Assessing the sustainability of agricultural systems is challenging because measurements frequently mix multiple dimensions, scales, and benchmarks. This study uses data envelopment analysis to develop benchmarks for comparing the sustainability of different agricultural systems and cropping techniques with a measure of sustainability called sustainable value . Incorporating DEA into the sustainable value approach expands upon the work of Figge and Hahn , who first introduced SV in the journal Ecological Economics. SV calculations integrate human, natural resource, and financial dimensions to generate a monetary measure of sustainability. The Loess Plateau provides a rich context to illustrate how sustainability measurements can be used to assess natural resource management trade-offs. It is a highly distressed region where intensive crop production is undermined by high soil erosion rates that threaten the long-term sustainability of the land and local food production . The Loess Plateau is one of the most severely degraded areas in the world, with over 60% of its land subjected to soil degradation and an average annual soil loss of 20–25 t/ha . Land use changes are often extreme. Much of the agricultural land has already been planted to trees through the Grain for Green program . However, converting from cropland to trees is an extreme conservation measure that generates little economic return to farmers who make a living mostly on their land. Many have migrated to urban areas in order to compensate for lost farm jobs, leading to other unintended consequences. In some cases child-care and other parental activities have been left to the elderly or older children . Rather than focusing on extreme land use changes,arándanos azules cultivo the analysis presented in this paper investigates how to balance environmental objectives with continued crop production.
Many frameworks have been proposed to measure sustainability. Macro scale proposals include the Green National Product , Ecological Footprints , and Genuine Savings . These methods are generally used to readjust Gross Domestic Product or Gross National Product calculations to account for net changes in environmental degradation. While ecologically minded organizations promulgate these alternative accounting approaches, the frameworks are not widely applied in China. Unlike the DEA/SV method proposed in this paper, these macro-level approaches do not explicitly map the value of meeting different environmental targets that are affected by local and regional agricultural production practices.SV is used to calculate net sustainable values, rather than the environmental burden imposed by natural resource use. First presented in Ecological Economics by Figge and Hahn , SV is based upon Capital Theory . SV assesses the sustainability of a proposed or existing system by comparing the opportunity cost of using capital in that system rather than a predetermined benchmark. SV can be used to evaluate the effectiveness of local or regional natural resource management decisions. For example, Van Passel et al. apply the SV approach to the Flemish dairy industry, finding the method to be useable and workable for smaller enterprises. Two years later the authors improved upon the method by applying a parametrically estimated efficiency frontier to provide individual benchmarks for each system . The research presented in this paper takes advantage of the SV approach to study cropping systems in the Loess Plateau, but refines it by creating discrete, customized individual benchmarks using a DEA. The DEA allows for comparison between similar systems and creates a discrete benchmark that is specifically comparable to that system. In contrast to previous studies that chose a single “best” benchmark, or a parametrically estimated frontier, the DEA method allows us to create a non-parametric frontier of benchmarks that takes into account the most efficient use of capital for each unique system .
The SV for over 2000 cropping systems reviewed for the Loess Plateau is recorded in a series of comparison matrices and organized by crop type, cropping technique or land type. These matrices are then used to determine which management practices, like rotation or terracing, have the greatest impact on sustainability. Although this would be the first known integration of the DEA with SV methods, the DEA method has been applied to other sustainability measures. For example, it has been used to compute environmental efficiency . Environmental efficiency is formulated in the same way as technical efficiency except that environmental impacts, rather than observed inputs, are calculated. One limit of environmental efficiency is that it becomes difficult to measure environmental impacts. Eco-efficiency, defined as the ratio of a created value over the environmental impact, is another popular indicator for measuring sustainability that used DEA . According to Van Passel et al. , the rebound effect is one major shortcoming of eco-efficiency. The rebound effect means that advances in environmental performance may be over stated because better eco-efficiency may also lead to growth and thus increase the use of environmental resources. The paper is organized, as follows: The methodology, presented in Section 2, updates Figge and Hahn’s SV approach by introducing frontier benchmarks estimated with DEA. Section 3 contains a description of the Loess Plateau study area and the simulation data, created from the Environmental Policy Integrated Climate model . This simulation is conducted on more than 2000 different cropping-system variations that are based upon Lu’s original work and a subsequent publication that uses approximately 500 of these cropping systems . The sustainability measurements evaluate different combinations of crop rotations, production situations, terracing techniques, tillage techniques, crop residue management techniques, mechanization levels, and land units. The empirical results are reported in Section 4. Discussion and conclusions are provided in Sections 5 and 6, respectively.This section provides a summary of steps for calculating SV and SE, which are applied to the Loess Plateau example.
According to Van Passel et al. , the SV and SE can be calculated in three steps. First, the scope of the analysis is determined. The data used in the presenting study employs 2006 distinct cropping systems as entities to create sustainable value. The entities all employ all three forms of capital. Different cropping systems are characterized by various technologies and practices such as crop rotations and terracing techniques. Second, relevant resources must be identified. In the context of sustainable development, the weight of relative importance of the capital forms used by a firm can be judged by the scarcity or degree of depletion of the capital . Over 60% of land in the Loess Plateau is subjected to soil degradation. Nitrogen loss is associated with soil loss. Thus, in this study soil and nitrogen are recognized as two forms of natural capital. Soil and nitrogen data are rare to observe at the farm level or national level, which strengthens the rationale for utilizing a SV measurement. Fortunately, the simulation model in Lu et al. , which is verified by experiments, provides extensive and realistic estimates of soil and nitrogen losses associated with various cropping practices. In addition to natural capital, financial capital and human capital are also taken into account through enterprise budgeting. Third, appropriate benchmarks must be determined. Four possible benchmarks were proposed by Van Passel et al. . First, the weighted average of a sample can be used. For example,maceta de 30 litros cropping systems with conservation practices can be chosen to calculate benchmarks. Second, a super-efficient firm that uses every single type of capital in the most efficient way can serve as the super-efficient benchmark. In practice, a super-efficient cropping system is highly unlikely. Third, a performance target can be used as a benchmark. A performance target example given by Van Passel et al. is 150 kg nitrogen per ha for the farm gate nitrogen surplus for dairy farms. Fourth, the unweighted average of all firms in the sample can be used as a benchmark.The benchmark choice reflects a normative judgment of sustainable development, and thus biases the way in which the SV is interpreted . Benchmarks should therefore be chosen with great care. Since the goal of this study is to identify the most sustainable cropping systems, the best performance benchmark is preferred. A performance target may also be appropriate, but it may not be easy to specify the reasonable target level. In this paper, many possible cropping systems for the Loess Plateau are considered, so a frontier is constructed for all the possible cropping systems. The frontier takes into account the most efficient use of capital for each unique system, rather than assuming that there is a single best system. Instead of using the parametric frontier benchmark proposed by Van Passel et al. , this study adopts a non-parametric DEA to determine benchmarks. Both parametric and non-parametric approaches have been proposed in the frontier literature . Data noise can be taken into account in the parametric approach, but specification error may arise from the choice of the functional form. The dataset incorporated in this study is simulated from the EPIC model, which is described in greater detail in Section 3.
Data noise is not expected to play a significant role in the estimation of the production frontier in this study because simulated data do not present sampling bias; that is, the simulated data can be readily replicated. The DEA method is also more computationally efficient, especially when multiple capital types are considered in the production process. Another advantage is that a unique frontier benchmark is specified for each cropping system through the consideration of each technology possibility. Lu et al. identified the cropping systems in Ansai County of the Loess Plateau. A summary of these systems is presented in Table 2. Their dataset includes 2006 cropping systems that are comprised of different combinations of 5 land units, 17 crop rotations, 3 production situations, 3 terracing techniques, 2 tillage techniques, 2 crop residue management techniques and 2 mechanization levels. Corresponding outputs of interest were simulated for each system using the Environmental Policy Integrated Climate model and validated with the experimental data as described by Lu et al. . EPIC is a comprehensive simulation model designed to predict the effects of management decisions on soil, water, nutrient and pesticide movements and their combined impact on soil loss, water quality and crop yield . It consists of weather, surface runoff, water and wind erosion, nitrogen leaching, pesticide fate and transport, crop growth and yield, crop rotations, tillage, plant environment control , economic accounting, and waste management. Lu et al. developed the comprehensive dataset regarding soil, weather, crop management, fertilizer and other parameters to meet the basic requirements to run the EPIC model. Hundreds of equations are applied in EPIC to then simulate processes such as crop growth and soil erosion. As described in Section 2, in order to apply the SV approach with DEA benchmarks, the value added and capital need to be specified. As previously defined, crop revenue minus intermediate consumption is specified as “value added” in the SV approach. To cope with multidimensionality, it is assumed that each cropping system uses all forms of capital to produce value. Typically, natural capital is difficult to measure. However, the EPIC model provides an opportunity to measure soil loss and nitrogen losses directly. Nitrogen losses are estimated in EPIC through “runoff and sediment, nutrient movement by soil evaporation, denitrification, ammonia nitrification and volatilization, mineralization, immobilization, biological-fixation, contribution of rainfall and irrigation, and NO3-N leaching Lu et al. .” Lu et al. note that most of the losses of N resulted from volatilization, runoff and soil erosion. Soil loss and nitrogen loss from the EPIC model are treated as natural capital inputs in the production process. Labor is viewed as human capital. Financial capital is calculated by aggregating all conventional inputs, including seeds, nutrients , biocides, irrigation if applicable, and farm equipment . Descriptive statistics of the data are given in Table 3. Revenue and cost data, except labor, are expressed in Chinese monetary units, the RMB. Natural capital, soil and nitrogen, are described in physical units. Financial capital and human capital are expressed in the RMB monetary units. On average, 5221 RMB/ha in revenue can be produced by a 3112 kg/ha soil loss and 15.3 kg/ha nitrogen loss, a cost of 1654 RMB/ha and 1390 RMB/ha for labor. Prices used to calculate aggregate value added and capital are taken from Lu et al. .