Dominos in the dairy: An analysis of transgenic maize in Dutch dairy farming

Ecological Economics 86 (2013) 107–116

Contents lists available at SciVerse ScienceDirect

Ecological Economics journal homepage: www.elsevier.com/locate/ecolecon

Analysis

Dominos in the dairy: An analysis of transgenic maize in Dutch dairy farming Rolf A. Groeneveld a,⁎, Justus Wesseler b, Paul B.M. Berentsen c a b c

Wageningen University, Environmental Economics and Natural Resources Group, Hollandseweg 1, 6706 KN Wageningen, The Netherlands Technische Universität München, Center of Life and Food Sciences Weihenstephan, Alte Akademie 8, 85354 Freising, Germany Wageningen University, Business Economics Group, Hollandseweg 1, 6706 KN Wageningen, The Netherlands

a r t i c l e

i n f o

Article history: Received 10 November 2011 Received in revised form 15 November 2012 Accepted 16 November 2012 Available online 17 December 2012 Keywords: Maize Herbicide resistance Genetic modification Coexistence Dairy farming European Union The Netherlands

a b s t r a c t EU member states require farmers growing transgenic maize to respect a minimum distance from fields with non-transgenic maize. Previous studies have theoretically argued that such minimum distance requirements may lead to a so-called ‘domino effect’ where farmers who want to grow transgenic maize are forced to grow the non-transgenic variety and in turn impose the same constraints on their neighbors. This article applies a spatially explicit farm model to a dairy region in the Southern Netherlands to assess how farmers growing non-transgenic maize limit other farmers' potential to grow transgenic herbicide-resistant maize. The results indicate that the minimum distance requirements can severely limit the benefits from herbicide resistant maize. Having different land use options in one farm, however, enables dairy farmers to grow transgenic maize despite having one or more neighbors growing non-transgenic maize. We also find that the share of the domino effect in the overall impact of minimum distance requirements decreases with the density of farmers not growing transgenic maize. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Transgenic introduction of herbicide resistance in agricultural crops has the potential to drastically reduce herbicide use in crops such as maize, oil seed rape, sugar beet, and cotton (Phipps and Park, 2002). Concerns over cross-pollination of non-transgenic crops by transgenic crops, however, have led to coexistence measures such as minimum distance requirements or buffer zones between transgenic and nontransgenic crops (Commission of the European Communities, 2009). EU member states, for instance, require farmers who want to grow a transgenic crop to maintain a distance from plots with the nontransgenic variety that varies from 25 m in The Netherlands to up to 600 m in Luxemburg for maize (Commission of the European Communities, 2009). The scientific literature on coexistence has so far focused on appropriate minimum distance requirements and the economic implications of coexistence. Soregaroli and Wesseler (2005) investigate the implications of minimum distance requirements on adoption, whereas Beckmann et al. (2006) expand the analysis by considering a set of coexistence policies and Beckmann et al. (2011) pay particular attention to transaction costs. These studies show minimum distance requirements discriminate against smaller farms. Beckmann and Wesseler (2007), Furtan et al. (2007), and Munro (2008) investigate the economic implications of coexistence from a property rights ⁎ Corresponding author. Tel.: +31 317 482009; fax: +31 317 484933. E-mail addresses: [email protected] (R.A. Groeneveld), [email protected] (J. Wesseler), [email protected] (P.B.M. Berentsen). 0921-8009/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ecolecon.2012.11.011

perspective. Beckmann and Wesseler (2007) show coexistence requirements provide incentives for regional agglomeration. Furtan et al. (2007) investigate the incentives for non-transgenic farmers to organize themselves in landscape clubs under the Canadian property right system. Munro (2008) stresses the importance of coexistence policies in homogeneous landscapes for achieving welfare improving results if consumers express a positive willingness to pay for non-transgenic food, while in heterogeneous landscapes coexistence might be possible without additional policies. A number of simulation studies have investigated the possibility for coexistence considering the specific pollination properties of plants as well as coexistence policies. Belcher et al. (2005) and Ceddia et al. (2007, 2009, 2011) use simulations to illustrate the possibility that under certain conditions transgenic crop introduction will make it impossible to maintain zero tolerance levels for non-transgenic crops. Considering the impossibility of avoiding adventitious presence many countries have introduced positive tolerance levels of approved transgenic crops for labeling of non-transgenic crops (Gruère et al., 2009; Scatasta et al., 2007). A disadvantage of many simulation models is the assumption of homogeneous landscapes. Coexistence considering threshold levels may be more achievable in heterogeneous landscapes. For instance, Gray et al. (2011) find that threshold levels of 0.9% can be maintained with 100% certainty in oilseed rape in Australia without any minimum distance requirements. Similarly, Ceddia et al. (2007) conclude in their simulation study for oil seed rape in a homogeneous landscape that planting density has to increase beyond 27% to cause a problem in meeting the 0.9% threshold level. Also Sanvido et al. (2008) cast serious doubt whether or not a 0.9% threshold level will

108

R.A. Groeneveld et al. / Ecological Economics 86 (2013) 107–116

be reached in heterogeneous landscapes. Nevertheless, many coexistence policies within the EU ask for mandatory minimum distance requirements (Beckmann et al., 2006). Demont et al. (2008b) study two types of spatial ex ante coexistence regulations for oilseed rape (OSR) in Central France: (1) rigid minimum distance requirements between fields with transgenic OSR and fields with non-transgenic OSR; and (2) flexible buffer zones where farmers grow non-transgenic OSR that is sold as transgenic OSR. The authors demonstrate that the impact of minimum distance requirements can be stronger than one may expect at first sight. When minimum distance requirements are combined with giving non-transgenic OSR priority over transgenic OSR, planting nontransgenic OSR in a particular plot of land may have consequences not only for other plots within the minimum distance from the non-transgenic plot, but also for plots further away. This is the case if the minimum distance forces users of the neighboring plots (who would want to grow transgenic OSR) to grow non-transgenic OSR, which in turn forces their neighbors to grow non-transgenic OSR, and so on. For example, the authors show in their study region (which comprises 1508 plots of 4233 ha in total) that when initially about 548 ha is intended for transgenic OSR and another 548 ha for non-transgenic OSR, a minimum distance rule of 50 m can limit the eventual adoption of transgenic OSR to 186 ha: a reduction of 362 ha, or 66% of the intended transgenic OSR area. Of this reduction, however, only 310 ha can be attributed to the direct impact from non-transgenic OSR plots on transgenic OSR plots. The remaining 52 ha, or about 14% of the total impact in hectare terms, is due to the impact from farmers who initially intended to grow transgenic OSR, but are forced to grow non-transgenic OSR, on other farmers. The authors refer to this phenomenon as the “domino effect”: the impact of a plot of a non-transgenic crop on adoption of transgenic varieties may be partly due not to its direct impact on neighboring plots, but to the land use choices of its neighbors, and their neighbors, and so on. Demont et al. (2009) further investigate the coexistence issue under different minimum distance requirements, different adoption rates, and different cropping densities. Their analysis shows that lower planting densities could lead not only to lower absolute, but also lower proportional impacts: when the area of transgenic and nontransgenic OSR is half that assumed by Demont et al. (2008b), the estimated impact is 91 ha, or 33%, of the intended 280 ha. Moreover, a smaller share of that impact (10 ha, or 11% of the total impact in hectare terms) is due to the domino effect. The ratio between transgenic and non-transgenic farmers also matters: when the initial allocation features a larger area of transgenic OSR for every hectare of non-transgenic OSR, the absolute impact of the minimum distance requirement will increase, whereas the proportional impact will decline. The share of the impact that can be ascribed to the domino effect will increase with the proportion of initial transgenic OSR farmers. The results found by Demont et al. (2008b, 2009) and authors using similar approaches (Belcher et al., 2005; Ceddia et al., 2007, 2009) depend on a number of assumptions caused by the data they had available and the specific case they analyze. Three of these assumptions, all of which are commonly made in spatial simulation models because of the poor availability of detailed farm level data, are particularly relevant for this paper. First, it is often assumed that the farms included in the analysis grow only one crop, and their only alternative for the transgenic variety is the non-transgenic variety of that crop. Farms, however, typically produce a mixture of crops. In landscapes dominated by dairy farms, smart spatial planning of fodder crops may avoid coexistence conflicts between transgenic and non-transgenic varieties of crops such as maize and fodder beet. Second, because of the aforementioned data problem, the spatial simulation models do not assume farm income maximization. This is relevant as also differences in transportation costs and the on-farm value of transgenic and non-transgenic crops affects field allocation decisions, particularly on dairy farms. Third, spatial simulation models

do not include the possibility of communication between farmers. Beckmann et al. (2011) discuss the implications of communication between farms from a theoretical point of view and show that this can reduce the direct effect as well as the domino effect while Skevas et al. (2010) present a case study showing this may indeed be the case. The objective of this article is to address the three aforementioned assumptions made in the literature by analyzing potential cultivation of transgenic herbicide-resistant maize in a Dutch dairy farming region. We focus on rigid minimum distance requirements, because the literature has so far shown that these regulations have the most serious impact on overall adoption of transgenic crops. First, we analyze how the impact of minimum distance requirements depends on the additional benefits of transgenic herbicide-resistant silage maize (as compared to non-transgenic silage maize). Moreover, we separate the direct effect of rigid minimum distances, which is the impact directly attributed to farmers growing non-transgenic maize in the initial allocation, from the domino effect, which is the impact that these farmers may have through their effect on other farmers' land use choices. This gives insight into how dairy farms' variety in fodder crops may limit the impact of minimum distance requirements and the role of the domino effect in this respect. We consider not only a scenario where several farmers refrain from growing transgenic maize, but also a scenario where only one farmer does so for comparison with the results by Demont et al. (2008b, 2009). This provides insight into how the share of the domino effect in the total impact of minimum distance requirements could vary with the density of farmers growing only the non-transgenic variety. Further, this allows us to identify the impact a single non-transgenic farmer may have not only on the crop choice of his direct neighbors but also, via the domino effect, on other farms in the region. This modification is of some interest as farmers can decide not to grow transgenic maize for various reasons, including low expectations from transgenic herbicideresistant maize (Areal et al., 2011), moral objections (Gaskell et al., 2000; Myskja, 2006), regulations for organic agriculture certification, or a price premium received for “GM-free” milk (Venus and Wesseler, 2012). Second, as we have access to detailed farm level data we identify the implications of the above-mentioned scenarios with a farm income maximizing model. Third, we assume farmers know about the intention of their neighbors and try to minimize potential conflicts by allocating crops to reduce the number of border lines between transgenic and non-transgenic crops. We focus on a region in the province of North-Brabant, The Netherlands, where we apply a spatially explicit dairy farming model to simulate land use decisions of 213 dairy farms. The model includes the feed demand of dairy cows as well as the farm's production, purchase, and sale of grass, maize, and feed concentrates. Similarly to Demont et al. (2008b), the model simulates farmers' land use decisions given their intentions and given the limits imposed by land use decisions of their neighbors. After all farmers have made their decisions given their knowledge about their neighbors' land use decisions, each farmer's knowledge about his or her neighbor's land use decisions is updated, and the model is run with the updated knowledge. This procedure allows us to trace the spatial effects of minimum distance requirements throughout the study area. This paper proceeds as follows. Section 2 explains the spatial effects of minimum distance requirements intuitively, in order to provide more insight into the mechanisms at work. Section 3 explains the structure of the model, the data used, the modeling procedure, and the scenarios in the analysis. Section 4 presents the results. Section 5 discusses the main assumptions of the analysis and compares the results to those found so far in the literature. Section 6 concludes. 2. Spatial Effects of Minimum Distance Requirements For the following we introduce abbreviations for readability and simplicity. Farmers not willing to cultivate transgenic maize, who

R.A. Groeneveld et al. / Ecological Economics 86 (2013) 107–116

therefore cultivate non-transgenic maize, will be indicated NTMFs (non-transgenic maize farmers). On the other hand, farmers willing to cultivate transgenic maize will be indicated TMFs (transgenic maize farmers). Note that we distinguish the two types of farmers by their growing intentions, and not by what they actually grow. So whereas we assume that an NTMF will not grow transgenic maize, it is possible that a TMF grows non-transgenic maize instead of transgenic maize if he is forced to do so by the minimum distance requirements. Further, we assume herbicide-resistant silage maize will be made available for famers. Herbicide-resistant crops offer a number of environmental benefits (Wesseler et al., 2011) and they are among the most popular transgenic crops among farmers (James, 2010), because they provide a substantial amount of non-pecuniary benefits to farmers (Marra and Piggott, 2006). Moreover, ex-ante studies show that they can generate farm level benefits among EU member states ranging from about €15 to €240 per ha per year for grain maize (Wesseler et al., 2007). Similar results have been reported by Venus et al. (2011) for silage maize and Demont et al. (2008a) for grain maize. We assume benefits from herbicide-resistant silage maize consist entirely of savings in herbicide use. To understand how a variety in land uses could limit the impact of minimum distance requirements, consider the configuration depicted in Fig. 1. In this example there are three farms with two plots each, one of which is located close to the farm (the ‘home plot’) and another further away (the ‘field plot’). For the sake of simplicity, assume that plots cannot be used partially. In other words, we assume farmers need to produce grass on one plot, and maize on another. Dairy farms prefer to produce grass on the plot located near the farm, and to produce maize on the other plot which is further away from the farm (Berentsen et al., 2000). This is plausible because grassland is harvested more frequently than maize, and when grassland is grazed by dairy cows farmers need to move the cows between the plot and the milking barn. Lastly, the minimum distance required is indicated by the dotted line in Fig. 1. We assume that any plot that lies fully or partly within the minimum distance from a plot where non-transgenic maize is grown is prohibited from growing transgenic maize. If the NTMF in Fig. 1 grows maize on the field plot, then TMF 1 cannot grow transgenic maize on the field plot. This effect will be referred to as the direct effect in the remainder of this article. TMF 1 has two options to deal with the direct effect: either grow non-transgenic maize on the field plot, or grow grass on the field plot and transgenic maize on the home plot. The first option will also affect TMF 2's ability to grow transgenic maize. We will refer to this effect as the domino effect (cf. Demont et al., 2008b, 2009) in the remainder of this article. Note, however, that the second option will allow TMF 2 to grow transgenic maize on both plots, so that no domino effect takes place. Another important observation is that TMF 1 is more likely to grow non-transgenic maize on the field plot if the benefits foregone

from not growing transgenic maize are lower than the costs from allocating transgenic maize production to a plot that would have been the preferred plot for grassland. This suggests that the domino effect, and hence the overall impact of minimum distance requirements, will be smaller when transgenic maize is more profitable compared to non-transgenic maize. Moreover, the share of the domino effect in the overall impact of minimum distance requirements will depend on the number of farmers not wanting to grow transgenic maize in the landscape, as illustrated in Fig. 2. In this figure NTMFs are indicated by black cells, TMFs are indicated by gray cells, and the minimum distance between transgenic and non-transgenic crops is indicated by the dotted line. TMFs who are prohibited directly from growing transgenic maize by an NTMF are indicated by dark gray cells. These cells thus indicate the direct effect. The light gray cells indicate the domino effect: like the dark gray cells, light gray cells indicate TMFs prohibited from growing transgenic maize by neighbors growing non-transgenic maize. Unlike the dark gray cells, however, light gray cells indicate TMFs who have no NTMF neighbors. They are forced to grow nontransgenic maize because their TMF neighbors are also forced to do so, either directly by an NTMF neighbor or indirectly through minimum distances imposed on TMF neighbors. In Fig. 2a, a single NTMF has a direct effect on eight neighboring plots, whereas the domino effect covers the rest of the landscape and thus has a larger share in the overall impact. In Fig. 2b, where we have five NTMFs, the entire landscape is also excluded from growing transgenic crops, but this time the majority of plots is affected by a direct effect rather than a domino effect. 3. Materials and Methods We analyze the impact of minimum distance requirements on the possibility of transgenic maize cultivation, and the mechanisms explained in Section 2, with a spatially explicit farm management model. The model is based on a dairy farm model developed by Berentsen and Giesen (1995), using additional insights from Nijssen and Van Scheppingen (1995) and Groeneveld et al. (2005), and updated for price levels of 2008 (ASG, 2008). The model is applied to dairy farms in a region in the Dutch province of North-Brabant, and run for different combinations of minimum distances and additional benefits from transgenic maize. 3.1. Model For each dairy farm f in the study area the model maximizes the gross margin, defined as the difference between revenues and variable costs. This definition of gross margin excludes fixed costs that are exogenous to the decision to cultivate transgenic maize, such as maintenance and depreciation of buildings. The objective function

a Home plot NTMF

Field plot NTMF

Home plot TMF 1

Field plot TMF 1

Home plot TMF 2

Field plot TMF 2

Fig. 1. Hypothetical configuration of one farmer wanting to grow conventional maize (the NTMF) and two farmers wanting to grow transgenic maize (the TMFs). The dotted line indicates the minimum distance to be kept between transgenic and non-transgenic maize.

109

b

Fig. 2. Direct effect (dark gray cells) and domino effect (light gray cells) caused by (a) a single NTMF (black cell); and (b) five NTMFs (black cells). The dotted line indicates the minimum distance between transgenic and non-transgenic crops.

110

R.A. Groeneveld et al. / Ecological Economics 86 (2013) 107–116

of the model for a given farm f, given the choices of all other farms, is therefore: (

)

max If ¼ rK f þ ∑ τ v T fv − ∑ ∑ ovp Avp −∑ pv P fv v

p∈Of v∈Q

v

∀f ;

ð1Þ

be produced on a plot that has designation ‘transgenic’. Likewise, if a feed type v is a member of Y, then that feed type can only be produced on a plot that has designation ‘non-transgenic’. Feed types without any coexistence concerns, which in this model concerns grass, can be cultivated on all plot designations, i.e. ‘transgenic’, ‘non-transgenic’, and ‘neutral’. This allows the model to combine grass and any kind of maize on a given plot, while separating transgenic from non-transgenic maize. This restriction is expressed as follows:

where If denotes the gross margin of farm f in € per year, r denotes net revenues (sales of milk and meat minus costs of health care and maintenance) in € per dairy cow per year; Kf denotes the number of dairy cows on farm f; τv denotes the selling price of feed type v in € per ton dry matter; Tfv denotes the sales of feed type v in ton dry matter per year; Of denotes the set of plots p used by farm f; Q denotes the set of feed types produced by dairy farms; ovp denotes the costs of producing feed type v on plot p in € per ha; Avp denotes the area of feed type v on plot p in ha; pv denotes the purchase price of feed type v in € per ton dry matter; and Pfv denotes the purchases of feed type v in ton dry matter. The model includes three feed types, namely grass, maize, and feed concentrates, of which grass and maize are grown, maize can be purchased as well as sold, and feed concentrates can only be purchased. Each farm is assumed to maximize its gross margin under restrictions regarding the cattle's feed requirements, available land, and possible restrictions on production of transgenic maize. The cattle's feed intake must supply sufficient energy, protein, and fiber. Moreover, the rumen degraded protein balance must at least be positive (Tamminga et al., 1994). These requirements are included in the following equation:

When considering minimum distances between transgenic and non-transgenic crops it is important to distinguish two situations: one where both plots involved are used by the same farm, and one where one plot is used by one farmer and the other plot by another. In the first case, the decision on land cover lies with one and the same person. We assume farmers keep a distance between transgenic and non-transgenic maize, even within their own farm, in order to maintain access to both silage maize markets. This means that a farmer faces the following restriction:

∑ niv Sfsv ≥δfis K f

Llp ≤1−Ll′ p′

v

∀f ; i; s;

ð2Þ

where niv denotes the nutritional value of feed type v with regard to criterion i; Sfsv denotes the supply on farm f in season s of feed type v in ton; δfis denotes the demand on farm f for feed nutritional element i in season s. There is a limit to how much feed a cow can digest, included through the following equation: ∑ ζ v Sfsv ≤ψs K f v

∀f ; s;

ð3Þ

where ζv denotes the fill value of feed type v; and ψs denotes the feed intake capacity per cow in season s in ton. Feed production is defined by Q fv ¼ ∑ dv Avp p∈Of

∀f ; v∈Q ;

ð4Þ

where Qfv denotes production on farm f of feed type v in ton per year; and dv denotes production of feed type v in ton per year per ha. Differences between production and supply must be resolved by either purchases or sales of feed: P fv −T fv ¼ ∑ Sfsv −Q fv s

∀f ; v:

ð5Þ

Only maize can be purchased as well as sold, whereas concentrates can only be purchased. The land area on a given plot p devoted to feed type v is restricted by the size of the plot, ap (in ha): ∑ Avp ≤ap

v∈Q

∀p :

ð6Þ

Considerations of coexistence between transgenic and non-transgenic maize are included as follows. Let X denote the set of transgenic feed types, in this case transgenic maize, and Y the set of their nontransgenic counterparts. Moreover, assume that any given plot must be designated either as ‘transgenic’, ‘non-transgenic’, or ‘neutral’. Denote these designations by the index l, and let Llp be a binary variable denoting whether plot p is used for land designation type l. Finally, let Vl be the set of feed types associated with land designation type l. Then if a feed type v is a member of X, then that feed type can only

∑ Avp ≤ap Llp

∀l∈X∪Y; p:

v∈Vl

ð7Þ

Land designation types are mutually exclusive: ∑ Llp ≤1 l

∀p:

ð8Þ

′

′

∀l∈X; l ∈Y; p∈Of ; p ∈Np ∩Of ;

ð9Þ

where Np denotes the set of plots p′ within a distance from plot p equal to or smaller than the minimum distance between transgenic and non-transgenic maize. In the second case, plot p is owned by farmer f, and plot p′ is owned by another farmer, so that the land cover type of p′ is exogenous to f's decision. Let parameter λl′p′ ∈ {0,1} reflect farmers' knowledge whether plot p′ is used for land use type l′. The restriction thus becomes Llp ≤1−λl′ p′

∀l∈X; l′ ∈Y; p∈Of ; p′ ∈Np 5Of :

ð10Þ

3.2. Data This analysis requires detailed spatial information on ownership of plots and distance between plots. The dataset used in this study was originally developed for a land reallotment scheme in 1996 (Schmitz, 1996) in the province of North-Brabant, The Netherlands (Fig. 3). Although the dataset is rather old, it describes a sufficiently realistic situation for the purpose of this study, which is to analyze farmers' land use decisions under different minimum distances. The dataset contains 3049 plots (10,023 ha), of which 2845 plots (7799 ha) are inside the study area. There are 713 farms in the study region, 241 of which have dairy cows. Other agricultural activities in the area include beef cattle farming, arboriculture and arable farming. Schmitz (1996) classifies the farms in the study area according to the fraction of farm size equivalents devoted to several agricultural activities. In this system, the fraction of agricultural activities in the total farm size determines the type of farm. This classification yields classes of farms with, for instance, 60–80% of farm size equivalents devoted to arable farming, 20–40% devoted to horticulture and less than 20% devoted to other, agricultural activities. The model assumes a rational profit-maximizing Dutch dairy farmer. This excludes farm types that do not focus on dairy farming, such as mixed farms, arable farms, and arboriculture, because these activities are not relevant to the topic of this paper. It also excludes farms that focus on dairy farming but are much smaller than a genuine Dutch dairy farm, because such farms are unlikely to be managed

R.A. Groeneveld et al. / Ecological Economics 86 (2013) 107–116

111

In model Not in model

Fig. 3. The study region. Black plots indicate plots included in the model; white plots indicate plots not used by dairy farms and hence excluded from the model.

on a commercial basis. We therefore confine our analysis to farms with more than half the farm size equivalents devoted to dairy farming and with more than 20 dairy cows, because these farms can reasonably be assumed to use all or most of their land for dairy farming. This selection includes 213 farms, together using 1235 plots with a total area of 4746 ha. Average farm size is 22.3 ha (standard deviation 9.1 ha). Data on market prices of feed, fertilizer, hired labor and other farm inputs come from ASG (2008). Maize production costs about €1261 per ha, €35 of which is spent on herbicides (ASG, 2008). Data on the average number of young cattle per dairy farm come from de Jong et al. (2006). Lastly, data on feed quality of grass and maize are taken from CVB (2008). Demarcation of the study area always raises the question what to assume with respect to the plots at its fringe. Moreover, some farms in the database own plots within as well as plots outside the study area, whereas the exact location of plots outside the study area is unknown. To arrive at a conservative estimate of the potential area of transgenic maize we assume these plots will not be used for transgenic maize, but are also too far from the study area to impose any restrictions on other plots. We also assume that plots not used by the farms in the database are not used at all for maize production. 3.3. Modeling Procedure Analyzing the impact of minimum distance requirements in the study region requires that the model be run in a number of phases: (1) the first phase to identify the spatial land use allocation if there were no restrictions on land use following from minimum distance requirements; (2) the second phase to identify the direct impact that the NTMFs will have on their neighbors given a certain minimum distance; and (3) the third phase to identify the impact that the NTMFs will have on other farms than their neighbors, which is the domino effect as it takes place via the neighbors. In the first phase we maximize the sum of all farms' gross margin (If) with λlp = 0 for all land use types and plots. Because each farm owns several plots, however, it is likely that a given farm can attain the same maximum gross margin by several different land use allocations. Typically, however, mixed-integer programming models like the model in this article give only one solution that maximizes the objective function, even if the same objective can be attained by many different solutions. Which of these many solutions is provided depends on such factors as the starting solution of the optimization and the algorithm used. This introduces an arbitrariness in the NTMFs' land use allocation that could also make NTMFs' impact on their TMF neighbors arbitrary.

In theory one could aim to identify all possible allocations that lead to the same objective function. It would then be possible, for instance, to draw randomly one allocation from that population for the next phase, and to repeat this procedure many times as a Monte Carlo analysis. With the number of farms and plots included in the model, however, this is not realistic. Even if only 50 of the farms had two profit-maximizing allocations, this would still yield 2 50 = 10 15 different possible land use allocations. Therefore, in this study we aim to arrive at a conservative estimate of the impact of the minimum distance requirement, so we select the land use allocation that minimizes the number of conflicts between TMFs and NTMFs, under the same gross margin of each individual farm. This is similar to assuming that farmers aim to reduce possible conflicts with their neighbors as long as it does not impede their maximum gross margin. If it is impossible to solve a conflict without one of the two farmers having to accept a lower gross margin, farmers let the law decide so the legal preference of non-transgenic over transgenic crops becomes relevant (Beckmann et al., 2011). 1 The mandatory registration of planting decisions for transgenic crops by farmers in publicly available databases or the information requirement asked for by regulators supports the empirical relevance of the approach (see e.g. Beckmann et al., 2011; Skevas et al., 2010). Mathematically this procedure is as follows. After finding the maximum gross margin that can be attained by each farm, we minimize the area of plots (M) with non-transgenic maize that have plots with transgenic maize within their vicinity:   min M ¼ ∑ ap C p s:t:If ¼ If p

∀f ;

ð11Þ

where Cp ∈ {0, 1} is a binary variable that denotes whether plot p (1) contains non-transgenic maize and (2) has transgenic maize within its minimum distance and hence is a plot with a potential conflict; and If⁎ denotes the gross margin of farm f in the first run. The value of Cp is determined as follows: C p ≥∑ Llp −F p l∈Y

∀p;

ð12Þ

where Fp denotes whether the vicinity of plot p is free of transgenic maize. Note that under the objective function in Eq. (11) the model 1 Judges might apply the “newcomer principle” as mentioned by Munro (2008) and non-transgenic famers might have a better chance at court. But there also exists doubts whether or not the “newcomer principle” will apply in the case of GM crops if they are viewed as a continuum of technical change (Rothbard, 1982).

112

R.A. Groeneveld et al. / Ecological Economics 86 (2013) 107–116

will strive to set Cp = 0 by either removing the non-transgenic maize from plot p (i.e. Llp = 0 for l ∈ Y) or by removing the transgenic maize from all plots within its minimum distance (i.e. Fp = 1). The value of Fp is determined as follows:

€35 per ha that farmers currently spend on herbicides. We therefore consider four different levels of cost savings, namely €5, €15, €25, and €35 per ha. 4. Results

F p ≤1−∑ Llp′ l∈X

∀p; p′ ∈Np :

ð13Þ

In the second phase, we update the farmers' knowledge with the land use allocation found in the end of the first phase. With this new knowledge we again maximize the farms' gross margin and select from all possible land use allocations with that gross margin the allocation with the smallest area of plots (M) with non-transgenic maize that have plots with transgenic maize within their vicinity. The result of the second phase gives the land use allocation and gross margin after the NTMFs' neighbors have adapted to the limitations imposed by the NTMFs. Hence, we can interpret the differences between the first and second phase as the direct effect of the minimum distance requirement. In the third phase we again update the farmers' knowledge with the land use allocation found in the end of the second phase, which now reflects the most conservative estimate of the conflicts between the NTMFs' TMF neighbors and other TMFs. With this new knowledge we again maximize the farms' gross margin and select the land use allocation with the lowest value of M. The third phase is repeated in order to trace the impacts of the TMFs on each other through the landscape until there are no more plots with transgenic maize within the minimum distance from a plot with non-transgenic maize. This procedure is similar to the procedure used in Demont et al. (2008b, 2009). Hence, we can interpret the difference between the second phase's result and the third phase's result as the estimated domino effect. 3.4. Scenarios We seek to analyze how the impact of minimum distance requirements, and the share of the direct effect and domino effect in this impact, will depend on the number of NTMFs in the study area, the extent of the minimum distances, and the private benefits from transgenic herbicide-resistant maize. With regard to the number of NTMFs in the study area we consider two scenarios: (1) the “Single NTMF” scenario, where one of the 213 dairy farms in the study area is not willing to grow transgenic maize; and (2) the “Many NTMFs” scenario, where 106 of the 213 farms in the study area are not willing to grow transgenic maize. In both scenarios the results depend on which farm exactly is an NTMF. Therefore, the “Single NTMF” scenario is analyzed for 213 different situations, in each of which another farm is the one NTMF in the study area. As regards the “Many NTMFs” scenario, considering all possible distributions of 106 NTMFs over 213 farms would require an unrealistically large number of model runs. Therefore, we consider 200 different random distributions of 106 NTMFs over 213 dairy farms. As regards minimum distance requirements, we focus on three different minimum distances between transgenic and non-transgenic maize applied in EU member states. The smallest minimum distance is 25 m, which is the minimum distance required in The Netherlands. The most widely used distance is 200 m, which is the minimum distance required in Lithuania, Latvia, Portugal, Romania, and Slovakia. The largest minimum distance found among EU member states is 600 m, which is the distance required in Luxembourg (Commission of the European Communities, 2009). As regards the private benefits of transgenic herbicide-resistant maize, we assume that its main benefit lies in savings in herbicide use. Under this assumption, farmers' net cost savings from transgenic herbicide-resistant maize (i.e. savings in herbicide expenses minus the mark-up for herbicide-resistant maize seed) will not exceed the

When no transgenic maize is available, the model predicts that the 213 farms in the model will generate a total annual gross margin of about €12.8 million per year, with an average of €59,995 per year per farm (standard deviation €21,034). The total area of maize production is 605 ha, or 12.7% of the total area. 90 farms will not grow any maize at all. These farms have a high cattle density and favorable parcellation, so that they use all their land for grassland. This grassland is then used for grazing in summer (which is economically preferable to feeding roughage) and for producing a minimum of grass silage for the winter season, and silage maize is bought from outside the farm. The farms that grow maize when no transgenic maize is available have on average 4.9 ha of maize. If no minimum distance requirements were in place while transgenic maize is introduced, and one of the 213 farmers in the model is an NTMF, the area of transgenic maize could be up to 550 ha, depending on the cost savings from transgenic maize (Table 1). Because we assume that plots outside the study area cannot be used for transgenic maize, there will still be a fair amount of non-transgenic maize grown outside the study area. As cost savings from transgenic maize increase, however, maize production outside the study area will be moved into the study area where growing transgenic maize is allowed. Total benefits of transgenic maize can be up to €17,699 per year, or about €83 per farm over all 213 farms. For individual farms, however, the annual cost savings can be much higher than that, namely up to €517. 4.1. Potential Effect of Minimum Distance Requirements A single NTMF can have a substantial effect on his or her neighbors through minimum distance requirements (Table 2). Because the impact of a single NTMF depends on the chosen NTMF among the 213 dairy farms in the region, Table 2 reports the average and largest 2 impact over all possible NTMFs. The results demonstrate that a single NTMF can limit the area of transgenic maize by up to 210.8 ha, depending on the cost savings from transgenic maize, the extent of the minimum distance requirement, and the location of the NTMF. The projected impact of a single NTMF on transgenic maize area increases with the minimum distance (which is to be expected), but decreases with cost savings. The average reduction of transgenic maize area is mostly about 10–20 times lower than the maximum reduction because about 90 farms do not grow any maize at all. Gross margin forgone because of minimum distance requirements is projected to be at most €6607 in total. In the scenario where the gross margin forgone is this high, 35 TMFs are affected, so that the average costs per affected TMF are about €189. At an individual farm, the losses can be up to €517. Both average and maximum impact of a single NTMF on gross margin increase with the minimum distance as well as cost savings from transgenic maize. The projected impact of a single NTMF on overall adoption of transgenic maize depends strongly on which of the 213 farms in the dataset is the single NTMF. First of all, remember that 90 farms will not grow maize at all: in the scenarios where one of these farms is the single NTMF adoption of transgenic maize will not be limited by the minimum distance regulations because there will be no non-transgenic maize to keep away from. In the other scenarios the NTMF does grow maize, which is then non-transgenic maize, but only few farms would have a serious impact on adoption of transgenic maize. Fig. 4 shows the number of farms for each impact range (expressed as the reduction in 2 Note that the smallest impact found of a single NTMF is zero, in all combinations of cost savings and minimum distance considered.

R.A. Groeneveld et al. / Ecological Economics 86 (2013) 107–116

113

Table 1 Total% area of grass and maize, and increase in total gross margin, in the absence of minimum distance requirements under four different levels of cost savings in the Single NTMF scenario (average and maximum over all possible NTMFs). No transgenic maize

Area transgenic maize (ha) Area non-transgenic maize (ha) Area grass (ha) Total gross margin (1000 €) Increase gross margin (€)

Average Maximum Average Maximum Average Maximum Average Maximum Average Maximum

Cost savings

– – 605.5 – 4140.8 – 12,779 – – –

benefits from transgenic maize vis-à-vis a situation where no minimum distance requirements are in place), for the three minimum distances considered and under herbicide cost savings of €35 per ha. Fig. 4 shows that between 105 and 143 farms, depending on the minimum distance requirements, could be the single NTMF in the region without having any impact on adoption of transgenic maize. An additional 20 to 60 farms will have an impact smaller than €250 per year: for all minimum distances considered this is the majority of farms. One should note, however, that under a minimum distance of 600 m about 15% of the farms considered would have an impact of more than €1500 per year if it were the single NTMF in the region. When interpreting the results from the Many NTMFs scenario (Table 3) it is important to keep in mind that under this scenario there are 106 NTMFs in the study area. This implies that the reductions in transgenic maize area and gross margin refer to the 107 other farms, which are TMFs. The highest income forgone found in the 200 scenarios considered is €10,364. In this scenario 64 TMFs will be affected by the minimum distance requirement, which corresponds to about €162 per affected TMF. Individually the gross margin forgone can again be up to €517. 4.2. Share of Domino Effect in Total Effect As discussed earlier, many farms can be an NTMF without limiting others' ability to grow non-transgenic maize, even under a minimum distance requirement of 600 m. When the minimum distance requirement does have an impact, the average share of the domino effect in the total impact on adoption of transgenic maize will be less than 50% when the minimum distance requirement is 25 m or 200 m and there is only one NTMF (Table 2). Even for these distances, however, the share of the domino effect can occasionally be much higher than that, and even larger than the direct effect. In the Many NTMFs scenario the share of the domino effect in the total impact

€5/ha

€15/ha

€25/ha

€35/ha

490.1 492.4 115.4 127.9 4140.8 4140.8 12,781 12,781 2426 2437

500.1 502.4 105.4 117.8 4140.8 4140.8 12,786 12,786 7398 7433

506.2 508.6 99.3 111.7 4140.8 4140.8 12,791 12,791 12,441 12,500

547.7 550.2 87.4 99.7 4111.2 4112.4 12,797 12,797 17,615 17,699

will be substantially lower than in the Single NTMF scenario (Table 3). Also note that the share of the domino effect in the total effect is higher under a minimum distance requirement of 200 m than under a minimum distance requirement of 25 m or 600 m. A possible explanation is that there are two mechanisms at work. The first mechanism is that a more stringent minimum distance requirement makes it more difficult for farms to grow transgenic maize, making it more likely that the impact of the minimum distance requirement is propagated throughout the study area. The second mechanism is that under a more stringent minimum distance requirement the share of the study area covered by the direct effect is also larger, leaving fewer plots potentially subject to the domino effect. The latter mechanism is particularly likely in the Many NTMFs scenario because this includes more NTMFs, and hence a larger area covered by the direct effect. 5. Discussion The intuitive analysis in Section 2 suggests that (1) the variety in land use types on dairy farms may limit the impact of minimum distance requirements between transgenic and non-transgenic maize on the overall area of transgenic maize; and (2) the share of the domino effect in the total impact of minimum distance requirements on farmers growing transgenic crops will decline with the number of farmers willing to grow only the non-transgenic variety. The modeling results support these expectations. The impact of minimum distance requirements on the area of transgenic herbicideresistant maize in the study region will decline with the potential cost savings from this maize variety. This suggests that as farmers have more to lose from switching to the non-transgenic variety, they will be more willing to allocate their land use such that it allows for the cultivation of the transgenic variety. Moreover, we see that the share of the domino effect will be generally smaller when there are many farmers growing non-transgenic maize than when only one farmer grows

Table 2 Total impact of minimum distance requirements on benefits from transgenic herbicide-resistant maize under different levels of net cost savings in the Single NTMF scenario (average and maximum over all possible NTMFs). The share of the domino effect in the total impact is given between brackets. Cost savings €5/ha Reduction in area transgenic maize (ha)

25 m 200 m 600 m

Gross margin forgone (€)

25 m 200 m 600 m

Average Maximum Average Maximum Average Maximum Average Maximum Average Maximum Average Maximum

4.8 59.2 8.2 91.2 49.3 210.8 25 297 43 471 244 1036

(27%) (98%) (35%) (97%) (61%) (100%) (28%) (98%) (34%) (97%) (61%) (99%)

€15/ha

€25/ha

€35/ha

2.9 48.6 5.8 64.3 24.3 188.2 51 794 99 1117 382 2856

2.6 (15%) 48.4 (99%) 5.0 (24%) 62.5 (99%) 22.7 (51%) 188.7 (99%) 74 (17%) 1267 (97%) 146 (25%) 1750 (98%) 597 (49%) 4740 (98%)

2.3 (12%) 45.9 (94%) 4.3 (18%) 61.6 (95%) 21.7 (48%) 194.3 (98%) 91 (13%) 1712 (94%) 173 (19%) 2362 (97%) 750 (45%) 6607 (98%)

(15%) (99%) (25%) (98%) (51%) (99%) (18%) (96%) (27%) (97%) (51%) (98%)

R.A. Groeneveld et al. / Ecological Economics 86 (2013) 107–116

Number of farms that would have the impact if they were the single NTMF

114

160 Minimum distance 25m

140

200m

120

600m

100 80 60 40 20 0

0

0 250

250 500

500 750

750 1000

1000 1250

1250 1500

1500

Impact of a single NTMF measured in forgone benefits from transgenic maize ( /year) Fig. 4. Number of farms for eight ranges of impact on overall adoption of transgenic maize (expressed as the reduction in benefits from transgenic maize vis-à-vis a situation where no minimum distance requirements are in place), for the three minimum distances considered and under herbicide cost savings of €35 per ha.

non-transgenic maize. This confirms earlier findings by Demont et al. (2009) that the share of the domino effect in the total effect of minimum distance requirements will increase with the planting density of transgenic crops. When interpreting the results it is important to consider a number of issues with respect to the data used and the assumptions made. Since the publication of the dataset in 1995 Dutch dairy farms have grown by about 50% in terms of cultivated area per farm and livestock per farm (van der Meulen et al., 2011). These farms achieve such growth by acquiring plots from farmers who leave the business, and through land reallotment schemes. In general, it can be expected that the upscaling process is likely to reduce the impact of minimum distance requirements. Previous studies have shown that large farms are better able to deal with minimum distance requirements (Consmüller et al., 2010; Skevas et al., 2010; Soregaroli and Wesseler, 2005). Moreover, as larger farms will try to consolidate their land, conflicts between plots belonging to different dairy farmers become less likely. In our analysis we assume the most stringent of ex ante spatial coexistence measures: transgenic maize is banned from an entire plot if any part of that plot lies within the minimum distance from a plot where non-transgenic maize is being grown. As Demont et al. (2008b, 2009) have shown, buffer strips can limit the spread of transgenic pollen at much lower costs than rigid minimum distance requirements. Moreover, we assume away the possibility that farmers partition plots in order to grow, for instance, grass within the minimum distance and transgenic maize just outside the minimum distance. Such partitioning, however, can only be made endogenous in mixed-integer programming models if one knows the exact location

of the line along which the plot may be partitioned. This means that the results of each phase in the modeling procedure described in this article should be fed into a GIS model to draw a new contour around plots with non-transgenic maize, identify the potential partitions, and feed that back into the optimization model. This would be a substantial extension of the model procedure that we leave for future work. We also assume that reduced expenses on herbicides are the only advantages of transgenic herbicide-resistant maize vis-à-vis non-transgenic maize. Although there could be more advantages to herbicide-resistant maize, such as a higher production and reduced externalities from herbicide use, the expenses saved on herbicides are the most important direct benefits. Hence cost savings could exceed the €35 per hectare considered in this paper. Among the 1814 plots in the dataset not used by any of the farms in the model, 1471 plots are used by farms for which we cannot fully rule out the possibility that they might grow maize. These plots are used by, for instance, arable farms, or farms with only a minor stake in dairy farming. If these plots were all used for non-transgenic maize production, the potential area of transgenic maize, even under a minimum distance of 25 m, would be at most 139 ha. Under a minimum distance of 600 m the presence of non-transgenic maize on these plots rules out all transgenic maize. It is much more likely, however, that maize is grown on none of these plots than that it is grown on all of them. As far as we can check, the model seems to represent the actual situation fairly well. When no transgenic maize is available, the model predicts that the 213 farms in the model will generate a total annual gross margin of about €12.8 million, with an average of €1206 per dairy cow (standard deviation €131). LEI (2010) suggests that farms

Table 3 Total impact of minimum distance requirements on benefits from transgenic herbicide-resistant maize under different levels of net cost savings in the Many NTMF scenario (average and maximum over 200 possible spatial distributions of NTMFs). The share of the domino effect in the total impact is given between brackets. Cost savings €5/ha Reduction in area transgenic maize (ha)

25 m 200 m 600 m

Gross margin forgone (€)

25 m 200 m 600 m

Average Maximum Average Maximum Average Maximum Average Maximum Average Maximum Average Maximum

72.0 176.8 147.2 240.7 216.4 297.4 405 930 766 1199 1089 1484

(13%) (52%) (17%) (44%) (9%) (27%) (12%) (47%) (15%) (41%) (8%) (27%)

€15/ha

€25/ha

€35/ha

56.8 (10%) 159.1 (54%) 123.8 (14%) 266.3 (39%) 215.0 (10%) 301.9 (31%) 1010 (10%) 2589 (50%) 2032 (12%) 3946 (35%) 3238 (8%) 4487 (29%)

54.0 149.2 110.2 158.7 206.1 290.5 1561 4081 3146 4537 5359 7435

54.4 (10%) 154.0 (60%) 109.2 (13%) 164.0 (35%) 221.7 (13%) 303.4 (33%) 2075 (10%) 5436 (54%) 4136 (11%) 6085 (33%) 7469 (10%) 10,364 (30%)

(11%) (57%) (14%) (39%) (11%) (29%) (10%) (52%) (12%) (36%) (9%) (29%)

R.A. Groeneveld et al. / Ecological Economics 86 (2013) 107–116

of the size class found in the study area had an average gross margin of €1894 per dairy cow in 2008, which is substantially larger than the model's estimate. This difference is mainly caused by a difference in revenues: the model predicts average revenues of €2353 (standard deviation €111), whereas the observed revenues are €3139 per dairy cow. Predicted costs (€1147 with standard deviation €87) are quite close to their observed counterparts (€1245). The costs are much more relevant to our analysis than the revenues, because revenues consist mainly of sales of milk and meat, which are fixed for the greater part because of the dairy quota system, whereas the costs depend mainly on the farm's land use decisions. As regards the land use decisions, about 12.8% of the dairy farming area in the model base run is used as maize land, which is close to the average figure among dairy farms in The Netherlands of 14.8% (LEI, 2010). As indicated in Section 3.3, it is likely that a farmer can attain the same gross margin with several different land use allocations. This introduces some arbitrariness in the land use allocations, which we minimize by focusing on a conservative estimate of the NTMF's impact on his neighbors. In reality, however, farmers may not strive to minimize conflict with their neighbors for several reasons, such as the transaction costs of doing so, or personal animosities between neighbors. Moreover, variables not included in the model, such as local variation in soil quality or the costs of switching from one land use type to another, may be important in farmers' land use allocations. 6. Conclusion Rigid minimum distance requirements for transgenic herbicideresistant maize can seriously limit the potential area of transgenic crops. Given the reductions in herbicide use possible with herbicideresistant crops, this effect may have environmental as well as farmlevel financial implications. Due to the spatial nature of minimum distance requirements, however, the scale of their implications may depend heavily on the spatial configuration of plots in an agricultural area. Our analysis shows that minimum distance requirements can substantially limit, or even completely eliminate, these benefits. This depends, however, on the number of farms that prefer to grow the non-transgenic variety, the cost savings from herbicideresistant maize, and the minimum distance between transgenic and non-transgenic maize. The impact of a minimum distance requirement can propagate throughout the landscape as farmers are prevented from growing transgenic maize and thereby impose similar restrictions on other farmers. Our results suggest that a single farmer who wants to grow non-transgenic maize may cause other farmers to forgo up to €517 per year in gross margin. The results presented in this paper suggest that the benefits from growing transgenic maize in Dutch dairy farms will be modest compared to the overall gross margin generated in this sector. In the study area the potential net benefits of herbicide-resistant maize will be between €2437 and €17,699 for the entire region, or €11 and €83 per farm, depending on how much farmers save in expenses on herbicides from using herbicide-resistant maize. For individual farms the net benefits can be much higher, but remain in the order of magnitude of a few hundreds of Euros per year. The non-pecuniary benefits as reported by Marra and Piggott (2006) have not been considered. These benefits may be substantially larger than the savings in herbicide costs and have to be added to the direct and domino effects reported. We also find that potentially a substantial part of the total impact on the gross margin generated in the area can be due to an indirect effect of the initial non-transgenic plot, which is also referred to as the ‘domino effect’. The importance of this effect depends on the number of farmers not willing to adopt transgenic maize, the minimum distance between transgenic and non-transgenic maize, and the cost savings from herbicide-resistant maize, but also on the land use allocations chosen by individual farmers. Because those choices

115

can be rather arbitrary, variables unobserved by researchers or policy makers may be important determinants of the eventual allocations, making the impact of minimum distance requirements rather unpredictable. There are two silver linings to this message. First, our results demonstrate that flexibility in land use decisions may reduce the impact of minimum distance requirements. Second, although the overall impact of minimum distance requirements will increase with the number of farmers not willing to grow non-transgenic maize, it may also become more predictable, because the direct effect becomes more prominent than the domino effect. The message to policy makers is that within heterogeneous landscapes land allocation flexibility may be able to resolve most, if not all, of the issues caused by minimum distance requirements. Flexible coexistence rules have been proposed in earlier studies (Demont et al., 2008b, 2009). Our study shows that regions dominated by farms with diversity in land uses may be better able to deal with minimum distance requirements than regions dominated by monocultures, even if only a minimum of communication takes place between farmers. This might explain why so far disputes among farmers cultivating GM crops in Europe with neighbors have not been reported and why farmers cultivating GM crops do not report coexistence policies as being a problem. Supporting the coordination of land use allocations has the potential to resolve coexistence conflicts entirely. Hence, the introduction of GM crops as well as GM-free labeling systems may not pose a problem for ensuring the coexistence of GM and non-GM crop cultivation but in some cases a single farmer can have a strong effect on the land use choice of other farmers. This may result in potential conflicts between neighbors. These conflicts will become more severe with an increase in the extra costs caused. The latest development on maize pests such as the Western Corn Root Worm and the European Corn Borer supported by global warming where transgenic maize offers a costeffective solution indicates that the costs of reduced access to the technology may increase. Policy makers can help reducing conflicts by also supporting communication and in particular information sharing about land use allocation among farmers. Acknowledgments This work was supported by European Framework VI Specific Target Research Project “Transcontainer” (023018) (http://www. transcontainer.wur.nl). Justus Wesseler acknowledges research leading to these results has received funding from the European Union's Seventh Framework Programme (FP7/2007-2013) under grant agreement nr KBBE-2011-5-289157 (EU-PRICE project: Practical Implementation of Coexistence in Europe, http://price-coexistence.com). References Areal, F.J., Riesgo, L., Rodríguez-Cerezo, E., 2011. Attitudes of European farmers towards GM crop adoption. Plant Biotechnology Journal 9 (9), 945–957. ASG, 2008. Kwantitatieve Informatie Veehouderij 2008–2009. Animal Sciences Group, Lelystad, The Netherlands. Beckmann, V., Wesseler, J., 2007. Spatial dimension of externalities and the Coase Theorem: implications for coexistence of transgenic crops. In: Heijman, W. (Ed.), Regional Externalities. Springer, Berlin, pp. 215–234. Beckmann, V., Soregaroli, C., Wesseler, J., 2006. Coexistence rules and regulations in the European Union. American Journal of Agricultural Economics 88 (5), 1193–1199. Beckmann, V., Soregaroli, C., Wesseler, J., 2011. Coexistence of genetically modified (GM) and non-modified (non GM) crops: are the two main property rights regimes equivalent with respect to the coexistence value? In: Carter, C., Moschini, G.C., Sheldon, I. (Eds.), Genetically Modified Food and Global Welfare. Emerald Group Publishing, Bingley, UK, pp. 201–224. Belcher, K., Nolan, J., Phillips, P.W.B., 2005. Genetically modified crops and agricultural landscapes: spatial patterns of contamination. Ecological Economics 53 (3), 387–401. Berentsen, P.B.M., Giesen, G.W.J., 1995. An environmental-economic model at farm level to analyse institutional and technical change in dairy farming. Agricultural Systems 49 (2), 153–175. Berentsen, P.B.M., Giesen, G.W.J., Renkema, J.A., 2000. Introduction of seasonal and spatial specification to grass production and grassland use in a dairy farm model. Grass and Forage Science 55 (2), 125–137.

116

R.A. Groeneveld et al. / Ecological Economics 86 (2013) 107–116

Ceddia, M.G., Bartlett, M., Perrings, C., 2007. Landscape gene flow, coexistence and threshold effect: the case of genetically modified herbicide tolerant oilseed rape (Brassica napus). Ecological Modelling 205 (1–2), 169–180. Ceddia, M.G., Bartlett, M., Perrings, C., 2009. Quantifying the effect of buffer zones, crop areas and spatial aggregation on the externalities of genetically modified crops at landscape level. Agriculture, Ecosystems and Environment 129 (1–3), 65–72. Ceddia, M.G., Bartlett, M., De Lucia, C., Perrings, C., 2011. On the regulation of spatial externalities: coexistence between GM and conventional crops in the EU and the ‘newcomer principle’. The Australian Journal of Agricultural and Resource Economics 55 (1), 126–143. Commission of the European Communities, 2009. Implementation of national measures on the coexistence of GM crops with conventional and organic farming, Commission Staff Working Document accompanying Report from the Commission to the Council and the European Parliament on the coexistence of genetically modified crops with conventional and organic farming. Commission of the European Communities, Brussels. Consmüller, N., Beckmann, V., Petrick, M., 2010. An econometric analysis of regional adoption patterns of Bt maize in Germany. Agricultural Economics 41 (3–4), 275–284. CVB, 2008. CVB Table Booklet Feeding of Ruminants, CVB-Series No. 43. Centraal Veevoederbureau, Lelystad, The Netherlands. de Jong, O., van Oostende, M., van Schie, T., 2006. Handboek Melkveehouderij: Nieuwe editie 2006. Roodbont Uitgeverij, Zutphen. Demont, M., Cerovska, M., Daems, W., Dillen, K., Fogarasi, J., Mathijs, E., Muška, F., Soukup, J., Tollens, E., 2008a. Ex ante impact assessment under imperfect information: biotechnology in new member states of the EU. Journal of Agricultural Economics 59 (3), 463–486. Demont, M., Daems, W., Dillen, K., Mathijs, E., Sausse, C., Tollens, E., 2008b. Regulating coexistence in Europe: beware of the domino-effect! Ecological Economics 64 (4), 683–689. Demont, M., Dillen, K., Daems, W., Sausse, C., Tollens, E., Mathijs, E., 2009. On the proportionality of EU spatial ex ante coexistence regulations. Food Policy 34 (6), 508–518. Furtan, W.H., Güzel, A., Weseen, A.S., 2007. Landscape clubs: co-existence of genetically modified and organic crops. Canadian Journal of Agricultural Economics 55 (2), 185–195. Gaskell, G., Allum, N., Bauer, M., Durant, J., Allansdottir, A., Bonfadelli, H., Boy, D., de Cheveigne, S., Fjaestad, B., Gutteling, J.M., Hampel, J., Jelsoe, E., Jesuino, J.C., Kohring, M., Kronberger, N., Midden, C., Nielsen, T.H., Przestalski, A., Rusanen, T., Sakellaris, G., Torgersen, H., Twardowski, T., Wagner, W., 2000. Biotechnology and the European public. Nature Biotechnology 18 (9), 935–938. Gray, E., Ancev, T., Drynan, R., 2011. Coexistence of GM and non-GM crops with endogenously determined separation. Ecological Economics 70 (12), 2486–2493. Groeneveld, R.A., Grashof-Bokdam, C.J., van Ierland, E.C., 2005. Metapopulations in agricultural landscapes: a spatially explicit tradeoff analysis. Journal of Environmental Planning and Management 48 (4), 527–547. Gruère, G.P., Carter, C.A., Farzin, Y.H., 2009. Explaining international differences in genetically modified food labeling policies. Review of International Economics 17 (3), 393–408.

James, C., 2010. ISAAA Report on Global Status of Biotech/GM Crops. International Service for the Acquisition of Agri-biotech Applications (ISAAA), Ithaca, New York. LEI, 2010. Farm accountancy data network. Agricultural Economics Research Institute. Marra, M., Piggott, N.E., 2006. The value of non-pecuniary characteristics of crop biotechnologies: a new look at the evidence. In: Alson, J.M., Just, R.D., Zilberman, D. (Eds.), Regulating Agricultural Biotechnology. Springer Press, New York, pp. 145–178. Munro, A., 2008. The spatial impact of genetically modified crops. Ecological Economics 67 (4), 658–666. Myskja, B., 2006. The moral difference between intragenic and transgenic modification of plants. Journal of Agricultural and Environmental Ethics 19 (3), 225–238. Nijssen, J.M.A., van Scheppingen, A.T.J., 1995. Verkaveling in de melkveehouderij, Publicatie. Experimental station for cattle production, Lelystad, The Netherlands. Phipps, R.H., Park, J.R., 2002. Environmental benefits of genetically modified crops: global and European perspectives on their ability to reduce pesticide use. Journal of Animal and Feed Sciences 11 (1), 1–18. Rothbard, M.N., 1982. Law, property rights, and air pollution. Cato Journal 2 (1), 55–99. Sanvido, O., Widmer, F., Winzeler, M., Streit, B., Szerencsits, E., Bigler, F., 2008. Definition and feasibility of isolation distances for transgenic maize cultivation. Transgenic Research 17 (3), 317–335. Scatasta, S., Wesseler, J., Hobbs, J., 2007. Differentiating the consumer benefits from labeling of GM food products. Agricultural Economics 37 (2–3), 237–242. Schmitz, I.M.J., 1996. Cultuurtechnische inventarisatie De Leijen-Oost en -West, Rapport. Staring Centrum, Wageningen, The Netherlands, p. 35. Skevas, T., Fevereiro, P., Wesseler, J., 2010. Coexistence regulations and agriculture production: a case study of five Bt maize producers in Portugal. Ecological Economics 69 (12), 2402–2408. Soregaroli, C., Wesseler, J., 2005. Minimum distance requirements and liability: implications for coexistence. In: Wesseler, J. (Ed.), Environmental Costs and Benefits of Transgenic Crops. Springer Press, Dordrecht, pp. 165–182. Tamminga, S., Van Straalen, W.M., Subnel, A.P.J., Meijer, R.G.M., Steg, A., Wever, C.J.G., Blok, M.C., 1994. The Dutch protein evaluation system: the DVE/OEB-system. Livestock Production Science 40 (2), 139–155. van der Meulen, H.A.B., de Bont, C.J.A.M., Agricola, H.J., van Horne, P.L.M., Hoste, R., van der Knijff, A., Leenstra, F.R., van der Meer, R.W., de Smet, A., 2011. Schaalvergroting in de land- en tuinbouw: Effecten bij veehouderij en glastuinbouw, LEI-rapport. LEI Wageningen UR, The Hague. Venus, T., Wesseler, J., 2012. Bereits doppelt so viel “ohne Gentechnik”- Milch als Biomilch in Deutschland: Welche Bedeutung hat GVO-freie Milch für unsere Milchwirtschaft? Deutsche Molkerei Zeitung 133 (2), 24–26. Venus, T., Casadamon, R., Soregaroli, C., Wesseler, J., 2011. Comparison of Bt and Non-Bt Maize Cultivation Gross Margin: A Case Study of Maize Producers from Italy, Spain and Germany. Futuragra, Rome. Wesseler, J., Scatasta, S., Nillesen, E., 2007. The Maximum Incremental Social Tolerable Irreversible Costs (MISTICs) and other benefits and costs of introducing transgenic maize in the EU-15. Pedobiologia 51 (3), 261–269. Wesseler, J., Scatasta, S., Fall, E.H., 2011. Environmental benefits and costs of GM crops. In: Carter, C., Moschini, G., Sheldon, I. (Eds.), Genetically Modified Food and Global Welfare. Emerald Group Publishing, Bigley, UK, pp. 173–199.