at any price for the period in question. This shape to the car capacity supply curve helps us understand the behavior of the freight car capacity market. Relatively small swings in demand can create dramatic changes in market prices for equipment. The supply curve we have constructed represents a fixed period in time. Of greater interest to the manager and planner is how this supply curve changes over time. In the very short run, the railroad can expand capacity only through basic maintenance. The supply curve resembles a backwards "L". At the limit, the amount of supply is perfectly fixed at the existing serviceable fleet level, regardless of the prevailing market price. In the long run, the railroad can respond to higher prices by performing major repairs and new car construction. Therefore, the longer the relevant time period, the greater the effect of a given change in price on the amount of capacity supplied. Over the very long run, when the railroad has complete flexibility, in the amount it will supply, the supply curve for freight car capacity will resemble the classic upward sloping supply curve. Our medium-run step function supply curve lies between these two extremes. Absent investment activity, the supply -curve "shifts left" over time as serviceable cars become unserviceable and cars of all types are scrapped and destroyed. While this "natural attrition" of supply puts a constant upward pressure on prices, we have seen in recent years that demand for a particular car type can contract far faster than the natural contraction rate for supply. The-more dynamic the transportation marketplace becomes in terms of its equipment needs, the more likely we are to see major discontinuities caused by market shifts away from existing capacity. Constructing a Demand Curve for Freight Car Capacity Determining the railroad's demand curve is a more difficult task both conceptually and practically. However, most railroads today have sufficient data available to permit a reasonable attempt at defining their internal demand curve for equipment, and the results of the effort can be enlightening. On the x-axis we measure the quantity of capacity required to support a given piece of business. On the y-axis, we estimate the net contribution that will be earned by supplying the required capacity. Both of these dimensions are explored in greater detail below, but first we need to spend time deciding how we will actually define "demand". What demand is out there? Clearly, the demand curve of interest in investment decisions should be forward-looking, and some thinking should be devoted to how intelligence gleaned from historical records should be adjusted for previously unserved demand and future adjustments in shipper requirements. That said, the remainder of this discussion focuses on the manipulation of historical traffic data to develop the demand curve. We should reiterate that we are looking at the railroad's internal • demand for cars, not shipper demand for transportation per se. We take as a given that the revenue received by the railroad for a particular movement is set through some combination of market and regulatory forces which are beyond the scope of this analysis. We further assume that the maximum price the railroad will pay for equipment is the full contribution before car costs derived from this revenue. The analysis presented here requires a data base that can identify shipper, origin, contribution, any specific car cost charges against contribution, and car days required (loaded and empty) to handle the traffic for individual traffic movements. We further make the heroic assumption that the assignment of empty days and empty transportation costs to individual data records is a reasonable representation of the true incremental empty costs generated by those movements. (How empty costs should be assigned to loaded moves is a major topic in itself.) Aggregating data around shipper/origin . points One simple approach to creating a demand curve would be to take all of the railroad's traffic records and sort them in descending order of contribution before car cost per day. While often' used, this approach is rarely effective. The least profitable traffic identified by such an approach tends to be thoroughly polluted with bad data, extraordinary operational foul-ups, and specific origin/destination pairs that are loss leaders necessary to hold other profitable business. This type of analysis stimulates extensive discussion of data quality problems and finger pointing over operational mistakes and pricing levels, but rarely decisive action over capacity investment We leap over many of these problems by shifting the focus from the profitability of individual carloads to the profitability of specific origin/shipper points. The rationale for this shift is as follows: • The shipper/origin point is the true logical unit of demand for a capacity invest-' ment decision. The railroad clearly has the ability to furnish or not furnish a car to a specific shipper/origin point It has ' virtually no operational control over how that car is subsequently used by the shipper. • Operational foul-ups are largely probability-driven phenomena. Their occurrences usually have little or nothing to do with the specific traffic movements with which they appear. Aggregating the data into larger units distributes these costs back into the overall traffic base as they should be. • Data errors exhibit similar characteristics to operational foul-ups. Costs and/or revenues have become detached from their appropriate home. Aggregating data according to some criteria of "likeness" improves the chances that the right costs and revenues will fall together in the analysis. Contribution per day before car costs Most railroads have automated costing systems for assigning to individual carloads an estimate of the variable costs associated with that movement These costs include train labor, fuel, roadway maintenance, locomotive costs, and equipment costs. The appropriate price for the analysis is contribution before car costs, the gross margin left after subtracting all variable costs except those associated with freight car ownership and maintenance from net revenue. This is the amount of incremental contribution which the company will earn if a freight car is provided. This contribution, will be compared to the incremental cost of obtaining a freight car, as indicated by the supply curve. Estimating capacity requirements Consistency with the supply analysis requires that contribution before car costs be translated into a contribution per unit time, usually per month or per day. This conversion process is not as simple as it sounds. In the denominator, we must be sure to include the incremental days needed for supporting empty movements as well as the loaded movement itself. Wherever possible, car day requirements which are movement driven (e.g. shipper detention) should be captured on the demand curve in conjunction with the appropriate traffic involved. Constructing the curve Construction of the demand curve begins at the left with the most profitable traffic and continues right in descending order of contribution per unit of capacity. While we can plot each shipper-origin point individually on this curve, for most purposes such detail is unnecessary. A simpler approach is to divide the traffic into larger blocks, say 10% increments, and plot the total capacity and average contribution of each increment. This simplifica• tion protects the mathematical integrity of the curve (for example, total contribution is the area under the curve under either the detailed or simplified approach) while making it easier to prepare and analyze. Observations Four examples of demand curves are shown in Figures C through F. These examples vividly illustrate how little an average contribution figure communicates about the profitability of the underlying traffic. Average contribution figures can mask the fact that 60% or more of the underlying traffic will not support the marginal cost of capacity. Demand and Supply Interaction The interaction of demand and supply determines the railroad's optimal car supply in a competitive market. Using historical data alone, there is a high risk that the first attempt to overlay supply and demand will be both surprising and disappointing. The two curves cross at or below the size of the existing fleet! (See Figure G.) In retrospect, this is not a surprising result at all. If the only demand reflected in the analysis is what was actually carried historically, and the supply curve reflects the current fleet size, the only feasible points of intersection are to the left of the current fleet size. Meaningful results are achieved only if we project future demand and supply. Demand must be adjusted up or down for anticipated changes in price and volume. Supply must be adjusted, almost always down, for anticipated losses in capacity over time, absent additions to the fleet With these adjustments, the analysis begins to help sharpen our understanding. (See Figure H.) What is the anticipated capacity shortfall? What is the contribution of the traffic at the margin? Is investment in capacity warranted, or should the marginal traffic be turned away? What is in the marginal traffic category? Working from the demand and supply curves and then- underlying data, we can intelligently review the options for bringing future capacity into line with future demand, and the probable economic consequences of each option. Dealing with the Un-simplifying Aspects of the Real-World Case The supply and demand curves created above ignore many of the more difficult issues associated with railroad interactions with one another and the broader •markets for car supply and transportation services. We identify below some of the major complications that arise in adjusting these curves to address real-world issues. External sources of car supply capacity Today there are a dazzling .array of options for obtaining cars on the secondary market, including short-term leases, medium-term leases, shareduse agreements, and car-hire leases. The general approach we have taken with these options is to attempt to define each one in sufficient detail to reasonably estimate the level of available capacity and effective cost of that capacity. Straight term Demand Curve Car Type A Dollars «° -g," per car perday *«-fi • fairly concentrated market $40 -f • large variability in contribution per day $35 -f • over half of this market has below average contribution $30 -t $25 -! $20 ~ ' ' Average Contribution = $17 $15 -f $10 $5- *°4—i -3.0i i 6.0i i 9.0i i 12.0i r 15.0i i 18.01 - 1 21.0i —i—r 0.0 24.0 Cumulative Equivalent Cars (OOOs) Figure C Demand Curve Car Type B Dollars &° f - per car per day $4o-i • highly concentrated market with few shipper/origins s35 ~\: • most of this market has above average contribution $30 -j $25 -| ^ / Average Contribution = $21 $20 -t $15 $10 -4 i $5 H 0.0 i i 3.0 i i 6.0 i i 9.0 i i 12.0 i i 15.0 n Cumulative Equivalent Cars (OOOs) Figure D i 18.0 i i 21.0 n i 24.0 Demand Curve Car Type C Dollars per car *so - •; \ • demand scattered across a number of shipper origins • a few highly profitable customers , $35 - \ while some traffic has profitability substantially below average : , Average Contribution = $22 •. •• \ v. •. s •••. : : ! s :: . rrn-. > x • ; p •> • • *"" >. ,. "" . :' :: : :: : : ! i 1 3.0 1 0.0 :: '•• :: % \ i 6.0 9,0 12.0 15.0 1E.O 21.0- 24.0 Cumulative Equivalent Cars (OOOs) . . Figure E Demand Curve Car Type D Dollars per car per day »°' • $40- few shipper/origins » a small number of profitable shipper origins distort average f f $35- contribution % : . ' f $30- •Average Contribution = $27 j $25- '• "• •. ; $20- '. f $15,_ % £10- •>•. - , .., , , o ^ Cumulative Equivalent Cars (OOOs) Figure F r. % . , ^ i 1 1 1 1 I 1 Demand & Supply Curve Car Type X Dollars per car per day Supply - Total Serviceable Fleet Demanc m w. P__Ir Heaviest Repair - Medium Repair - Light Repair Basic Maintenance 0.0 3.0 6.0 9.0 12.0 15.0 18.0 21.0 24.0 27.0 30.0 33.0 36.0 39.0 Cumulative Equivalent Cars (OOOs) Figure G Demand & Supply Based on Historical Data Demand & Supply Curve Car Type X Dollars * per car per day *« • — Heaviest Repair — Medium Repair - Ugh! Repair 0.0 3.0 6.0 9.0 12.0 1S.O 10.0 21.0 24.0 27.0 30.0 Cumulative Equivalent Cars (OOOs) Figure H Demand & Supply Based on Forecasted Data 10 33.0 36.0 39.0 Basic Maintenance leases under which cars function as if they were owned by the lessee railroad are relatively easy to incorporate. Reloading foreign equipment is probably the most difficult, and is discussed in detail below. As a practical matter, many railroad data systems fail to make this linkage. Absent significant data enhancement, it is necessary to treat off-line time, as a stand alone demand for capacity. If nothing else, this treatment does allow the analyst to see how off-line earnings compare with the marginal cost of the capacity generating those earnings. The general objective is to insert these capacity options on the supply curve at the appropriate point given by their cost. Hence, one source of leased equipment may fall between two categories of repair options. In this case and in the absence of other factors, the leased cars are a more attractive capacity option than the second set of cars to be repaired. Strategies for Improving Management Decision Making Despite these limitations, the demand and supply curves can provide valuable insight to the asset manager. Non-system equipment A major source of equipment on most railroads is foreign equipment that can be reloaded before it is returned to the owner% As a practical matter, this is a difficult type of capacity to incorporate into the model. Using the supply curve By definition, the supply curve indicates the amount of capacity that the railroad should be' willing to supply at any given price. As such, the data also reflect the relative cost of different options for expanding, or contracting the fleet Several fundamental questions may then be addressed. In estimating the volume of available foreign capacity we need to look at how much effective, system-car equivalent capacity can be produced, a number considerably differentfrom the actual number of car days on-line. Adjustments must be made to refiectdiscretionary car days used to meet on-line needs and involuntary .time created by inbound loads. Part of the involuntary time is the unavoidable days required to dispose of the car if it is not used for discretionary purposes. This source of capacity may also be more or less efficient in terms of general levels of utilization and empty transportation costs. Simply put, it is not safe to assume that a foreign car with a nominal cost of $15 per day has the same cost per unit output as a system car that prices out to $15 per day. For example, has the asset manager chosen the most economical alternative for changing the fleet size? In some instances, management may choose other than the least cost option for intangible or "strategic" reasons. The supply curve will then indicate how much management is paying for these "strategic" benefits. Second, how does the current marketrate for freight car capacity (in the form of short and long term lease rates) compare to the marginal cost of the existing fleet size? InFigurel, the prevailing marketrate (M) is higher than the railroad's marginal cost of capacity (C). Therefore, assuming that external demand exists that will support rate M, the railroad shouldrepaircarsup topointB and lease all capacity not needed internally to the external market Off-line time Most railroads see their equipment go off-line in support of forwarded movements. The investment required to support this movement needs to be incorporated in our analysis. A third area of constructive inquiry is how fast the railroad can respond to any significant increase in demand. When the supply of potential light repairs and short term lease options is high, management can take a "wait and see" approach to projected demand. However, if short lead time options are limited, the railroad may need to start placing orders Ideally, off-line time associated with individual movements shouldbe tied back to those movements in the traffic data base. Hence, the demand for capacity created by a specific movement and the contribution produced includes the off-line portion of that movement 11 Enhanced Supply Curve Car Type B Dollars per car per day New Cars Heaviest Repair Medium Repair Short-term lease Light Repair < |- Basic Maintenance i 0.0 i r I T i n i r i i i i i t 11 j i I i i 3.0 ' 6.0 9.0 12.0 1S.O 18.0 21.0 A B 27.0 30.0 I I I i i 310 36.0 3S.O Cumulative Equivalent Cars (OOOs) Figure I and/or performing heavy repairs before they are actually needed. than marginal cost. The railroad in Figure J should therefore expand .capacity to point E using light repairs and short-term leases, where the contribution of the marginal traffic is greater than or equal to the cost of the additional capacity. Using the demand curve With a few enhancements, the demand curve is a powerful tool for stimulating management discussion. First, add a vertical line (C-D)' for the forecasted level of total capacity for the relevant time period. Barring additions to the fleet, the railroad must forego part of its projected traffic base. Second, superimpose horizontal lines representing the cost of various supply options. We now . have a powerful one-page summary of the economics surrounding a capacity investment decision. Replacing the supply curve with the horizontal lines masks some of the details around the supply sideissues.butmakes the chartfar more readable and comprehendible to the layman. We can also see the relative importance of "car ownership" and "track ownership" to the railroad's economics. We will assume that the short term lease rate Erie represents the' current market, rate for equipment For most analyses, this line represents the economic cost of the equipment used by the railroad. The contribution below this line less the operating costs of car ownership is the net contribution created by car ownership. The contribution above this line is created by other factors, primarily by the ownership of the railroad's track network. We assert that a major problem in the financial analysis of railroad capacity investments is the failure to separate, and not double count, these two pools of contribution. The downward sloping demand curve presumes that-the least profitable traffic will be the first to be turned away. (Again, management may have "strategic" reasons for sacrificing other than the least profitable markets.) Basic economics tells us thatit will be profitable formanagement to expand capacity whenever marginal revenue is greater Economic analysis suggests that car ownership is not required to capture the contribution above the line, nor is railroad ownership required to capture the contribution below the line. A freight car investment analysis thatiricludes contribution above 12 Enhanced Demand Curve Car Type x . Dollars per oar per day D ' *« ~| ... D eman( $40 - forecasted serviceable fleet M $35 - New Cars $30 — X $25 — ' $20 — ?%%%%{ J^ Off-line | ] On-line [ .. $15 — % $10 — s -; - „-; i v $5 — . $ ° I 0 I 5,000 ' „, 1 10,000 ' ( ( '(///////A < .' -™ j- . - Medium Bepate . n ,' ',',',' '.'.'.'.' y//MW 'jjjfa Short-term leases • w%%. I 15,000 Heavy Repairs 1 C . HI E . UghJ BeRaV1* 6asfc Maintenance 1 30,000 Cumulative Equivalent Cars Figure J the line as a return on' that investment grossly overestimates the true economic return to the railroad from car ownership (as contrasted with the options of short term leasing or shipper supplied equipment). Similarly, an analysis of network economics thatfails to deduct the full market value of equipment from contribution figures can substantially overstate the economic earnings of the network. References Allen, W. Bruce.The Demand for Freight Transportation: A Micro Approach". Transportation Research. Volume 11, pp. 914,1977. Borts, G. 'Production Relations in the Railway Industry". Econometrica, Volume 2.0, pp. 7179,1952. Braeutigam, R.R., Daughety, AP., and Tumquist, MA. "The Estimation of a Hybrid CostEunctionforaRailroadFirm".J?evfe>v ofEconomics and Statistics, Volume 64, pp. 394-404,1982. Conclusion Despite the complexities of the freight car market, the basic microeconomic concepts of demand and supply can be applied effectively to railroad freight car capacity investment decisions. The results of such an analysis are invariably informative, often surprising, and should be considered essential to making informed investment decisions. .Daughety, AP. and Inaba, F.S. "Estimation of Service Differentiated Transport Demand Functions". Transportation Research Record, Volume 688, pp. 23-30,1978. Daughety, A.F. "Freight Transport Demand Revisited: A Microeconomic View of Multimodal, Multicharacterisric ServiceUncertainty and the Demand for Freight Transportation". 13 Transportation Research, Volume 13B, pp. 281-288,1979. Acknowledgements The authors would also like to acknowledge the following individuals whose own work contributed directly or indirectly to the concepts included in this paper: Joel Szabat, Ralph Tang, F. William Barnett, Lee Broesch, R.M. Hebbeler, David Levy, and R.A. Sewell. Daughety, AJP., Tumquist, M.A. and Griesbach, S.L. "Estimating Origin-Destination Specific Railroad Marginal Operating Cost Functions". TransportationResearch, Volume 17A, No. 6, pp. 451-462,1983. Daughety, A. F. "Transportation Research on Pricing'and Regulation: Overview and Suggestions for Future Research". Transportation Research, Volume 19A, No. 5/6, pp. 471-487, 1985. Felton, John Richard. The Economics of Freight Car Supply. University of Nebraska Press, Lincoln, Nebraska, 1978. Harker, Patrick T. "Research Directions in Transportation Regulation and Pricing". Transportation Research, Volume 19A, No. 5/6, pp. 489-491,1985. Keeler, TJ3. "Railroad Costs, Returns to Scale, and Excess Capacity". Review of Economics and Statistics, Volume 56, pp. 201-208,1974. Martland, Carl D. "Overcoming Fundamental Problems in the Car Hire/Car Service System". Proceedings - Twenty-third Annual Meeting, Transportation Research Forum, Volume 23, No. 1, pp. 549-560, Oxford, Indiana, 1982. Wilson, George ~W.Economic Analysis of Intercity ' Freight Transportation. Indiana University Press, Bloomington, Indiana, 1980. Winston, C. "The Demand for Freight Transporta' tion: Models with Applications". Transportation Research, Volume 17A, pp.419-427, 1983. Winston, C. "Conceptual Developments in the Economies of Transportation: An Interpretive Survey". Journal of Economic Literature, Volume 23(1), pp. 57-94,1985. . 14