Bio-based and Applied Economics 8(2): 211-234, 2019 ISSN 2280-6180 (print) © Firenze University Press ISSN 2280-6172 (online) www.fupress.com/bae Full Research Article DOI: 10.13128/bae-8931 Determinants of Farm Households’ Willingness to Accept (WTA) Compensation for Conservation Technologies in Northern Ghana EvElyn DElali ahialE1,*, KElvin BalcomBE2, chittur SrinivaSan2 1 Department of Agricultural Economics and Extension Education, University of Education, Winneba, Ghana 2 School of Agriculture, Policy and Development, University of Reading, UK Abstract. This paper examines the determinants of farmers’ willingness-to-accept (WTA) in a hypothetical payment scheme for the adoption of stone and soil bunds in northern Ghana using contingent valuation data from 305 farm households. Bayesian estimation of the interval-data regression model is employed to obtain farmers’ WTA and its determinants. Besides farmer and household characteristics, farm characteris- tics and socioeconomic and institutional variables such as soil fertility and previous participation in conservation projects increase and decrease willingness-to-accept respectively. The results suggest that costs of soil and water conservation payment schemes may be significantly decreased by careful targeting of households more ready to accept compensation. Keywords. Conservation technology/practice, interval-data regression model, will- ingness-to-accept, Contingent valuation method, Payment for Environ- mental Services. JEL Codes. Q18, Q56, Q57. 1. Introduction Land degradation is one of the world’s environmental concerns today. It can be regarded as a process that includes soil degradation and erosion. The main processes that lead to land degradation are soil erosion by water and wind; chemical changes such as acidification, salinization, and nutrient loss; and physical degradation through pressures such as compaction (Eswaran et al., 2001; UNCCD, 2013). There is no consensus on the exact extent and severity of land degradation in the African region, there is however con- sensus that it is severe and widespread. Analyses of global land degradation indicate that Africa is especially susceptible to land degradation and is the most severely affected part of the world (Lal, 1995; Obalum et al., 2012). An estimate of two-thirds of Africa’s pro- ductive land is affected by land degradation and almost all the land area is susceptible to *Corresponding author: abedella@yahoo.com 212 Evelyn Delali Ahiale, Kelvin Balcombe, Chittur Srinivasan soil and environmental degradation (FAO, 2011; Jones et al., 2013; UNCCD, 2013; Vlek et al., 2008). In sub-Saharan Africa (SSA), in excess of 320 million hectares of land have been made unsuitable for agricultural purposes due to soil erosion, deforestation, over- grazing and mismanagement of land resources (Sant, 2001). Nabhan (1997) also reports that 67% of agricultural lands are affected by land degradation, with close to 490 million hectares displaying signs of erosion and declining fertility. The issue of land degradation is of immense importance in Africa as majority of its population’s livelihood is heavily reliant on natural resources. Agricultural productivity in the region is stagnating or declining, largely due to land degradation. Land degradation in Africa has thus been immensely detrimental to agricultural ecosystems and crop pro- duction consequently leading to increasing levels of food insecurity, loss of farm incomes, poverty, high mortality rates, other social vulnerabilities, migration and conflict (Gomiero, 2016; Hamdy & Aly, 2014; Hemant & Padmini, 2013; UNCCD, 2013). Land degradation thus has socioeconomic implications for African countries. Soil and land degradation in Ghana was recognized decades ago, as since the 1930s, it has attracted considerable attention and concern (Agyepong, 1987; Benneh & Agyepong, 1990). Land degradation is affecting all parts of Ghana, however, the northern regions placed within the Guinea and Sudan Savannahs are the most vulnerable zones and the most degraded area of the country (Asiedu et al., 2016; World Bank, 2006). Ghana had 35% of its land threatened by desertification particularly in the northern regions (Upper East, Upper West and Northern Regions) since the 1960s (Adanu et al., 2013; Kenwor- thy, 1995). Land degradation in the northern regions of Ghana has thus rendered large tracts of croplands which were once fertile currently unproductive as such contributing to depleting farm income and food sources. As a result of land degradation, grasslands, woodlands and forests are being lost while natural water bodies are drying up due to prolonged droughts and deposition of sediments into water courses (Adanu et al., 2013). Land degradation in Ghana, which is mainly as a result of soil erosion and soil nutri- ent depletion, has negative impact on farm productivity and environmental quality. The human-associated drivers of long-term soil and vegetation degradation in Ghana include unsustainable farming practices, removal of vegetation cover (including deforestation and overgrazing), mining activities, and urbanization and industrial activities caused by increased population growth pressures. Agriculture remains an important sector in the Ghanaian economy contribut- ing about 22% to the country’s gross domestic product (GDP) and providing 44.7% of employment in 2013 (Aryeetey & Baah-Boateng, 2015). Agriculture also remains the main source of livelihood for many subsistence smallholder farmers living in rural Gha- na. The agricultural activities of these smallholders is cited to be a key factor in promot- ing land degradation through the use of environmentally unsustainable cultural practices (Asiedu-Amoako et al., 2016; Boardman et al., 2003; Diao & Sarpong, 2007; Helming et al., 2006; Senayah et al., 1998). As agriculture is the major user of rural land, its rele- vance is not only in relation to its economic significance, but also its influence over the use of land in rural Ghana and its environmental health in general. With the relationship between land degradation, agricultural productivity and poverty well understood (Das- gupta & Mäler, 1995; Gomiero, 2016; Hamdy & Aly, 2014; Heath & Binswanger, 1996; Hemant & Padmini, 2013; Shetty et al., 1995; World Bank, 1992), it is clear that land 213Determinants of Farm Households’ Willingness to Accept Compensation in Northern Ghana degradation is a threat not only to national and household food security but the overall welfare of many households in Ghana. In order to maintain agricultural productivity, reduce food insecurity and poverty, and improve environmental conditions, the Government of Ghana (GoG), international donor agencies and Non-Governmental Organizations (NGOs) have promoted soil and water conservation practices and technologies including soil and stone bunds. This has been done for several decades particularly in the Northern, Upper-East and Upper-West regions because they collectively constitute the most degraded part of the country. Adoption of the promoted technologies has arguably been unsuccessful due among others to weak regulatory institutions which have restricted the ‘command and control’ interventions (Wunder, 2008). Farmers’ inability to adopt soil and water conservation measures is mainly as a result of constraints resulting from market failures which lead to externalities like degradation. When externalities are present, government interven- tion has the potential to internalise these externalities. One potential intervention is Pay- ment for Environmental Services (PES) in which incentive payments are made to resource managers in return for the adoption of conservation practices/technologies. Such external financial incentives may be crucial in ensuring that socially desirable levels of environ- mental services/goods (ES) are supplied and maintained since poor smallholder farmers may not be able to afford to maintain healthy environmental quality especially when large opportunity costs occur when conservation technologies/practices are adopted. Soil and water conservation technologies (e.g., soil and stone bund) are technologies that preserve the integrity of soils and their water content, and they offer a number of on-farm and off-farm ecosystem services of value to society as well as on-farm productiv- ity improvements. Stone and soil bunds are stone or soil walls built across a slope (along a contour) to act as a barrier to prevent run-off, therefore helping in reducing soil ero- sion and increasing water retention capacity of soil. They are often appropriate for gen- tle slopes (2-5%) (Diao & Sarpong, 2007). Ecosystem services from stone and soil bunds include: substantial flood and erosion control, substantial reduction in sedimentation of water bodies and its consequent improvement in water quality and aquatic life; reduction in leaching and deposition of fertilizers, herbicides and pesticides, i.e. generally improved landscape quality, etc. (Bingham et al., 1995; Holland, 2004; Webb et al., 2001). The adop- tion of soil and water conservation (SWC) technologies is aimed at returning a landscape to a condition where it can again provide the ES enumerated above after a period of deg- radation. Farmers can therefore be paid/compensated for the adoption of such technolo- gies as soil and stone bund per unit area to produce the socially beneficial ES mentioned. Payments can be in the form of money, in kind, and access to resources and markets. In order to know the optimum rate of public investment, the required level of com- pensation (WTA) necessary for encouraging agricultural households to adopt a conserva- tion technology which produces ES must be ascertained. In addition, knowledge of the factors determining WTA also informs policy implementation by enabling the direction of payments towards those that are the most predisposed towards adopting the proposed technologies. The current study therefore uses the contingent valuation (CV) method to estimate farmers’ WTA for adopting stone and soil bunds in a hypothetical conservation plan/ valuation scenario context. The method of elicitation within the CV employed allows 214 Evelyn Delali Ahiale, Kelvin Balcombe, Chittur Srinivasan for uncertain responses so as to maximise respondent’s engagement with the survey. It employs an interval regression model to estimate WTA and determines the factors influ- encing their WTA, having adapted this model to allow for uncertain responses. Extensive literature exists on PES as an alternative intervention for environmental conservation. However, much of this has focused on parts of the world other than Africa leading to a dearth of knowledge on environmental values and the main factors influenc- ing WTA for PES conservation practice/technologies for Africa and specifically for PES schemes in Ghana. This study fills this gap by building on previous studies, with a specific aim to determine the manner in which various factors influence WTA compensation for stone and soil bunds in northern Ghana. The manner in which various factors influence WTA for conservation technologies may be location and conservation practice specific. This study therefore serves to analyse these factors, so they can be understood in a way that enables better designed interventions and decision-making in Ghana. The paper proceeds by first reviewing literature on valuation of the welfare impact of adoption of soil and stone bunds and the factors that influence farmers’ preferences for or WTA for conservation practices in Section 2. The interval data regression model speci- fication and estimation is presented in Section 3 while Section 4 discusses the field sur- vey, data and variables. The results are presented in Section 5 and Section 6 discusses the results. The conclusions and policy implications is given in Section 7. 2. Literature review 2.1 Valuation of the welfare impact of adoption of soil and stone bunds The contribution of a resource to human welfare forms the basis of the economic approach to the valuation of resources. An economic value is measured by the variations in welfare related to the variation in the quantity or quality of goods or services. Varia- tions in environmental service flows can influence the welfare of individuals in complex ways and through both marketed or non-marketed activities (Shiferaw et al., 2005). Inter- ventions like adoption of soil and stone bund by farm households that lead, for example, to reduction in soil erosion apparently change the welfare of different members of the society. Welfare economics suggests that welfare values or changes are determined by indi- vidual preferences and measured by their personal assessment of changes in well-being (Bockstael et al., 2000) or the extent to which they are willing to make trade-offs between scarce resources to obtain or preserve something. Investments in soil and stone bund provide multiple economic and environmental ben- efits to different groups of people beside the adopting smallholder farm households. An impact evaluation of the interventions should therefore take into account any non-mar- keted ecosystem goods and services along with marketed economic benefits (Baker, 2000; Shiferaw et al., 2005). The welfare gains from investments in soil and stone bund include the direct economic benefits (e.g., yield gains) and environmental benefits (e.g., sustain- ability benefits and ecosystem services) that have both use and non-use values to people. Indirect welfare benefits obtained from environmental improvements are justifiable compo- nents of the welfare changes related to any conservation interventions, and must be meas- ured in impact evaluation (Shiferaw et al., 2005). Total welfare benefit to people, therefore, 215Determinants of Farm Households’ Willingness to Accept Compensation in Northern Ghana is the sum of the direct economic and indirect environmental benefits. Hence, the benefits accruing to soil and stone bund can be assessed as those captured privately by the farm household, which include the value of yield loss averted and/or yield gains which may be felt on-site by the farm household, and those external to the farm household that are cap- tured publicly, whose value include the improvement in ecosystem services. The valuation of changes in ecosystem services as a result of the adoption of soil and stone bund by farmers “needs to take into account both intended and unintended out- comes as different individuals may attach values for such changes because of the use ben- efits they derived, or any expected or conceived non-use welfare benefits” (Shiferaw et al., 2005). The concept of total economic value (TEV), the usual and most appropriate frame- work for aggregating the value of non-market ecosystem goods and services and meas- uring welfare changes is a vital part of economic valuation (Pearce, 2002; Philcox, 2007). Economic values reflect the services of an ecosystem and not the economic value of that ecosystem (Nijnik & Miller, 2017). The potential welfare changes or impacts as a result of soil and stone bund on groups of individuals differ, i.e., the farm household’s welfare change is different from the welfare impact on the consumers of the ecosystem services accruing from soil and stone bund. Assessing the economic value of soil and stone bund can thus be done in two ways. First, the measurement “of how much better or worse-off a person is due to the variation in the quantity or quality of the service flow” and second, “the addition of the individual wel- fare variation (gains and losses or WTP/WTA) to assess the value of this variation for the entire society (Shiferaw et al., 2005). The former is the focus of the current paper. Total welfare gains include the direct economic benefits (e.g., yield gains) and indirect environmental benefits (e.g., ecosystem services) that have both use and non-use values to people (Shiferaw et al., 2005). The benefits accruing to the adoption of soil and stone bunds can be assessed as those captured privately by the farm household, which include the value of yield loss averted and/or yield gains which may be felt on-farm by the farm household, and those external to the farm household (off-farm) that are captured publicly, whose value include improvement in ecosystem services. Direct use values comprise con- sumptive uses including the potential yield increase that may be associated with the adop- tion of the conservation technology and indirect use value is improvement in landscape quality. The non-use value of soil and stone bund includes improved and preservation of aquatic life, erosion control, and the reduction of the deposition of soil and agricultural chemicals into water bodies. For this study, the WTP or WTA indicates how the adoption of stone and soil bund impact on the welfare of participating farm households. The benefits accruing to farmers by adopting soil and stone bunds are often below the total benefits created once public good values have been accounted for, leading to below optimal levels of resource supply. In the presence of high public good values, incen- tive payments for resource conservation may be necessary. There is dissimilarity between the average gross margin of adoption and non-adoption of conservation technologies/ practices that result in conservation opportunity costs for farmers (Krishna et al., 2013). The opportunity costs for farmers adopting soil and stone bunds include: loss of valua- ble cropping land to bunds (Ludi, 1997; Wyatt, 2002) which for farmers is an important issue when land is scarce and which imposes revenue loss to farmers; additional labour requirements of household for construction and annual maintenance (Shiferaw & Holden, 216 Evelyn Delali Ahiale, Kelvin Balcombe, Chittur Srinivasan 2001; Stocking & Abel, 1989). PES schemes should pay for the farmers’ opportunity costs. The stated preference (SP) of farm households’ stated WTA compensation for adopting soil and stone bunds can be employed as an appropriate measure of the opportunity cost of adopting such technologies. The minimum compensation needed to motivate a farm household to accept a PES contract involving the construction of soil or stone bunds on a unit area of land is presumed to indicate the farmer’s real opportunity cost per unit area of soil or stone bunds adoption. 2.2 Factors determining farmers’ willingness to accept/preferences for conservation practices/ technologies The factors influencing preferences and WTA compensation for conservation prac- tices and technologies have generally been categorised into: farm characteristics; farmer and household characteristics; socioeconomic; and, institutional factors by previous stud- ies (see, e.g., Ayuba et al., 2011; Cooper & Keim, 1996; Matta et al., 2009; Minten, 2003). Key farmer and household characteristics have generally been thought to include gender, age, level of education of household head, own labour, labour sufficiency of the household, and wealth status of the household. For example, Thurston (2006) observes that females have higher WTA than males in valuing environmental conservation. Ste- phen (2015) and Wang et al. (2019) also estimate higher WTA for females relative to males. Sangkapitux et al. (2009), PRESA/ICRAF (2010), Minten (2003), Stephen (2015), and Wang et al. (2019) all find a positive relationship between age and WTA compen- sation. However, this relationship is found to be significant by Sangkapitux et al. (2009), Minten (2003), Stephen (2015), and Wang et al. (2019) and insignificant by PRESA/ ICRAF (2010). Feng et al. (2018) however observe a negative effect of age on WTA. Edu- cation has usually been found to have a positive influence on WTA for conservation/sup- ply of environmental services (Ninan et al., 2007; Wang et al., 2018; Wang et al., 2019; Xu et al., 2015). Minten (2003) explains that more educated households, who have a higher reservation wage, prefer to put more effort in off-farm earnings and hence, prefer to prac- tice agriculture in a more extensive manner. In contrast, Xiong and Kong (2017) and Yu and Cai (2015) observe a negative influence of education on WTA. Household size has been used as an index of the farm household’s access to labour in most studies. PRESA/ ICRAF (2010) find a significant negative effect of household size on WTA compensation for watershed services whilst Minten (2003) observes an insignificant positive influence of household size on WTA to give up slash and burn agriculture (‘tavy) and an insignificant negative effect of household size on WTA to give up forest use in Madagascar. Stephen (2015), Xiong and Kong (2017), and Wang et al. (2019) all find a significant positive effect of household size on WTA. Sangkapitux et al. (2009) observe that poorer farmers have a higher willingness to engage in a compensation scheme for providing better ecologi- cal services, probably indicating a lower WTA for the supply of environmental services. Farmers who rely on income from farm and aquatic products have higher WTA (Stephen, 2015; Xiong & Kong, 2017). Key farm characteristics include total farm size, level/severity of erosion on farm/plot, level of soil fertility of farm/plot, slope of plot, and location/region. Sukic (2001), Xiong and Kong (2017) and Wang et al. (2019) find statistically significant positive impact of 217Determinants of Farm Households’ Willingness to Accept Compensation in Northern Ghana land size on WTA compensation for conservation practices whilst PRESA/ICRAF (2010) discovers an insignificant influence. Previous findings suggest that the higher the level of erosion and the lower the soil fertility, the higher the willingness to participate in a pay- ment scheme (Sangkapitux et al., 2009). This in turn suggests that farmers with severe farm erosion and low soil fertility are likely to demand less compensation for conservation practices than those with less severe erosion on their fields and more fertile soils. Farm location heavily influences WTA (Minten, 2003; Stephen, 2015; Xiong & Kong, 2017; Yu & Cai, 2015). Minten (2003) reports that households with more lowland (which are more flat) are willing to accept less for compensation, though the estimates are statistically insignificant. By contrast farming in highlands (likely to be steeper) is a significant deter- minant of WTA compensation. Monthly income, adoption status, and previous participation of the household in a con- servation programme/project are institutional and socio-economic factors that have been associated with WTA compensation. A positive relationship between income and WTA is observed by PRESA/ICRAF (2010) and Sukic (2001), though the effect is statistically sig- nificant and insignificant respectively. Xu et al. (2015), Yu and Cai (2015), and Wang et al. (2018) find the opposite, that is a negative influence of household income on WTA. For non-market valuation, respondents may be unable to give their true preferenc- es because they have had little prior experience with the item in question and so have trouble establishing their minimum WTA during a single survey (Cummings et al., 1986). Household knowledge of the good being valued can be proxied by identifying households who have previously adopted the conservation practice of interest or taken part in wider conservation programmes or projects. The literature is, however, silent on the direction of influence on WTA of households’ prior adoption of a practice/technology and participa- tion in previous conservation projects. However, various studies have found farmers’ envi- ronmental awareness and knowledge affect WTA positively (Feng et al., 2018; Wang et al., 2018; Xu et al., 2015; Yu & Cai, 2015). 3. Estimation methodology CV studies require questions as to whether respondents are prepared to pay or accept specified monetary amounts in the light of changes that will impact upon them (e.g. the adoption of a technology). Valuation studies often assume that respondents know their preferences with certainty, i.e. they know how much they would be willing to accept for ES provision. However, empirical evidence in the SP literature indicates that respondents are uncertain about their responses (Akter & Bennett, 2013; Akter et al., 2008; Alberini et al., 2003; Champ et al., 1997; Ready et al., 1995). This is mainly because respondents use a heuristic mode while processing information provided in any of several CV formats which tends to dominate over more systematic ways of information processing for decision- making (Bateman et al., 2004). Unsurprisingly, the CV literature has spawned multiple forms of elicitation which are aimed at minimising or mitigating biases. One factor that leads to potential bias is that respondents may not know or be able to state with certainty their underlying preferences and forcing them to do so can induce bias (Akter & Bennett, 2013; Ariely et al., 2003; Poe, 2016; Ready et al., 2010). Uncertainty is an important aspect of many public goods, especially environmental goods (Mitchell & Carson, 1989) such 218 Evelyn Delali Ahiale, Kelvin Balcombe, Chittur Srinivasan as produced by SWC technologies like stone and soil bunds. Preference uncertainty is a stochastic error term which comes about in hypothetical valuation scenarios as individu- als do not know their true values of a good with certainty (Li & Mattsson, 1995). In the current study, the LB and UB are obtained based on the expansion approach of Broberg and Brännlund (2008), which takes uncertainty into consideration. This method can be termed the ‘multiple-bounded uncertainty choice’ (MBUC) approach. It has been argued that this approach is more intuitive, better fits the data, estimates mean and median WTP with better precision, is less sensitive to distributional assumptions, and it is better suited for policy analysis than other approaches (Broberg & Brännlund, 2008). The elicitation method employed here (and outlined in Section 4) is to give respondents a ‘payment card’ (a series of possible ranges) but allowing them to indicate whether they would pay for that amount with certainty or with some level of uncertainty. However, this approach also requires an adaptation to the standard methods used to model WTA/WTP. A standard model for dealing with the case where the dependent variable is only known within a range is the interval data model (Stewart, 1983). Hanemann et al. (1991) observed that the interval-data model improves the statistical efficiency of WTP estimates by reducing the variance and point estimates of WTP models relative to single-bound models. Alberini (1995) explored the efficiency and biases of the estimates obtained from the bivariate probit and interval-data models and observed robust estimates for mean/ median WTP and concludes that in the absence of perfect correlation as is the case in many CV studies, the interval-data model might be appropriate. However, the interval- data model assumes that answers provided by respondents reflect WTP/WTA which are known with certainty (Alberini, 1995). However, as discussed above, the elicitation approach outlined in Section 4 is more general in that it allows for uncertain responses. Because MBUC responses do not translate directly into the statistical models convention- ally employed to model stated-preference responses, assumptions about the interpreta- tion of the responses by the researcher are necessary. The literature provides a number of empirical ways to convert MBUC CV data to easily estimable forms. Intervals are obtained by assigning, LB