This paper was written while I was a Ph.D. student at the University of Pennsylvania, Philadelphia, PA. Written for Systems Engineering 566, "Natural Resource Systems".
Full text of article. This was written in Nov of 94 so it contains some references to current events of that time and may be otherwise slightly out of date. All results are preliminary so take them with a grain of salt and please contact me if you want to quote them somewhere. Unfortunately, I couldn't include the references and figures at this time but I hope to do so later.
Full title: Reducing Energy Consumption by Redistributing Production:
a Test Case using the Raising and Shipment of Fruits & Vegetables
ABSTRACT
This study tests the hypothesis of potential energy savings due to redistribution
of production so as to reduce transport distances, for the case of fruit
and vegetable distribution in the United States. Part I outlines basic
features of the agricult ural and transportation sectors in the US which
pertain to the calculations made. In part II, the report estimates energy
consumption for growing produce in different states and for transportation
per unit grown and shipped. Two comparisons, one for a c omprehensive transfer
of production to minor producing states in the Eastern U.S., and one for
the marginal energy benefit of moving one ton of production to each of
three minor producing states, both upheld the hypothesis, although weakly
in the case of conservative assumptions for energy efficiency of produce
agriculture in the east. With these conservative assumptions, efficiency
improvements of 5% to 30%, and an overall improvement of 10% aggregated
across 7 major items were deemed possible, while wi th more liberal assumptions
about efficiency, the ranges were 30 to 40% and 32% respectively. Caveats
at the end of the report outline simplifications to the study and ways
that its accuracy could be improved, and the conclusions discuss possible
opportu nities to make such a transition in the political, economic, and
natural resource context.
This paper is divided into two parts. Part I is an explanation of my hypothesis and an overview of the agricultural and transportation setting in which I will test it. Because of the central nature of the hypothesis and the "levels of energy consumpti on" to the discussion, the first subsection has been kept almost verbatim from the interim report which I submitted on October 21. The rest of Part 1 is a revised and expanded version of the interim report. Part II develops a basic model of energy use i n the raising and transporting of major fruit and vegetable commodities in the U.S., and an estimate of savings from redistributing this energy use in order to test the hypothesis.
Part I: Explanation of hypothesis and overview of trends
1. Explanation of hypothesis
Statement: "By redistributing crops and growing regions so that fruits and vegetables are grown closer to where they will be consumed, the United States can substantially reduce the overall energy consumption in the life cycle of these activities, primari ly by reducing the energy consumption in transportation."
General Description: Without taking up the specific case of agriculture, one can describe the general life cycle of many products consumed by individuals as a progression of production (procurement of raw materials, processing, etc), transportation (movem ent of required raw materials and finished products, etc.), and consumption (storage in local warehouses, marketing, storage in domestic setting, etc). All energy consumption in this cycle can be designated to one of the three steps, and thereby one can calculate an overall total of gross energy consumed in the cycle per year or else, dividing by the quantity produced, an energy per unit. Since energy in transportation is generally a function of distance moved, if the lifecycle can be reorganized so tha t it happens over a smaller geographic area, the savings in transportation may be larger enough to offset any potential increase in consumption in the production or consumption phase, resulting in a net overall increase in life cycle energy efficiency. I n certain cases, the "regionalizing" of the transportation input might lead to efficiency improvements in the production phase as well.
Transportation context for this approach to reducing energy consumption: >From an air quality and global warming perspective, the United States may benefit from a reduction in energy consumed by freight transportation. The redistribution of production to reduce transportation that I am studying is one possible approach to this goal. To put this hypothesis in context, I have compiled five general descriptions of energy reduction policies that spans a range of specific solutions to this challenge (alth ough some specific solutions may lie outside the bounds of these descriptions). They are listed below, in order of broadening scope and increasing time horizon required for implementation:
1. Improving the technology: freight transportation networks would continue to carry commodities with the same modal shares as they do now, but more efficient vehicles would be substituted for the current technology. This approach would have the greatest effect with regard to the two most energy intensive modes, airfreight and truck.
2. Pack vehicles more efficiently: currently, a large fraction of load capacity in the average freight vehicle goes unused. In this scenario, the transportation network would remain fixed, but by increasing the load per vehicle, there would be fewer vehic le trips and therefore energy consumption would be reduced.
3. Shift from less- to more-efficient modes: the national freight transportation infrastructure would continue to carry the same volume of commodities, but some freight currently carried by truck and air freight would be carried by rail, barge, and possib ly pipeline instead. This approach essentially would require reversing the current trend towards increased use of truck and air, and therefore would require a sustained national commitment. Nevertheless, the increased use of container-on-flatcar (COFC) service is perhaps a first step in this direction, since COFC provides a level-of-service closer to that of trucks (e.g. regular departures, direct connections, guaranteed timeliness, smooth ride) than any service previously provided by the railroads.
4. Reorganize production so that trip lengths are shorter (the option being studied here): where possible, produce goods closer to where they will be consumed, so that overall travel distances are shorter. This approach runs against the current trend tow ard increased trip distances, especially due to the globalization of production.
5. Decrease overall demand for energy intensive goods, especially those that are carried long distances, by fulfilling the same human need with a substitute which is less energy intensive. As an example, one could substitute a vegetarian dish for a meat dish (perhaps some fish product that was caught in the Aleutian Islands and flown into the continental United States); and if the substitute dish is in some way equally tasty, then the need for a high quality meal can be met while reducing overall energy consumption. A similar option (We might call it 5A) would be the outright elimination of the demand with no replacement: in the example, the consumer would reduce their meal from fish, rice, and vegetables to rice and vegetables only. Of the two approa ches, 5 may be more agreeable to the consumer than 5A since they would not make as great a sacrifice. Also, although either approach might be the farthest reaching approach to decreasing energy consumption among the list 1 through 5A, it is probably also the most difficult to pursue. Not only does it require changes at the educational and societal level, which are inherently long-term measures, but in either case the replacement of energy-intensive and profitable practices with less-intensive practices contradicts one of the fundamental underpinning of our economic system, namely the pursuit of increased profit through increased economic activity.
Some overlap exists between the five levels of energy reduction. For instance, if a consumer substitutes a locally produced food item for one that previously came from California, the two products may not be exactly identical, and if the taste is som ewhat different, the consumer is pursuing option 4 and option 5 simultaneously. Still, one can in general see from this list that option 4, the redistribution of production, is a relatively long-range option and one that is more difficult to pursue than options 1-3, although perhaps less difficult than number 5 since the consumer need not always perceive the change of origin. Literature searches up to this time have uncovered relatively little research related to this option, especially compared with th e large volumes of literature on improving the technology. (See fig.1 for a visual representation)
2. Overview of transportation and its impact on the environment
What motivation is there for looking at a more complicated and longer time-frame approach such as redistributing production if an extensive research initiative is already underway to look at the most direct solutions such as developing a more efficien t truck or plane? To answer this question, we need to examine recent trends in energy consumption and also in air quality and greenhouse gases.
Fig.2 shows trends in overall energy consumption in the United States up to the year 1989, as measured in million tons of oil equivalent. The three curves together represent all possible categories of energy consumption except for non-combustive uses of fuels, such as chemical feedstocks; these uses, however, represent less than 1% of the total. The category of "other" includes household, commerce, and agriculture. ______The graph shows how industrial (Note: in the interim report I erroneously stat ed that this sector includes electric power generation, but in fact electric power is distributed among the three main sectors as used) energy use dipped during the 1982 recession while transportation use rose continuously over the entire time period, so that since the recession transportation has been the leading consumer of electricity. The most recent estimation, which is not included on the graph, shows that in 1991 transportation consumed 479 MTOE, industry consumed 411.5 MTOE, and the "other" categ ory consumed 416.68 MTOE, confirming that transportation is still the leading user [17, p.177].
The majority of this transportation is accounted for by automobile travel and not by freight transportation. However, in recent years, automobile efficiency has been improving rapidly, thanks in part to the corporate average fuel economy or CAFE stand ards. With no similar standards in place for commercial transportation, efficiency improvements have not been so significant in trucks and railroads, although commercial aviation has been forced to make improvements by the rising cost of jet fuel. As an example, the average fuel efficiency of tractor trailers only increased 16% between 1970 and 1990, from 4.8 MPG to 5.5 MPG; during the same period the efficiency for cars approximately doubled [15]. In any case, because of the different rates of efficie ncy improvements, fuel consumption in commercial transportation is increasing faster than in private, and it may surpass private transport by the year 2005 (See Figure). Although the figure for commercial transportation includes passengers transported by public means, under certain scenarios, freight could surpass passenger as the leading user of transportation energy by the year 2030 [9, p.304].
As the dominant energy user among all modes of freight transportation, truck makes a measurable contribution to the overall quantity of anthropogenic emissions in the United States. Again this contribution is currently less than from light-duty vehicl es: for the two most significant pollutants, NOx and PM, trucks emit 10% and 5% of the national totals, respectively, compared to 21% and 11% for light-duty vehicles [OECD, Choosing an alternative Transportation Fuel, Table 4]. Nevertheless, because so m any cities are in non-compliance for ozone levels and since NOx is a precursor for ozone, it seems justifiable to expand the campaign to reduce NOx from the cars-only focus pursued at present to include reducing emissions from trucking. As long as cleane r automotive technology continues to infiltrate the market while truck technology remains at a standstill as is happening now, the relative importance of focusing on truck will only continue to grow.
Along with concern about emissions of airborne compounds with direct health effects, the U.S. also must consider the emissions of greenhouse gases, or those compounds whose accumulation in the atmosphere may lead to a rise in the average temperature o f the atmosphere. According to the World Resources Institute, recent research indicates that "by far the largest contributor (about 50%) is energy consumption, mostly from fossil fuels [9, p.2]." Equally daunting are the estimates of how much we would n eed to reduce the emissions of greenhouse gases in order to stabilize their concentrations in the atmosphere. The process is as yet not well understood, and estimates vary widely, but one recent study indicated that while the United States emits 4900 kg of CO2 per capita per annum and Sweden emits 2200 kg, the level that would be required for stability is 300 kg [source: European Conference of Ministers of Transport, Transport Policy and Global Warming]. While such estimates might imply that urgent acti on is necessary immediately, not everyone agrees that global warming should be combatted: some sources suggest that it would be more practical and cost effective to prepare for the consequences of global warming, for example by making wildlife refuges rea dy for the mass movement of wildlife habitats, or by anticipating the development of some new agricultural regions and the abandonment of others [3, p.242].
At these point I do not believe that the scientific community can definitively choose between these two alternative courses of action. However, in this study I will assume that the "prevent" strategy is preferable to the "cope" strategy, so as to avo id a potential catastrophe. This assumption places a premium on reducing energy consumption on every level of human activity, including freight transportation. The assumption also favors the exploration of the broadest possible approach to efficiency im provements, especially approaches #4 and 5, because the magnitude of reduction of fossil fuel combustion required is so large (over 90% for the United States from the previously quoted study). In this sense, the reduction of carbon dioxide emissions may be fundamentally different from the control of NOx, PM, and other toxic pollutants: for example, while NOx control is largely a function of the technology used, and in fact below a certain temperature such as with fuel cells combustion processes produce n o NOx at all, the generation of CO2 from fossil fuels is a function of economic activity, and unless the economic activity is reorganized, adjusting the technology alone (as in approach #1) will have limited positive impact.
3. Brief overview of agriculture in the US today:
Along with the introductory remarks about transportation, some comments about agriculture will help to provide a basis for this study. Agriculture's contribution to overall US energy consumption appears to be slight. According to the OECD, agricultu re consumed 13.37 MTOE out of a national total of 1366.55 MTOE, or roughly 1%. An additional 9.86 MTOE was expended on food and tobacco processing. While this total does not include aspects such as the manufacturing of agricultural equipment or the hous ehold energy requirements of those engaged in agriculture, it seems fair to assume that since neither the volume of equipment or the population in agriculture is very large, these amounts of energy consumption would not be very large either.
Among activities which might consume energy in agriculture, irrigation is a possible candidate for scrutiny. One study indicated that 49 million acres were irrigated in 1982 out of 326 million total planted, or 15 %; and that 84% of the irrigated ac res were in 17 arid & semi-arid western states [27, p.1]. However, this statistic does not tell the whole story. The great majority of acreage is cultivated for producing grains, while all of the domestic fruit and vegetable market can be supplied with only a small fraction of the total acreage ññ even a leading producer of fruits and vegetables such as California uses only 16% of its acreage for these items [24]. The production of many fruits and vegetables is concentrated in the southwest, and in the se cases the farmers are 100% dependent on either aquifer or river water sources, since this land has been reclaimed from the desert. Therefore, the prominence of irrigation in energy consumption for fruits and vegetables is likely higher than the 15% fi gure suggests.
Farmers dependent on irrigation are faced with a difficult dilemma. If they unilaterally reduce the amount of water they are taking from the groundwater source or the irrigation channel, their neighbor may take even more, so that collectively the far mers cannot reduce their utilization rate of the resource ññ in fact, they are under pressure to maximize their output in the short run lest the surrounding farmers put them out of business, so that the water resource is used up that much faster [22, p.2] . Utilities, in turn are often under pressure from agriculture to supply as much water as possible to the farmers while at the same time reclaiming water behind dams to generate electricity for pumping out of aquifers [1, p.100]. Because farmers in the Northeast and Midwest do not face the same difficulties with irrigation as in the southwest, they may be able to produce fruits and vegetables with a lower rate of energy consumption.
One other comparison of major and small-scale agricultural producers is in order for this study. Basic economics suggests that positive returns to scale in agriculture should exist up to some point. Hallam investigated this point for two groups of fa rms in California, one larger than 700 acres and one smaller, and found no appreciable difference in the cost of production [8, p.205]. Therefore if one were to compare very large farms in California with small farms in the Northeast, for example, one m ight expect the California farms to have a cost and also energy efficiency advantage due to their size. However, if the farms in the northeast were medium-size or larger, this advantage might disappear.
In general, this study compares a high-transport and low-transport on two different levels, one at the comprehensive level (subsection 2), and one at the marginal level (subsection 3). In order to compare the existing system with the proposed alterna tive, it was necessary to create a framework for comparison. Details of this framework appear in the Appendix.
1.. Overall assessment for potential for energy savings: national evaluation of comprehensive change
The word "overall" is used here to describe measuring the total potential for energy savings by considering the most abundantly produced fruits and vegetables on a national level, singling out those which are concentrated in a very few states, and th en reassigning them to more localized regions where some production capacity already exists. For example, in 1991 California was the leading producer of peaches, but New Jersey also grew and shipped some number of peaches, so that if one could increase p each production in the latter then you could in turn reduce production in the former. It is important to note that this approach seeks to find an upper ceiling for potential energy savings: I am ignoring the question of how to find space for such a larg e-scale expansion of fruit and vegetable growing in states which today are minor producers until subsections 5 and 6.
The model in this section consists of three major regions in the continental United States: two major producing regions, the south (centered around growing in Florida) and the west (centered around growing in california), and one region which is in the base case a net consumer, the northeast-midwest. The populations of the three regions according to the 1990 census are 32.3 million, 79.1 million, and 137 million, respectively [4, p.364]. Since the majority of the production happens in the first two r egions but the majority of the population lives in the third region [See appendices], a potential exists for very long freight carrying distances. To continue the above example, a shipment of peaches moving from Los Angeles to Trenton, NJ, would travel 2 714 miles by truck [20].
Not all fruits and vegetables move by truck; the national rail network serves most parts of the country, and boat and for certain shipments airfreight are also available for many shipments. On examining the level of shipments by state for all the vari ous modes in 1991, truck was found to account for 502276 standard units of shipment out of a total of 548,750 units, or 91.5% of all shipments [26; also see table below]. This market dominance occurs because truck has the best combination of flexibility with reasonable cost. Boat service is limited to a few waterways and ports, and for example cannot carry products rapidly across the continent. Air freight incurs a high cost and therefore must be limited to high value commodities, such as shipping priz e fruits from Washington to Japan. Even rail, which is cost competitive and has a national network, cannot compete well with truck because of service problems. For example, if a rail shipment of a perishable product is held up in a large rail yard betwe en its trajectory endpoints, much of the shipment may spoil; this is especially troublesome because if the railcar must be refrigerated, supplying power to the car while it stands in the yard poses a complicated problem. With truck, the shipment generall y moves continually from origin to destination and stays connected to the truck tractor so that power supply is not a problem. Because of the market dominance that truck enjoys thanks to this technological advantage, this study will ignore shipments made with other modes and focus only on the effects of shifting long-distance truck shipments to regional production.
Table 1. energy consumption by mode, using aggregate averages [15]
mode total loading avge length intensity est.tot.En.used
_____[1000 cwt]______[miles]____[BTU/tm]___ [BTU]
rail / 25093 / 726 / 411 / 3.74 e+11
piggyback* / 18088 / 726 / 411 / 2.70 e+11
truck / 502276 / ca.400 / 3357 / 3.37 e+13
air / 1758 / -- / -- / --
boat / 1535 / 743 / 403 / 2.30 e+10
*separate data for rail piggyback service not available, so same data used as for rail. Also, data not available for airfreight, but considered negligible due to low total loading.
Focusing on truck shipment data, I selected the most commonly shipped products, the leading fruit and vegetable production states, and the portion all shipments accounted for by these states as a preliminary indication of how much potential there might be to redistribute production and so reduce energy consumption (See Appendix, table 1). Among 50 or 60 possible products I chose the 13 most commonly shipped items and also 20 other items which had at least 1000 cwt of truck shipment per annum. Among t he group of 13, the four leading originating states were California, Florida, Washington, and Arizona. For warm-weather crops and apples, these states had captured the bulk of the market, accounting for a range of 51% to 100% of domestice production. Fo r cool weather produce (i.e. potatoes, onions) and watermelons, these states were less dominant, accounting for only 16% to 48% of the market. This dichotomy between "centralized" and "decentralized" produce suggested that the hypothesis test be narrowed to those items for which shipping distances were likely to longest, e.g. the warm-weather crops and apples. By contrast, the states of Idaho, Colorado, Wisconsin, Michigan, Washington, and California are all major producers of potatoes, and 17 other sta tes have measurable amounts of potato shipment by truck (the leading shipper, Idaho, has only 20.7% of the total market), so that the model is already largely decentralized and no obvious model to localize production further presented itself.
Before calculating levels of energy consumption, I set up three simple models of "current" and "redistributed" production and distribution systems, one for apples, one for warm weather crops, and one for tomatoes. In each of the following scenarios, the city pairs represent a movements within a region and to the other region. Since Washington dominates the apple market, the current apple model consists of a node in Spokane connected to a node in Denver, Washington DC, and Atlanta. The redistributed mode has a Spokane-Denver link and an unconnected Lansing-Louisville, Lansing-Atlanta network, so that east coast apple production would substitute for Washington apples in the market in the South and NorthMid regions. By similar reasoning, the warm wea ther crop model converts from an LA-Denver/LA-Lansing/LA-Atlanta model to an LA-Denver/Orlando-Lansing/Orlando-Atlanta model for broccoli and strawberries, and an LA-denver/Lansing-Atlanta/Lansing-Lousville model for carrots, celery, grapes, and lettuce ( I judged these crops to be too sensitive to heat to grow in Florida. The tomato redistribution required a special model since tomatoes currently have two domestic centers depending on the time of year: California in the summer and Florida in the other th ree seasons ññ this model can be understood by observing Table 3 in the appendix.
The study next required two types of energy estimates, one for transportation and one for agricultural production. For transportation I used the estimate of average energy consumption per ton-mile in 1990 of 3357 BTU (actual energy consumption may be higher since fruit and vegetable trucks often have an additional refrigeration load.)[15]. Unfortunately, for production the literature search no estimates of energy per ton of production by specific crop and by state, although aggregate energy for each state did exist [Source 28: David Torgerson/USDA, Energy and US Agriculture, Lipp. S494.5.e5 U54]. To develop a life cycle model for agricultural processes for each crop and the energy consumption of each activity, so as to estimate an energy intensity (i.e. BTU/ton of broccoli grown) would have been extremely time consuming, and subject to great inaccuracy. I therefore assigned some portion of the total energy expenditure by state to fruit and vegetable production and then divided this number into the total shipment of fruits and vegetables from that state so as to derive an aggregate estimate of energy per general unit of fruit or vegetable produced (See explanation in appendix). The average value for the 4 leading states was 102 million BTU per 100 0 cwt of produce; for four representative NorthMid states (Arkansas, Michigan, New Jersey, and Virginia) the average was 350 million BTU.
Based on the results of these energy calculations, a tradeoff emerges between higher energy consumption in the smaller producing states versus energy expenditure on long shipping distances in the leading states, which happen to be in the geographic co rners of the U.S. landmass. In fact, one can state the tradeoff for a unit of 1000 cwt of production in the form of an equation:
loss incurred by producing in the NorthMid = gain by reducing shipping distance by 1490 miles
Since moving production destined for NorthMid markets will typically reduce shipping distances by more than 1490 miles (i.e LA-Lansing = 2233 miles), the calculations for most of the items supported the hypothesis. Before making a final assesment, I redu ced the total energy savings by 50% for carrots, celery, grapes, and lettuce, since I reasoned that even with some storage these crops could only be produced in the NorthMid for 6 months of the year. The typical energy saved on the production-transportat ion process was then about 4.5% as compared to the current model for these 4 items, a modest improvement. Energy savings for apples was about 6.4%, but energy savings for broccoli and strawberries was about 40% since they could be grown in Florida where the estimated energy per 1000 cwt was 109.6 BTU. The overall energy reduction for redistributing all 7 of these items was 10.9%, so that the hypothesis of energy savings was weakly upheld.
The estimate of 350 mBTU/Kcwt in the previous case reflects the possible inefficiency of production in small fruit and vegetable producing states such as New Jersey or Arkansas. However, if the average energy consumption of a larger producer such as Michigan is used (49.3 mBTU/Kcwt) then the savings are more impressive. Michigan is notably more efficient in some aspects of the agricultural operation, for example in irrigation where it spends about 1% of its energy budget versus 18% for California [2 8]; it therefore has a lower average energy consumption per unit output than the average of the four leading states. In this situation, the transportation-production tradeoff dissappears since moving production saves both in production and in transportat ion to market in the NorthMid region. Table 4 looks at the savings: they all fall in the 30-40% range even taking into account the assumed 6 month growing season, and the overall energy savings from shifting all seven items is around 32%. If all produci ng states in the region could achieve this level of efficiency, the savings might be quite substantial.
2. Marginal assessment for potential energy savings of shifting one ton of nationally shipped tomatoes to in-state production.
The previous section address only how much energy could be saved by a wholesale transformation, and ignores the question of whether land is available, let alone whether the economics would be feasible. In this section I will look at the specific cas e of tomatoes and study whether shifting a single ton of production from California or Florida to one of the three NorthMid states (AR, MI, NJ) for which some tomato production currently occurs would uphold the hypothesis. Making such a marginal shift do es not require major changes in infrastructure; it can be as simple as a single current grower deciding to plant tomatoes in a currently fallow field, and then his or her successfully displacing one ton of California tomatoes from the instate market by ec onomic competition. Historically, these three states have all lost market to the big producers in the past [26, 1968 edition] and New Jersey in particular has reduced the total amount of land in agriculture from 1 million acres in the 1940's to 500,000 a cres today [quote from presentation by Prof. Donn Derr, 10th Int'l Conference on Solid Waste Management]. If these states at one time produced more tomatoes than they do now, one can reason that they in the future might again increase their output.
In each of these three cases, the in-state production and consumption of the product implies an estimated average shipping distance of only 50 to 100 miles, so that energy consumption in shipping is very low (Appendix, table 5). Michigan also has a p roduction efficiency advantage over California and Florida, so that the marginal effect on energy is a 78%-85% improvement. Arkansas is at a disadvantage in terms of energy efficiency in production, but because the shipping distance is so much shorter (c ompare to 1791 miles LA-Little Rock or 946 mi Orlando-Little Rock) the marginal change is a net gain of 30-50% in energy efficiency over the current case. Only in the case of New Jersey is the change a net loss in energy efficiency because of the very hi gh estimate of energy per unit of agricultural production.
3. Evaluation of hypothesis
At the beginning of this paper we set out to explore the hypothesis of whether the U.S. could reduce energy consumption by redistributing the production of fruits and vegetables so that they would be grown closer to where they are consumed. To reiter ate the results from this section, we found that in all cases except for the marginal effect of moving production of tomatoes to New Jersey, the hypothesis was upheld. This hypothesis test did not consider an economic evaluation of such a change; technol ogical and economic issues are considered in later subsections.
4. Caveats:
The original intent of this project was to take into account many of the ways in which the states which today are large producers in the south and west and the states in the northeast and midwest are different: population density, type of infrastructu re, amount of congestion on that infrastructure, and nature of agriculture in the respective states. One lesson from the project is in the difficulty of pursuing such a broad scope, for in the end, there were only time and resources available to pursue i nformation on amounts of production, energy intensities, and distances. The following list of caveats is therefore intended in part as a list of areas in which the study could be made more accurate and reliable by bringing in more and better information.
Caveats related to the data and methodology:
1. The available truck shipments may differ from the actual level of production in the states: as stated earlier, on average 8% of the total shipment not moved by truck has been excluded from the study. Some of this freight moves by rail and boat which a re much less energy-intensive than truck, so that it is more difficult to improve energy efficiency by decreasing shipping distance. More importantly, the USDA tabulates this data as "available truck", suggesting that an unknown portion of the cargo area of the trailer may travel empty. One way to establish a more accurate quantity in shipment would be to compare shipped values [26] with harvested values [24, 25]. Lastly, the models have assumed certain patterns in the movement of trucks; actual data o n patterns of movement for trucks carrying agricultural products throughout the U.S. might help these models to reflect reality much better.
2. Ignored international shipments: For the approximately 548,000,000 cwt of fruits and vegetables from domestic shippers in the United States in 1991, an additional 150,000,000 cwt arrived from foreign countries [26]. This portion of the market was igno red to simplify the study, especially given that distances to origins, routes of travel, and energy per ton-mile are difficult to estimate. However, the models also separate the domestic and import portions of the market into two sub-models which have no influence on eachother, i.e. the demand for domestic produce can be estimated without considering the effect of foreign competition. This assumption must by its nature lead to some errors. For example, the model assumes that a California-grown product is spread evenly around the country; but in fact the California item may be distributed in the west only while an import from Brazil or Chile serves the East. It is possible that if the U.S. wanted to pursue the implications of the hypothesis, imported f ruits and vegetables would offer the best possibilities for energy gains.
3. Didn't consider refrigeration in shipment: Refrigeration in trucking may somewhat increase the energy consumption above the 3357 BTU/ton-mile average given. In addition, the energy consumption for warehousing has also been ignored.
4. Ignored energy consumption in post-processing of fruits and vegetables before sale to consumer: although energy in this sector might be the same regardless of where the produce was grown, relying on smaller producers might also lead to smaller processi ng plants, thereby affecting energy consumption. The literature search revealed a life-cycle energy comparison for food production only for the case of a loaf of bread in the United Kingdom in 1976, in which 20% of the total energy was in growing the whe at, 9.8% was in transport (distances in the U.K. are presumably shorter), and the rest was in processing, packaging, and marketing [11, p.28]. Therefore, changes in these processing aspects could have a large impact on the overall life cycle.
5. Didn't consider energy impact of changing population in farming: Human ecologist Albert Punti has developed an interesting methodology for comparing the energy intensity before and after a transition not only of the farming method but also of the lifes tyles of the persons farming [19, p.79-85]. One of his conclusions was that in some instances, the energy consumption in the farming technique may increase by switching to a new technology, but the overall energy consumption may decrease because of a lif estyle change or because fewer people will be working in the field. There is perhaps a danger that small local farms may be similarly efficient in the activity but inefficient overall.
6. Did not take into account changing energy demand for irrigation: The USDA estimated in 1981 that groundwater levels were dropping at the rates of between 0.5'/year (Arkansas) and 6.6'/yr. (California) [22, p.14]. Since energy consumption for irrigatio n is directly proportional to the head height being pumped, the study could have derived a more accurate estimate of current energy consumption by adjusting the irrigation component from the 1978 data up to the present.
Caveats related to underlying economic and technological assumptions
1. Economic obstacles: Energy costs are only one component of the overall economics of agricultural production. If additional energy inputs (fertilizers, locating in the Imperial Valley) can allow a farmer to decrease overall costs, then the economic adv antage outweighs other prerogatives. Indeed, if U.S. agriculture is currently concentrated in Florida and California, then this probably came about because it was economically advantageous to locate there. On the other hand, at the time that agriculture in these two states was expanding, government in general believed that subsidizing infrastructure such as water reclamation (i.e. American Aqueduct) and transportation (i.e. Interstate system) would lead to the long-term enhancement of the economy. At p resent the argument appears to favor removing subsidies from all aspects of the economy so that firms will not pursue counterproductive behavior. Nevertheless, in order to make redistribution economically attractive, either consumers will need to volunta rily pay more for produce or else existing subsidies must be removed to give other states more of a chance.
2. Political obstacles: Intimately linked to the economic obstacles is the political intransigence of the national government, which must be a part of any large change. However, big growers in California or Washington along with the American Trucking Ass ociation would likely oppose such changes.
3. Dominance of truck may diminish in the future: this study is based on the dominant nature of trucking in the fruit and vegetable shipment market, which because of its high energy intensity lends itself to energy cutting measure. However, the railroads have been making advances in the areas of logistics, so that trains reach their destinations more reliably, and power supply technology, so that perishable shipments can be kept refrigerated throughout the journey. In particular, the advances in trailer -on-flatcar and container-on-flatcar, both of which can be refrigerated, may help rail to recapture some of the market. If such a shift occurred, the incentive for localizing agriculture might decrease (This is approach #3 described in the introduction t o the paper.).
4. Technology in future may make truck less energy intensive: In the short run, saturating the market with state-of-the-art truck technology could increase overall efficiency for trucks up to 75% [10]. In the long run, replacing internal combustion engin es with advanced technology such as fuel cells may provide the answer to eliminating intractable problems such as excess ozone in the large urban areas of this country. Fuel cells in conjunction with non-fossil fuel sources could also help to stem global warming. Researchers have proposed a network of solar-powered hydrogen formation stations in a desert region connected to a fuel distribution network in an urban area for fuel-cell powered cars; the system might cost $100 billion in the year 2010 and th e equivalent of a gallon of gas might cost $2.50 to $3.50 per gallon [9, p.209]. If a similar system provided hydrogen to fuel-cell powered trucks, the incentive to reduce greenhouse gas emissions by reducing trucking would no longer exist.
5. Conclusion: factors that might encourage a shift to local agriculture to take place
While the previous subsection may have illustrated some powerful impediments to changing agriculture, there are also some positive factors, other than the possible energy savings, that may bring about a shift of fruit and vegetable production away fro m where it is currently concentrated. These possibilities arise from current trends in transportation and irrigation costs, labor inputs, and available water supply.
As mentioned earlier, some studies suggest that road freight transportation is underpriced [14]. Ketcham's calculations suggest that of the total highway transportation bill of approximately $800 billion, trucking accounts for roughly $267 billion. However, he has also estimated that there are an additional $105 billion in indirect costs incurred by trucking that are not payed by the trucker or the shipping client, for aspects such as uncompensated wear and tear on the highways ññ trucks may be resp onsible for 95% of all highway damage [ibid, p.6] ññ and medical and lost productivity costs from accidents, incurred congestion, and air pollution due to trucks. If government policy raised the cost of trucking by 40% to account for these external costs , some farming operations which rely on cross-continental produce shipments by truck would no longer be economically viable, so that the either the grower would switch to rail shipment or else the seller in the destination market would find a source for t he product closer to that market (i.e. buying Arkansas tomatoes instead of California ones.).
One small case study for pig farming suggests that this type of cause-and-effect relationship can occur in agriculture. When energy and transportation costs rose rapidly from 1978 to 1982, pig farming that was located far from the northern midwest fe ed supplies and "fattening" centers, such as in the Ozarks or southeastern states, declined by 15-30%, while production in Wisconsin, Iowa, and Nebraska grew by 15-22% [12, p. H12]. "Of course, the reasons for the shifts are transportation related. It's expensive to transport pibs to the Midwest from distant points in Arkansas and Tennessee," according to the author. Moving livestock is a special operation, so its cost rise during the early 1980's may have been higher than the overall average for truck ing. However, if trucking rates were raised significantly across the board, the phenomenon that occurred in raising pigs could also occur in fruits and vegetables.
Along with adjusting truck rates, removing subsidies from water supplies as well might bring about greater efficiency in fruit and vegetable production. Although many sources suggest that western farmers do not pay the true price for irrigation water [e.g. 6], the exact level of this subsidy was unclear. What was more explicit was that the construction of the irrigation infrastructure had been subsidized. Goldsmith and Hildyard state that "Even in the San Joaquin Valley, in California, some 60% of farmers do not have adequate drainage facilities. Instead of tackling salinization at the source, many farmers simply grow shallow-rooting salt-tolerant crops. [Ibid, p.145]" The authors go on to conclude that "If the true cost of drainage were include d, then many water projects would appear to be wholly uneconomical. [ibid, p.146]" These comments suggest that irrigation farmers in the San Joaquin valley and elsewhere in California are even today benefitting from a subsidized project and that furtherm ore they are engaging in an unsustainable practice of not dealing with the salinization of the soil, because it would be too costly to install proper drainage. Correcting this type of behavior might well force some portion of California's fruit and veget able production to move to eastern and southeastern states.
In the case of the southwest, underpriced labor is another factor that may have allowed growers to carve out an artificially high portion of the overall market. Gerber has documented how the earliest grain farmers in California in the 1850's were abl e to succeed in the national market in the face of a great labor shortage (Most settlers were earning high wages panning for gold) in part by hiring native Americans for subsistence wages [5, p.40-57]. In fact, the role of the laboring class that the ear liest Spanish farmers developed before California became a US state has persisted to the present day:
The persistence of Hispanic labor relations by US settlers in California bridges the gap between the Mission period [when farmers coerced native Americans to work for them] and the post-bellum era when successive waves of Chinese, Japanese, Philipino, and Mexican workers made agriculture in California so profitable. [p.57]
If today's immigrant farmworkers in the southwest were to achieve wage parity with other types of work, or if the current anti-immigrant sentiment in California forced farmers to hire non-Mexican workers at higher rates of pay, then a market rationalizati on might come about in which southwestern fruit and vegetable operations might no longer be viable and so the production could move to the east.
Finally, a shift away from irrigated agriculture may come about due to the scarcity of water, in which case the transition will be purely economic and political events will not have any influence. For instance, the USDA has predicted that irrigated ag riculture on the high plain of western Texas may decline by 57% in area due to aquifer depletion [22, p.9]. The author goes on to state that some shifting of crops would be inevitable,although "crop production adjustments resulting from declining water l evels (the case of moving crops from California to Arkansas is mentioned here) will occur very gradually. Salinization of water supplies is another factor that could eventually lead to a redistribution of agriculture: according to Goldsmith and Hildyard, "the Colorado and Rio Grande are becoming increasingly saline" [6, p.157]. With not just certain croplands but the source rivers themselves suffering from increasing salinity, widespread problems may occur in the future with obtaining usable irrigation water at a reasonable price.
One group of agriculturalists has suggested another generation of infrastructure building to cope with the problem of water shortages, aquifer depletion, and salinization, known as the North American Water and Power Alliance or NAWAPA. This scheme wou ld build a network of waterways into the northwest, the Canadian rockies, and perhaps even to Alaska, so that southwestern farmers would have an inexhaustible supply of good water with which they could among other things continually flush their fields so as to keep down the salinity levels. According to Goldsmith, the North American Water and Power Aliance may be the "only way to achieve salt balance necessary for the long-term health of Western agriculture, on which the entire US and indeed much of the world depends. "[6, p.159] Other intermediate solutions short of such a drastic and expensive plan exist. Allen et al have suggested that western utilities would stand to benefit by subsidizing those who irrigate to use water more efficiently: "Potentia l subsidies paid to irrigators by energy utilities as a result of increased hydropower generation or decreased pumping load due to lower consumptive-use requirements could be considered to offset lower profit margins of alternative, lowconsumptive-use cro ps. [1, p.100]" In the end, if approaches like Allen's were not sufficient to bring western agriculture onto a more sustainable footing, then the market for agricultural goods would be faced with a choice between building NAWAPA and encouraging states li ke Arkansas, Michigan, and Virginia to grow more of our fruits and vegetables. At that point, the latter option might indeed have a large economic advantage as well as energy advantage and so the redistribution of agriculture tested in this study might i ndeed come to pass.