Conservation contracts for supplying Farm Animal Genetic
Resources (FAnGR) conservation services in Romania
1,2WARWICK WAINWRIGHT, 2KLAUS GLENK, 2FAICAL AKAICHI AND 3DOMINIC
1 Grant Institute, School of Geosciences, University of Edinburgh, Kings Buildings, West Mains
Road, Edinburgh, EH9 3JW
2 Land Economy, Environment and Society Group, SRUC, Kings Buildings, West Mains Road,
Edinburgh, EH9 3JG
3 Global Academy of Agriculture and Food Security, The Royal (Dick) School of Veterinary
Studies, The Roslin Institute Easter Bush Campus, Midlothian, EH25 9RG
Correspondence: Warwick Wainwright, Land Economy, Environment and Society Group, SRUC,
Kings Buildings, West Mains Road, Edinburgh EH9 3JG
Keywords: conservation contracts; choice experiment; farm animal genetic resources; agri-
This paper describes a choice experiment (CE) administered to explore farmer preferences for
conservation agreements to conserve rare breeds among a sample of 174 respondents in Transylvania
(Romania). The study site was chosen due to the prevalence of small-scale and extensive farm systems
threatened by a changing policy environment that is increasing the scale and intensity of production
units. Agreement attributes included length of conservation contract (5 or 10 years); scheme structure
(community or individual managed conservation programme), and scheme support (application
assistance or farm advisory support). A monetary attribute that reflects compensation for scheme
participation allows the assessment of farmers’ willingness to accept (WTA) for different contracts.
Results suggest 89% of respondents would be willing to farm with rare breeds; cattle and sheep being
the most popular livestock option; 40% of farmers were reportedly farming with endangered breeds.
However, only 8% were likely to qualify for funding support under current requirements. WTA
estimates reveal minimum annual compensation values of €167 and € 7 per year respectively, for bovine
and ovine farmers to consider enrolling in a contract. These values are comparable to Romanian Rural
Development Programme (RDP) support offered to farmers keeping rare breeds of € 200 and € 10 per
year for bovine and ovine farmers respectively. Our estimates of scheme uptake, calculated with
coefficient values derived from the CE, suggest rare breed conservation contracts are considered
attractive by Romanian farmers. Analysis suggests meeting farmer preferences for non-monetary
contractual factors will increase participation.
Farm Animal Genetic Resources (FAnGR) diversity underpins resilient agricultural systems and
need to be part of any sustainable intensification (SI) strategy to meet rising demand for livestock
products (Eisler et al., 2014). However, concentration on elite breeding lines has reduced genetic
variation in many commercial breeds whilst marginalising traditional breeds whose value is often
poorly understood (Ahtiainen and Pouta, 2011; FAO, 2015).
SI strategies should include investments to maintain genetic variation across a range of breeds
(including rare breeds) to ensure adaptive capacity in livestock systems. This is particularly important
when considering profound demographic and environmental changes facing the agri-food sector
including population growth, land scarcity and climate change (FAO, 2017). Equally important, but less
often articulated in decision making, are the cultural and heritage attributes embodied in rare breeds
(Gandini and Villa, 2003; Zander et al., 2013). Markets often fail to reflect these values, which can be
substantial but difficult to measure. Breed genetic diversity is therefore undersupplied by markets and
there is a need to explore policy interventions to counter market failure.
While contractual schemes for rare breed conservation are present in Europe, many are often poorly
targeted (Kompan et al., 2014; Bojkovski et al., 2015). Targeting incentives towards small-holder and
extensive farm systems may improve scheme efficiency and uptake, given their lower opportunity cost
of conservation (Naidoo et al., 2006). This paper explores rare breed conservation contracts in
Transylvania (Romania), where the average farm size is only 3.4 ha and the economic efficiency per
farm (as measured by standard monetary output of agri-products per holding) is significantly lower
than the European Union (EU) average (Popescu et al., 2016).
Traditional farm systems in Transylvania are under pressure from development of more intensive
farm systems that are changing the scale and nature of practices (Sutcliffe et al., 2013, 2015). A focus
on improved efficiency is at the expense of the supply of public goods, including breed diversity. Some
42% of livestock breeds in Romania are classified as ‘at-risk’, according to the United Nations Food
and Agricultural Organisation (FAO) definition of an ‘at-risk’ breed (Draganescu, 2003). This figure
may be an underestimate since population estimates for many Romanian breeds are unknown (FAO,
2018). There is therefore a need to develop targeted policy responses that aid conservation by balancing
an intensification agenda with incentives for the supply of other non-market goods and services.
Farm scale drivers of diversity loss are often assumed to relate solely to the lower productivity of
traditional livestock breeds (Cicia et al., 2003). While income forgone is a key factor to establish the
cost of incentive-based schemes, other factors also motivate farm business decisions, and may be
particularly relevant in a semi-subsistence farming context. Such non-financial motives may include
tradition, community relations, professional pride and independence (Gasson, 1973; Ilbery, 1983;
Burton et al., 2008). It is therefore necessary to identify how such attributes might influence the design
of conservation programmes and farmer willingness to supply diversity. Other potential technical and
institutional barriers-to-entry (i.e. requirements for breed genealogical records) also warrant exploration
in this context.
We used a choice experiment (CE) survey to elicit farmer preferences for supplying (rare breed)
conservation under alternative contracts forms. CEs are a stated preference technique where individual
preferences for attributes of a good or service are elicited using surveys that mimic hypothetical
scenarios – in this case conservation contracts (Louviere et al., 2000). The paper adds to the literature
on farmers’ willingness to participate in incentive-based schemes (Ducos et al., 2009; Ruto and Garrod,
2009; Broch and Vedel, 2010; EspinosaGoded et al., 2010; Greiner, 2015; Lienhoop and Brouwer,
2015) but focuses on the neglected issue of the cost of conserving FAnGR in small-holder and extensive
farm systems. The paper aims to investigate farmer preferences for rare breed conservation contracts,
including the minimum compensation required for enrolment in a conservation scheme. We explore
whether some of the heterogeneity associated with contractual choices is systematically associated with
farm or farmer characteristics.
The paper is structured as follows. Section 2 presents background to the CE design and case study
site. Section 3 reports the analysis of choice data. Section 4 provides discussion of the design of rare
breed conservation programmes, and Section 5 provides conclusions.
2.1 Case study: Romania
As an EU member state, Romania’s agricultural policy is structured and supported in an agreed
Rural Development Programme (RDP 2014-2020), which includes a support measure (M10.2, art 28)
for rearing endangered livestock breeds under EU Regulation 1305/2013 (MARD, 2014). Uptake for
this RDP option is anticipated to be low due to farmer difficulties in meeting EU standards to qualify
for subsidy payments (Page, 2015, personal communication). Data on uptake rates are not yet available,
but previous work has found that 70% of Romanian farmers experienced difficulties meeting EU
environmental standards for payments under the Common Agricultural Policy (CAP) (Fischer et al.,
2012). It is therefore important to explore whether such barriers persist for farmers in small-scale and
extensive systems, as this could reduce participation. Equally important is to measure whether voluntary
agri-environmental stewardship (AES) measures, specifically M10.2, match farmer preferences and
expectations for scheme design and rewards.
Much of the study site (Figure 1) is situated in the foothills of the Carpathian Mountains and features
an undulating topography with low nutritional pastures (Mikulcak et al., 2013). Part of the area (Tarnava
Mare) is classified as high nature value (HNV) farmland. Traditional agricultural practices are common
in this area, as is the presence of many small scale and semi-subsistence farms (Page et al., 2011).
Mechanised systems are the mainstay for medium to large farms, though are much less common. The
site is characterised by high levels of rural poverty, with average household incomes below the national
average (Gherghinescu, 2008).
We surveyed livestock keepers across 5 counties (Sibiu, Brasov, Mures, Cluj and Alba). The
sampling frame was based on local farmer information held by village mayors, with further random
sampling of farms. The survey was administered from June to August (2015).
Figure 1: Land cover map of the survey area with inset map of Romania. Sampling locations are shown
by yellow stars.
2.2 Questionnaire design and administration
The survey consisted of four sections. The first asked about the farm business including livestock
species and breeds, farm size, and traits farmers deem most important when considering choice of breed.
In the second, respondents were asked if they receive AES payments and whether they were aware of
financial support for rare breeds and ever considered applying for this support. The third part of the
questionnaire included the CE. Two CE versions were created - one for ovines and one for bovines.
Farmers answered either one or both depending on whether they were keeping ovines, bovines, or both.
After the CE tasks were completed, respondents were asked to state their motivations for their choices
in the CE, and this information was used to identify genuine choices from protest bids; the latter
subsequently being removed from the analysis. Respondents were also asked about their preference
concerning scheme remittance (i.e. individual or community payment). The fourth section collected
socio-economic information including respondent age, gender, educational attainment and household
2.3 Choice experiment design
In CEs, respondents are asked to repeatedly choose from a number of options that differ in their
attributes or characteristics following an experimental design. The CE elicited individual preferences
using hypothetical contract choice sets requiring farmers to upkeep rare breeds from a list of breeds
proposed by the Romanian Government for support under the 2014-2020 RDP measure (see Appendix
2 for list of eligible breeds). Farmers were advised that the breeding of animals must be pedigree to
qualify for further subsides on offspring (i.e. non-random mating). Each choice task consisted of two
alternative contracts and a ‘none’ option to embody the voluntary nature of the conservation scheme.
Attributes and their levels used to describe the conservation contract were determined in a multi-stage
process involving literature review, expert consultations and pilot testing.
Each contract option consisted of four attributes (Table 1). The first three attributes described
contract length (CL); scheme support (SS); and structure of scheme (SOS). Choice of attributes drew
on empirical work suggesting their importance in AES scheme design (Ruto and Garrod, 2009;
Christensen et al., 2011; Greiner, 2015). A final monetary attribute (COS) represented an annual
payment to farmers (per animal) and took four different levels. The monetary attribute in local currency
(Lei per year) was based on a percentage (10%, 30%, 60% and 100%) of the proposed monetary reward
outlined in the RDP; the premise being that some farmers may be willing to accept (WTA) a lower
reward, depending on contract design. The choice tasks were differentiated based on the livestock
species. For bovine (cattle, horses and buffalo) and ovine farmers (sheep and goats) the choice tasks
were similar except for the value of the monetary attribute, which reflected the relative support normally
given to different species under current RDP conditions.
Table 1: Attributes and attribute levels used in the CE including relevant coding and a prioir
- 5 years
+ 10 years
- Basic assistance to complete the scheme
+ Additional advisory support throughout the
scheme (e.g. additional training for animal
Structure of scheme
- Individually managed conservation scheme
+ Community managed conservation scheme
- Bovines = 90; 270; 530; 890 Lei / year
- Ovines = 5; 15; 25; 45 Lei / year
Choice set design was optimised according to prior information on the distribution of random
parameters to improve statistical efficiency - i.e. reduction in sample size needed to achieve statistical
significance (Crabbe and Vandebroek, 2011). Prior information concerning the parameter coefficients
was estimated from results of the pilot data that was collected in situ to ensure the attributes were
relevant to participants. A D-efficient experimental design optimised for the random parameter logit
(RPL) model was formulated using NGene (Metrics, 2012). The final CE comprised 16 choice sets
which were blocked into 4 blocks of four choice tasks each in a bid to reduce the cognitive burden for
respondents (Hensher, 2006). Figure 2 shows a typical choice task presented to respondents.
support (e.g. extra
(per animal / per year)
I prefer: Option A Option B Nothing
❑ ❑ ❑
Figure 2: A typical choice task shown to respondents
2.4 Econometric specification of choice models
Respondent choices were modelled with reference to Lancaster's theory of value (Lancaster, 1966)
and Random Utility Theory (McFadden, 1973; Luce, 2005). For a general description see (Holmes et
al., 2017). The multinomial logit (MNL) model (McFadden, 1973) was used in the first iteration of this
analysis. This assumes the random component of the utility of the alternatives is independent and
identically distributed (i.i.d.). A key limitation of the MNL is that preferences for attributes of different
alternatives are assumed to be homogenous across individuals. Subsequently, the RPL model was
employed in the second iteration because the approach is more advanced and takes into account
heterogeneity of the parameter values among respondents. The RPL relaxes key assumptions that
constrain the use of conditional logit models, namely independence of irrelevant alternatives - iia
(Hensher et al., 2005). Under a RPL specification, the utility a respondent i derives from an alternative
j in each choice situation t is given by:
Where Uijt is a utility maximising individual, Xijt is a vector of observed attributes associated with
each contract option (i.e. contract length, scheme support, structure of scheme and price) plus the socio-
economic characteristics of respondents, and εijt is the random component of the utility that is assumed
to have an iid value distribution. Conditional on the individual specific parameters βi and error
components εi the probability that individual i chooses alternative j in a particular choice task n is
Note, choices for bovine and ovine farmers were modelled separately to explore preference
heterogeneity between both groups. The empirical model was estimated using the econometric software
NLOGIT 5.0. For a full description of the model specification, see Appendix 3.
3.1 Respondent characteristics
A total 174 respondents were surveyed - 116 were bovine farmers and 81 were ovine farmers (note
45 respondents kept both ovines and bovines). The means and standard deviation of multiple individual
specific variables is outlined in Table 2. There were later used as interaction terms in the choice model
to determine significant covariates that help to explain respondent choice. The mean age of participants
was from 40-49 years, with highest education levels of either secondary school or college. Fewer female
respondents featured in our sample as more males are generally employed in agriculture (European
Commission, 2012). Average monthly household income was reported to be in the range of €181 to
€362; lower than the national average but anticipated at the sample site (Page et al., 2011). The primary
income for most farmers was EU subsides, while sale of milk and meat products were generally
secondary and tertiary sources, respectively. Some 40% of farmers claimed to be farming with a rare
breed from a list of ‘at risk’ breeds, while 32% were enrolled in AES measures. Only 21% of respondents
were aware of RDP support for rare breeds whilst only 8% actually met the EU’s criteria to qualify for
Table 2: Summary of individual specific variables (with means) and relevant interpretation
1, if male, 0 otherwise
Categorical (1=<20, 2=20-29, 3=30-39, 4=40-49, 5=50-
59, 6=60-69, 7=over 70 years)
Categorical (1=secondary, 2=college, 3=degree &
Categorical (1=<€45, 2=€45-€90, 3=€91-€181, 4=€181-
€362, 5=€362-€678, 6=>€679)
Categorical (1=1-2 ha, 2=3-6 ha, 3=7-20 ha, 4=>20 ha)
1, if farming with rare breeds, 0 otherwise
1, if farmer would consider farming with rare breed in the
future, 0 otherwise
1, if farmer is currently enrolled in an agri-environment
scheme (AES), 0 otherwise
1, if farmer aware of RDP support for rare breeds, 0
Categorical (1=if farmer prefers 100% individual cash
benefits from a conservation programme, 2=50% cash
benefit, 50% community in-kind benefit, 3=100%
community in-kind benefit)
1, if farmer is registering livestock in a genealogic
register, 0 otherwise
1, if farmer is keeping cross breeds for yield
improvement, 0 otherwise
References: a(National Institute of Statistics, 2013) b(National Institute of Statistics, 2015) c(Popescu et al., 2016)
3.2 Farm characteristics
To determine how intensification may threaten traditional farming systems and breed diversity,
respondents were asked to detail how their farming practices have changed over the preceding 10 years
(Figure 3). Increases to dairy cattle herd size were reported by 52% of respondents. Of the 20% of our
sample that reported manual hay cutting, 74% reported this to be either stable or increasing; a clear
response to EU incentives that reward small-holders for the activity. Mechanical hay cutting was
reported to be increasing (67% of respondents) and some 54% of farmers also stated their sheep herd
size was increasing.
Figure 3: Reported change in farming practices over the last 10 years from respondents.
To investigate whether willingness to participate in a (rare breed) conservation programme was
linked to preferences for farm animal species, respondents were asked both livestock species kept and
their interest in joining a conservation scheme. Pigs were the most frequently kept farm animal followed
by cattle and sheep (Table 3). The highest number of breeds reported was for pigs, while buffalo had
the least. The prevalence of breed diversity varied across species. For instance, the main breed kept for
each farm species ranged from 83% (Romanian Buffalo) to 37% (Large White pig). Across the sample,
89% of farmers registered interest in joining a rare breed conservation programme, of which cattle
(52%) and sheep (39%) were the most popular species. Least popular species were goats (11%); horses
(13%) and buffalo (14%). Of interest is the low preference for conserving rare horse breeds given their
popularity in the Romanian farming context. This may suggest rare horse breeds do not match farmer
preferences for horse breed characteristics and hence are undersupplied.
Table 3: Sample summary of farm animal characteristics, breed abundance and farmer interest in
farming with a rare breed
Incidence of farm
animal in sample
Most popular breed
Farmers stating interest
in farming with rare
Large White (37%)
Romanian Buffalo (83%)
Baltata Romanesca (61%)
Unknown mix (51%)
* Percentage abundance was calculated as the number of farm animals in our sample that correspond to a specific breed
Livestock-keepers in different countries prefer different breed attributes. Respondents were
asked to rank livestock attributes by importance for breed selection. In Figure 4 radar charts indicate
different preferences between rare breed and commercial breed keepers for some attributes. Here,
farmers were asked to rank multiple breed attributes in terms of importance on a 1-8 scale (1 being most
important, 8 being least). The proportion of farmers selecting each attribute (for ranks 1, 2 and 3) is
shown. Yield was the most important attribute for both rare breed and commercial breed keepers.
Adaptability was ranked 2nd for farmers keeping rare breeds, while disease and parasitic resistance was
ranked 3rd. For commercial breed keepers, yield was also ranked 2nd and adaptability 3rd. This suggests
productive traits are considered most important by both farmer groups, but they differ in perceived
importance of non-productive traits. This supports work suggesting rare breed adaptability
characteristics play an important role within the livestock sector not matched by commercial breeds
(Leroy et al., 2018).
Figure 4: Radar charts showing ranked importance of livestock attributes according to farmer preference.
The charts reveal the proportion (%)of farmers who chose each attribute in 1st 2nd and 3rd rank. Key, CT =
cultural tradition; DPR = disease and parasitic resistance; VB = veterinary bills; MH = management and
handling; PQ = product quality
3.3 Choice Models
The choice models investigate whether some of the heterogeneity associated with contractual
choices is systematically associated with farm or farmer characteristics. Initial results from the MNL
are provided in Appendix 3 to provide an overview of the basic model estimation. Results from the
more sophisticated RPL model for bovine and ovine farmers are reported separately in Table 4. Both
models delivered a good statistical fit (i.e. the model is a good estimator of respondent choice) as
indicated by McFadden pseudo R2 values
of 0.33 (bovines) and 0.38 (ovines).
Table 4: RPL model output of estimated marginal utilities for both ovine and bovine farmers for all
CE attributes and significant covariate interaction terms
[CL] Contract Length
[SS] Scheme Support
[SOS] Structure of Scheme
[N0] Nothing option
Standard deviations of random parameters
[CL] Contract Length
[SS] Scheme Support
[SOS] Structure of Scheme
[N0] Nothing option
Covariates (socio-economic variables)
Note the McFadden pseudo R2 can be interpreted very much like a regression R2 value but the goodness of
fit will always be much lower in CE modelling (typically between 0.2 to 0.4).
No of observations
Prob > Chi square
McFadden Pseudo R2
Note: ***; ** indicates significance at 1% and 5% respectively. SE=standard error.
Socio-economic parameter definitions: AES (whether the respondent is enrolled in an
agri-environment scheme), BEN (farmers with preferences for receiving either
individual or community benefits from the scheme).
The N0 (non-contract option) is positive and significant in both models meaning most farmers have
preferences for the status quo option which follows economic theory (Greiner, 2015). This is perhaps
because there are some variables, not included in the model, which induce farmers to prefer to not join
the offered contract alternatives. The subsidy attribute is positive in both models meaning higher
conservation payments increased likelihood of enrolment. Contract length (bovines and ovines) is
significant and negative meaning respondents prefer a shorter contract. Scheme support was not
significant for both bovine and ovine farmers. Structure of scheme was negative and significant for
bovine farmers meaning they prefer individually managed conservation schemes. For ovine farmers
structure of scheme is positive and significant, suggesting they prefer community managed conservation
Significant standard deviations of the normally distributed coefficients indicate there is
heterogeneity in farmers’ preferences for some attributes. The standard deviations were significant for
all attributes accept contract length and subsidy (bovines only) and scheme support and subsidy (ovines
Additionally, we also tested for significant relationships between respondent preferences for
different contract attributes and various individual specific covariates. The significant covariate
interactions for both models are listed in Table 4. For both models, a negative, significant relationship
was obtained by interacting farmers currently enrolled in AES schemes (AES) with subsidy (COS)
suggesting farmers enrolled in AES measures typically require less subsidy support. Conversely,
farmers not enrolled in AES schemes demanded higher subsidy payments. The N0 interacted with AES
was positive and significant suggesting farmers currently enrolled in AES schemes were more likely to
select the non-contract option. Education level did not influence likelihood of enrolling into a contract
and farmer age did not affect preferences for contract length (both non-significant).
For bovine farmers, interacting respondents wishing to receive community benefits from the scheme
(BEN) with COS was significant and positive, indicating farmers looking to receive community based
(in-kind) rewards require a higher equivalent subsidy reward. For ovine farmers, interacting BEN with
structure of scheme (SOS) is negative and significant meaning farmers preferring individual benefit
schemes also prefer individually managed conservation programmes (i.e. consistency in our results).
Interacting BEN with COS was also negative and significant suggesting ovine farmers preferring
individual payment schemes are WTA lower subsidy premiums.
3.4 Willingness to accept estimates
For WTA estimates (Table 5) the positive value for the N0 of €167 year-1 and €7 year-1 for bovine and
ovine farmers, respectively, can be interpreted as the starting value needed for farmer participation in
the contractual scheme relative to the baseline contract (Christensen et al., 2011); where baseline refers
to a shorter contract length, scheme application support only and an individually managed conservation
breeding programme. Changing from a 5 to 10 year contract would cost around €72.8 year-1 and €3.3 year-
1 for bovines and ovines respectively. To move from an individual to a community managed
conservation scheme would cost an additional €48.6 year-1 for bovine farmers while conversely for ovine
farmers it would cost an additional €5 year-1 to enrol them in an individual scheme.
Table 5: WTA results (€ year-1) derived from the RPL model for both ovine and bovine farmers
[CL] Contract Length
-33.1 to -144.7
-1.4 to -7.3
[SS] Scheme Support
40.7 to -37.6
1.4 to -2.3
[SOS] Structure of Scheme
-8.3 to -121.8
6.0 to 3.1
[N0] Nothing option
198.3 to 109.8
67.6 to 5.9
Note, ***; ** indicates significance at 1% and 5% respectively
3.5 Estimating contract participation
Contract participation was estimated according to different payment and contract scenarios to
determine how projected uptake by farmers varied according to contract attributes. Coefficient means
from the RPL model were used for calculating probabilities under two alternative scenarios; optimal
and non-optimal contracts, where optimal refers to contract attributes that meet farmer preferences
elicited in the CE while ‘non-optimal’ contracts do not. For instance, for bovines this would be a 5 year
contract that is individually managed. The subsidy premium took consistent values across both
scenarios, ranging from 10% to 100% of remuneration offered in the RDP scheme option. This allowed
exploration of how scheme uptake might vary with different contract options to gauge the importance
of monetary and non-monetary attributes in farmer decision making.
As expected, non-optimal contracts were estimated to receive lower participation relative to optimal
contracts (Figure 5). Participation estimates ranged from 4% (€20 year-1) to 70% (€200 year-1) for bovines
and 2% (€1 year-1) to 78% (€10 year-1) for ovine farmers under the non-optimal scenario. Conversely, in
the optimal scenario participation estimates ranged from 38% (€20 year-1) to 97% (€200 year-1) for bovines
and 71% (€1 year-1) to 99% (€10 year-1) for ovine farmers. Recalling that subsidy premiums are comparable
across both contract scenarios, our estimates show the difference in participation (between the two
contract scenarios) ranges from 27% to 58% for bovine farmers and 22% to 84% for ovine farmers.
We find a non-linear relationship between participation and financial reward, suggesting a one unit
change in subsidy does not necessarily equate to a mirrored change in participation (i.e. there are other
factors exogenous to our model influencing farmers willingness to participate). Respondents presented
with optimal contract designs were much more likely to enrol in a conservation programme even at
lower premiums. Ovine farmers were less likely to enrol in a contract that did not match their
preferences for non-monetary attributes at lower subsidy premiums (though this was not the case with
higher premiums). For both farmers groups (non-optimal contracts) there appears to be a tipping point,
before which contract enrolment is relatively static.
Figure 5: Probability of contract participation according to ‘non-optimal’ and ‘optimal’ contract
scenarios for different subsidy premiums (bovine and ovine farmers). ‘Optimal’ refers to contract
attributes that meet the preferences of agents.
4.1 Contract preferences
Results suggest farmers demonstrate a clear willingness to participate in conservation programmes
for rare breeds. Participation may be reduced by up to 84% if farmer preferences for non-financial
attributes are not taken into consideration Within the model, the N0 may capture the dis-utility of
enrolling in a voluntary subsidy scheme that is not linked to contract attributes, but potentially other
factors not included in our model (e.g. family tradition or mistrust in authorities). It may also reflect a
general reluctance to join a voluntary incentive scheme (Christensen et al., 2011). However,
heterogeneity across farmers in our sample (as shown by significant standard deviation of non-random
parameters) complicates interpretation of the N0.
Farmers revealed a tendency to value flexibility in contracts as demonstrated through a
preference for shorter contract durations, a common finding in similar studies (Christensen et al., 2011;
Tesfaye and Brouwer, 2012; Santos et al., 2015). While bovine farmers preferred individually managed
conservation programmes ovine farmers preferred community managed schemes. This seems logical in
post-communist Romania, which has seen a shift from collective to individual ownership rights across
agriculture (Tudor and Alexandri, 2015). On the other hand an enduring communal herd grazing regime
among sheep farmers may explain the alternative preference. The significance of the standard deviation
for this attribute further complicates interpretation. Although scheme support for a conservation
programme was not considered important by both farmer groups similar attributes were significant in
other studies (Ruto and Garrod, 2009). For instance, work by Christensen et al. (2011) has shown
farmers are able to place a monetary value on being released from certain administrative burdens and
that the use of farm advisors for schemes might make farmers willing to accept a lower payment for
enrolling in a scheme. In developing countries like Romania, where rural populations are generally less
educated than the wider population (FAO, 2001) application support for schemes may in-fact be
paramount to securing farmer participation.
A number of covariates help explain heterogeneity in both models. We did not find that farmers
keeping rare breeds were WTA less for supplying conservation services, perhaps suggesting other non-
monetary motives were driving their decisions regarding the contract options. Both farmer groups
enrolled in AES schemes were WTA less compensation for supplying conservation services, thus
providing a means for conservation agencies to target least cost service providers. However, farmers
enrolled in AES schemes were also more likely not to select a contract option, suggesting overlap with
existing contractual schemes may deter farmers from participating. In addition, farmers already enrolled
on AES programmes are more likely to harbour pro-environmental attitudes (Heyman and Ariely, 2004)
that may improve compliance with contractual schemes.
In both models community (in-kind) based support is associated with higher cost than those
preferring cash based payments; implying the use of in-kind rewards will increase overall scheme cost.
However, in-kind payments have been shown to be more effective than cash payments in stimulating
conservation effort (Gorton et al., 2009) and may provide longer term infrastructure benefits to
communities supplying public goods. In addition, Narloch et al. (2017) argue collective payments to
community groups may effectively ‘crowd-in’ compliance, thus reducing monitoring costs and
improving conservation outcomes. The additional costs of community schemes must therefore be
weighed against (potentially) improved social and farm animal diversity outcomes.
4.2 Contract participation
Contract participation estimates reveal a trade-off between non-monetary attributes and financial
incentives. For instance, if RDP subsidies paid € 120/ animal year-1 and € 6/ animal year-1 for bovine and
ovine farmers in an ‘optimal’ contract scenario then uptake rates could be as high as 86% and 98%,
respectively. This contrasts with enrolment of just 28% and 25% for identical price premiums but with
‘non-optimal’ contracts for bovine and ovine farmers, respectively. The higher uptake rates associated
with ovine farmers in optimal contracts may reflect that performance differences between rare and
commercial breeds are larger for bovines than ovines, though this supposition requires further evidence.
These participation estimates are still well above actual participation rates of 15% for an AES
scheme in Northern Italy (Defrancesco et al., 2008). Empirical work by Wossink and van Wenum,
(2003) suggests participation of up to 60% might be achieved in a hypothetical Dutch field margin
programme, suggesting the scheme proposed here is indeed considered attractive by farmers. However,
while strategies were employed to prevent hypothetical bias (e.g. cheap talk statement) it nonetheless
must be considered that the high participation rates found in our work may be exaggerated by such bias
(i.e. the hypothetical nature of a CE may induce respondents to overstate their desire to enrol in a
contract option). That said, farmers in our sample were generally poorer than the national average which
may be an underlying factor driving an increased desire to participate.
Contrary to expectations, farm size, education level and age did not have a significant effect on
participation. These findings confirm conflicting results found in the literature concerning the influence
of education (Dupraz et al., 2002; Defrancesco et al., 2008; Greiner, 2015), age (Wossink and van
Wenum, 2003) and farm size (Christensen et al., 2011; Adams et al., 2014) on participation in
contractual conservation schemes. The hypothesis that farmers keeping rare breeds would be more
likely to participate in a conservation scheme was not supported. This may be because a high number
of farmers were keen to participate in the scheme, irrespective of whether they were currently farming
with a rare breed. Although few studies have directly assessed farmer willingness to participate in rare
breed conservation programmes, work by Pattison et al. (2007) suggests that farmers keeping rare breed
pigs in Mexico were willing to participate in a community conservation breeding programme even
without financial incentives.
4.3 Barriers to uptake
Some have been critical of RDP approaches to rural policy (Shortall, 2008; Milcu et al., 2014). This
study suggests there are clear barriers to entry for smallholder farmers wishing to participate in some
RDP options. This is apparent where RDP eligibility requires a minimum parcel size of 0.3 ha to be
entered into agreements and a cumulative field size of 1 ha or more (Mikulcak et al., 2013). The average
farm size in our sample was 3-6 ha and discussion by Page et al. (2011) stresses this is a major obstacle
for small-scale farmers in Eastern Europe wishing to enrol land into incentive schemes (Gorton et al.,
2009). Herd or flock-book registration of livestock is a requirement to qualify for RDP support for
rearing local livestock breeds in danger of extinction (MARD, 2014) yet only 8% of farmers in our
sample reported having animals registered in this way revealing a major barrier-to-uptake.
Implementing alternative mechanisms, or proxies, to identify the genetic merit of farm animals has been
identified as an important consideration by Pattison et al. (2007) and novel approaches developed by
Bhatia et al. (2010) may serve as a way to surpass such barriers through phenotypic identification of
EU rural development policy needs be more clearly communicated. In our sample, only 21% of
farmers were aware of RDP funding support for farmers rearing endangered breeds. Surveys by
Mikulcak et al. (2013) suggest funding measures are often poorly communicated to small-scale farmers
and local mayors in Transylvania, emphasising the importance of using local communication channels.
In Transylvania, Fundatia ADEPT (a local conservation NGO) are meeting this need by helping small
scale farmers through workshops on the CAP and RDP measures; developing milk collection points in
local villages and facilitating cooperative bids for farm applications to AES options where, individually,
farmers would be ineligible to apply (Fundatia ADEPT, 2014). These factors have culminated in better
support for small-scale farm incomes in Transylvania while maintaining the high levels of public goods
that arise from these production systems.
Farm intensification is a trend across Romania and Central and Eastern Europe (Henle et al., 2008;
Popescu et al., 2016) threatening breed diversity. Sustaining this diversity makes an important
contribution to the delivery of SI objectives given the high option value that arises from breed diversity,
through greater adaptive capacity (Hoffmann et al., 2014). This adaptability, in addition to breed
cultural heritage, is considered important by farmers in Transylvania, particularly those keeping rare
This analysis supports the findings of other work (e.g. Greiner, 2015; Permadi et al., 2018) that
suggest contract length and the structure of schemes, in addition to monetary rewards, are important
determinants of participation rates in conservation programmes. But we also acknowledge that the
monetary values farmers place on accepting specific contractual schemes are case specific (Christensen
et al., 2011). As a consequence, the robustness of these results needs to be addressed in further work
exploring cost-effectiveness of FAnGR conservation programmes in similar contexts. Moreover, this
work has not explored how farmer WTA a contract might vary depending on breed options as part of
the scheme. Indeed, work by Zander and Drucker, (2008) suggests farmer do possess heterogeneous
preferences for breed attributes and breeds themselves. Exploring the importance of alternative breed
and attribute combinations in contracts appears warranted and may further affect farmer willingness to
participate in schemes and their WTA a conservation contract.
We found that the average bovine farmer (in Transylvania) needs to be paid €122 per annum per
animal extra in order to enrol in a 10 year community managed conservation contract. For ovines, an
additional price incentive of €8.3 would be required for farmers to enrol in a 10 year individually
managed conservation contract. A key question is whether the conservation and genetic diversity benefit
of a longer contract that either includes a collectively or individually managed conservation breeding
scheme will exceed the additional costs.
We acknowledge NERC E3 DTP studentship (NE/L002558/1) and the support offered by
Operation Wallacea throughout the project and funding that made fieldwork possible. We thank
Fundatia ADEPT for advice on fieldwork planning and Marcela Man for her work in survey
implementation. Finally, thanks are extended to Frazer Christie for GIS mapping.
Adams, V.M., Pressey, R.L., Stoeckl, N., 2014. Estimating landholders’ probability of participating in a
stewardship program, and the implications for spatial conservation priorities. PLoS One 9, e97941.
Ahtiainen, H., Pouta, E., 2011. The value of genetic resources in agriculture: a meta-analysis assessing existing
knowledge and future research needs. Int. J. Biodivers. Sci. Ecosyst. Serv. Manag. 7, 27–38.
Bhatia, A.K., Jain, A., Sadana, D.K., Gokhale, S.B., Bhagat, R.L., 2010. Phenotypic identification of farm animal
genetic resources using computer learning with scoring function. Comput. Electron. Agric. 73, 37–43.
Bojkovski, D., Simčič, M., Kompan, D., 2015. Supports for local breeds in the European region–an overview.
PoljoPrivreda 21, 7–10.
Broch, S.W., Vedel, S.E., 2010. Heterogeneity in landowners’ agri-environmental scheme preferences, in:
Proceedings of the Economic Council Conference on Environmental Economics.
Burton, R., Kuczera, C., Schwarz, G., 2008. Exploring Farmers’ Cultural Resistance to Voluntary Agri‐
environmental Schemes. Sociol. Ruralis 48, 16–37.
Christensen, T., Pedersen, A.B., Nielsen, H.O., Mørkbak, M.R., Hasler, B., Denver, S., 2011. Determinants of
farmers’ willingness to participate in subsidy schemes for pesticide-free buffer zones—A choice experiment
study. Ecol. Econ. 70, 1558–1564.
Cicia, G., D’Ercole, E., Marino, D., 2003. Costs and benefits of preserving farm animal genetic resources from
extinction: CVM and Bio-economic model for valuing a conservation program for the Italian Pentro horse.
Ecol. Econ. 45, 445–459. https://doi.org/10.1016/S0921-8009(03)00096-X
Crabbe, M., Vandebroek, M.L., 2011. Using appropriate prior information to eliminate choice sets with a
dominant alternative from D-efficient designs.
Defrancesco, E., Gatto, P., Runge, F., Trestini, S., 2008. Factors affecting farmers’ participation in agri‐
environmental measures: A Northern Italian perspective. J. Agric. Econ. 59, 114–131.
Draganescu, C., 2003. Romanian strategy for a sustainable management of farm animal genetic resources.
Ducos, G., Dupraz, P., Bonnieux, F., 2009. Agri-environment contract adoption under fixed and variable
compliance costs. J. Environ. Plan. Manag. 52, 669–687.
Dupraz, P., Vanslembrouck, I., Bonnieux, F., Van Huylenbroeck, G., 2002. Farmers’ participation in European
agri-environmental policies. Zaragoza (Spain) 28, 31.
Eisler, M.C., Lee, M.R., Tarlton, J.F., Martin, G.B., Beddington, J., Dungait, J.A., Greathead, H., Liu, J., Mathew,
S., Miller, H., 2014. Steps to sustainable livestock. Nature 507, 32.
Espinosa Goded, M., Barreiro Hurlé, J., Ruto, E., 2010. What do farmers want from Agri-environmental scheme
design? A choice experiment approach. J. Agric. Econ. 61, 259–273.
European Commission, 2012. European Commission DG Agriculture and Rural Development. Brussels.
Eurostat, 2010. Romania agricultural census.
FAO, 2018. Domestic Animal Diversity Information System (DAD-IS). URL http://www.fao.org/dad-is/en/
FAO, 2017. The future of food and agriculture – Trends and challenges. Rome, Italy.
FAO, 2015. The Second State of The Worlds Farm Animal Genetic Resources Report. Rome, Italy.
FAO, 2001. Farming Systems and Poverty. Improving farmers’ livelihoods in a changing world. Rome, Italy.
Fischer, J., Hartel, T., Kuemmerle, T., 2012. Conservation policy in traditional farming landscapes. Conserv. Lett.
Fundatia ADEPT, 2014. Fundatia ADEPT Transilvania 10 year report 2004-2014. Sighisoara, Romania.
Gandini, G.C., Villa, E., 2003. Analysis of the cultural value of local livestock breeds: a methodology. J. Anim.
Breed. Genet. 120, 1–11.
Gasson, R., 1973. Goals and values of farmers. J. Agric. Econ. 24, 521–542.
Gherghinescu, O., 2008. Poverty and social exclusion in rural areas: Romania.
Gorton, M., Hubbard, C., Hubbard, L., 2009. The folly of European Union policy transfer: why the Common
Agricultural Policy (CAP) does not fit Central and Eastern Europe. Reg. Stud. 43, 1305–1317.
Greiner, R., 2015. Factors influencing farmers’ participation in contractual biodiversity conservation: a choice
experiment with northern Australian pastoralists. Aust. J. Agric. Resour. Econ.
Henle, K., Alard, D., Clitherow, J., Cobb, P., Firbank, L., Kull, T., McCracken, D., Moritz, R.F.A., Niemelä, J.,
Rebane, M., 2008. Identifying and managing the conflicts between agriculture and biodiversity conservation
in Europe–A review. Agric. Ecosyst. Environ. 124, 60–71.
Hensher, D.A., 2006. How do respondents process stated choice experiments? Attribute consideration under
varying information load. J. Appl. Econom. 21, 861–878.
Hensher, D.A., Rose, J.M., Greene, W.H., 2005. Applied choice analysis: a primer. Cambridge University Press.
Heyman, J., Ariely, D., 2004. Effort for payment: A tale of two markets. Psychol. Sci. 15, 787–793.
Hoffmann, I., From, T., Boerma, D., 2014. Ecosystem Services Provided By Livestock Keepers and Breeds , With
Special Consideration To The Contributions Of Small-Scale Livestock Keepers and Pastoralists. Rome.
Holmes, T.P., Adamowicz, W.L., Carlsson, F., 2017. Choice experiments, in: A Primer on Nonmarket Valuation.
Springer, pp. 133–186.
Ilbery, B.W., 1983. Goals and values of hop farmers. Trans. Inst. Br. Geogr. 329–341.
Kompan, D., Klopcic, M., Martyniuk, E., Hiemstra, S.J., Hoving, A.H., 2014. Overview and assessment of support
measures for endangered livestock breeds: Subsibreed: Final project report. European Regional Focal Point
for Animal Genetic Resources.
Lancaster, K.J., 1966. A new approach to consumer theory. J. Polit. Econ. 132–157.
Leroy, G., Baumung, R., Boettcher, P., Besbes, B., From, T., Hoffmann, I., 2018. Animal genetic resources
diversity and ecosystem services. Glob. Food Sec.
Lienhoop, N., Brouwer, R., 2015. Agri-environmental policy valuation: Farmers’ contract design preferences for
afforestation schemes. Land use policy 42, 568–577.
Louviere, J.J., Hensher, D.A., Swait, J.D., 2000. Stated choice methods: analysis and applications. Cambridge
Luce, R.D., 2005. Individual choice behavior: A theoretical analysis. Courier Corporation.
MARD, 2014. National Rural Development Programme for the 2014 – 2020 period. Bucharesti, Romania.
McFadden, D., 1973. Conditional logit analysis of qualitative choice behavior.
Metrics, C., 2012. Ngene 1.1 User Manual and Reference Guide. Sydney, Aust. ChoiceMetrics.
Mikulcak, F., Newig, J., Milcu, A.I., Hartel, T., Fischer, J., 2013. Integrating Rural Development and Biodiversity
Conservation in Central Romania. Environ. Conserv. 40, 129–137.
Milcu, A.I., Sherren, K., Hanspach, J., Abson, D., Fischer, J., 2014. Navigating conflicting landscape aspirations:
Application of a photo-based Q-method in Transylvania (Central Romania). Land use policy 41, 408–422.
Naidoo, R., Balmford, A., Ferraro, P.J., Polasky, S., Ricketts, T.H., Rouget, M., 2006. Integrating economic costs
into conservation planning. Trends Ecol. Evol. 21, 681–687.
Narloch, U., Drucker, A.G., Pascual, U., 2017. What role for cooperation in conservation tenders? Paying farmer
groups in the High Andes. Land use policy 63, 659–671.
National Institute of Statistics, 2015. Press Release No. 135 of June 5th 2015. Household income and expenditure
in 2014 Family Budget Survey. Bucharest, Romania.
National Institute of Statistics, 2013. Press Release No. 159 of July 4th 2013 of the final results population and
housing census 2011.
Page, N., Popa, R., Gherghiceanu, C., Balint, L., 2011. Linking High Nature Value Grasslands to Small-Scale
Farmer Incomes: Târnava Mare, Romania. Mt. Hay Meadows Hotspots Byodiversity Tradit. Cult. Ghimeş.
Pattison, J., Drucker, a. G., Anderson, S., 2007. The cost of conserving livestock diversity? Incentive measures
and conservation options for maintaining indigenous Pelón pigs in Yucatan, Mexico. Trop. Anim. Health
Prod. 39, 339–353. https://doi.org/10.1007/s11250-007-9022-4.
Permadi, D.B., Burton, M., Pandit, R., Race, D., Walker, I., 2018. Local community’s preferences for accepting
a forestry partnership contract to grow pulpwood in Indonesia: A choice experiment study. For. Policy Econ.
Popescu, A., Alecu, I.N., Dinu, T.A., Stoian, E., Condei, R., Ciocan, H., 2016. Farm Structure and Land
Concentration in Romania and the European Union’s Agriculture. Agric. Agric. Sci. Procedia 10, 566–577.
Ruto, E., Garrod, G., 2009. Investigating farmers’ preferences for the design of agri-environment schemes: a
choice experiment approach. J. Environ. Plan. Manag. 52, 631–647.
Santos, R., Clemente, P., Brouwer, R., Antunes, P., Pinto, R., 2015. Landowner preferences for agri-
environmental agreements to conserve the montado ecosystem in Portugal. Ecol. Econ. 118, 159–167.
Shortall, S., 2008. Are rural development programmes socially inclusive? Social inclusion, civic engagement,
participation, and social capital: Exploring the differences. J. Rural Stud. 24, 450–457.
Sutcliffe, L., Akeroyd, J., Page, N., Popa, R., 2015. Combining approaches to support high nature value Farmland
in southern Transylvania, Romania. Hacquetia 14, 53–63.
Sutcliffe, L., Paulini, I., Jones, G., Marggraf, R., Page, N., 2013. Pastoral commons use in Romania and the role
of the Common Agricultural Policy. Int. J. Commons 7.
Tesfaye, A., Brouwer, R., 2012. Testing participation constraints in contract design for sustainable soil
conservation in Ethiopia. Ecol. Econ. 73, 168–178.
Tudor, M.M., Alexandri, C., 2015. Structural Changes in Romanian Farm Management and their Impact on
Economic Performances. Procedia Econ. Financ. 22, 747–754.
Wossink, G.A.A., van Wenum, J.H., 2003. Biodiversity conservation by farmers: analysis of actual and contingent
participation. Eur. Rev. Agric. Econ. 30, 461–485.
Zander, K.K., Drucker, A.G., 2008. Conserving what’s important: Using choice model scenarios to value local
cattle breeds in East Africa. Ecol. Econ. 68, 34–45. https://doi.org/10.1016/j.ecolecon.2008.01.023.
Zander, K.K., Signorello, G., De Salvo, M., Gandini, G., Drucker, A.G., 2013. Assessing the total economic value
of threatened livestock breeds in Italy: Implications for conservation policy. Ecol. Econ. 93, 219–229.