Econometric models, in particular the logit, probit, tobit and multinomial logit models, have been widely used to determine the composition of explanatory variables influencing the adoption process of new technologies by farmers (Shields, Rauniyar and Goode, 1993; Jansen, 1992). Literature suggests that the farm, farmer and institutional factors drive farmers to adopt new technologies (De Francesco, Gatto, Runge and Trestini, 2008; Rehman, McKemey, Yates, Cooke, Garforth, Tranter, Park, and Dorward, 2007; Hattam, 2006). Factors such as the financial and socio-economical impacts of new technologies, effects of new technologies on the risk of the farm, available resources, and technology transfer programmes also have an effect on the decision of the farmer to adopt new technologies (Feder et al., 1985).
When the objective is to identify the socio-economic variables that influence both adoption and intensity of adoption, the probit and the tobit models are preferred (Nichola and Sanders, 1996; Adesina and Zinnah, 1993; McDonald and Moffit, 1980). Different approaches towards adoption models that were used in the past are described by Nichola (1994). There are many econometric studies dealing with economic and environmental aspects of conversion to more sustainable farming systems such as organic farming. It is clear that the majority of the reviewed econometric studies are oriented towards supporting policy making (Feder et al., 1985).
A study by Workneh and Parikh (1999) used probit and ordered probit to examine both the significance of the impact of farmers‘ perception in adoption decisions of new technology and how perceptions are influenced by the decision to adopt new technology. The probit approach is used to analyse the adoption decision, while farmer perceptions are modelled using the ordered probit methodology since there is an ordering to the categories associated with the dependent variable (Calatrava and Gonzales, 2008; De Cock, 2005; Albisu and Laajimi, 1998). The ordered probit model assumes that there are cut off points which define the relationship between the observed and the unobserved dependent variables (Verbeek, 2008; Pindyck and Rubinfield, 1981).
Belknap and Saupe (1988) used maximum likelihood to estimate a probit model relating variables to the probability that a farm operator used conservation tillage. Farmers were defined as having adopted conservation tillage if conservation tillage was used on part of the farm. Independent variables were classified as being the physical characteristics of the farm, farm business characteristics and human resources characteristics. Unlike Rahm and Huffman (1984) human capital variables were included in the adoption model to approximate psychological cost of adoption, attitudes and management objectives. Other authors that have used this methodology include Isin, Cukur and Armagan (2007) and Hattam and Holloway (2004) for the estimation of conversion to organic certification and to establish the factors affecting the adoption of the organic dried fig agriculture system in Turkey respectively. Sinja, Karugia, Mwangi, Baltenweck and Romney (2004)
investigated farmers‘ perception of technology and its impact on adoption of legume forages in central Kenya highlands by estimating the ordered probit model to assess relative importance of each attribute to the farmer.
Lohr and Salomonsson (2000) focused on analysing the factors that determine whether a subsidy is required to motivate organic conversion by using a utility difference model with Swedish data. From these results Lohr and Salomonsson concluded that services rather than subsidies may be used to encourage conversion to organic agriculture. Pietola and Oude-Lansink (2001) focused on analysing the factors determining the choice between conventional and organic farming technology in Finland using a Bellman equation. The choice probabilities were estimated in a closed form by an endogenous Probit- type switching model using Maximum Likelihood Estimation (MLE).
Logistic regression was used by van Vuuren,Larue and Ketchaba (1995) to determine the impact tenant, contract and land characteristics have on adoption of farm practices that enhance productivity and environmental husbandry on rented land. The logit model was also successfully used by Parra and Calatrava (2005) to identify factors related to the adoption of organic farming in Spanish olive orchards. Rigby and Young used logit model to establish why some agricultural producers abandon organic production systems. Wynn, Crabtree and Potts (2001) aimed to model the entry decisions of farmers and the speed of entry to Environmentally Sensitive Areas (ESA) in Scotland. A multinomial logit model was used for modelling entry decisions and a duration analysis was made to quantify the relative speed at which the farmers joined the ESA scheme. They concluded that the logit and duration models were reasonably successful in explaining the probability and speed of entry to the scheme respectively.
Using discriminant analysis, Thompson (1996) identified and ranked the partial effects of the variables that distinguish lessors and lessees in KwaZulu-Natal. The results showed that the most important variable distinguishing lessors from lessees was farm size followed by liquidity. On the other hand, Cooper (1997) made an attempt to estimate the minimum incentive payments a farmer would require in order to adopt more
environmentally friendly ―best management practices‖ (BMPs), using contingent valuation method (CVM). Table 2.1 present other studies that have analysed organic farming adoption and its determinants using various models and methodological approaches. These empirical modelling studies show the importance of incentives and agricultural policy. They provide an understanding of the factors influencing a certain dependent variable example the factors influencing the conversion to more sustainable farming systems and the effect of different policies on the decision making of farmers. The ordered probit model has been applied in this study because of its suitability in modeling categorical dependent variables. It is an especially useful and informative approach to understand the farmers decision on their organic farming status represented by fully-certified organic, partially-certified organic and non-organic.
Table 3.2: Studies that analyse organic farming adoption and its determinants
Study Sample size Method of analysis
Organic Conventional
Acs, Berentsen and Huirne (2007) Dynamic linear programming
Albisu and Laajimi (1998) 97 125 Probit model
Anderson, Jolly and Green (2005) 28 118 Multinomial Logit model
Calatrava and Gonzales (2008) 254 Ordered probit model
Darnhofer et al. (2005) 9 12 Decision tree modeling
De Cock (2005) 93 190 Ordered Probit model
Fairweather (1999) 16 27 Decision tree modeling
Gardebroek and Jongeneel (2004) 16 - Bayesian approach
Genius, Pantzios and Tzouvelekas (2006) 44 118 Ordered Probit model Hanson, Dismukes, Chambers, Greene
and Kremen (2004)
61 - Focus group(Qualitative)
Hattam and Holloway (2004) 47 186 Probit model
Isin et al. (2007) 20 107 Probit model
Kerselaers, De Cock, Lauwers and van Huylenbroeck (2007)
- 685 Linear programming
Lohr and Salomonsson (2000) 234 316 Probit model
Musshoff and Hirschauer (2008) Investment under
uncertainty
Parra and Calatrava(2005) 161 161 Logit Model
Pietola and Oude Lansink (2001) 169 779 Switching-type probit
Rigby and Young (2000) 86 35 Logit model
Wossink and Kuminoff (2005) 80 167 Option theory
Cisilino and Madau (2007) 115 114 Data envelopment analysis
Klepper, Lockeretz, Commoner, Gertler, Fast, O’Leary and Blobaum (1977)
14 14 Basic statistics
OECD(2000) -- Basic statistics
Oude Lansink and Jensma (2003) 29 571 Profit maximization model
Zhengfei, Oude- Lansink, Wossink, and Huirne(2005)
3.8 Chapter summary
Technologies play an important role in economic development and technological change has been a major factor shaping agriculture in the last 100 years. There is a strong belief in the ability of agricultural technology to continue to provide farmers with the needed strategic and tactical options to address food security while addressing environmental concerns. The literature on innovation is diverse and has developed its own vocabulary. In this chapter the basic concepts and theoretical foundation for technology adoption and diffusion is explored with the definition by Rogers (2003) and other authors referred to. A distinction is made between individual and aggregate adoption. The induced innovation as outlined by Hicks (1932) is examined and it has been tested in many countries and industries. The categorization of adopters into innovators, early adopters, early majority, late majority and laggards are illustrated and the cumulative adoption is described as an S-shaped curve resulting from the fact that few farmers adopt the new technology in the early stages of the diffusion process and the essential differences among farmers can explain this phenomenon.
A non exhaustive selection of empirical research in trying to understand the determinants of farmer‘s decisions to be organic certified are reviewed. The review reveals that adoption or organic farming by farmers is influenced by personal attributes of the farmer, farming systems and resource characteristics, institutional, infrastructure and environmental factors, attitudes and opinions. In explaining the mode and sequence of agricultural technology adoption two approaches are common in literature: the adoption of the whole package or sequential adoption. Various arguments are given for the different approaches to adoption. The barriers to adoption of organic farming are explored and highlighted as (i) perceptions; (ii) access to technical and financial information; (iii) institutional barriers; and (iv) social barriers. Finally the approaches of analyzing technology adoption and diffusion are examined which the analysis revealing that econometric models, in particular the logit, probit, tobit and multinomial logit models, have been widely used to determine the composition of explanatory variables influencing the adoption process of new technologies by farmers.