|
An Application To Intensive Dutch Dairy Landscapes Economics Essay
荷兰奶牛密集景观的应用经济学论文
关于多功能性的概念，政治和学术领域产生了很多的辩论并且发现了不同的解释。然而，没有既定的术语来描述它的关键要素（OECD，2001年）。多功能性是指一个经济活动（通常是指是农业）可能有多个输出，鉴于这个，它可能有助于实现一系列的社会目标。多功能可以被视为一个没有必然内在价值的活动的特征（积极看法）或作为活动的一个理想的目标（规范性观点）。在这项研究中，我们采用的是第二个角度，因为我们考虑到多功能农业潜在的社会价值，第二个原因是因为农业活动的目标是不断尝试着满足社会的需求。从上世纪80年代开始，社会对农业的需求有了新的增加，比如说可持续农业、在农业实践的过程中追求环保以及负责任地管理自然资源。这些想法是在生态、技术和社会经济方面更广泛的可持续发展观（哈伍德，1990），并且也在含蓄地暗示农业的多功能性。
The concept of multifunctionality has generated ample debate in political and academic spheres and different interpretations can be found. However, there is no established terminology to describe its key elements (OECD, 2001b). Multifunctionality refers to the fact that an economic activity (usually agriculture is referred to) may have multiple outputs and, by virtue of this, may contribute to several social objectives at once. Multifunctionality can be seen either as a characteristic of an activity without a necessarily intrinsic value (positive view) or as a desirable objective of the activity (normative view). In this study we adopt the second perspective since we consider multifunctional agriculture potentially socially valuable and therefore a goal for agricultural activities is to try and satisfy social demand. From the 1980s onwards new demands of society to agriculture increased, related to concepts such as sustainable agriculture, environmentally friendly agricultural practices and responsible management of natural resources. These ideas refer to the ecological, technological and socioeconomic dimensions of the broader concept of sustainable development (Harwood, 1990), and implicitly allude to the multifunctional nature of agriculture.
At present, awareness among rural and urban citizens of the positive and negative effects of agriculture beyond commodity production is growing and governments are increasingly looking for ways to ensure that the non-commodity outputs of agriculture correspond in quantity, composition and quality to those demanded by society (OECD, 2003). Delivering to growing public demands raises two main questions for farmers and policy makers: (1) what might the public actually want (Hall et al., 2004); (2) how to integrate preferences of citizens in the evaluation of the multifunctional performance and the sustainability of alternative land use options. An increasing number of studies address integrated and holistic evaluation of the economic, environmental and social impacts of human activities, in general, and of land use and agricultural production in particular (e.g. Parker et al., 2002; Osinski et al., 2003; Van Calker et al., 2006 and 2007; Rossing et al., 2007; Van Keulen, 2007; Van Ittersum et al., 2008). Economic studies can be divided into those that focus on economic valuation of landscapes and their elements, and those that integrate economic aspects within land-use modelling (Osinski et al., 2003), roughly corresponding to the distinct and sometimes conflicting approaches of Environmental Economics and Ecological Economics, respectively (Parra-López et al., 2004). Turner et al. (2000) suggested that integrating key elements in an approach that combines economic valuation, integrated modelling, stakeholder analysis, and multi-criteria evaluation can provide complementary insights into sustainable and welfare-optimizing management and policy. In this paper we investigate these suggestions.
The main objective of this paper is to propose and apply a methodological framework to integrate the social demand for multifunctionality of agriculture in the evaluation and design of more sustainable agro-landscapes composed of dairy farming systems. The methodology will be illustrated in a case study of an intensively managed, ecologically and historically valuable agricultural landscape in the Northern Friesian Woodlands (The Netherlands). The description of the case study is provided in section 2. The methodological framework (Figure 1) consists of three interconnected components which are introduced in sections 3 to 5. In section 3 social preferences for agricultural non-market functions are prioritized using a novel application of the multi-criteria decision analysis techniques of Quality Function Deployment and Analytic Network Process (QFD/ANP). In section 4 the social preferences for non-market functions are combined with benefits from market functions to arrive at social net benefit based on the neoclassical concept of utility. Finally in section 5, the Landscape IMAGES model is used to generate and evaluate alternative landscapes and to reveal trade-offs between market and non-market functions. The results (section 6) demonstrate the alternative development options for the case study landscape and highlight opportunities and pitfalls for further improvement of social net benefit.
The case study: Northern Friesian Woodlands, The Netherlands
The case study focuses on an intensively managed agricultural landscape in the Northern Friesian Woodlands (The Netherlands). This region is characterized by a small scale landscape on predominantly sandy soils with dairy farming as the prevailing land-use activity. On some farms a limited proportion of up to 5% of the area is used for forage maize production, while the rest of the area is occupied by permanent grassland, rotationally grazed and mown. The fields with an average size of 2 ha are often surrounded by hedgerows and frequently border on ponds. The average grazing season lasts 6 months from May to October. Grazing systems range from day and night grazing to restricted and zero grazing. The bio-physical farm and field characteristics and the social demands as articulated in regulations to maintain landscape and land-use have limited the possibilities to convert to large scale agriculture in the past. On the other hand, the region offers ample opportunities to provide non-productive amenities, the remuneration of which has recently been argued to sustain farming in the area (Berentsen et al., 2007).
In the 1990s, the farmers were confronted with strict regulations to reduce emissions of ammonia and nitrate to the environment. Environmental cooperatives VEL (Vereniging Eastermar's Lansdouwe) and VANLA (Vereniging Agrarisch beheer Natuur en Landschap in Achtkarspelen) emerged in the Northern Friesian Woodlands as a response to predominantly generic and means-oriented policy interventions. The cooperatives developed activities to reach the aims of the proposed policies with context-specific measures that were acceptable for farmers. In addition the farmers committed themselves to maintaining the historical landscape which is the basis for a strong local identity of its inhabitants and the cooperatives organized activities related to nature and landscape management by farmers (Renting and Van der Ploeg, 2001; Wiskerke et al., 2003; Anonymous, 2005).
Here we focus on three key non-market functions that are supported by the activities of the environmental cooperatives:
Landscape quality (LQ): This function refers to variation in number of plant species in pasture and to irregularity in the hedgerow pattern, referred to as half-openness of the landscape, and thus pertains to the spatial scales of field and landscape.
Nature value (NV): This function refers to high species diversity in the grass swards and hedgerows (number of species per ha). This function is relevant at the field scale.
Environmental health (EH): Low nitrogen loss from agriculture, here also interpreted at the field scale.
In this paper we will explore opportunities to satisfy both the non-market and the market functions by adapting agricultural land use and land management in an area of 232 ha, comprising three farms. As indicator of the market function we use landscape gross margin (GM), which is defined as the total revenues minus all variable costs, at the landscape scale. The impact of land use and land management is expressed in variable costs and will become apparent in changes in gross margin rather than total economic results, which also include fixed costs (Ondersteijn et al., 2003). Although we distinguish farms to evaluate technical constraints related to ratio between grazed and mown herbage and the maximum allowed fertilizer application rates, we focus on landscape gross margin rather than the farm gross margin for evaluation of on market function to avoid excess detail in disaggregating individual farm gross margin. This aspect is however required for policy design at farm level which will be covered in subsequent research of the present authors.
Prioritizing agricultural non-market functions based on social preferences
The market function of agriculture is assumed to be valued in the market, which reflects preferences and demands of consumers in a monetary value. In contrast, non-market functions are not valued in a market and therefore lack a monetary value, but could affect the welfare of society. In such case preferences of all citizens may be used as a proxy to reflect potential demand. In this section, two methodologies (QFD and ANP) will be combined to estimate social preferences for non-market functions of agro-landscapes.#p#分页标题#e#
The QFD/ANP methodologies
Quality Function Deployment (QFD) is an analytical tool first conceptualized in Japan in the late 1960s (Akao, 1997) within psychology and marketing, with the aim of allowing firms to incorporate the preferences of consumers in the design stage of the product planning. Following the seminal paper by Kogure and Akao (1983) QFD spread to multiple applications especially in the USA. For an extensive literature review of general applications of QFD, see Chan and Wu (2002), and for specific agri-food case studies see Benner et al. (2003). Despite the broad application of QFD, its implementation in public planning of agriculture is missing from the literature.
QFD proposes a core scheme for strategic planning (e.g., Bergquist and Abeysekera, 1996; Govers, 1996), which can be represented by a decisional structure named the “House of quality” (HoQ) (Figure 2). The aim is to translate what a customer needs, or the WHATs (vector wP in Figure 2), equivalent with the social preferences for agro-landscapes in our application, into strategic or technical requirements or HOWs, i.e. how can these needs be satisfied (vector wF in Figure 2), equivalent with the relative priorities of the non-market functions of agriculture in our case study. This is done on the basis of a relationship matrix WF,P between social preferences (WHATs) and non-market functions (HOWs). Inner relationships among WHATs (WP,P) and among HOWs (WF,F) may be incorporated in the analysis to fine-tune the results (Partovi, 2006). The matrices of relationships are usually elicited from experts. Shortcomings of the classical application of QFD have been pointed out, and relate to the scale of measurement of the relationships among decisional elements (Wasserman, 1993), and the treatment of inner relationships (Partovi and Corredoira, 2002). Recently it was suggested that the Analytic Network Process provides a scientific basis to overcome these limitations (Partovi, 2001; Partovi and Corredoira, 2002; Karsak et al., 2003; Partovi, 2006).
ANP, Analytic Network Process (Saaty, 1996) is a multi-criteria decision-making tool, which represents a decision problem as a network of components, denoted as elements and clusters of elements, where every element can have an influence on itself or some or all the other elements of the system (Niemira and Saaty, 2004). The ANP representation of a QFD model is shown in Figure 3. Its structure consists of three clusters (after Karsak et al., 2003). The Cluster of the Goal (CG) consists of one element, the final aim of the QFD analysis, in our case to determine the priorities of the agricultural non-market functions according to social preferences. The Cluster of WHATs (CW) consists of the WHATs (social preferences), and the elements of the Cluster of HOWs (CH) represents the HOWs (non-market functions).
The ANP network is represented in a super matrix (Table 1). Each cell describes the contribution of element i to the achievement of element j, or in ANP jargon, the dominance of element j over element i. This dominance is represented by an arrow from element j to element i (Figure 3). A cluster is related to another cluster (outer dependence) or to itself (inner dependence) if at least two elements of the cluster(s) are related through a dominance relationship. Outer and inner dependences are the two types of interdependencies among elements. In a QFD network (Figure 3), the cluster of WHATs is dominated by the cluster of HOWs, and each is also inner dependent. To specify the magnitude of the relationships (wi,j), WHATs must be evaluated with respect to their contribution to each HOW, WHATs with respect to each WHAT, and HOWs to each HOW. This evaluation is carried out on the basis of pairwise comparisons of the relative contributions (or priorities) of the dominated elements in one cluster with respect to the element that dominates them (Saaty, 1980). Dominated elements are compared by pairs, and their relative contribution is calculated once all pairs of elements have been compared. This pair-wise comparison can be done on three different scales: numerical, graphical, and verbal (e.g., Forman and Selly, 2001) and usually involves judgement of experts or stakeholders. ANP provides an algorithm to calculate the vector with the relative contributions of the dominated elements i in a cluster to each element j (e.g. Saaty, 1994), e.g. the vector of contributions of the functions F1 to F3 to preference P1 (wF1,P1, wF2,P1, wF3,P1)T in Table 1. In this way the so called unweighted super matrix is constructed (see Table 1).
To calculate the priorities of the non-market agricultural functions (the HOWs in the general case) a matrix manipulation procedure has been proposed (Saaty and Takizawa, 1986; Lee and Kim, 2000; Karsak et al., 2003; Kahraman et al., 2006). Firstly, the vector of priorities of the social preferences is calculated considering the inner dependencies according to . Secondly, the matrix of priorities of the non-market functions as determined by the social preferences is calculated considering the inner dependences among the non-market functions: . Finally, the priorities of the functions considering all the interdependencies are calculated: . A vector wF can be obtained for each expert or decision agent (thus we should call it as wF(e), to refer to the specific expert e), being possible to aggregate them as shown in the next section.
In summary, the combination of QFD and ANP methodologies allows to translate the preferences of citizens into priorities of the different non-market functions of the analysed agro-landscapes.
Prioritizing non-market functions of agriculture in the case study
The set of stakeholders affected by a change in the provision of non-market goods consists potentially of all citizens of the world. However, the impact on different groups could be different depending among others on the spatial proximity to the source of the non-market goods. In this context, the need to properly establish the spatial delimitation of the problem arises: local, national or international level. For practical reasons we consider the Dutch national population, since the impact of the non-market functions of the small analysed region at the international level is considered to be negligible. Demands of Dutch citizens to agriculture are probably not homogeneous and are related to the citizens’ proximity to the analysed agro-landscapes. We suppose that the preferences of all Dutch citizens are equally important and thus used the average demand of Netherlands.
Average preferences of Dutch citizens for non-market functions of agriculture were derived from information in the Eurobarometer (EB) [1] . This indicator summarizes opinions and attitudes of European citizens about general and sectorial topics of interest for the European Union, both in a periodical and consolidated manner twice a year (Standard EB), and in an ad hoc manner (Special EB). A Special EB (EC, 2006) deals with perceptions and opinions of citizens in the specific field of agricultural policy. We selected the preferences of Dutch citizens towards agriculture (Table 2), distinguishing twelve categories of preferences (vector wP). We measure the preferences of citizens through the relative priorities (weights) that they give to different aspect related to the agricultural policy
This national demand must be translated to the specific functions at level of the analysed agro-landscape level. We used expert knowledge to describe the outer and inner relationships between Dutch citizens’ preferences and the non-market functions of the analysed agro-landscapes as described by the Landscape IMAGES model. Ten experts on sustainable farming systems and with knowledge of the case study situation were interviewed individually following a structured questionnaire. We use average expert assessment which we consider more reliable than individual assessments since it removes individual biases and lack of knowledge on some topics. Such analysis of the mean opinion is common in group decision-making literature (Saaty, 1989). Each of them filled out all correlation matrices, resulting in one super matrix for each expert. Sub-matrices WF,P and WF,F were defined according to the pair-wise procedure.
On the contrary, the matrix with relationships between preferences, WP,P, was calculated by direct elicitation (e.g., Larichev et al., 1995; Bottomley and Doyle, 2001). When the number of elements is high as for the social preferences (usually 7-9 is recommended as maximum in ANP; in this case we have 12), the number of pair-wise comparisons can increase excessively, and a ‘direct rating’ weighting method (Bottomley and Doyle, 2001) is more reasonable. The influence of one element on another was directly elicited using a rating scale, ranging from 1 (very weak relationship) to 9 (very strong relationship). Direct elicitation is equivalent to a rating scale in ANP where scale point 9 is 9/1 times greater than scale point 1, 9/2 times greater than 2, and so on. In the case of no influence of an element i on another j, a 0 is assigned (wi,j=0).
In this way, we obtained one super matrix for each interviewed expert and the associated vector of interdependent priorities of the non-market functions (wF(e)) of the Friesian dairy systems. The individual priorities were aggregated to the group by Aggregation of Individual Priorities (AIP) (Ramanathan and Ganesh, 1994), averaged: , where e is the expert e, G is the number of experts, and .
Assessment of social net benefits of agro-landscapes
Market net benefit
#p#分页标题#e#
Market net benefit is defined as a change in utility for society as result of a change in the equilibrium point in the market of agricultural products. Utility is assumed equivalent to the neoclassical concept of surplus (e.g. Varian, 1999), and can be measured in monetary units. Changes in utility will be measured with respect to the current situation. Assumptions for market net benefit assessment in our case study were:
Alternative land uses are based on the same fixed inputs as in the current situation, but may require different amounts of variable inputs.
Prices of inputs and outputs are assumed constant over the short-time horizon considered in the study.
Market net benefit (?UM) is composed of market net benefits of farmers, consumers and government:
Market net benefit for farmers (?UM,FARM): The surplus of farmers is the gross margin (GM). Thus, a change in GM is equivalent to a change in utility for farmers, and, since fixed costs are assumed constant, to a change in their profit as producers [2] . So, we have: ; GM=R+S-VC, where GM is gross margin, R revenue from market, S subsidies and VC variable costs. In our case study, the market function of agriculture is production of milk.
Market net benefit for consumers (?UM,CONS): An approximation of the surplus of consumers for small changes in supply at constant demand curve is (P2-P1)*(Q2+Q1)/2, where 1 and 2 indicate two equilibrium points, and P and Q the price and the quantity of the product in the market, respectively. Since in the case study price of the output (milk) is assumed constant, that is, it is not affected by a change in the supply, we obtain: .
Market net benefit for government (?UM,GOV): Utility for the government increases if support for agriculture decreases. We assume the relation to be: , where S is the public support.
A general relation for market net benefit of a change from the present situation (situation 0) for all stakeholders is: . For our case study this implies: .
Non-market net benefit
The non-market functions in our application are landscape quality (F1=LQ), nature value (F2=NV) and environmental health (F3=EH). We define a change in the utility of an agro-landscape for society (dUNM) as:
(1)
where dFi/Fi is the change in the non-market function Fi relative to the current performance; ?Fi is the relative importance that society attaches to such a change; and n is the number of non-market functions of agriculture. We assume that preferences of society for the non-market functions as determined previously by QFP/ANP refer to the relative weights ?Fi. That is, priorities calculated in section 3.2 for the group of experts, wFi(group), equal the relative importance that society attaches to each function, ?Fi, as defined in this section. This assumption, and therefore the functional form of Eq. 1, is based on the observation that human beings generally perceive relative change, that is, gains and losses in relation to the level from which the change starts (Lootsma, 1996). It is the result of psycho-physical research on the relationship between the intensities of physical stimuli and sensory responses [Weber’s law, Fechner’s law, Brentano proposal, and Stevens power law; cited by Lootsma (1996)], and a common assumption in the definition of a practical measure of utility in Economics.
Integrating Eq. 1, the non-market net benefit of a change of agriculture from the present situation (0) to a given situation (s) following a change in its non-market performances results in:
(2)
Our proposal of utility is additive and must satisfy the basic requirements according to the Multi-Attribute Utility Theory (MAUT). Of these conditions, the requirement of additive independence (Keeney and Raiffa, 1976) is the strongest: risk attitude of a person for a given attribute does not depend on the levels of the other attributes (Ananda and Herath, 2005). If components are close to be additively independent, then the simple additive utility function can be applied (Van Calker et al., 2006). In our case, this independence is guaranteed by the implementation of the ANP, which takes interdependencies into account in the calculation of relative priorities of the non-market functions (?Fi). Moreover, our proposal is additive (sum of components). In an additive functional form it is advisable that (Ananda and Herath, 2005), as in our case. An additive utility function gives consistent and reliable results if non-linearities in the utility functions are adequately captured (Stewart, 1996), as in our expression (Eq. 2) which is a Cobb-Douglas function. In any case, it has been shown that an additive form should be at least a good approximation of overall value (Kwak et al., 2002), even if aforesaid theoretical conditions are violated (Ananda and Herath, 2005).
Social net benefit
We define a change of social benefit (dUS) as:
(3)
where RUM and RUNM are the ranges of possible market and non-market net benefits, respectively, for the set of potential agro-landscapes. In Eq. 3, market and non-market net benefits are standardized by these utility ranges (similarly to Qiu, 2005). This implies that improving market performance of the agro-landscape from its worst state to the best one is considered as important as improving its non-market performance from its worst value to the best one.
The social net benefit of a change in the agro-landscape from the present situation (0) to a given situation (s) is obtained by integrating Eq. 3:
(4)
Model-based generation, evaluation and selection of agro-landscape alternatives using the Landscape IMAGES framework
According to OECD (2001a), reconciling food production and environmental goals can sometimes be achieved simply by changing the level, type and location of agricultural production. This section describes how such hypothesis can be tested quantitatively with Landscape IMAGES (Groot et al., 2007), a static modelling and optimizing framework for exploration of the potential contribution of agricultural land-use and landscape management to the improvement of economic and environmental performances at field, farm and landscape levels.
Exploration methodology
The assessment of the performance of a landscape was based on market (GM) and non-market (LQ, NV and EH) functions. Performance is determined by the arrangement of two types of land-use activities. The first type concerns a field with pasture and its fertilizer and harvesting management regimes. The second concerns the field borders, each of which may or may not contain a hedgerow. Market and non-market performances may be affected by interaction between land-use activities on two or more spatial units. The allocation of discrete packages of land-use activities (Groot et al., 2007) to the landscape makes the problem of finding the trade-off between the performance criteria ‘NP hard’: no algorithm exists that guarantees that the exact trade-off surface is obtained under all circumstances, because the dimensionality of the problem, and therefore the computational difficulty, grows faster than any polynomial in the number of decision variables. Heuristic techniques such as genetic algorithms and evolutionary strategies can be employed to obtain approximations of the trade-off surfaces in a population of solutions (Bergey and Ragsdale, 2005; DeVoil et al., 2006; Groot et al., 2007).
In this study we are not only interested in the trade-off between performance criteria, but also in the shape of the solution space to assess both utopia and dystopia. The solution space was explored in three steps: (1) Determination of the extremes for individual performance criteria by single objective optimization; (2) Exploration of the trade-off frontiers between performance criteria by (2a) constrained single objective optimization, followed by (2b) multi-objective optimization; (3) Homogeneous spreading of solutions within the solution space. The subsequent steps use the input of the previous step.
A multi-objective design problem can be generally stated as follows:
Max F(x) = ( F1(x),...,Fk(x) )T (5)
x = (x1,...,xn)T (6)
Subject to i constraints:
gi (x) ≤ hi (7)
where, F1(x),...,Fk(x) are the objective functions that are simultaneously maximized, and (x1,...,xn) are the decision variables that represent land use activities allocated to the n spatial units. The decision variables can take on values from a predefined array, x ? S, where S is the solution or parameter space. The problem evolves to a single objective optimization problem when k=1. Constraints (Eq. 7) arise from the problem formulation, for instance by limitations on the inputs or outputs related to the activities.
The evolutionary strategy of Differential Evolution (Storn and Price, 1995) was applied to a population of solutions to improve its average performance criteria generation by generation (Bergey and Ragsdale, 2005). During this iterative process, solutions are selected for each new generation if they perform better than other solutions (for single-objective optimization, in steps 1 and 2a) or on the basis of Pareto optimality or presence in a less crowded area of the solution space (for multi-objective optimization, in step 2b). In step 2a, additional constraints were imposed by setting limits to the non-optimized performance criterion.#p#分页标题#e#
In multi-objective optimizations, the solutions can be ranked using the Pareto concept. A set of Pareto optimal solutions consists of solutions that are not dominated by other solutions, when all objectives F1(x),...,Fk(x) are considered. Ranking of solutions follows the procedure proposed by Goldberg (1989). First the Pareto optimal subset is established. This subset receives the highest rank and is removed from contention. This procedure is repeated, and each next subset receives a lower rank, until all solutions have been ranked. Subsets with higher ranks have an increased probability of being maintained in the next generation. To avoid clustering of solutions and to stimulate spreading of solutions across the trade-off surface, a crowding metric (Deb et al., 2002) was applied. The crowding metric measures the Euclidian distance of a solution to the nearest alternative solution. This metric was used as the only selection criterion in step 3 (with k=2) to homogenize the distribution of solutions within the solution space by maximizing the distances between solutions.
In summary, the exploration methodology allows to define the set of potential production alternatives available in the short term. It gives us information about the potential farming practices available for farmers as well as the economic and environmental impacts of them, which will allow us to detect the best and worst farm practices.
Application to the case study
An agro-ecological engineering approach was used to design land use activities, which are defined as the cultivation of a crop or vegetation and/or management of a herd in a particular physical environment, completely specified by its inputs and outputs (Van Ittersum and Rabbinge, 1997). Grassland activities including their fertilizer and harvest regimes were allocated to the fields, and field borders could be occupied by a hedgerow or remain unoccupied. The inputs (soil fertility, fertilization level and harvesting regime) and outputs (production of net energy for lactation, species diversity and nutrient emissions) of the land use activities were calculated from established empirical agro-ecological relations. This approach was applied to an area of 232 ha enclosed by roads, comprising three farms with an average area of 42 ha and buffer fields belonging to other land users. Random land-use was allocated to the buffer fields, which were not included in the evaluation of the performance criteria.
The indicator of market performance, landscape gross margin (GM) was calculated as the sum of revenues from milk and animal sales plus subsidies from nature conservation packages minus variable costs. The applicability of conservation packages to individual fields was assessed on the basis of plant species abundance, and harvesting and fertilization regimes. Costs per field were separated into costs related to production (harvesting by grazing or mowing and fertilizer) and transport costs. The revenues were fixed at farm level by the size of the milk quota and the herd size, which was used to calculate the returns from animal sales, with an equation derived from farm data from the dataset used by Groot et al. (2006). The costs at farm level depended on the amount of grass produced on the farm, since the required supplementary feed imports and related costs were calculated from the difference between the amount of milk produced from grass and the allowed production according to the quota. Costs for veterinary care, breeding and contracting were related to quota volume on the basis of farm data (dataset Groot et al., 2006). Farm level constraints were imposed to the proportion of grass dry matter that should be grazed (dependent on the grazing system) and the maximum allowed nitrogen fertilizer application rate (see Groot et al., 2007).
Species abundance in the grass swards and hedge rows was used as an indicator for nature conservation value (NV). The relationship between nutrient availability and average species presence in grasslands was derived on the basis of data of Oomes (1992). Landscape quality (LQ) was described as variation in the landscape and calculated by adding (1) the sum over all fields of the variance of the species number in a field and its neighbours, and (2) the degree of half-openness of the landscape, i.e. 50% of the borders occupied by hedges, calculated as the squared deviation from a completely open or closed landscape (0 or 100% occupation of the borders). Half-openness is seen as a highly typical landscape characteristic in the case area (Renting and Van der Ploeg, 2001). Environmental health (EH) was defined as the avoidance of emission of nutrients, implemented mathematically as the inverse of the emission of nutrients. The emission of nutrients was calculated from the difference between uptake of N by grass and availability of N from natural soil fertility and fertilizer application.
Results
Priorities of the non-market functions for Dutch citizens
The results of expert elicitation are illustrated in Table 3 by the super matrix for expert 01, determined according to the QFD/ANP methodology. The matrices for the other experts are available upon request. After calculating the priorities of the non-market functions of the farming systems in Northern Friesian Woodlands based on each super matrix and averaging, results indicate that ‘Environmental health’ is the most valued non-market function of the farming systems of the region for Dutch citizens followed by ‘Landscape quality’ and ‘Nature value’ (Table 4). These priorities determine the non-market benefit that society may obtain from a change of land use.
Market and non-market performances of the production alternatives
Using Landscape IMAGES the entire solution space, defined by relative changes in market and non-market benefits, was revealed. The evolutionary algorithm found 1261 land use alternatives which together approximated the contours of the solution space. Results are shown in Figure 4, where the current situation is placed in the origin of the graph. Points above the line indicating indifference in ‘social benefit’ (?US=0, see Eq. 4) represent social gains, and points below this line entail social losses. By imposing restrictions on the desired improvements of market, non-market and social benefits, we selected a number of alternatives that could serve as prototypes of desired or undesired situations of the agro-landscape. These prototypes are indicated in Figure 4, and they are described in the next section.
Utopian and dystopian landscape prototypes
Our analysis focused on entire prototypes of land use alternatives at landscape level. The prototypes were selected as ‘icons’ to follow or avoid depending on their particular market and non-market performances. In Figure 5 we show the landscapes of the different prototypes for the case study. In Table 5, the main land use characteristics and the performance of the selected landscape alternatives are summarised.
The results indicate that the current state of the region is closer to the social optimum than to the socially least preferred prototype (Figure 4). Further approaching the social optimum would involve increasing the market net benefits but slightly decreasing the non-market net benefits of the system. Market net benefit could be increased by a higher gross margin for farmers of almost 2%, even with 83% lower subsidies from agri-environmental schemes. However, non-market net benefit would be decreased since ‘environmental health’, ‘landscape quality’ and ‘nature value’ performances of the current landscape are even slightly beyond social optimum. If we assume that a decrease in non-market net benefit would not be acceptable even when associated with an increase in market net benefit, the search for better alternatives than the current situation is limited to the upper right quadrant of Fig. 4. The socially optimal landscape is then the ‘Win-win optimum’ alternative (Fig. 4), increasing both the market and non-market net benefits of the system. This alternative would represent slightly lower ‘environmental health’ than the current situation and lead to 7% increase in nitrogen emission, and slightly lower ‘nature value’. However, these losses are compensated by the higher ‘landscape quality’ of the system resulting in a higher non-market value.
The ‘Maximum non-market net benefit’ prototype demonstrates the largest changes compared to the current situation, especially for the ‘environmental health’ objective due to a drastic reduction in application and leaching of nitrogen. However, to achieve this requires a substantial sacrifice in terms of market performance of the system. Overall, the change would result in a slight increase of social net benefit.
The ‘Maximum market net benefit’ prototype is characterised by a high application of nitrogen at the expense of a large reduction in non-market net benefit, especially due to a loss in ‘landscape quality’. The absence of hedgerows contributes to this loss of landscape value (Figure 5). Overall the change would entail a negative social net benefit.
The ‘Worst social’ prototype represents dystopia: the alternative with the smallest social net benefit in comparison to the current situation. Each market and non-market indicator would deteriorate except ‘environmental health’ and ‘nature value’. This prototype is characterised by low application of nitrogen and low emission of nitrogen but also low landscape quality and very high subsidies.#p#分页标题#e#
Discussion
This paper addressed social demand for the multiple functions of agriculture as part of evaluation and design of more sustainable agro-landscapes. The approach set out to answer three subsequent questions: (1) what functions the public want from agriculture, (2) which functions can agriculture supply to the public, and (3) how can public preferences be integrated in the evaluation of agro-landscape performance. Social net benefit was introduced as the indicator which allowed linking the stated preferences in the Eurobarometer to indicators that were evaluated for given agro-landscapes using the agro-ecological landscape model Landscape IMAGES. By using regional preferences, the result of the approach is contextual. The methodology, on the contrary, is broadly applicable and although involving expert opinion, transparent by drawing on the QFD/ANP approach, multi-attribute utility theory and agro-ecological modelling.
In our case study, non-market functions are essentially environmental. Similar relative changes of market and non-market (environmental) benefits thus have the same influence on social welfare. This assumption is in agreement with social demand in Europe: 85% of people in the EU-25 and 75% in the Netherlands think that, on key issues, political decision-makers should pay the same degree of attention to environmental concerns as to economic and social factors (EC, 2005a). Moreover, according to “The Lisbon Agenda” (EC, 2005b) 63% of citizens both in the EU-25 and in the Netherlands give priority to protecting the environment over economic competitiveness, compared to 24% who disagree. Other studies concerned with the multi-attribute sustainability of Dutch dairy farming systems (Van Calker et al., 2006; 2007), using different utility functional forms, elicitation and aggregation methods, obtain similar priorities to our research. Van Calker et al. (2006; 2007) evaluated the preferences of particular groups of stakeholders (producers, consumers, industrial producers, and policy makers) for different economic, social and ecological functions, and aggregate their preferences. Obtained priorities for economic criteria, once normalized, are 0.55, and for ecological plus landscape quality, 0.45.
The results of our study indicate that there is only limited scope for improvement of the current situation in terms of social net benefit (Fig. 4). It may be that the strict environmental policies of the last decade (e.g., Henkens and Van Keulen, 2001) have been effective to reach low inputs and emissions (Table 5; cf. Groot et al., 2006). To satisfy public demand the new challenge appears to be a shift in policy focus to a more landscape-oriented emphasis. Following public demand apparently is not necessarily equivalent to pursuing long-term environmental policy goals. These severe implications of accommodating public opinion should be interpreted with care due to the small scale of the selected case study area as well as due to the choice of case study region where landscape is an important characteristic and farmers have developed high-profile activities to demonstrate their environmental engagement. Application to a larger part of the region and to different regions is desirable. Moreover, the results raise the question how accurate the outcome is in view of uncertainty and variation in inputs and relations. Such uncertainty and variation occurs not only in the definition of public demand, but also among experts on the relation between indicators as quantified in the model and public preferences aimed at the medium and long-term. Uncertainties are inherent to any methodology. The methodology we propose addresses uncertainties by aiming for as much transparency as possible. Uncertainty analysis (Rossing et al., 1994) would yield information on the relation between sources of variation and their contribution to uncertainty in the conclusions. Time series analyses would reveal to which extent the definition of public demand using the Eurobarometer as a proxy, results in consistent recommendations that can be used for consistent policies aimed at the medium and long-term.
A strong point of the methodology is its ability to demonstrate trade-offs between market and non-market benefits of land use alternatives as well as to reveal utopian and dystopian alternatives. The current situation of a landscape consisting of several farms can be compared with alternatives, including the ‘social optimum’, that is, the land use alternative that maximizes the social value of the landscape given the current technical and agronomic restriction at field and farm scales, and the economic and social environment. The best and worst options can serve as ‘icons’ to follow or avoid.
The unit of analysis in this study has been the landscape, aggregating market benefits across farms. Farms may differ in their market benefits and aggregation does not address equity among farms, which constitutes an important element for policy design in modern economies. Equity may be addressed by considering market benefits of individual farms or, in larger regions, categories of farms in terms of objectives to be optimized in the multi-dimensional Pareto-objective problem. This increases the dimensionality of the problem and renders the evolutionary algorithm less efficient. Recently, methods to overcome such problems have been put forward by Di Pierro et al. (2005) and Brockhoff and Zitzler (2006). We address this issue of aggregation in more detail in a subsequent paper.
This paper set out to investigate the suggestions of Turner et al. (2000) to assess options for sustainable and welfare-optimizing management and policy using economic valuation, integrated modelling, stakeholder analysis and multi-criteria evaluation as key methodological components. We used Landscape IMAGES and its Pareto-based multi-objective optimization approach as the integrated modelling framework, QFD/ANP to analyze stakeholder priorities and multi-attribute utility theory as a basis for multi-criteria evaluation. The methodology illustrated here at landscape level can also be applied at another level of analysis (field, farm, regional and national) and for other farming systems. The limits of application concern the availability of information on the performance of farming systems at the level and land use of interest. Rather than assessing the monetary equivalents of non-market functions as an input to the assessment, our approach reveals the economic value of non-market functions as a result, expressed as the trade-off between market and non-market net benefits. The economic value of non-market commodities thus emerges as the consequence of the stated preferences of citizens for different functions of agriculture. In view of the methodological objections put forward to various ways of measuring economic value as an input to evaluation, this approach provides an alternative that merits further investigation.
Acknowledgments
The authors thank anonymous reviewers and the editor for their constructive comments which have substantially improved the final version of this paper. We would also like to express our thanks to interviewed experts for their time and interest in this research. Finally, we would like to recognize financial support from the Spanish National Institute for Agricultural Research and Technology (INIA) through MULTIPREF project (RTA2006-00055).
|
 |
|||
| 网站地图 |
