ARTICLE pubs.acs.org/est
Inventory Development and Input-Output Model of U.S. Land Use: Relating Land in Production to Consumption Christine Costello,*,† W. Michael Griffin,‡,§ H. Scott Matthews,†,‡ and Christopher L. Weber†,|| †
Department of Civil and Environmental Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, United States Department of Engineering and Public Policy, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, United States § Tepper School of Business, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, United States Science and Technology Policy Institute, Washington, D.C. 20006, United States
)
‡
bS Supporting Information ABSTRACT: As populations and demands for land-intensive products, e.g., cattle and biofuels, increase the need to understand the relationship between land use and consumption grows. This paper develops a production-based inventory of land use (i.e., the land used to produce goods) in the U.S. With this inventory an input-output analysis is used to create a consumption-based inventory of land use. This allows for exploration of links between land used in production to the consumption of particular goods. For example, it is possible to estimate the amount of cropland embodied in processed foods or healthcare services. As would be expected, agricultural and forestry industries are the largest users of land in the production-based inventory. Similarly, we find that processed foods and forest products are the largest users of land in the consumption-based inventory. Somewhat less expectedly this work finds that the majority of manufacturing and service industries, not typically associated with land use, require substantial amounts of land to produce output due to the purchase of food and other agricultural and wood-based products in the supply chain. The quantitative land use results of this analysis could be integrated with qualitative metrics such as weighting schemes designed to reflect environmental impact or life cycle impact assessment methods.
1. INTRODUCTION Approximately 29% of the Earth’s land surface has been converted to agricultural or built-up areas to support humanity, and projections indicate an additional 33% of the land surface could be converted over the next 100 years.1 To meet future demands for goods and services it will be important to know how land is utilized in the production of all goods and services in order to assess trade-offs among competing land-intensive goods. This paper presents a straightforward method using transparent data sources and assumptions to clearly demonstrate how U.S. land types are currently utilized to meet U.S. domestic consumption and exports. Previous efforts to quantify land use (LU) in terms of consumption can be broadly grouped into the following three categories. First, the ecological footprint (EF) method compares countries by “equivalent number of earths required” to meet each country’s consumption.2 Criticism of EF includes the use of a hypothetical land area (e.g., acres of trees needed to offset greenhouse gases (GHG) emitted to produce a good), lack of specificity with regard to types of consumption or land type, and failure to assign any spatial reference. More importantly, EF is not constrained by the quantity or quality of land actually available3,4 making it only a useful metaphor for overconsumption. r 2011 American Chemical Society
Second, partial, e.g. the Food and Agricultural Policy Research Institute (FAPRI)5,6 and general, e.g. the Global Trade Analysis Project (GTAP),7 equilibrium models use economic relationships to relate land use to consumption.79 These models are relatively coarse in their representations of economies and only assign LU to agricultural sectors. For example, GTAP has a total of 57 sectors including 8 crop sectors resulting in considerable aggregation. This 434-sector model includes 15 crop sectors. The approach taken in this work includes developed land, an often overlooked, but increasingly important land use category.10,11 Third, authors have incorporated LU into input-output analysis frameworks.4,12,13 Lenzen and Murray created a LU inventory based on land used for the production of goods in Australia and conduct an EIO-LCA model to relate land used for production to consumed goods similar to the work done herein.4 Other authors have tracked physical land used in trade by determining a country’s “virtual” LU resulting from imports.3,12 Similar methods have been applied to assess virtual water use through Received: December 17, 2010 Accepted: April 25, 2011 Revised: April 18, 2011 Published: May 11, 2011 4937
dx.doi.org/10.1021/es104245j | Environ. Sci. Technol. 2011, 45, 4937–4943
Environmental Science & Technology
ARTICLE
agricultural trade.14,15 The concept of environmental impacts being “embodied” in traded goods is often described in climate change research.1618 For some time LCA researchers have been attempting to incorporate LU into an LCA framework, particularly the environmental quality impacts associated with LU, for example, human appropriation of net primary productivity,11 biodiversity, specifically vascular plant species,19 species diversity,1922 and soil organic matter in parallel with biodiversity indicators.23 Defining qualitative metrics that are meaningful at a large-scale or across ecosystems has been difficult due to varying spatially explicit factors that lead to differing environmental impacts.24 The aim of this paper is not to weigh in on this debate but instead to provide unweighted, quantitative estimates of the land occupied by industries to support U.S. consumption. The results of this analysis could then easily be integrated with life cycle impact assessment methods or weighting schemes designed to reflect the extent of impact associated with the land use occupied. A U.S.-centric economic input-output life cycle assessment (EIO-LCA, also referred to as environmentally extended inputoutput analysis) framework is used to relate current U.S. land used for production to the goods and services consumed in the U.S. Economic input-output analysis was originally formalized by Leontief and represents a linear model of all interindustry or intercommodity transactions in a national economy.25,26 EIOLCA has been used since the 1930s, and more recently has been used to capture environmental impacts associated with the production of a particular good or service across the entire supply chain of production.4,11,27 A production-based inventory of LU was developed and incorporated with the most recent (2002) U.S. Department of Commerce tables.25 The production-based inventory includes land area used for production in each sector of the economy. For example, area occupied by corn is assigned to the Corn farming sector. A consumption-based inventory of LU is created using EIOLCA. The consumption-based inventory provides insight into the quantity of land required to support domestic consumption, including U.S. exports. The LU associated with demand for imports was evaluated by approximating the amount of land that would have been required to produce imports domestically. Challenges associated with creating the inventory are discussed, including data availability and allocation of data to sectors. No U.S. study to our knowledge has attempted to create production or consumption land use inventories for all sectors of the economy.
2. METHOD Input-Output Modeling. EIO-LCA was employed to relate the production-based LU inventory to goods and services creating a consumption-based inventory LU. The power of EIO-LCA is its ability to delineate upstream environmental impacts as a result of including all levels of supply, i.e., the suppliers of the suppliers without cutoff error, a serious issue otherwise in LCA. As initially described by Leontief, the total output of an economy, x, can be expressed as the sum of intermediate consumption, Ax, and final consumption, y, as follows
x ¼ Ax þ y
ð1Þ
where A is the direct requirements matrix, representing interindustry transactions. The direct requirements matrix used for this work is the most recent, 2002, U.S. Department of Commerce 428-sector benchmark commodity-by-commodity input-
output matrix.25 Four agricultural sectors were disaggregated into ten sectors, for a net total of 434 sectors, to provide greater detail in agricultural sectors;28,29 details are in provided in the Supporting Information (S.I.). When solved for total output, x, this equation yields x ¼ ðI-AÞ1 ðy pe þ y pi þ y exp þ y imp þ y govt Þ
ð2Þ
Total supply chain requirements, x, for a specified final demand, y, can be estimated using eq 2. The expression (I-A)-1 represents the total economic supply chain requirements, also called the Leontief inverse.30 Final demand, y, is separated into five categories in this analysis: ype for demand by personal expenditure or households, ypi for private fixed investment and changes in private inventories, yexp for exports, yimp for imports, and ygovt for government purchases. Private fixed investment is defined as investment by industries in equipment, software, and structures.31 The consumption-based inventory of LU, f, associated with demand for any good or service can be calculated using the following linear relationship f ¼ F x ¼ FðI-AÞ1 y
ð3Þ
The vector F, millions of hectares per million U.S. dollars (Mha/ $M), is created using the LU production inventory, described below, and economic value of output for each sector obtained from the U.S. Department of Commerce. To demonstrate the importance of assessing the entire supply chain the results are presented in terms of direct and total supply chain LU. Direct LU is defined as Ay and represents land associated with direct purchases made by individual sectors. Upstream LU represents all subsequent purchases made throughout the entire supply chain. The model is balanced and linear ensuring that the total area included in the production inventory matches the area in the consumption inventory. The U.S. imports goods and services to meet domestic demand, and these goods and services require the use of land in the countries of origin. In lieu of expanding the EIO-LCA to include each trade partner, each with different land use efficiencies, the amount of U.S. LU avoided through imports was estimated.32 Additional details regarding EIO-LCA modeling are included in the S.I. Production Inventory. The production-based inventory, required to create the LU vector (F), was assembled from several data sources. Available sector-specific data (e.g., in hectares) were generally limited to crop and animal farming sectors. Where land area data were not available a top-down approach was used. This is particularly relevant to sectors located on developed land, e.g., manufacturing. The USDA Major Uses of Land Report (MULR) for 2002 assigns all 920 Mha of U.S. land (including Alaska and Hawaii) to six major categories: grassland/pastureland, 238; forest, 263; cropland, 179; special uses (e.g., roads, rail, parks, recreational lands), 120; miscellaneous (e.g., rural residential areas, marshes, deserts), 92; and urban land, 24 Mha.33 The MULR is used to ensure that estimates derived from other sources are reasonable and to avoid double-counting. In the case of timberland, road and air transportation, and commercial and manufacturing industries, the MULR estimates were either used directly or allocated to a number of sectors using additional data. Mining industries were assigned areas based on data from the 1992 U.S. Geological Survey (USGS) National Land Cover Data set (NLCD).34 4938
dx.doi.org/10.1021/es104245j |Environ. Sci. Technol. 2011, 45, 4937–4943
Environmental Science & Technology
ARTICLE
Table 1. Summary of Production-Based Land Use Inventoryb NAICS code
NAICS code description
million hectares
number
number of sectors
range in land use
average value
(Mha)
of sectors
with nonzero values
multiplier (F) (ha/$M)c
of F (ha/$M)c
11
agriculture, forestry, fishing and hunting
650
24
21
587000
2000
111
crop production
140
15
15
654300
1500
112
animal production
310
4
4
587000
2000
113
forestry and logging
200
2
2
-
5700
21
mining
1
11
9
0.840
20
22
utilities
0.3
3
1
-
3
23 3133
construction manufacturing
0 4.1
7 280
0 280
0.2970
1.2e2
42
wholesale trade
0.3
1
1
-
9.3e3
4445
retail trade
1.7
1
1
-
4.5e2
2
4849
transportation and warehousing
15
10
9
2.2e 140
30
51
information
1.4
14
14
7.3e30.7
0.2
52
finance and insurance
0.3
6
6
1.4e31.1e2
5.3e3
53
real estate and rental and leasing
0.5
6
6
4.0e30.2
5.8e2
54 55
professional, scientific, and technical services management of companies and enterprises
0.1
